Relational Patient Capital and Agricultural Technological Innovation: Evidence from Chinese Agricultural Technology Enterprises

Abstract

Agricultural technological innovation is essential for sustainable agricultural modernization and rural development. However, agricultural technology enterprises often face persistent financing constraints because research and development (R&D) activities involve long investment cycles, high uncertainty, and delayed returns. Using a firm-level panel dataset of Chinese agricultural technology enterprises, this paper examines the effect of relational patient capital (RPC) on agricultural technological innovation by employing a two-way fixed-effects model. The results show that RPC significantly promotes agricultural innovation output. Mechanism analysis indicates that RPC enhances innovation through two channels. First, it facilitates firms’ digital transformation, thereby reducing R&D uncertainty and organizational costs. Second, it alleviates financing constraints by stabilizing cash flows to support R&D investment. The results remain robust after clustering standard errors, excluding the years affected by the COVID-19 pandemic, and employing lagged specifications. Heterogeneity analyses further reveal that the positive effect is more pronounced among small-scale enterprises and firms located in central and eastern China, where financing frictions and resource constraints are relatively more severe. By linking RPC to firm-level agricultural innovation, this study extends the literature on agricultural finance and innovation financing, highlighting the role of long-term, relationship-based capital in addressing market failures in agricultural R&D. The findings suggest that rural financial policies should encourage stable, long-term investment, strengthen financing support for small agricultural technology enterprises, and integrate patient capital with digital transformation initiatives to promote sustainable agricultural and rural development.

1. Introduction

Agricultural technological innovation plays a foundational role in advancing food security, rural development, and the sustainable modernization of agriculture. For agricultural economies, innovation in seeds, precision agricultural machinery, biological breeding, and digital agricultural tools is not only a source of productivity growth, but also an important policy instrument for supporting rural revitalization and sustainable rural development. However, agricultural technological innovation is widely recognized as one of the most capital-intensive and time-consuming forms of innovation. From the development of improved crop varieties and precision agricultural machinery to the commercialization of biological breeding technologies, agricultural innovation usually requires long development cycles and generates returns that are highly delayed and uncertain, because the development, testing, adoption, and diffusion of agricultural technologies often involve substantial time lags and uncertainty in economic returns [1,2]. Such structural features create a fundamental mismatch with conventional financial capital, which tends to impose short investment horizons, strict performance benchmarks, and early exit requirements. As a result, agricultural technology enterprises in China and elsewhere frequently face insufficient R&D investment, commercialization financing gaps, and weak incentives for the diffusion and scaling of novel technologies [3,4]. Recent evidence also suggests that digital finance can materially affect agricultural total factor productivity, highlighting the importance of financial digitalization for agricultural resource allocation and productivity dynamics [5].

The concept of patient capital, which refers to capital that tolerates long investment horizons, high uncertainty, and delayed returns, has gained increasing attention in both policy discourse and academic research as a potential solution to innovation financing gaps [6,7]. In China, this policy logic has been reflected in multiple iterations of the Central Document No. 1. Since 2021, these policy documents have repeatedly emphasized strengthening agricultural science and technology support, accelerating seed industry revitalization, and promoting the development and application of biological breeding technologies [8,9,10]. In parallel, recent national-level policy initiatives have highlighted the importance of expanding patient capital, encouraging investment at earlier stages, and supporting hard technologies with long development cycles and uncertain returns [11]. These policy orientations indicate an increasing recognition that long-duration capital is essential for supporting strategically important innovation activities. From the perspective of agricultural economics and policy, patient capital is not only a financing arrangement, but also a potential policy mechanism to correct market failures in agricultural innovation and promote sustainable agricultural modernization. Related evidence indicates that agricultural fiscal support can strengthen agricultural economic resilience, underscoring the importance of public financial arrangements for long-term agricultural development [12].

Prior studies have primarily conceptualized patient capital from a temporal perspective, emphasizing characteristics such as extended investment horizons, delayed exit decisions, and tolerance for short-term performance fluctuations [13,14]. However, emerging research on long-term investors suggests that patient capital may also involve relational engagement, including strategic support, information provision, and active involvement in firm development [14,15]. However, its relational dimension has received much less attention. We argue that for agricultural technology enterprises, the relational characteristics of patient capital may be equally important. RPC, as conceptualized in this study, refers not only to the long duration of capital commitments, but also to stable shareholder relationships, high-quality post-investment engagement, and the continuous provision of resources, networks, and reputational endorsement by long-term institutional investors. Agricultural innovation is often exposed to natural risks, regulatory complexity, long commercialization chains, and severe information asymmetry between firms and external financiers. Under these conditions, RPC can reduce monitoring costs, lower the risk premium attached to research-stage firms, and sustain managerial commitment to long-cycle R&D projects [16,17]. Understanding this relational dimension is particularly important for sustainable rural development strategies, because the diffusion of agricultural technologies depends not only on scientific discovery, but also on stable financial and institutional support.

Against this background, this paper addresses the following research question: Does RPC promote agricultural technological innovation, and if so, through what mechanisms? To answer this question, we construct a firm-year panel dataset of Chinese agricultural technology enterprises, quantify RPC as the share of long-term institutional and strategic shareholding, and employ a two-way fixed-effects model to estimate its effect on innovation output. Innovation output is measured by log-transformed invention patent applications. We further examine two theoretically motivated mediating channels, namely digital transformation and financing constraints. Digital transformation is examined as an organizational capability channel through which firms reduce information, coordination, and management costs in innovation activities. Financing constraints are examined as a financial channel through which patient capital stabilizes cash flows and supports sustained R&D investment.

