Impact of Economic Policy Uncertainty on Agribusiness Technology Innovation: Evidence from 231 Listed Firms in China

Abstract

The role of enterprise technological innovation in driving high-quality economic development has become increasingly prominent in a dynamic environment of uncertainty. This study examines the impact of economic policy uncertainty on enterprise technological innovation and its mechanism of action using panel data of agricultural listed firms in China 2007–2021. We obtain the following results. (1) Economic policy uncertainty significantly promotes technological innovation in agricultural firms, and this finding holds under various robustness checks, and works mainly through two paths: tightening financial flexibility and highlighting managerial capabilities. (2) From the heterogeneity results, the incentive effect of economic policy uncertainty on technological innovation in agricultural firms is more pronounced in state-owned agricultural firms, basic agricultural firms, and high adjustment cost agricultural firms. (3) Government subsidies strengthen the role of economic policy uncertainty in promoting technological innovation in agricultural firms. The results of this study provide data support for agricultural firms technology innovation decisions and have important implications for the guidance of national economic policy formulation.

1. Introduction

Macroeconomic fluctuations have gradually become the main theme of global economic development in recent years, and agriculture is more susceptible to external shock effects such as economic changes, natural disasters, and food safety [1]. According to the data published by the Ministry of Agriculture and Rural Affairs, as many as 1000 policies and regulations related to agricultural enterprises were published from 1995 to 2021 (as obtained by combing policies and regulations published on the official website of the Ministry of Agriculture and Rural Affairs, https://zwfw.moa.gov.cn/#/lawsList, accessed on 4 May 2023), but the economic environment is rapidly changing and there is an information asymmetry between economic policymakers and agricultural enterprises [2]. Agricultural listed enterprises are the key carriers for the realization of the modernization of agriculture with Chinese characteristics [3]. In order to cope with the severe challenges to agriculture from multiple changes in the external environment, technology, and resources, the production of agricultural enterprises must carry out technological innovation [4]. The quality of their technological innovation is directly related to the process of agricultural industrialization [5]. Policies are the “rules of the game” for business operations [6], and uncertainties in economic policies will have a profound impact on the technological innovation of enterprises [7].

Agribusiness is a new type of market operation subject that meets the basic characteristics of a modern enterprise, and its innovative behavior is affected by economic fluctuations and policy adjustments. However, compared with other enterprises, agricultural enterprises are “micro-profit” and naturally weak. On the one hand, the product characteristics and industry features of agribusiness production make their technological innovation behavior more sensitive to changes in economic policies. On the other hand, the production cycle of agribusiness is affected by climate, economy, and other comprehensive factors, so the characteristics of weakness will be mapped to the behavior of agricultural enterprises [8], and the technological innovation cycle is longer. In addition, the public production characteristics of agribusiness make technological innovation more dependent on economic policies, and policy supply becomes an important factor affecting the innovation process in the agricultural industry [9]. Therefore, it becomes an important theoretical question whether agribusiness can adapt to economic policy uncertainty and how policy uncertainty will affect agribusiness. However, most studies on the impact of economic policy uncertainty on agriculture have focused on the macro level. For example, past models have emphasized the role of business investment in increasing productivity under uncertainty [10,11]. In recent years, they have started to explore the impact of economic policy uncertainty on micro-agricultural firms, including firm investment, firm performance, and business financing. For example, Some scholars input uncertainty shocks induced by COVID-19 into the disaster impact estimation model developed by Nicholas and Stephen et al. [12]. but technological innovation in agricultural firms has been neglected by much of the literature.

The relationship between economic policy uncertainty and technological innovation in agribusiness has important theoretical implications. In economic theory, risk implies opportunity. According to resource dependence theory, agribusiness innovation requires high natural resources and production factors, and has high R&D adjustment costs, requiring continuous input from enterprises to obtain innovation output results. From the perspective of financial frictions, when economic policy uncertainties increase, the external financing constraints of agribusinesses are limited, and enterprises can only turn to “inward” investment in R&D and innovation in order to seek breakthroughs and profits. According to rational decision theory, when economic policy uncertainty increases, on the one hand, agricultural enterprises’ innovation behavior has greater technological uncertainty and longer construction period, which cannot be solved by postponing investment, and companies can only improve their success rate by gaining more experience through early research and development; at the same time, agricultural enterprises can enhance their self-development ability and competitive advantage through technological innovation to obtain market power and excess profits.

Existing researches on the impact of economic policy uncertainty on firm technological innovation can be broadly divided into two perspectives. One view is that economic policy uncertainty inhibits firm technological innovation [13,14,15]. Increased economic policy uncertainty increases uncertainty at the firm level, leading to an increased sense of “risk” on the part of firm management, which discourages firms from investing in innovation and thus reduces the level of firm technological innovation. The other view is that economic policy uncertainty promotes firm technological innovation [16,17,18,19,20]. Uncertainty is generally regarded as a source of profit in economics, and an increase in uncertainty increases the expected rate of return on firms’ innovation investment [21]. In addition, studies have shown that government subsidies play a moderating role in agricultural firms’ R&D investment [13]. Therefore, in studying the relationship between economic policy uncertainty and agricultural firms’ technological innovation, this paper also focuses on the moderating role of government subsidies.

This paper constructs a theoretical framework of “policy change-internal growth-technology innovation”, tries to theoretically explain the potential causal relationship between economic policy uncertainty and technological innovation in agricultural enterprises, and tries to expand the previous literature to include government subsidies in the study of the relationship and explore their moderating effects. The main contributions of this paper are reflected in the following three aspects: First, in the context of economic change and firm innovation, economic policy uncertainty is incorporated into agribusiness technological innovation, and this paper enriches the research in the field of firm technological innovation and economic policy by providing reliable empirical evidence that economic policy uncertainty promotes agribusiness technological innovation. Second, considering the heterogeneity of agricultural enterprises in terms of property rights nature, industry attributes, and R&D input adjustment costs, this paper integrates the heterogeneous impact of economic policy uncertainty on agricultural enterprises’ technological innovation. Third, unlike previous studies on a single mechanism of technological innovation, this paper introduces new variables from multiple perspectives to study the technological innovation behavior of agricultural enterprises under the influence of economic policy uncertainty. First, based on the perspective of agribusiness internal growth, this paper systematically analyzes the transmission paths from two channels: financing constraints and organizational dynamics, and tests them through two paths: financial flexibility and managerial competence. Second, government subsidies are introduced to explore their moderating effects on economic policy uncertainty and agribusiness technology innovation, extending the research on government innovation subsidies. Finally, this paper identifies the impact of environmental uncertainty on agribusiness technology innovation. Micro-environmental uncertainty is both significantly different from macroeconomic policy uncertainty and provides a useful complement to exploring the externalities of agribusiness technology innovation.

The remainder of this paper is organized as follows. Section 2 presents the theoretical analysis and research hypotheses. Section 3 provides our empirical research design. Section 4 presents a series of empirical results. Section 5 presents the results of further discussions. Section 6 summarizes the conclusions and policy implications.

2. Theoretical Analysis and Research Hypotheses

2.1. Analysis of the Impact of Economic Policy Uncertainty on Technological Innovation in Agribusiness

According to the endogenous growth theory, technological innovation is the driving force for firms to maintain their competitive advantage. According to uncertainty theory [22,23], rising uncertainty may increase R&D costs and operational risks for agricultural enterprises, which should consider uncertainties in various aspects of economic policies in order to make rational decisions on technological innovation. Economic policy uncertainty refers to the difficulty of forming stable policy expectations because economic agents cannot accurately grasp information when the economic environment is uncertain [24,25]. The current economic policy uncertainty in China mainly stems from three aspects: first, the uncertainty of policy formulation, i.e., the subject of policy formulation, the content of policy formulation, etc.; second, the uncertainty of policy implementation, i.e., the uncertainty arising from implementation and the uncertainty of post-implementation effects, etc; and third, other factors of economic policy uncertainty, such as the outbreak of COVID-19 and the Russia-Ukraine conflict, etc. When economic policies change, companies with high levels of uncertainty of their own are more likely to be affected by economic policy uncertainty [26].

Increased economic policy uncertainty makes market information more and more uncertain, so how can agricultural enterprises carry out technological innovation under uncertainty? Innovation investment itself is an investment behavior, but R&D investment has special characteristics. First, capital investment is irreversible, and managers can choose to wait for the resolution of uncertainty to make investment decisions, while for R&D, investment may not be irreversible, such as for intellectual capital and experimental costs, and other R&D investments can seek similar products to switch the direction of capital investment in due course. However, due to the objective existence of the R&D race, managers often do not have more waiting options when making R&D decisions [27,28]. For example, in the patent race, firms engage in technological innovation among themselves for the purpose of acquiring patents. As a new form of modern business competition, the patent race is an important way for firms to gain a competitive advantage, and its competitive and profit-seeking nature drives firms to make timely R&D investment decisions and technological innovations. Therefore, in general, the increased uncertainty of economic policies will prompt agricultural enterprises to increase their R&D investment.