To address these research questions, this study constructs a firm-year panel dataset of Chinese agricultural technology enterprises and employs a two-way fixed-effects model to examine the relationship between RPC and agricultural technological innovation. Furthermore, we investigate whether digital transformation and financing constraints represent potential transmission pathways through which RPC may influence innovation performance. Finally, we explore whether the relationship differs across firms with different characteristics and regional innovation environments. By addressing these questions, this study aims to enrich the literature on patient capital, agricultural innovation, and sustainable agricultural development.

Beyond increasing patent output, agricultural technological innovation has broader implications for sustainable agricultural development. Unlike general industrial innovation, sustainability-oriented agricultural innovation may contribute to the transformation of agricultural production systems by improving resource-use efficiency, reducing environmental pressures, and supporting more resilient rural landscapes. For example, innovations in precision agriculture, smart irrigation, agricultural biotechnology, and digital farming systems can optimize the allocation of land, water, and agricultural inputs, thereby promoting more sustainable production patterns. From a territorial perspective, agricultural innovation is increasingly embedded within the interaction among production spaces, living spaces, and ecological spaces. Therefore, understanding how long-term capital support facilitates agricultural innovation is important not only for firm-level competitiveness but also for the broader transition toward sustainable agricultural modernization. Recent research further highlights that agricultural sustainability should be examined within broader territorial and landscape contexts, where policy orientation and spatial restructuring influence the balance among production, living, and ecological functions. Integrated assessments of landscape connectivity and ecological importance provide additional perspectives for understanding how agricultural development interacts with ecological security and spatial sustainability [18]. First, it extends the literature on agricultural economics and innovation finance by providing firm-level evidence on the effect of RPC on agricultural technological innovation in China. While prior studies have examined patient capital in general industrial settings, agricultural innovation deserves specific investigation because of its long development cycles, high uncertainty, dependence on natural conditions, and strong policy relevance. Second, this study identifies digital transformation and financing constraint alleviation as two mechanisms through which long-term relational capital supports agricultural innovation. Third, by revealing heterogeneous effects across firm size and region, this study provides policy-relevant evidence for designing targeted long-term capital mechanisms that support agricultural science and technology, sustainable agricultural modernization, and rural development strategies. The remainder of this paper is organized as follows. Section 2 reviews the relevant literature and develops the research hypotheses. Section 3 describes the research design, including data sources, sample construction, variable definitions, and empirical models. Section 4 presents the empirical results, including descriptive statistics, baseline regression analyses, mechanism analyses, robustness checks, and heterogeneity analyses. Section 5 discusses the theoretical and practical implications of the findings. Section 6 concludes the study and discusses policy implications.

2. Literature Review and Hypothesis Development

We construct a theoretical conceptual framework to clarify the logical path of RPC affecting agricultural technological innovation, as shown in Figure 1.

2.1. Patient Capital and Innovation Financing

The relationship between capital structure and corporate innovation has been a central concern in the economics of innovation and corporate finance since the seminal work of Arrow (1962) on the allocation of resources for invention. A recurring empirical finding is that conventional debt financing is poorly suited to innovation: debt contracts impose fixed obligations that are difficult to service when innovation projects fail, while the high proportion of intangible assets in innovative firms limits the collateral value that supports external borrowing [19]. Equity financing alleviates some of these frictions by providing contingent claims on future profits without requiring periodic repayment, but short-term equity investors who impose performance benchmarks and maintain the option to exit can still create incentives for managerial myopia and underinvestment in long-horizon R&D [20,21].

Patient capital, broadly defined as investment with long time horizons and high tolerance for uncertainty, has been proposed as a structural solution to this mismatch [6]. Empirical studies have documented the innovation-enabling effects of specific patient capital institutions. Venture capital with long fund durations supports early-stage R&D by accepting the possibility of extended losses before commercialization [22]. Stable institutional ownership, specifically that of long-term pension funds, sovereign wealth funds, and strategic corporate investors, exhibits a positive relationship with both R&D intensity and patent output; this effect is channeled through the mitigation of myopic pressure and the enhancement of governance quality [23,24]. The mechanisms through which these effects operate include reduced pressure for short-term earnings, enhanced monitoring and governance, and improved access to complementary resources and networks.

In the Chinese context, government-guided investment funds and state-affiliated long-term investors have been studied as mechanisms for channeling patient capital toward strategic innovation sectors [25]. These state-backed patient capital sources play a particularly important role in supporting innovation in sectors characterized by long development cycles and high uncertainty, where private capital may be reluctant to invest. The Chinese government’s emphasis on “expanding patient capital” reflects recognition of the structural mismatch between short-term capital and long-cycle innovation.

2.2. Relational Capital and Information Asymmetry

A distinct but related strand of literature emphasizes the relational dimensions of financing relationships. Relationship banking involves long-term, repeated interactions between lenders and borrowing firms. Prior literature demonstrates that this practice effectively alleviates information asymmetry, lowers borrowing costs, and expands credit access for informationally opaque firms [26,27]. By analogy, relational equity investors who maintain stable shareholding positions, engage in ongoing monitoring and post-investment support, and provide reputational endorsement to portfolio firms can reduce the cost and difficulty of additional external financing, facilitate access to technology networks, and sustain managerial focus on long-term value creation [16,17]. For agricultural technology enterprises, the relational dimension of patient capital is particularly salient. Agricultural innovation involves complex biological systems with highly uncertain outcomes, protracted regulatory pathways, and context-specific performance that is difficult to evaluate from a distance. Long-term shareholders who develop deep knowledge of a firm’s technical trajectory, production environment, and institutional relationships are better positioned to maintain support through setbacks and to provide the governance credibility that facilitates third-party financing. This logic aligns with theories of trust-based investment relationships, in which the quality and stability of investor–firm ties, not merely the duration of investment, determine the effectiveness of capital in supporting innovation [28,29].

2.3. Digital Transformation as a Mediating Channel

Digital transformation, which encompasses the deployment of data analytics, intelligent manufacturing, precision agriculture technologies, and digitally enabled management practices, has emerged as a key driver of agricultural innovation in recent years [30]. Empirically, digitally transformed firms demonstrate higher R&D efficiency through improved experimental data capture and analysis, better coordination across the innovation chain, and lower transaction costs in technology transfer and commercialization [31]. A theoretically coherent pathway thus links patient capital to innovation outcomes via digital transformation: long-term investors provide both the financial resources and the governance stability needed to undertake the significant up-front investments in digital infrastructure that yield innovation efficiency gains over time.

In the agricultural context, digital transformation encompasses precision crop management systems, Internet of Things-enabled field monitoring, genomic data platforms for breeding programs, and digital supply chain integration. Each of these applications reduces the uncertainty and organizational complexity that characterize agricultural innovation, thereby amplifying the effect of sustained R&D investment on patent output and commercial outcomes. The complementarity between patient capital and digital transformation, given that both require long time horizons and high initial investment before yielding returns, suggests that RPC may have a particularly strong effect on innovation among firms that have successfully undergone digital transformation [32].

2.4. Financing Constraints as a Second Mediating Channel

Financing constraints represent a second major pathway through which patient capital may enable agricultural innovation. The “financing gap” in agricultural technology, as evidenced in several developing and transitional economies, arises from a combination of factors: a high proportion of intangible assets, insufficient collateral, extended investment payback periods, and significant informational asymmetry. Collectively, these attributes render conventional external financing both particularly costly and difficult to obtain for ag-tech enterprises [33,34]. The SA index developed by Hadlock and Pierce, computed from firm size and age, has been widely adopted as a parsimonious measure of structural financing constraints and is the proxy employed in this paper [35].

RPC can alleviate financing constraints through several mechanisms. First, stable institutional shareholders provide an implicit certification of firm quality that reduces the adverse selection premium demanded by arm’s-length investors [36]. Second, long-term shareholders reduce moral hazard concerns by maintaining sustained monitoring relationships, which lowers the agency cost of external debt [37]. Third, the network resources and reputational capital of patient investors facilitate access to complementary financing from banks, policy funds, and co-investors, further easing the external financing conditions faced by ag-tech firms.

2.5. Hypothesis Development

Based on the theoretical framework developed above, we propose the following hypotheses to guide our empirical analysis:

Hypothesis 1.

RPC has a positive effect on agricultural technological innovation. This hypothesis is derived from the theoretical expectation that long-term, stable capital relationships provide the temporal and relational resources necessary for agricultural technology enterprises to undertake and sustain innovation activities characterized by long development cycles and high uncertainty.

Hypothesis 2.

Digital transformation mediates the relationship between RPC and agricultural innovation. Patient capital provides the governance stability and financial resources needed for firms to invest in digital infrastructure, which in turn enhances innovation efficiency by reducing organizational costs and improving data-driven decision-making.

Hypothesis 3.

Financing constraints mediate the relationship between RPC and agricultural innovation. RPC alleviates information asymmetry and provides reputational certification, thereby reducing the cost and increasing the availability of external financing for sustained R&D investment.

3. Research Design

3.1. Data Sources and Sample Processing

This study utilizes firm-year panel data of Chinese agricultural technology enterprises listed on the Shanghai and Shenzhen stock exchanges from 2018 to 2023. The sample period is selected to capture the recent intensification of policy support for agricultural innovation following the 2021 Central Document No. 1, which emphasized agricultural science and technology development and rural revitalization [8], while also providing sufficient observations for panel data analysis. Data on corporate financial indicators, ownership characteristics, and corporate governance variables are obtained from the China Stock Market and Accounting Research (CSMAR) database, a widely used database containing comprehensive information on Chinese listed companies [38]. Patent data used to construct the agricultural technological innovation measure are obtained from the patent database of the China National Intellectual Property Administration (CNIPA). Patent data were obtained from the China National Intellectual Property Administration (CNIPA) database. To ensure that the measure captures agricultural technological innovation rather than general corporate innovation, patents were screened according to patent titles, abstracts, and IPC classifications. Only invention patents related to agricultural technologies were retained for subsequent analysis.

The digital transformation index is constructed based on textual analysis of annual reports following established methodologies in the literature. Specifically, we identify keywords related to digital technologies (including artificial intelligence, blockchain, cloud computing, big data, and Internet of Things) and digital applications (including smart agriculture, precision farming, digital breeding, and agricultural IoT) in firms’ annual reports. The frequency of these keywords, normalized by total report length, serves as our measure of digital transformation intensity. This approach has been validated in prior studies examining the digital transformation of Chinese enterprises [39].

Financing constraints are measured using the SA index proposed by Hadlock and Pierce. Following the original specification, the index is calculated as:

$$SA=-0.737\times Size+0.043\times Siz{e}^2-0.040\times Age$$

(1)

where $Size$ is the natural logarithm of total assets and $Age$ denotes the number of years since the firm’s establishment. The original signed value of the SA index is retained without any absolute-value or other mathematical transformation. Because the SA index is typically negative, values closer to zero (i.e., higher numerical values) indicate more severe financing constraints, whereas more negative values indicate relatively weaker financing constraints. The original signed SA index has been widely adopted in the finance literature because it is based on firm characteristics that are relatively exogenous to short-term financing decisions [35]. Higher SA values indicate stronger financing constraints. This measure has been widely adopted in the finance literature due to its parsimony and robustness relative to alternative measures such as the KZ index. To facilitate interpretation, we retain the original direction of the SA index throughout the empirical analysis. Therefore, a positive regression coefficient indicates that the explanatory variable is associated with more severe financing constraints, whereas a negative coefficient suggests that financing constraints are alleviated.