According to the theory of practice option, innovation input is an investment activity, which increases the value of practice option when uncertainty increases and leads to an increase in firms’ expectation of return on future innovation output. Coupled with the long production cycle and high adjustment costs of agricultural enterprises, when economic policy uncertainty increases, enterprises will increase innovation input in the present to obtain the return on future innovation output. Furthermore, from the perspective of the external environment in the innovation ecology theory [29], the enterprise innovation ecosystem is a whole, and the linkage effect of the external environment will have an impact on agricultural enterprises. While economic policy uncertainty has an impact on agribusiness, information asymmetry makes it difficult for agribusiness to determine whether competitors are increasing their innovation investment. If competitors make innovation investments before themselves and the agricultural R&D cycle is long—for example, the average R&D cycle of seed companies is 8–9 years—the rational strategy for agribusinesses is to make innovation investments as early as possible. The theory of dynamic organizational capabilities [29,30] also suggests that firms can only gain a foothold in competition if they have the dynamic ability to adapt to changes in the environment [31]. Therefore, when economic policies change, firms can seize the opportunity to increase their investment in technological innovation by “moving first”.

Therefore, based on the above theoretical analysis, we propose the first hypothesis:

H1. 

Economic policy uncertainty contributes to technological innovation of agricultural enterprises.

2.2. Analysis of the Influence Mechanism of Economic Policy Uncertainty on the Impact of Technological Innovation in Agricultural Enterprises

2.2.1. Financial Flexibility

When economic policy uncertainty occurs, the financial dynamic adjustment ability of agricultural enterprises is an important factor in their technological innovation. Financial flexibility refers to the ability of agribusinesses to obtain funds in a timely and cost-effective manner and to prevent or exploit potential uncertainties in economic policies to alleviate financing constraints. The financial flexibility of agribusinesses will be further tightened when economic policy uncertainty increases. From the perspective of financing constraints, when economic policy uncertainty increases, the financial flexibility of agricultural enterprises will inevitably deteriorate due to the strengthening of financing constraints faced by agricultural enterprises, so agricultural enterprises can only rely on their own technological innovation to seek improvements and obtain future cash inflows. At this time, agricultural enterprises under the rational broker hypothesis will hope for their own technological innovation. Therefore, the increased uncertainty of economic policies will tighten the financial flexibility of agribusinesses, which will shift the hope of profitability to their own internal innovation. Based on this, we believe that economic policy uncertainty positively impacts technological innovation activities by tightening the financial flexibility of agricultural enterprises.

H2. 

Economic policy uncertainty leads to tighter financial flexibility for agribusinesses, which in turn drives them to rely more on their own technological innovation.

2.2.2. Managerial Capacity

When economic policy uncertainty increases, managers’ capabilities will have a significant impact on agribusiness technology innovation. According to the theory of dynamic organizational capabilities, when agribusinesses face economic policy uncertainty, they need to make internal organizational adjustments to adapt to such changes, and managers’ capabilities are an important factor influencing agribusiness innovation behavior. Economic policy uncertainty contains both opportunities and challenges, and if managers are competent and make the right decisions, they will invest in innovation in the right direction. On the contrary, if managers are not competent, there is a risk of innovation failure due to a lack of innovation investment or deviation from innovation investment, resulting in a series of problems such as “illusion of control” and “difficulty effect”. At the same time, when the economic environment becomes uncertain, a certain proportion of front-line farmers in agribusinesses are more eager to have a competent manager lead the agribusiness, which is conducive to stabilizing people’s confidence, and the manager’s ability is more prominent at this time. Therefore, economic policy uncertainty affects the technological innovation of enterprises through the capability of managers in agricultural enterprises, so we establish the theoretical framework of “increased economic policy uncertainty, manifestation of managerial capability, enhancement of technological innovation capability”.

Based on this, we believe that when economic policy uncertainty increases, the competence of agribusiness managers becomes more pronounced, and proper managerial competence will motivate agribusinesses to enhance their sense of independent innovation and improve technological innovation.

H3. 

There is a mediating effect of managerial competence between economic policy uncertainty and agribusiness technology innovation.

2.3. Moderating Effect of Government Subsidies on Economic Policy Uncertainty and Technological Innovation in Agribusiness

Government subsidies have been increasing year by year with increasing national attention to technological innovation in listed companies. Government subsidies support the technological innovation activities of agribusinesses, which are conducive to stimulating their innovation enthusiasm and prompting them to increase their innovation investment [32,33,34,35]. The role of government subsidies in influencing the technological innovation of agribusinesses is mainly reflected in three aspects: financial support, signaling, and resource acquisition. First, government subsidies bring direct cash inflows to agribusinesses and support innovation inputs such as research and development activities, new variety breeding, new technology promotion, etc. These subsidies, unlike debt and equity financing, do not need to be repaid by enterprises, so enterprises will have no worries about investing government subsidies in risky technological innovation. Secondly, the support of government subsidies for technological innovation by agribusinesses will send signals to the outside world that innovative agribusinesses receiving government subsidies have good technological innovation capability and innovation projects. Finally, based on the signaling effect of government subsidies, the technological innovation activities of agribusinesses will attract R&D talents, partners, and other social resources. Agribusinesses engaged in innovation activities that are inherently risky are more likely to enjoy the multiplier effect of government subsidies, thus enhancing agribusinesses’ confidence in the face of policy uncertainty and motivating them to innovate technologically and resist the uncertain economic environment in times of economic policy uncertainty [36].

Based on this, we believe that government subsidies give positive support to agribusinesses, thus promoting their technological innovation as a whole (Figure 1).

H4. 

When economic policy uncertainty increases, government subsidies can play a moderating role in promoting technological innovation in agribusiness.

3. Empirical Research Design

3.1. Sample Selection and Data Sources

In 2007, the state implemented new accounting standards and adjusted certain accounting indicators. To ensure the consistency of financial data, this paper selects agricultural enterprises in the Shanghai and Shenzhen A markets from 2007 to 2021 as the sample for analysis. We selected 231 agricultural enterprises according to the selection criteria of the SFC’s “Industry Classification Guidelines for Listed Companies” (revised in 2012) and finally obtained a sample of 2580. The 231 listed agricultural enterprises were selected mainly based on the following considerations.

The sample of listed agricultural enterprises plays a key role in China’s agricultural economy and innovation and is highly representative. First, these listed agricultural enterprises include traditional agricultural enterprises in the primary industry, agricultural product processing enterprises in the secondary industry, and agricultural product service enterprises in the tertiary industry. They cover nine categories of enterprises, including agriculture, forestry, animal husbandry, fishery service industries, agro-food processing industries, and food manufacturing industries, reflecting the overall business situation of Chinese agricultural enterprises. Second, these 231 listed agricultural enterprises can largely reflect the overall innovation level of Chinese agricultural enterprises. Economic policy uncertainty not only has an important impact on fluctuations in the macroeconomic and agricultural fields but also has a considerable impact on microentities of agricultural enterprises in general. Economic policy uncertainty directly affects the innovation behavior of these 231 enterprises. At present, the technological innovation of Chinese agricultural enterprises mainly relies on the economic policies formulated by the government. The key areas of science and technology innovation investment funds and the quantity and quality of innovation output results of these 231 listed agricultural enterprises are directly affected by economic policy uncertainty, and their R&D investment, transformation of scientific and technological achievements, and innovation environment are all closely related to economic policy uncertainty.

Considering the stability and reliability of the sample data, and in order to make the data more representative and refer to existing studies [37,38,39], this paper treats the data according to the following principles: first, the sample of listed companies in the financial industry is excluded; second, the sample of companies treated by ST and *ST in the sample period is excluded; third, the sample of companies with abnormal key financial indicators is excluded. The data on innovation inputs and outputs of agricultural enterprises are obtained from the China Research Data Services Platform (CNRDS), and policy uncertainty is measured by the Economic Policy Uncertainty Index (hereinafter referred to as the “EPU”) jointly published by Stanford University and the University of Chicago, which is based on the South China Morning Post (SCMP). Data on financial indicators of listed companies from the CSMAR (China Stock Market Accounting Research) database. In order to avoid the influence of outliers on the regression results, all continuous variables in this paper are treated with 1% and 99% tailing.