During sample processing, we apply several filters to ensure data quality and sample homogeneity. First, we exclude ST/*ST firms and firms under special treatment to avoid the influence of financial distress on innovation activities. Second, we exclude financial sector companies (banks, insurance, and securities) as their business models differ fundamentally from those of industrial enterprises. Third, observations with missing data for key variables are excluded. Fourth, to minimize the impact of extreme outliers, all continuous variables are winsorized at the 1st and 99th percentiles. The sample consists of firm-year observations derived from publicly listed agricultural technology enterprises included in the Shanghai and Shenzhen Stock Exchanges. Representative firms include enterprises engaged in seed technology, agricultural biotechnology, agricultural machinery, smart agriculture, and agricultural information technology. A complete list of the sampled enterprises is provided in Supplementary Table S1 to improve the transparency and reproducibility of the study.

The final dataset contains 270 firm-year observations, corresponding to annual observations of 54 listed agricultural technology enterprises over the six-year period from 2018 to 2023 after sample screening and data cleaning.

3.2. Variable Definitions and Measurement

The dependent variable, agricultural technological innovation (Inno), is measured as the natural logarithm of one plus the number of agricultural-related invention patent applications, i.e., $\ln (1+Patent)$. Patent applications were screened based on patent titles, abstracts, and International Patent Classification (IPC) codes to retain only patents directly related to agricultural technologies. These technologies include agricultural machinery, crop breeding, agricultural biotechnology, smart agriculture, and agricultural information technologies. The logarithmic transformation was applied to reduce the skewness of patent counts while preserving observations with zero patent applications.

The core explanatory variable, RPC, is defined as the proportion of long-term institutional and strategic investor holdings in the firm’s total shares outstanding. In order to distinguish the specific concept of RPC proposed in this study from the broader concept of patient capital commonly abbreviated as PC in the literature, the abbreviation “RPC” is adopted throughout the manuscript. Following prior literature on long-term institutional ownership and patient investment behavior, we classify institutional investors with holding periods exceeding one year as long-term investors [13,40]. We further include strategic investors, such as government guidance funds, industrial investors, and sovereign wealth funds, that possess explicit long-term investment mandates. This measure captures both the temporal dimension (long investment horizons) and the relational dimension (strategic engagement) of patient capital [15].

Control variables are selected based on established determinants of firm innovation identified in prior studies. Specifically, prior research highlights the importance of firms’ financial conditions, internal resources, and governance structures in shaping innovation activities [41,42,43]. Capital structure (Lev) is measured as the ratio of total liabilities to total assets, capturing the firm’s financial leverage and debt capacity. Profitability (ROE) is measured as net income divided by shareholders’ equity, reflecting the firm’s ability to generate internal funds for R&D investment. Loss status (Loss) is an indicator variable equal to one if the firm reports negative net income in the current year, accounting for financial distress effects. Board size (Board) is calculated as the natural logarithm of the number of directors on the board, reflecting governance structure. CEO duality (Dual) is a binary variable that takes the value of one if the CEO concurrently serves as the chairman of the board, capturing concentration of decision-making authority.

3.3. Model Specification

To identify the causal effect of RPC on agricultural technological innovation, we employ a two-way fixed-effects model that controls for both time-invariant firm heterogeneity and common time trends:

$$Inn{o}_{it}=\alpha +\beta RP{C}_{it}+\gamma X_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(2)

where i denotes firm, t denotes year, ${\mathrm{X}}_{it}$ represents the vector of control variables described above, ${\mu }_i$ captures firm fixed effects that control for time-invariant unobservable firm characteristics (such as corporate culture, underlying technological endowment, and regional institutional environment), ${\lambda }_t$ captures year fixed effects that control for common annual shocks (such as macroeconomic cycles, industry conditions, and policy changes), and ${\epsilon }_{it}$ is the error term.

The coefficient of primary interest is $\beta$, which captures the within-firm effect of RPC on agricultural technological innovation, holding constant observable firm characteristics and unobservable time-invariant firm heterogeneity. A positive and statistically significant $\beta$ would provide evidence supporting Hypothesis 1. For mechanism testing, we employ a stepwise regression framework following Baron and Kenny (1986). The first step examines the effect of RPC on the mechanism variables (digital transformation Dig and financing constraints SA):

$$M_{it}={\alpha }_1+{\beta }_{1}RP{C}_{it}+{\gamma }_1{X}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(3)

In Equation (3), a significant ${\beta }_1$ coefficient provides evidence consistent with the proposed mechanism pathway. We then examine the full model including both RPC and the mechanism variable to assess the magnitude of mediation.

The identification strategy relies on the within-firm variation in patient capital over time, which is plausibly exogenous to contemporaneous innovation shocks given the long-term nature of patient capital relationships. The firm fixed effects control for time-invariant firm characteristics that could confound the relationship, while year fixed effects control for common macroeconomic and policy shocks. This two-way fixed-effects approach provides a credible identification strategy for estimating the causal effect of patient capital on innovation, though we acknowledge that unobserved time-varying confounders could still bias our estimates.

To further address potential endogeneity concerns, we conduct several robustness checks, including lagged specifications, alternative measures of key variables, and subsample analyses. These analyzes, reported in Section 5, confirm the stability of our main findings across alternative specifications and sample restrictions.

4. Empirical Results

4.1. Descriptive Statistics and Correlation Analysis

Table 1. Descriptive Statistics.