3.2. Variables

3.2.1. Dependent Variables

The existing research measured enterprise technological innovation from two aspects: innovation input and innovation output. Since the data in this paper are right-skewed and contain zero values, drawing on the study of Ya, et al. [37], the natural logarithm is taken after adding 1 to R&D input to measure innovation input (Specifically, R&D inputs refer to research and development expenses, which are for generating patents or new products, including expenses for basic research, applied research, and experimental development. For example, the R&D inputs of seed enterprises include material costs, material fees, testing fees, validation test costs of varieties, equipment costs, travel of R&D personnel, intellectual property input, cooperative R&D costs, and costs of purchased varieties). Drawing on the study of Yang. et al. [20], the number of patent applications is added by 1 and then taken as the logarithm to measure innovation output.

3.2.2. The Core Independent Variable

The core independent variable is the economic policy uncertainty. In this paper, we use the Economic Policy Uncertainty Index jointly published by Stanford University and the University of Chicago to measure economic policy uncertainty. Baker et al. [40,41,42,43] used the South China Morning Post, the largest circulation newspaper in Hong Kong founded in 1903, as a sample for textual analysis and identified and counted the number of stories containing the keywords “China”, “economy”, “uncertainty”, and “policy” in the newspaper each month as a proportion of the total number of reports in the newspaper in that month, so as to determine the monthly data on China’s economic policy uncertainty. In this paper, the annual average of the monthly economic policy uncertainty index is used as a measure.

3.2.3. Moderating Variable

In this paper, the intensity of government subsidies is selected as a moderating variable. The main measures in current literature [38] are the logarithm of total government subsidies, the proportion of government subsidies to main business income, and the proportion of government subsidies to total assets [44]. This paper uses the proportion of government subsidies to main business income to measure it.

3.2.4. Other Control Variables

Referring to the studies of Fang Xianming [45], Li Jiuchin [46], and Zhang Feng [13], the technological innovation of agricultural enterprises may be influenced by factors such as governance structure, profitability, and corporate age. This paper selected the ratio of the shareholding of the first largest shareholder, the growth rate of operating income, the percentage of minority shareholders’ equity, the ratio of tangible assets, the ratio of cash flow, corporate value, the number of years in the market, the size of the board of directors, return on total assets, return on net assets, and corporate cash flow as control variables.

The higher shareholding ratio of the first largest shareholder represents the more centralized decision-making and implementation power of the enterprise, which may be less conducive to technological innovation. The higher operating income growth rate indicates that the enterprise focuses more on short-term operating income and less on technological innovation activities. The higher minority interest ratio indicates that the equity structure of the enterprise tends to be balanced, which is more conducive to technological innovation. The higher tangible assets ratio means that the intangible assets of the enterprise ratio is lower, indicating that the enterprise’s technological innovation activities are limited. The higher cash flow ratio may be caused by the enterprise’s selling innovation results into cash flow in order to maintain a large amount of cash flow [14,19], so the less technological innovation results on the books. Enterprise value is measured by Tobin’s q value, which makes up for the lack of financial indicators from a capital market perspective [47]. The larger the Tobin’s q value, the greater the incentive for capital market liquidity arbitrage than for technological innovation, and the less the enterprise invests in technological innovation. The longer a firm has been listed, the more conducive it is to innovation. The larger the board of directors and the more reasonable the governance structure, the more likely it is to promote the technological innovation activities of the firm. The larger return on total assets, the more profitable the enterprise is and the more conducive it is to technological innovation. A larger return on net assets reflects a stronger ability to obtain net income from an enterprise’s own capital and a stronger ability to innovate technologically. The greater the proportion of enterprise cash flow, the more favorable it is to the enterprise’s capital investment in technological innovation (

Table 1. Variable Definitions.

Type	Name	Symbols	Definition
Dependent Variables	R&D investment	lnR&D	Add one to the amount of R&D investment and take the natural logarithm
	Patent output	lnPat	The patents filed in the current year are added by one and the natural logarithm is taken.
The core independent variable	Economic Policy Uncertainty	epu	Economic Policy Uncertainty Index, which converts monthly indicators into annual indicators through arithmetic averages.
Moderating Variable	Government subsidies	sub	Ratio of government subsidies to revenue from main business
Control variables	Shareholding ratio of the first largest shareholder	top1	Number of shares held by the first largest shareholder as a percentage of the total number of shares
	Operating income growth rate	growth	Ratio of current year operating income to prior period operating income of listed agricultural enterprises minus 1
	Minority interests as a percentage of	mino	Minority interests as a percentage of owners’ equity
	Tangible assets ratio	tangibility	Tangible assets as a percentage of total company assets
	Cash Flow Ratio	cashflow	Corporate cash flow as a percentage of total assets
	Enterprise Value	tq	Ratio of market capitalization to total assets of a company
	Number of years on the market	lnage	Take logarithm of the number of years the company has been listed
	Board Size	board	The number of directors of the company is taken as a logarithm
	return on total assets	roa	Net profit as a percentage of total assets
	Return on Net Assets	roe	Net income as a percentage of shareholders’ equity
	Corporate Cash Flow	cfo	Cash flows from operating activities as a percentage of assets
Dummy Variables	Nature of ownership	soe	Dummy variable, 1 for state-owned enterprises, 0 for non-state-owned enterprises
	Industry Properties	ind	Dummy variable, 1 for basic agribusiness, 0 for manufacturing agribusiness
	Adjustment costs	adjust	Dummy variable, R&D investment above the industry average is 1, otherwise it is 0

).

3.3. Descriptive Statistics

Table 2. Results of variables descriptive statistics.

Variables	Mean	Sd	Min	Med	Max
lnR&D	16.830	2.409	0	17.04	21.430
lnPat	2.441	1.460	0	2.565	6.161
epu	5.861	0.649	4.594	5.899	6.674
sub	38.630	1107	0	0.011	355.00
top1	0.370	0.144	0.085	0.363	0.779
growth	0.236	0.488	0.001	0.156	13.960
mino	0.047	0.075	−0.074	0.016	0.406
tangibility	0.934	0.061	0.561	0.951	1
cashflow	−1.002	13.230	−54.78	−0.088	45.860
tq	2.466	1.749	0.891	1.931	16.920
lnage	2.013	0.951	0	2.138	3.367
lnboard	2.119	0.203	1.609	2.197	2.773
roa	0.071	0.072	−0.263	0.062	0.324
roe	0.096	0.119	−1.195	0.089	0.487
cfo	0.085	0.088	−0.300	0.079	0.515
soe	0.343	0.475	0	0	1
ind	0.177	0.382	0	0	1
adjust	0.472	0.499	0	0	1

presents the results of descriptive statistics for the main variables. In general, the R&D investment and patent output of Chinese listed agricultural enterprises are not high, with mean values of 16.83 and 2.441, respectively, and large differences between the maximum and minimum values, which indicate that the level of technological innovation of listed agricultural enterprises varies significantly during the sample period. The mean and median of China’s economic policy uncertainty index for 2007–2021 are 5.861 and 5.899, respectively, with a standard deviation of 0.6498. The mean value of government subsidies is 38.63, and the difference between the maximum and minimum values is huge, which indicates that the government subsidies of different agribusinesses are highly variable. From the control variables, the mean values of operating income growth rate, minority interest ratio, enterprise value, return on total assets, return on net assets, and enterprise cash flow are all greater than the median, showing a right skew, indicating large overall differences. The mean value of the shareholding ratio of the first largest shareholder is 37%, the proportion of tangible assets is 93%, the mean value of the cash flow ratio is −1.002, the average number of years listed is 2.013, and the size of the board of directors is 2.119. The mean values of the dummy variables show that among the listed agricultural enterprises, state-owned enterprises account for 34.3% of the total sample, basic agricultural enterprises account for 17.7%, and listed agricultural enterprises with high R&D adjustment costs account for 47.2%.

3.4. Empirical Model

Specifically, we use the following model to investigate the impact of economic policy uncertainty on technological innovation in agribusiness:

$${innovation}_{i,t}={\beta }_0+{\beta }_1{epu}_{i,t}+{\beta }_j{\sum }_j{controls}_{i,t}+{\eta }_t+{\eta }_{ind}+{\epsilon }_{i,t}$$

(1)

To explore the mechanisms by which economic policy uncertainty affects technological innovation in agribusiness, we construct the following mediating effects model:

$${med}_{i,t}={\beta }_0+{\beta }_1{epu}_{i,t}+{\beta }_j\sum {controls}_{i,t}+{\epsilon }_{i,t}$$

(2)

$${innovation}_{i,t}={\alpha }_0+{\alpha }_1{epu}_{i,t}+{\alpha }_2{med}_{i,t}+{\alpha }_j{\sum }_{j}controls+{\epsilon }_{i,t}$$

(3)

where med_i,t is the mediating variable, including financial flexibility fanancial_i,t and managerial ability manager_i,t; When the coefficient β₁ of Equation (1) is significant, if the coefficient β₁ of Equation (2) and the coefficient α₂ of Equation (3) are significant, it indicates that there is a mediating effect; when the coefficient α₁ of Equation (3) is not significant, there is a full mediating effect at this time; when α₁ is significant, if the coefficient β₁ × α₂ has the same sign as α₁, there is a partial mediating effect; if the sign is different, there is a masking effect.

$${innovation}_{i.t}={\beta }_0+{\beta }_1{epu}_{i,t}+{\beta }_{2}sub+{\beta }_3{epu}_{i,t}\ast {sub}_{i,t}+{\beta }_j{\sum }_j{controls}_{i,t}+{\eta }_t+{\eta }_{ind}+{\epsilon }_{i,t}$$

(4)

Model (4) is used to examine the moderating effect of government subsidies on the impact of economic policy uncertainty on technological innovation at listed agricultural firms, focusing on the epu × sub coefficient of the cross-product of government subsidies and economic policy uncertainty. In the model, innovation_i,t represents the R&D input and patent output; epu_i,t is the economic policy uncertainty index; sub_i,t is the government subsidy; controls_i,t are the control variables; η_t denotes the time fixed effect; η_ind denotes the industry fixed effect; ε_i,t is the random error term.