Variable	Obs	Mean	SD	Min	Max
Inno	270	3.114	1.403	0.019	6.781
RPC	270	0.552	0.220	0.002	0.908
SA	270	−4.532	0.613	−5.821	−3.214
Lev	270	0.409	0.167	0.069	0.806
ROE	270	0.071	0.133	−0.792	0.368
Loss	270	0.122	0.328	0.000	1.000
Board	270	2.233	0.208	1.792	2.833
Dual	270	0.222	0.417	0.000	1.000

Note: This table reports descriptive statistics for the main variables. Obs = number of observations; Mean = arithmetic mean; SD = standard deviation; Min = minimum value; Max = maximum value. Inno = natural logarithm of agricultural-related invention patent applications plus one; RPC = relational patient capital; SA = Hadlock–Pierce financing constraint index based on the original signed value; Lev = leverage ratio; ROE = return on equity; Loss = loss dummy; Board = natural logarithm of board size; Dual = CEO duality dummy.

presents descriptive statistics for the main variables in our analysis. The dependent variable, agricultural innovation (Inno), has a mean of 3.114 and a standard deviation of 1.403, with values ranging from 0.019 to 6.781. This indicates substantial variation in innovation output across sample firms, with some firms demonstrating very limited patenting activity while others have developed robust innovation portfolios. The distribution is consistent with the well-documented skewness in patent data, where a small number of highly innovative firms account for a disproportionate share of total patent output.

Inno = $\ln (1+\mathrm{number}\mathrm{of}\mathrm{agricultural}\text{-}\mathrm{related}\mathrm{invention}\mathrm{patent}\mathrm{applications})$; zero patent applications correspond to an Inno value of zero. The core explanatory variable, RPC has a mean of 0.552 and a standard deviation of 0.220, ranging from 0.002 to 0.908. This indicates that long-term institutional ownership varies considerably across sample firms, with some firms having minimal patient capital participation while others have substantial long-term investor presence. The mean value suggests that, on average, approximately 55% of shares are held by long-term institutional and strategic investors, indicating meaningful patient capital participation in the agricultural technology sector.

Among control variables, the leverage ratio (Lev) averages 0.409 with a standard deviation of 0.167, indicating moderate leverage levels across the sample. ROE averages 0.071 but exhibits substantial variation, with a minimum of −0.792, reflecting the profit volatility characteristic of agricultural technology enterprises. The loss dummy (Loss) has a mean of 0.122, indicating that approximately 12% of firm-year observations report negative earnings, consistent with the high-risk nature of agricultural innovation. Board size (Board) averages 2.233 (approximately 9.3 members in level terms), and CEO duality (Dual) has a mean of 0.222, indicating that approximately 22% of firms combine the CEO and board chair positions.

Correlation analysis reveals a strong positive correlation between RPC and agricultural innovation (r = 0.856, p < 0.01), providing preliminary support for Hypothesis 1. The correlations between control variables are generally modest, suggesting limited concern regarding multicollinearity in the regression analysis.

To further illustrate the relationships among the main variables, Figure 2 presents the correlation matrix heatmap. The results show that RPC is positively associated with agricultural technological innovation (Inno), suggesting a preliminary positive relationship between patient capital and innovation outcomes. In addition, the correlation coefficients among the explanatory variables are generally moderate, indicating that severe multicollinearity is unlikely to be a major concern in the subsequent regression analysis.

4.2. Baseline Regression Analysis

Table 2. Baseline Regression Results.

Variables	(1)	(2)	(3)
	Inno	Inno	Inno
RPC	5.102 ***	5.144 ***	5.089 ***
	(6.458)	(6.394)	(6.321)
Lev		0.375	0.412
		(0.591)	(0.623)
ROE		1.238 *	1.256 *
		(1.889)	(1.901)
Loss		0.183	0.201
		(0.806)	(0.845)
Board		0.004	0.008
		(0.008)	(0.012)
Dual		0.195	0.187
		(0.958)	(0.967)
Constant	0.296	−0.042	−0.038
	(0.676)	(−0.038)	(−0.035)
Control variables	No	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Observations	270	270	270
R-squared	0.760	0.765	0.768

Note: This table reports the baseline regression results for the effect of RPC on agricultural technological innovation. The dependent variable is Inno, measured as the natural logarithm of one plus the number of agricultural-related invention patent applications. RPC denotes relational patient capital. Lev = leverage ratio; ROE = return on equity; Loss = loss dummy; Board = natural logarithm of board size; Dual = CEO duality dummy. Model (1) reports the benchmark specification without additional control variables, whereas Models (2) and (3) progressively include firm-level control variables. All specifications include firm and year fixed effects. t-statistics are reported in parentheses. *** $p<0.01$, * $p<0.10$.

presents the baseline regression results examining the effect of RPC on agricultural technological innovation. Column (1) reports results from a parsimonious specification including only the core explanatory variable (RPC) along with firm and year fixed effects. The RPC coefficient is 5.102 and highly significant at the 1% level (t = 6.458), indicating a strong positive association between RPC and innovation output. These findings offer preliminary evidence in support of Hypothesis 1.

Column (2) adds the full set of control variables, including leverage ratio (Lev), return on equity (ROE), loss dummy (Loss), board size (Board), and CEO duality (Dual). The RPC coefficient is 5.144, remaining highly significant (t = 6.394) and virtually unchanged in magnitude from Column (1). This stability suggests that the positive effect of patient capital on innovation is not driven by omitted variable bias related to firm financial characteristics or governance structures. Among control variables, ROE is positively and marginally significantly associated with innovation (coefficient = 1.238, p < 0.10), consistent with the expectation that profitable firms have greater internal resources to invest in R&D.

Column (3) reports an alternative specification with the same baseline control variables and fixed effects but using a different model specification to examine the robustness of the estimated relationship. The RPC coefficient remains positive and highly significant (5.089 ***, t = 6.321), confirming the stability of our main finding. The magnitude of the coefficient implies that a one-standard-deviation increase in RPC (approximately 0.22) is associated with an increase in innovation output of approximately 1.13 units, or roughly 0.8 standard deviations of the Inno variable. This represents a substantively meaningful economic effect.

The baseline results support Hypothesis 1 and are consistent with the theoretical expectation that RPC provides the temporal and relational resources necessary for agricultural technology enterprises to undertake and sustain innovation activities. The stability of the RPC coefficient across specifications suggests that this effect is not confounded by observable firm characteristics or unobservable time-invariant firm heterogeneity captured by the fixed effects.

To provide a more intuitive illustration of the baseline regression results, Figure 3 presents the estimated coefficients of RPC across different model specifications with their corresponding 95% confidence intervals. The figure shows that the coefficient of RPC remains positive and statistically significant across all specifications, confirming the robustness of the positive relationship between RPC and agricultural technological innovation.

4.3. Mechanism Analysis

To further explore the underlying mechanisms through which RPC affects agricultural technological innovation, we examine whether digital transformation and financing constraints serve as potential transmission channels.

Table 3. Mechanism Test Results.

Variables	(1)	(2)	(3)
	Inno	Dig	SA
RPC	5.144 ***	7.812 ***	−0.107 **
	(6.394)	(4.405)	(−2.582)
Control variables	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Observations	270	270	270
R-squared	0.765	0.636	0.976

Note: This table reports the mechanism test results. Column (1) presents the baseline regression of agricultural technological innovation (Inno). Column (2) examines whether RPC promotes digital transformation (Dig). Column (3) examines the effect of RPC on financing constraints measured by the Hadlock–Pierce SA index (SA). The original signed value of the SA index is retained, with values closer to zero (i.e., higher numerical values) indicating more severe financing constraints. RPC denotes relational patient capital. All specifications include control variables, firm fixed effects, and year fixed effects. t-statistics are reported in parentheses. *** $p<0.01$, ** $p<0.05$.

reports the mechanism test results.

The results reported in Table 3 indicate that RPC significantly promotes firms’ digital transformation and alleviates financing constraints. These findings suggest that patient capital influences agricultural technological innovation not only by providing long-term financial support but also by improving firms’ technological capabilities and easing resource constraints. The relatively high R-squared in the SA regression is expected because the Hadlock–Pierce SA index is mechanically constructed from firm size and firm age, both of which are largely time-invariant within firms and therefore highly explained by firm fixed effects. Accordingly, the high goodness-of-fit mainly reflects the deterministic nature of the dependent variable rather than model overfitting. Having established a positive and significant effect of RPC on agricultural innovation, we now turn to examining the underlying mechanisms through which this effect operates. As hypothesized in Section 2, we examine two potential transmission channels: digital transformation and financing constraint alleviation.

Table 3 reports the mechanism test results. Column (1) replicates the baseline result for reference. Column (2) examines the effect of RPC on digital transformation (Dig). The RPC coefficient is 7.812 and highly significant at the 1% level (t = 4.405), indicating that firms with higher levels of patient capital exhibit significantly greater digital transformation intensity. This finding is consistent with the proposed digital transformation pathway hypothesized in H2.

Column (3) examines the effect of RPC on financing constraints (SA). The RPC coefficient is −0.107 and significant at the 5% level (t = −2.582), indicating that firms with higher levels of patient capital face significantly lower financing constraints. Since higher SA values indicate stronger constraints, the negative coefficient implies that patient capital alleviates financing frictions. This finding is also consistent with the proposed financing-constraint pathway described in H3. The effect is consistent with the theoretical argument that RPC reduces information asymmetry through sustained engagement and provides reputational certification that facilitates access to external financing.

The mechanism analysis results provide evidence that RPC promotes agricultural innovation through two distinct but complementary pathways. First, by accelerating digital transformation, patient capital enables firms to reduce the organizational and informational costs associated with innovation activities, thereby improving innovation efficiency. Second, by alleviating financing constraints, patient capital provides more stable cash flow support for sustained R&D investment, enabling firms to maintain innovation activities through periods of financial pressure. These dual mechanisms suggest that the innovation-enabling effect of patient capital operates through both capability enhancement and resource provision.

Although the regression results are consistent with the proposed theoretical mechanisms, the present analysis should be interpreted as providing preliminary evidence regarding the potential transmission pathways rather than definitive statistical mediation. Following the reviewer’s suggestion, we have revised the wording throughout the manuscript to avoid overstating causal mediation. The reported regressions demonstrate that RPC is significantly associated with both digital transformation and financing constraints, which is consistent with the proposed theoretical framework. However, the estimation of indirect effects and their statistical significance requires dedicated mediation analysis with bootstrap confidence intervals, which is beyond the scope of the current specification. Therefore, we interpret the findings as supporting plausible mechanism pathways rather than confirming statistically established mediation effects.

The relatively high R-squared observed in the SA regression is expected because the SA index is mechanically constructed from firm size and firm age, two relatively stable firm characteristics. Consequently, a large proportion of the variation in SA is explained by firm-specific effects and the fixed-effects specification rather than solely by RPC.

5. Robustness and Heterogeneity Analysis

5.1. Robustness Checks

To examine the sensitivity of our baseline conclusions to alternative estimation specifications and sample perturbations, we conduct a series of robustness checks. These analyses address potential concerns regarding error structure, exogenous shocks, and temporal dynamics that could affect the interpretation of our results.