4. Empirical Results

4.1. Analysis of Baseline Regression Results

Table 3. The results of baseline regression.

	(1)	(2)	(3)	(4)
Variables	lnRDi,t	lnPatenti,t	lnRDi,t+1	lnPatenti,t+1
epu	0.673 ***	0.498 ***	0.853 ***	0.346 ***
	(4.46)	(4.82)	(3.45)	(3.03)
sub
epu × sub
top1	0.118	−0.112	−0.142	−0.261
	(0.29)	(−0.41)	(−0.28)	(−0.78)
growth	−0.037	−0.053 ***	−0.037	−0.057 ***
	(−1.63)	(−3.62)	(−0.83)	(−2.77)
mino	2.897 ***	1.960 ***	3.194 ***	1.574 **
	(3.36)	(3.73)	(2.58)	(2.35)
tangibility	−1.484	0.501	−0.573	1.026
	(−1.35)	(0.87)	(−0.40)	(1.52)
cashflow	0.002	−0.014 ***	0.002	−0.014 ***
	(0.28)	(−3.47)	(0.24)	(−2.84)
tq	−0.033	−0.077 ***	−0.065	−0.079 **
	(−0.83)	(−3.52)	(−1.14)	(−2.39)
age	0.040	0.241 ***	0.074	0.261 ***
	(0.41)	(5.01)	(0.58)	(4.23)
board	−0.162	0.302	0.255	0.165
	(−0.49)	(1.63)	(0.54)	(0.75)
roa	3.959 *	2.781 **	9.581 ***	1.802
	(1.83)	(2.18)	(3.74)	(1.04)
roe	0.411	0.327	−0.842	1.249
	(0.38)	(0.53)	(−0.54)	(1.29)
cfo	1.959 *	2.461 ***	0.791	3.224 ***
	(1.78)	(3.90)	(0.47)	(4.13)
Constant	14.092 ***	−3.674 ***	11.144 ***	−2.786 **
	(8.73)	(−4.03)	(5.07)	(−2.55)
n	1504	1534	1040	1054
R-squared	0.159	0.223	0.163	0.219
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

reports the regression results of economic policy uncertainty on agribusiness technology innovation. From columns (1) and (2), whether it is R&D input measured by lnR&D or patent output measured by lnPat_i,t, the regression coefficients of economic policy uncertainty on agribusiness innovation input and output after controlling for relevant influencing factors are significantly positive at the 1% level after adding industry fixed effects and time fixed effects in the model, indicating that economic policy uncertainty significantly increases the level of technological innovation of agricultural enterprises, and H1 is verified. Meanwhile, considering that there may be a lag in the impact of economic policy uncertainty on technological innovation in agricultural enterprises, this paper further analyzes whether economic policy uncertainty affects technological innovation in agricultural enterprises in the next period, as shown in columns (3) and (4). The regression results do not change significantly.

4.2. Heterogeneity Analysis

4.2.1. Heterogeneity of Agribusiness Nature of Property Rights

For agricultural enterprises, property rights heterogeneity is an important factor affecting technological innovation in agribusiness. In order to verify whether there is property rights heterogeneity due to economic policy uncertainty on technological innovation in agricultural enterprises, dummy variables are set with the nature of enterprise property rights: 1 for state-owned enterprises and 0 for non-state-owned enterprises. The results of regression analysis in

Table 4. The results of the heterogeneity analysis of the nature of property rights in agribusiness.

	Innovation Input		Innovation Output
Variables	(1) State Enterprise	(2) Non-State Enterprise	(3) State Enterprise	(4) Non-State Enterprise
	lnR&D	lnR&D	lnPatent	lnPatent
epu	0.772 **	0.686 **	0.620 ***	0.434 ***
	(2.44)	(2.60)	(3.99)	(2.63)
top1	1.465 *	−0.585	−0.789 *	0.339
	(1.74)	(−0.86)	(−1.90)	(0.92)
growth	−0.037	−0.032	−0.278 **	−0.043 ***
	(−0.12)	(−0.76)	(−1.98)	(−4.11)
mino	2.084	3.536 ***	1.902 ***	2.371 ***
	(1.28)	(2.93)	(2.75)	(3.38)
tangibility	−6.358 ***	2.516 *	0.045	1.823 **
	(−3.09)	(1.74)	(0.05)	(2.44)
cashflow	−0.001	0.003	−0.008	−0.015 ***
	(−0.10)	(0.29)	(−1.28)	(−2.96)
tq	−0.136 *	0.043	−0.039	−0.069 **
	(−1.84)	(0.73)	(−1.29)	(−2.40)
age	−0.060	0.121	0.267 ***	0.264 ***
	(−0.30)	(0.98)	(3.22)	(4.23)
board	0.391	−0.299	0.269	0.479 *
	(0.65)	(−0.62)	(0.95)	(1.93)
roa	5.927 *	2.296	4.120 ***	0.632
	(1.79)	(0.80)	(3.23)	(0.41)
roe	0.151	0.328	−0.673	1.303 *
	(0.10)	(0.23)	(−1.20)	(1.74)
cfo	1.600	3.033 **	0.934	3.215 ***
	(0.82)	(2.03)	(1.06)	(3.80)
Constant	16.426 ***	10.842 ***	−3.603	−5.242 ***
	(4.84)	(3.57)	(−1.66)	(−2.73)
n	638	866	642	892
R-squared	0.294	0.141	0.383	0.222
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES
Chow Test	6.49 ***		2.56 ***

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

show that the epu coefficients of both column (1) SOE group and column (2) non-SOE group are statistically significant at the 5% level in terms of innovation inputs, and with the help of the Chow test, the difference between the groups passes the significance test at the 1% level, indicating that economic policy uncertainty does not have the same impact on agricultural enterprises with different property rights nature. The coefficient of economic policy uncertainty is 0.772 for the SOE group and 0.686 for the non-SOE group. In terms of innovation output, the regression coefficient of patent output is 0.62 for the SOE group in column (3) and 0.43 for the non-SOE group in column (4), both of which are significant at the 1% statistical level. With the help of the Chow test, the difference between groups passes the significance test at the 5% level. This indicates that economic policy uncertainty has a significant promotion effect on both state-owned and non-state-owned agricultural enterprises, but the “incentive” effect of economic policy uncertainty on technological innovation is more pronounced for state-owned agricultural enterprises than for non-state-owned agricultural enterprises.

Possible reasons for this are the heterogeneity of the economic environment for agribusiness due to economic policy uncertainty, which is reflected in two aspects: access to financing and the natural “blood” relationship with the government. First, state-owned enterprises have easier access to external financing than non-state-owned enterprises, while private enterprises and others rely mostly on endogenous financing in the early stages of innovation investment. Especially when economic policy uncertainty increases, SOEs have easier access to external financing than non-SOEs because of their special status, while non-SOEs may face more financing constraints and a smaller credit supply due to the uncertain economic environment. Second, SOEs are the firstborn of the Chinese economy and play an important strategic support role, especially in the fields of food security, science and technology innovation, and national livelihood. By virtue of their “blood” relationship with the government, SOEs have a natural advantage over non-SOEs in terms of access to information. By keeping abreast of economic policy trends, SOEs can interpret the uncertainties in time and even grasp the information in advance so as to make decisions and adjustments that are conducive to enterprise technological innovation [37], while non-SOEs lag behind in information acquisition and innovation decision-making, resulting in heterogeneity between the two types of enterprise technological innovation in the face of economic policy uncertainties.