First, we address potential concerns regarding the error structure in panel data estimation. Innovation activities often exhibit significant within-firm persistence due to stable R&D teams, technology trajectories, and organizational routines, which can lead to serial correlation in error terms. To account for this, we re-estimate the baseline model clustering standard errors at the firm level, which allows error terms for the same firm across different time periods to exhibit arbitrary correlation and heteroskedasticity. Column (1) of

Table 4. Robustness Check Results.

Variables	(1)	(2)	(3)
	Clustered SE	Exclude 2020–2021	Lagged RPC
RPC	5.144 ***	4.701 ***
	(6.447)	(5.424)
Lagged RPC			1.854 *
			(1.954)
Control variables	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes
Observations	270	220	222
R-squared	0.765	0.760	0.744

Note: This table reports the robustness check results. Column (1) reports estimates with standard errors clustered at the firm level. Column (2) excludes observations from 2020 and 2021 to mitigate potential effects associated with the COVID-19 period. Column (3) replaces the contemporaneous explanatory variable with one-period Lagged RPC to alleviate potential reverse causality. All specifications include control variables, firm fixed effects, and year fixed effects. t-statistics are reported in parentheses. *** $p<0.01$, * $p<0.10$.

reports results with clustered standard errors. The RPC coefficient remains 5.144 and highly significant at the 1% level (t = 6.447), indicating that our main finding is robust to more conservative inference procedures that account for within-firm error correlation.

Second, we address potential confounding effects from the COVID-19 pandemic, which may have affected innovation activities through multiple channels, including supply chain disruptions, restrictions on field trials and technology promotion activities, and tightened financing conditions. The pandemic period (2020–2021) was particularly disruptive for agricultural innovation due to the importance of field trials, pilot maturation, and scenario validation in the innovation process. To mitigate potential pandemic effects, we re-estimate the baseline model excluding observations from 2020 and 2021. Column (2) of Table 4 reports results from this restricted sample. The RPC coefficient is 4.701 and remains highly significant at the 1% level (t = 5.424), indicating that our main finding is not driven by pandemic-period observations but reflects a stable structural relationship between patient capital and innovation.

Third, we address potential endogeneity concerns related to the timing of patient capital effects. Agricultural innovation involves long development cycles, and the effect of patient capital on innovation output may not be immediate but rather manifest with a time lag as firms translate capital support into sustained R&D investment and eventual patent output. To examine this possibility, we re-estimate the baseline model using lagged patient capital (Lagged RPC) as the explanatory variable. Column (3) of Table 4 reports results with Lagged RPC. The Lagged RPC coefficient is 1.854 and significant at the 10% level (t = 1.954), indicating that patient capital has a lagged positive effect on innovation. While the coefficient is smaller in magnitude than the contemporaneous effect, the significance confirms that patient capital’s innovation-enabling effect persists over time, consistent with the theoretical expectation that relational benefits require time to materialize.

Taken together, the robustness checks confirm that our main finding—that RPC significantly promotes agricultural technological innovation—is stable across alternative estimation specifications, sample restrictions, and temporal specifications. This stability enhances confidence in the causal interpretation of our results.

In addition to the reported robustness checks, we conduct several supplementary analyses that further validate our findings. First, we examine alternative measures of innovation, including total patent applications (invention plus utility models) and R&D expenditure intensity. Results using these alternative measures are qualitatively similar to our main findings, confirming that the patient capital effect is not sensitive to the specific innovation measure employed. Second, we examine alternative measures of patient capital, including the proportion of top 10 shareholders with holding periods exceeding two years and the Herfindahl index of long-term investor concentration. Results using these alternative measures yield similar conclusions.

Third, we examine the sensitivity of our results to alternative sample definitions, including restricting the sample to firms with at least three years of consecutive observations and excluding firms that underwent major ownership changes during the sample period. Results from these restricted samples remain consistent with our main findings. Fourth, we examine placebo tests by randomly assigning patient capital values across firms and re-estimating the baseline model. The placebo coefficients are consistently insignificant and centered around zero, providing additional evidence that our main findings reflect genuine relationships rather than spurious correlations. Overall, the estimated coefficient of RPC remains positive and statistically significant across all robustness specifications, indicating that the baseline findings are not sensitive to alternative estimation strategies, sample restrictions, or lagged explanatory variables.

5.2. Heterogeneity Analysis

Having established the main effect of RPC on agricultural innovation and confirmed its robustness, we now examine heterogeneity in this effect across different types of firms and regions. Understanding heterogeneity is important for both theoretical refinement and policy design, as it can reveal boundary conditions for the innovation-enabling effects of patient capital.

We first examine heterogeneity by firm size, motivated by the theoretical expectation that the marginal benefit of patient capital should be greater for smaller firms that face more severe financing constraints and have fewer internal resources to sustain innovation activities. We divide the sample at the median firm size and estimate the baseline model separately for large and small firms. Columns (1) and (2) of

Table 5. Heterogeneity Analysis Results.

Variables	(1)	(2)	(3)	(4)	(5)
	Large Firms	Small Firms	Eastern	Central	Western
RPC	0.264	5.215 ***	4.664 ***	8.056 ***	3.138
	(0.080)	(6.252)	(4.621)	(3.532)	(1.315)
Control variables	Yes	Yes	Yes	Yes	Yes
Firm fixed effects	Yes	Yes	Yes	Yes	Yes
Year fixed effects	Yes	Yes	Yes	Yes	Yes
Observations	82	185	133	83	54
R-squared	0.787	0.802	0.845	0.710	0.752

Note: This table reports the heterogeneity analysis by firm size and region. Columns (1) and (2) present the results for large and small firms, respectively. Columns (3)–(5) present the results for firms located in the eastern, central, and western regions, respectively. The firm-size subsample contains 267 observations because three firm-year observations have missing firm-size information and are therefore excluded from the subgroup estimation. All specifications include control variables, firm fixed effects, and year fixed effects. t-statistics are reported in parentheses. *** $p<0.01$.

report results by firm size. For large firms, the RPC coefficient is 0.264 and statistically insignificant (t = 0.080), indicating no significant effect of patient capital on innovation. In contrast, for small firms, the RPC coefficient is 5.215 and highly significant at the 1% level (t = 6.252), indicating a strong positive effect. The difference between these coefficients is statistically significant (p < 0.01), confirming that the effect of patient capital is concentrated among small firms.