4.2.2. Heterogeneity of Agribusiness Industry Attributes

Chinese agricultural enterprises are involved in a wide range of businesses, including both primary industries engaged in basic agriculture and secondary and tertiary industries engaged in manufacturing services. To test the heterogeneity of the impact of economic policy uncertainty on the technological innovation of agricultural enterprises, this paper divides the sample agricultural enterprises into basic agricultural enterprises and manufacturing agricultural enterprises and sets dummy variables with industry codes to classify enterprises belonging to the primary industry as basic agricultural enterprises and those belonging to the secondary and tertiary industries as manufacturing agricultural enterprises.

Table 5. The results of the heterogeneity analysis of industry attributes in agribusiness.

Variables	Innovation Input		Innovation Output
	(1) Basic Agribusiness	(2) Manufacturing Agribusiness	(3) Basic Agribusiness	(4) Manufacturing Agribusiness
	lnRDi,t	lnRDi,t	lnPatenti,t	lnPatenti,t
epu	2.019 **	0.510 ***	0.690 ***	0.459 ***
	(2.38)	(3.53)	(3.17)	(3.64)
top1	2.346	0.390	−1.272 **	0.235
	(1.59)	(0.91)	(−2.09)	(0.74)
growth	0.001	−0.040 *	−0.071	−0.053 ***
	(0.00)	(−1.74)	(−0.46)	(−4.08)
mino	8.205 ***	1.261	2.713 ***	1.611 **
	(3.93)	(1.40)	(3.42)	(2.54)
tangibility	−12.422 ***	0.782	0.342	1.026
	(−3.15)	(0.75)	(0.28)	(1.55)
cashflow	0.037 *	−0.004	−0.007	−0.015 ***
	(1.73)	(−0.60)	(−0.74)	(−3.36)
tq	−0.049	−0.038	−0.029	−0.100 ***
	(−0.50)	(−0.91)	(−0.53)	(−4.14)
age	−0.467 *	0.144	−0.369 ***	0.358 ***
	(−1.77)	(1.35)	(−2.86)	(6.88)
board	1.325 *	−0.176	0.836 **	0.302
	(1.94)	(−0.48)	(2.35)	(1.37)
roa	−15.503	4.856 **	−7.419 **	4.117 ***
	(−1.65)	(2.34)	(−2.30)	(3.19)
roe	12.334 **	−0.816	4.829 ***	−0.223
	(2.38)	(−0.82)	(2.89)	(−0.38)
cfo	−3.330	3.758 ***	−0.342	3.114 ***
	(−1.18)	(3.71)	(−0.29)	(4.29)
Constant	13.564 **	12.689 ***	−3.815 *	−5.177 ***
	(2.07)	(4.62)	(−1.76)	(−3.17)
n	321	1183	325	1209
R-squared	0.358	0.100	0.315	0.204
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES
Chow Test	7.29 ***		8.06 ***

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

reports the estimation results of each parameter of the model. From columns (1), (2), (3), and (4), it can be found that economic policy uncertainty has a significant impact on technological innovation regardless of whether R&D input or patent output is used as the dependent variable; with the help of the Chow test, the differences between the two groups pass the significance test at the 1% level, indicating that economic policy uncertainty does not have the same impact on agricultural enterprises with different industry attributes.

The estimated coefficient of economic policy uncertainty is larger in the sample of basic agribusinesses, indicating that basic agribusinesses are more motivated to engage in technological innovation activities and outcomes when the risk of economic policy uncertainty increases compared to manufacturing agribusinesses. The possible reason is that the first impact of economic policy changes is generally on basic agricultural enterprises, such as agriculture, forestry, animal husbandry, fisheries, and other production-oriented agricultural enterprises, which are generally more affected by economic policies and more obviously affected by the direct impact of changes in national policies on the three rural areas and economic policies; while manufacturing-oriented agricultural enterprises are in the downstream of basic agriculture, changes in economic policies are more likely to play through basic agricultural enterprises. For example, manufacturing agribusinesses such as the agro-food processing industry, the food manufacturing industry, and the wine, beverage, and refined tea manufacturing industries are much less sensitive to policy changes than basic agribusinesses. At the same time, basic agribusinesses have long R&D cycles and slow iterative updates and need longer waiting periods to test innovation results compared with manufacturing agribusinesses, so they generally carry out technological innovation activities in time or even in advance when economic policies are uncertain.

4.2.3. Heterogeneity of Agribusiness Adjustment Costs

Innovation input costs for agricultural enterprises are an important heterogeneous factor of economic policy uncertainty affecting technological innovation in agricultural enterprises. In order to examine whether there is a difference in the impact of economic policy uncertainty on the adjustment cost of agricultural enterprises, firstly, the dummy variable is set for the growth of R&D input of listed agricultural enterprises, and the current year’s input is higher than last year’s defined as growth, and the dummy variable is taken as 1, otherwise it is taken as 0; then the dummy variable is compared with the industry mean, and the value higher than the industry mean is taken as 1, otherwise it is taken as 0.

Table 6. The results of the heterogeneity analysis of adjustment cost in agribusiness.

Variables	Innovation Input		Innovation Output
	(1) High Adjustment Cost Companies	(2) Low Adjustment Cost Companies	(3) High Adjustment Cost Companies	(4) Low Adjustment Cost Companies
	lnR&D	lnR&D	lnPat	lnPat
epu	2.986 ***	4.148 *	1.348 ***	0.687
	(4.09)	(1.76)	(3.01)	(1.42)
top1	0.871 **	−1.098	1.281 ***	−1.397 ***
	(1.98)	(−1.25)	(2.98)	(−2.68)
growth	−0.025	−0.975 *	−0.143	0.150
	(−0.35)	(−1.70)	(−1.16)	(0.61)
mino	1.793	4.051 ***	3.317 ***	0.673
	(1.43)	(3.37)	(2.90)	(0.96)
tangibility	−0.116	−0.746	2.348 **	−0.933
	(−0.13)	(−0.58)	(2.31)	(−1.12)
cashflow	−0.008	−0.002	−0.011	−0.013 **
	(−1.08)	(−0.23)	(−1.52)	(−2.47)
tq	−0.032	0.065	−0.111 **	0.020
	(−0.63)	(1.31)	(−2.34)	(0.56)
age	0.485 ***	−0.228	0.528 ***	0.092
	(5.07)	(−1.38)	(6.10)	(1.17)
board	−0.252	−0.263	0.384	0.388
	(−0.47)	(−0.58)	(1.21)	(1.20)
roa	4.390	2.451	2.474	1.792
	(1.26)	(0.65)	(0.65)	(0.91)
roe	0.917	−0.364	1.238	−0.269
	(0.39)	(−0.25)	(0.44)	(−0.32)
cfo	1.051	2.763 *	1.400	1.123
	(0.61)	(1.68)	(1.09)	(1.24)
Constant	−1.781	−7.000	−11.572 ***	−3.054
	(−0.45)	(−0.48)	(−3.51)	(−0.93)
n	499	555	499	555
R-squared	0.276	0.177	0.282	0.245
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES
Chow test	6.11 ***		8.28 ***

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

shows the regression results for the grouping of high-level R&D input adjustment costs and low-level R&D input adjustment costs. With the help of the Chow test, the differences between both groups pass the significance test at the 1% level, indicating that economic policy uncertainty does not have the same impact on agribusinesses with different adjustment costs.

The results show that in terms of innovation output, the epu coefficient of 1.348 for high adjustment cost agribusinesses in column (3) is significant at the 1% level, while the epu coefficient of low adjustment cost agribusinesses in column (4) does not pass the test, which indicates that economic policy uncertainty has a more prominent role in promoting technological innovation in high adjustment cost agribusinesses compared to low adjustment cost agribusinesses. This is because when agribusinesses increase their innovation inputs, they will incur upward adjustment costs such as capital and human inputs, while when they abandon their innovation inputs, they will incur downward adjustment costs such as abandoning farm machinery and equipment and discharging R&D manpower. Since the future benefits of crop production and animal growth are in sharp contrast to the present loss probability of direct abandonment, the rational decision of agribusinesses is to adjust innovation costs upward when the benefits of agricultural innovation lag behind, and the sticky characteristics of innovation costs are crucial to the innovation costs of agribusinesses. Thus, when economic policy uncertainty increases, agribusinesses with higher innovation adjustment costs perform better in terms of innovation inputs and outputs compared to agribusinesses with lower innovation adjustment costs.

However, in terms of innovation inputs, the epu coefficient of column (1), high R&D adjustment cost agribusinesses (2.986), is significant at the 1% level, and the epu coefficient of column (2), low R&D adjustment cost agribusinesses (4.148), is significant at the 10% level. This result indicates that the role of economic policy uncertainty in promoting technological innovation in high adjustment cost agribusinesses is not very prominent relative to low adjustment cost agribusinesses, probably because when faced with economic policy uncertainty shocks, high adjustment cost agribusinesses have difficulty raising sufficient funds to invest in innovation, and financing constraints are somewhat stronger relative to low adjustment cost agribusinesses.