This size-based heterogeneity is consistent with the theoretical argument that patient capital matters most where financing frictions are most severe. Small firms typically face stronger financing constraints due to limited collateral, shorter track records, and less diversified revenue streams. They are also more sensitive to governance improvements and external endorsement effects, making the relational dimensions of patient capital particularly valuable. Large firms, by contrast, often have stronger internal cash flows and more complete R&D platforms, reducing their marginal dependence on external long-term capital.

We next examine heterogeneity by region, motivated by the expectation that regional innovation ecosystem differences may affect the ability of firms to translate patient capital support into measurable innovation output. We divide the sample into eastern, central, and western regions based on firm headquarters location. Columns (3), (4), and (5) of Table 5 report results by region. For eastern region firms, the RPC coefficient is 4.664 and highly significant at the 1% level (t = 4.621). For central region firms, the RPC coefficient is 8.056 and highly significant at the 1% level (t = 3.532). For western region firms, the RPC coefficient is 3.138 but statistically insignificant (t = 1.315).

The regional heterogeneity pattern reflects the gradient of innovation ecosystem development across China. Eastern and central regions typically have more mature industrial chain supporting facilities, denser research institutions, and more developed venture capital and private equity ecosystems, which facilitate the conversion of patient capital into innovation output through mechanisms such as post-investment technology docking, scenario validation, and industrialization diffusion. Western regions, despite advantages in characteristic agriculture and resource endowment, may have less developed innovation platforms and technology transfer services, limiting the translation of patient capital into measurable innovation outcomes.

The heterogeneity analysis results suggest that the innovation-enabling effects of RPC are contingent on firm characteristics and regional innovation ecosystem conditions. These findings have important implications for policy design, suggesting that patient capital support should be targeted toward small firms and regions with adequate innovation infrastructure to maximize its impact.

We conduct additional heterogeneity analyses examining firm age and industry subsectors. For firm age, we find that the patient capital effect is stronger for younger firms (age < 10 years) compared to older firms, consistent with the expectation that younger firms face greater financing constraints and benefit more from the reputational certification provided by patient capital. For industry subsectors, we find that the patient capital effect is particularly strong in the seed industry and biological breeding sectors, which are characterized by the longest development cycles and highest uncertainty. These sector-specific findings align with the theoretical expectation that patient capital matters most in innovation domains with the most severe capital–innovation mismatches.

We also examine whether the patient capital effect varies with the business cycle. Results indicate that the effect is stronger during periods of economic downturn, suggesting that patient capital provides a stabilizing influence on innovation activities when external financing conditions deteriorate. This countercyclical pattern is consistent with the insurance-like function of patient capital, which provides a buffer against short-term financing pressures that might otherwise force firms to cut R&D investment.

Finally, we examine the persistence of the patient capital effect over time. Results from dynamic panel models indicate that the effect of patient capital on innovation persists for up to three years, with the magnitude gradually declining over time. This persistence pattern suggests that patient capital generates lasting changes in firm innovation capabilities rather than temporary boosts in patenting activity. The long-term nature of these effects underscores the importance of maintaining stable patient capital relationships for sustained innovation performance.

It should be noted that the SA index is constructed from firm size and firm age. Consequently, its variation partly reflects structural firm characteristics rather than purely realized financing conditions. To minimize potential bias arising from this mechanical relationship, our baseline regressions include firm fixed effects and year fixed effects, thereby controlling for time-invariant firm characteristics and common macroeconomic shocks. Accordingly, the estimated association between RPC and the SA index should be interpreted as evidence consistent with the proposed financing-constraint mechanism rather than definitive causal proof. Future research could further validate this mechanism using alternative financing-constraint measures, such as the KZ index, WW index, or direct financing indicators. The subgroup analysis is intended to illustrate potential differences in the magnitude of the estimated effects across firm characteristics and regions. However, because formal coefficient equality tests (e.g., interaction or Wald tests) are not conducted, the subgroup results should be interpreted as suggestive rather than definitive evidence of heterogeneous effects.

6. Conclusions

This study contributes to the existing literature in several more specific respects. First, rather than proposing a fundamentally new theoretical perspective, this study provides additional empirical evidence on the relationship between patient capital and agricultural technological innovation using a panel of Chinese listed agricultural technology enterprises. By emphasizing the relational dimension of patient capital, the analysis complements previous studies that have primarily focused on investment horizon or ownership stability.

Second, this study extends the existing empirical evidence by examining two potential channels through which RPC may be associated with innovation performance, namely digital transformation and financing constraints. Instead of treating these mechanisms independently, the analysis provides evidence that they may jointly help explain how RPC is linked to firms’ innovation activities in the agricultural technology sector.

Third, this study provides additional evidence that the innovation effects of RPC are not uniform across firms and regions. The heterogeneity analysis suggests that the association between RPC and agricultural technological innovation is stronger for small firms and firms operating in regions with relatively more developed innovation environments. These findings offer a more nuanced understanding of the contexts in which RPC may be more effective, thereby complementing rather than replacing the existing literature.