4.3. Robustness Checks

4.3.1. Discussion of the Endogeneity Issue

The previous empirical study shows that economic policy uncertainty enhancement helps agribusiness technology innovation. To address the possible omitted variables problem, this paper adopts a fixed-effects model to mitigate the impact of the endogeneity problem caused by the omitted variables. Since economic policies are national macro-level policies, it is difficult for the micro-level behaviors of agribusinesses to influence macro-level policies, so the probability of reverse causality between agribusiness technology innovation activities and economic policy uncertainty is low. Nevertheless, in order to avoid possible reverse causality, drawing on the practice of existing studies [13,18,42,48], this paper selects the U.S. economic policy uncertainty index as the instrumental variable and estimates it using two-stage least squares. One of the main reasons is that the U.S. and China have similar levels of economic and agricultural development. The U.S. and China rank first and second in the world in terms of economic aggregates, respectively, while the U.S. is the top agricultural power and China is a global agricultural power. The similarity of economic and agricultural development levels makes the two countries have some commonality in the formulation and implementation of economic policies. Second, trade exchanges between China and the U.S. are characterized by both cooperation and conflict. In 2022, bilateral trade between China and the U.S. reached a record high of $690.6 billion, with trade volumes far exceeding those of China’s other trade partners. Despite the close trade exchanges, trade disputes do occur from time to time. The U.S. government also formulates trade policies against China, thus influencing the formulation and implementation of China’s trade and economic policies. This trade relationship of both cooperation and conflict makes the two countries influence each other.

Table 7. Robustness-check results I: IV estimation.

Variables	Innovation Input: lnR&D			Innovation Output: lnPat
	Baseline Return 1	Phase I	Phase II	Baseline return	Phase I	Phase II
lnx1	0.761 ***		0.502 ***	0.527 ***		0.522 ***
	(6.95)		(3.53)	(10.08)		(7.27)
iv		1.707 ***			1.717 ***
		(41.77)			(42.84)
top1	0.153	0.010	0.133	−0.131	−0.009	−0.308
	(0.38)	(0.12)	(0.26)	(−0.47)	(−0.10)	(−1.17)
growth	−0.028	−0.011	−0.035	−0.050 ***	−0.009	−0.048 **
	(−1.18)	(−1.55)	(−0.87)	(−3.17)	(−1.37)	(−2.28)
mino	2.883 ***	−0.330 **	2.731 ***	1.776 ***	−0.423 ***	1.104 **
	(3.28)	(−2.08)	(2.89)	(3.43)	(−2.85)	(2.41)
tangibility	−1.425	−0.169	−1.468	0.437	−0.197	0.363
	(−1.31)	(−0.88)	(−1.28)	(0.77)	(−1.05)	(0.63)
cashflow	−0.000	−0.001	−0.002	−0.014 ***	−0.001	−0.017 ***
	(−0.02)	(−0.88)	(−0.28)	(−3.53)	(−1.09)	(−4.44)
tq	0.000	−0.025 ***	−0.002	−0.065 ***	−0.025 ***	−0.081 ***
	(0.01)	(−3.29)	(−0.04)	(−3.07)	(−3.37)	(−3.55)
age	0.052	0.084 ***	0.095	0.244 ***	0.076 ***	0.231 ***
	(0.54)	(5.20)	(0.97)	(5.02)	(4.94)	(4.82)
board	−0.248	−0.363 ***	−0.419	0.214	−0.354 ***	0.154
	(−0.74)	(−5.93)	(−1.12)	(1.14)	(−6.24)	(0.86)
roa	2.837	0.035	3.128	2.618 *	0.006	2.921 ***
	(1.32)	(0.09)	(1.44)	(1.96)	(0.02)	(2.66)
roe	0.417	−0.312 *	0.280	0.527	−0.302 *	0.459
	(0.39)	(−1.78)	(0.27)	(0.82)	(−1.75)	(0.86)
cfo	2.514 **	1.158 ***	2.998 **	2.405 ***	1.191 ***	2.940 ***
	(2.28)	(5.93)	(2.53)	(3.82)	(6.32)	(4.96)
Constant	13.032 ***	−2.207 ***	14.795 ***	−3.328 ***	−2.194 ***	−2.219 ***
	(9.19)	(−7.00)	(8.85)	(−4.46)	(−7.24)	(−2.67)
Observations	1504	1504	1504	1534	1534	1534
R-squared	0.141	0.589	0.137	0.197	0.588	0.153
Industry	YES	YES	YES	YES	YES	YES
Cragg-Donald Wald F		1787.330			1835.452

¹ The benchmark regression introduces time and industry fixed effects, and the introduction of instrumental variables may have multicollinearity problems with time fixed effects, so only industry fixed effects are controlled for here. Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

reports the regression results with U.S. economic policy uncertainty as the instrumental variable. In the first stage regression, the U.S. economic policy uncertainty index shows a significant positive correlation with the Chinese economic policy uncertainty index, while the Cragg-Donald Wald F-statistic values of the U.S. economic policy uncertainty index on innovation inputs and innovation outputs are significantly larger than the values of Stock and Yogo’s [49] proposed value of 10% maximal IV size of 16.38, which passed the weak instrumental variables test. Comparing the results of the baseline regression and the instrumental variables regression, the effect of the economic policy uncertainty on firms’ technological innovation in the instrumental variables method is significantly smaller than the results of the baseline regression, indicating that the effect of economic policy uncertainty on agricultural firms’ technological innovation decreases after the elimination of endogeneity but remains significant. The above results suggest that endogeneity and estimation bias may have an effect on the coefficient of economic policy uncertainty, but the regression coefficients of economic policy uncertainty on technological innovation of listed agricultural firms are all significantly positive, indicating the robustness of the benchmark regression results.

4.3.2. The Core Independent Variable Replacements

Distinguishing from the statistical approach of the articles on economic policy uncertainty in the South China Morning Post, an English newspaper in Hong Kong, Davis et al. [50] developed a set of indices of economic policy uncertainty in China based on two major newspapers in mainland China, People’s Daily and Guangming Daily, which divided the sample into three periods based on domestic economic development: the central planning era (1949–1978), the reform and opening-up period (1979–1999), and the era of globalization (2000 onward). The index reflects the characteristics of China’s economic policies and is equally representative and relevant for research. This paper re-runs the regression using this indicator as the independent variable (

Table 8. Robustness-check results II: Replacement of the core independent variable.

Variables	(1)	(2)	(3)	(4)
	lnR&Di,t	lnPati,t	lnR&Di,t+1	lnPati,t+1
epu	0.715 ***	0.530 ***	0.920 ***	0.374 ***
	(4.46)	(4.82)	(3.45)	(3.03)
top1	0.118	−0.112	−0.142	−0.261
	(0.29)	(−0.41)	(−0.28)	(−0.78)
growth	−0.037	−0.053 ***	−0.037	−0.057 ***
	(−1.63)	(−3.62)	(−0.83)	(−2.77)
mino	2.897 ***	1.960 ***	3.194 ***	1.574 **
	(3.36)	(3.73)	(2.58)	(2.35)
tangibility	−1.484	0.501	−0.573	1.026
	(−1.35)	(0.87)	(−0.40)	(1.52)
cashflow	0.002	−0.014 ***	0.002	−0.014 ***
	(0.28)	(−3.47)	(0.24)	(−2.84)
tq	−0.033	−0.077 ***	−0.065	−0.079 **
	(−0.83)	(−3.52)	(−1.14)	(−2.39)
age	0.040	0.241 ***	0.074	0.261 ***
	(0.41)	(5.01)	(0.58)	(4.23)
board	−0.162	0.302	0.255	0.165
	(−0.49)	(1.63)	(0.54)	(0.75)
roa	3.959 *	2.781 **	9.581 ***	1.802
	(1.83)	(2.18)	(3.74)	(1.04)
roe	0.411	0.327	−0.842	1.249
	(0.38)	(0.53)	(−0.54)	(1.29)
cfo	1.959 *	2.461 ***	0.791	3.224 ***
	(1.78)	(3.90)	(0.47)	(4.13)
Constant	14.256 ***	−3.553 ***	11.298 ***	−2.724 **
	(8.94)	(−3.96)	(5.20)	(−2.53)
n	1504	1534	1040	1054
R-squared	0.159	0.223	0.163	0.219
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

), and the results of both columns (1) and (2), R&D input and patent output, remain highly significant. Further, considering the time lag of economic policy impact, the impact of economic policy uncertainty on technological innovation in agricultural enterprises remains significant by lagging the independent variable by one period, and the conclusion remains unchanged.

In addition, referring to the existing literature [21], we use a weighted moving average method [51] to re-measure the economic policy uncertainty index based on the characteristics of quarterly changes in economic policy uncertainty within the year in China. First, We use a weighting approach to assign weights to the economic policy uncertainty index for each of the three months in each quarter, and then convert them to annual indicators. The robustness results for column (1) innovation inputs and column (2) innovation outputs in

Table 9. Robustness-check result III: Moving weighted measures of core independent variable.

	(1)	(2)	(3)	(4)
Variables	lnR&Di,t	lnPati,t	lnR&Di,t+1	lnPati,t+1
epu	0.697 ***	0.517 ***	0.886 ***	0.360 ***
	(4.46)	(4.82)	(3.45)	(3.03)
top1	0.118	−0.112	−0.142	−0.261
	(0.29)	(−0.41)	(−0.28)	(−0.78)
growth	−0.037	−0.053 ***	−0.037	−0.057 ***
	(−1.63)	(−3.62)	(−0.83)	(−2.77)
mino	2.897 ***	1.960 ***	3.194 ***	1.574 **
	(3.36)	(3.73)	(2.58)	(2.35)
tangibility	−1.484	0.501	−0.573	1.026
	(−1.35)	(0.87)	(−0.40)	(1.52)
cashflow	0.002	−0.014 ***	0.002	−0.014 ***
	(0.28)	(−3.47)	(0.24)	(−2.84)
tq	−0.033	−0.077 ***	−0.065	−0.079 **
	(−0.83)	(−3.52)	(−1.14)	(−2.39)
age	0.040	0.241 ***	0.074	0.261 ***
	(0.41)	(5.01)	(0.58)	(4.23)
board	−0.162	0.302	0.255	0.165
	(−0.49)	(1.63)	(0.54)	(0.75)
roa	3.959 *	2.781 **	9.581 ***	1.802
	(1.83)	(2.18)	(3.74)	(1.04)
roe	0.411	0.327	−0.842	1.249
	(0.38)	(0.53)	(−0.54)	(1.29)
cfo	1.959 *	2.461 ***	0.791	3.224 ***
	(1.78)	(3.90)	(0.47)	(4.13)
Constant	13.955 ***	−3.776 ***	10.960 ***	−2.861 ***
	(8.55)	(−4.08)	(4.91)	(−2.58)
n	1504	1534	1040	1054
R-squared	0.159	0.223	0.163	0.219
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

are significant, and the results for columns (3) and (4) lagging the explanatory variables by one period are also significant.

4.3.3. The Dependent Variables Replacements

Drawing on the existing literature, current scholars have selected two main measures of R&D investment: absolute [45] and relative [52]. To distinguish from the amount of R&D innovation in the previous section, we use R&D innovation intensity: R&D investment as a share of total assets and R&D investment as a share of operating income to replace the explanatory variables, respectively. The estimated results are shown in columns (1) and (3) of

Table 10. Robustness-check results IV: Replacements of dependent variables.

Variables	R&D Innovation Intensity1		R&D Innovation Intensity2		Replacement of Sample Data
	(1) R&D1i,t	(2) R&D1i,t+1	(3) R&D2i,t	(4) R&D2i,t+1	(5) R&Di,t	(6) R&Di,t+1
epu	0.003 ***	0.002 ***	0.009 ***	0.007 ***	0.005 ***	0.003 ***
	(7.77)	(2.81)	(11.17)	(7.58)	(10.99)	(4.20)
top1	−0.004 *	−0.005 **	−0.006 **	−0.008 **	−0.003 *	−0.005 **
	(−1.87)	(−2.05)	(−2.18)	(−2.25)	(−1.80)	(−1.97)
growth	−0.000 *	−0.000 *	−0.000	−0.000	−0.000 *	−0.000
	(−1.88)	(−1.75)	(−0.62)	(−0.24)	(−1.71)	(−1.31)
mino	−0.001	−0.001	−0.011 ***	−0.008	−0.000	−0.001
	(−0.26)	(−0.26)	(−2.83)	(−1.57)	(−0.18)	(−0.22)
tangibility	−0.006	−0.003	−0.002	0.006	−0.007 *	−0.003
	(−1.62)	(−0.81)	(−0.34)	(0.94)	(−1.83)	(−0.84)
cashflow	−0.000	−0.000	−0.000	−0.000 *	−0.000	−0.000
	(−1.22)	(−0.10)	(−0.81)	(−1.76)	(−1.35)	(−0.37)
tq	0.000 **	0.000	0.000	0.000	0.000	0.000
	(2.02)	(1.31)	(1.48)	(0.85)	(0.73)	(0.98)
age	−0.002 ***	−0.002 ***	−0.002 ***	−0.002 **	−0.002 ***	−0.002 ***
	(−4.59)	(−3.78)	(−3.87)	(−2.54)	(−4.77)	(−3.97)
board	−0.002 *	−0.002	−0.000	0.002	−0.002	−0.002
	(−1.69)	(−1.08)	(−0.15)	(0.58)	(−1.52)	(−1.11)
roa	0.006	0.013	−0.000	0.011	0.008	0.013
	(0.93)	(1.38)	(−0.05)	(0.86)	(1.17)	(1.43)
roe	0.003	0.003	0.004	−0.001	0.003	0.003
	(1.13)	(0.71)	(1.00)	(−0.22)	(1.02)	(0.81)
cfo	0.009 **	−0.003	−0.002	−0.001	0.010 **	−0.001
	(2.08)	(−0.45)	(−0.26)	(−0.18)	(2.47)	(−0.16)
Constant	0.005	0.010	−0.018 **	−0.022 **	−0.005	0.003
	(0.90)	(1.36)	(−2.37)	(−2.38)	(−0.95)	(0.40)
n	1486	1033	1446	1011	1534	1054
R-squared	0.186	0.178	0.232	0.244	0.206	0.194
Industry	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

. Meanwhile, since there are some missing values of R&D investment, the missing values are taken as 0 here and put into the regression to obtain the new sample data, and the estimated results are shown in column (5) of Table 10. All of them show that the estimated coefficients are robust and reliable. At the same time, the above three tests are lagged by one period in the regression, and the robustness results in columns (2), (4), and (6) are still significant.

4.3.4. Replacement of the Regression Method

Considering that there are a large number of “0” and “1” values in the R&D investment and patent applications of agricultural enterprises, which are imputed data. For such restricted dependent variables, Existing studies [13] suggest that Tobit estimation should be used. Our regression results using the Tobit method to test the impact of economic policy uncertainty on technological innovation in agribusiness are shown in

Table 11. Robustness-check results V: replacement of the empirical method.

Variables	(1)		(2)
	lnRDi,t	t-Value	lnPati,t	t-Value
epu	0.680 ***	2.94	0.746 ***	5.11
top1	0.116	0.22	−0.118	−0.36
growth	−0.037	−0.88	−0.057 **	−2.26
mino	2.953 ***	3.08	2.588 ***	4.42
tangibility	−1.569	−1.35	0.239	0.33
cashflow	0.002	0.32	−0.016 ***	−3.43
tq	−0.034	−0.73	−0.113 ***	−3.68
age	0.034	0.34	0.257 ***	4.22
board	−0.166	−0.44	0.406 *	1.74
roa	3.977 *	1.81	3.053 **	2.17
roe	0.419	0.40	0.725	1.04
cfo	1.909	1.60	2.906 ***	3.89
Constant	14.151 ***	7.24	−5.613 ***	−4.55
Observations	1504		1534
Controls	control		control
Industry	YES		YES
Year	YES		YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

, which are consistent with the results estimated from the fixed effects model.

5. Further Discussion

5.1. Mechanism Analysis

We have previously verified that economic policy uncertainty promotes technological innovation in agribusiness, so what is the mechanism by which economic policy uncertainty has a “crowding-in” effect on R&D investment? Macroeconomic shocks first bring about the restriction of external financing constraints, leading to the tightening of the financial flexibility of agribusinesses, which then turn to independent innovation to find a way out. At the same time, in the face of economic policy uncertainty, the organizational structure of agribusinesses becomes more visible in terms of managerial capacity, as the advantage of strong managerial capacity largely enhances the technological innovation level of agribusinesses.

Drawing on Zeng et al.’s [53,54,55,56,57] study, this paper uses the sum of cash flexibility and debt flexibility to measure financial flexibility (In the context of China, the issuance of stocks by enterprises is strictly regulated by the Securities Regulatory Commission, and the options related to substantive issues such as qualification assessment, timing of issuance, and amount of financing do not belong to enterprises, so equity flexibility is not included in the calculation of financial flexibility) (Financial), where cash flexibility is equal to the cash ratio of the firm minus the average cash ratio of the industry in which the firm is located, and debt flexibility is the greater of the difference between the average debt ratio of the industry in which the firm is located minus the debt ratio of the firm compared to zero. Drawing on the way Hayward et al. [58] measure managerial capability, the stronger the managerial capability, the higher the pay is relatively, so this paper uses the ratio of the highest paid executive pay to total executive pay to measure managerial capability (Manager).

5.1.1. Financial Flexibility

The regression results of financial flexibility as a transmission mechanism between economic policy uncertainty and agribusiness technology innovation are shown in

Table 12. The results of mediating-mechanism analysis of financial flexibility.

Variables	Innovation Input			Innovation Output
	(1)	(2)	(3)	(4)	(5)	(6)
	lnR&Di,t	Financiali,t	lnR&Di,t	lnPati,t	Financiali,t	lnPati,t
epu	0.673 ***	−0.055 ***	0.631 ***	0.455 ***	−0.055 ***	0.456 ***
	(4.46)	(−3.20)	(4.25)	(4.35)	(−3.20)	(4.35)
Financial			−1.341 ***			−0.770 ***
			(−2.81)			(−3.32)
top1	0.118	0.034	0.142	−0.089	0.034	−0.086
	(0.29)	(0.84)	(0.35)	(−0.33)	(0.84)	(−0.31)
growth	−0.037	−0.001	−0.037 *	−0.054 ***	−0.001	−0.053 ***
	(−1.63)	(−0.20)	(−1.66)	(−3.60)	(−0.20)	(−3.61)
mino	2.897 ***	−0.050	2.781 ***	1.910 ***	−0.050	1.921 ***
	(3.36)	(−0.71)	(3.25)	(3.61)	(−0.71)	(3.63)
tangibility	−1.484	0.447 ***	−0.866	0.863	0.447 ***	0.844
	(−1.35)	(5.03)	(−0.81)	(1.48)	(5.03)	(1.45)
cashflow	0.002	0.003 ***	0.006	−0.012 ***	0.003 ***	−0.012 ***
	(0.28)	(4.85)	(0.87)	(−2.96)	(4.85)	(−2.97)
tq	−0.033	0.005	−0.027	−0.073 ***	0.005	−0.073 ***
	(−0.83)	(1.38)	(−0.65)	(−3.35)	(1.38)	(−3.36)
age	0.040	−0.066 ***	−0.048	0.188 ***	−0.066 ***	0.191 ***
	(0.41)	(−8.76)	(−0.48)	(3.70)	(−8.76)	(3.76)
board	−0.162	−0.057 **	−0.271	0.249	−0.057 **	0.258
	(−0.49)	(−2.01)	(−0.82)	(1.34)	(−2.01)	(1.39)
roa	3.959 *	1.645 ***	6.195 ***	4.074 ***	1.645 ***	4.047 ***
	(1.83)	(9.83)	(2.75)	(3.13)	(9.83)	(3.12)
roe	0.411	−0.622 ***	−0.435	−0.145	−0.622 ***	−0.152
	(0.38)	(−7.77)	(−0.40)	(−0.24)	(−7.77)	(−0.25)
cfo	1.959 *	−0.171 *	1.702	2.283 ***	−0.171 *	2.329 ***
	(1.78)	(−1.88)	(1.57)	(3.63)	(−1.88)	(3.72)
Constant	14.092 ***	0.327 **	14.395 ***	−3.409 ***	0.327 **	−3.422 ***
	(8.73)	(2.21)	(8.99)	(−3.76)	(2.21)	(−3.77)
Observations	1504	1534	1504	1532	1534	1534
R-squared	0.159	0.229	0.167	0.232	0.229	0.234
Industry	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

. After introducing financial flexibility in the baseline regression, the estimated coefficient of economic policy uncertainty in column (2) and the estimated coefficient of the mediating variable Financial in column (3) are both significantly negative, thus corresponding to the mediating effect models (2) and (3); β₁∗α₂ is the same sign as α₁, and the same in terms of innovation output. This suggests a partial mediating effect of financial flexibility between economic policy uncertainty and agribusiness technological innovation, and the effect of this transmission path is that when economic policy uncertainty increases, the financial flexibility of agribusiness tightens in the opposite direction, thus prompting firms to pay more attention to technological innovation, increase R&D investment, and improve patent output. Research hypothesis H2 is tested.

5.1.2. Managerial Competence

Table 13. The results of mediating-mechanism analysis of managerial competency.

Variables	Innovation Input			Innovation Output
	(1)	(2)	(3)	(4)	(5)	(6)
	lnR&Di,t	Manageri,t	lnR&Di,t	lnPati,t	Manageri,t	lnPati,t
epu	0.673 ***	0.028 **	0.712 ***	0.498 ***	0.028 **	0.536 ***
	(4.46)	(2.58)	(4.71)	(4.82)	(2.58)	(5.13)
Manager			−1.599 ***			−1.338 ***
			(−2.69)			(−5.00)
top1	0.118	0.046 *	0.198	−0.112	0.046 *	−0.050
	(0.29)	(1.67)	(0.50)	(−0.41)	(1.67)	(−0.18)
growth	−0.037	0.010 ***	−0.021	−0.053 ***	0.010 ***	−0.040 ***
	(−1.63)	(3.31)	(−0.91)	(−3.62)	(3.31)	(−3.07)
mino	2.897 ***	0.106 *	3.017 ***	1.960 ***	0.106 *	2.102 ***
	(3.36)	(1.92)	(3.54)	(3.73)	(1.92)	(4.23)
tangibility	−1.484	−0.107	−1.679	0.501	−0.107	0.358
	(−1.35)	(−1.62)	(−1.58)	(0.87)	(−1.62)	(0.65)
cashflow	0.002	0.001 **	0.003	−0.014 ***	0.001 **	−0.013 ***
	(0.28)	(2.02)	(0.45)	(−3.47)	(2.02)	(−3.29)
tq	−0.033	0.008 ***	−0.021	−0.077 ***	0.008 ***	−0.066 ***
	(−0.83)	(3.24)	(−0.51)	(−3.52)	(3.24)	(−3.18)
age	0.040	−0.003	0.035	0.241 ***	−0.003	0.237 ***
	(0.41)	(−0.64)	(0.36)	(5.01)	(−0.64)	(4.95)
board	−0.162	−0.183 ***	−0.462	0.302	−0.183 ***	0.057
	(−0.49)	(−9.11)	(−1.38)	(1.63)	(−9.11)	(0.29)
roa	3.959 *	−0.078	3.907 *	2.781 **	−0.078	2.677 **
	(1.83)	(−0.45)	(1.81)	(2.18)	(−0.45)	(2.27)
roe	0.411	0.069	0.487	0.327	0.069	0.420
	(0.38)	(0.73)	(0.45)	(0.53)	(0.73)	(0.75)
cfo	1.959 *	−0.113 *	1.766	2.461 ***	−0.113 *	2.310 ***
	(1.78)	(−1.79)	(1.58)	(3.90)	(−1.79)	(3.72)
Constant	14.092 ***	0.786 ***	15.425 ***	−3.674 ***	0.786 ***	−2.623 ***
	(8.73)	(7.98)	(9.85)	(−4.03)	(7.98)	(−2.81)
Observations	1504	1534	1504	1534	1534	1534
R-squared	0.159	0.163	0.164	0.223	0.163	0.236
Industry	YES	YES	YES	YES	YES	YES
Year	YES	YES	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

reports the regression results for managerial capability as a mediating mechanism between economic policy uncertainty and agribusiness technology innovation. The results show that there is a “masking effect” between managerial competence, economic policy uncertainty, and technological innovation in agribusiness. In terms of innovation inputs, the regression coefficient β₁ (0.028) of economic policy uncertainty on the mediating variable in column (2) is significantly positive at the 5% level, indicating that when economic policy uncertainty increases, agribusinesses will be more willing to believe in the competence of managers. The estimated coefficient of economic policy uncertainty in column (3), α₁ (0.712), is significantly positive at the 1% level, but the estimated coefficient of the mediating variable Manager, α₂ (−1.599), is significantly negative when β₁*α₂ has a different sign from α₁; on the other hand, the total effect of economic policy uncertainty on innovation inputs, β₁ (0.673), is smaller than the direct effect of economic policy uncertainty on lnR&D, α₁ (0.712).

There is also a masking effect in terms of innovation output. The coefficient α₁ (0.536) of the effect of economic policy uncertainty on agribusiness technology innovation in column (6) is significantly positive at the 1% level, but the estimated coefficient α₂(−1.338) of the mediating variable Manager is significantly negative, at which point β₁∗α₂ has a different sign from α₁; also, the total effect of economic policy uncertainty on innovation output β₁ (0.498) is smaller than the direct effect of economic policy uncertainty on innovation output α₁ (0.536). This shows that the effect of economic policy uncertainty on agribusiness technology innovation is masked when managerial capacity is not controlled for, and the effect of economic policy uncertainty on agribusiness technology innovation is swiftly expanded once managerial capacity is controlled for.

5.2. The Moderating Effect of Government Subsidies

To verify the moderating effect of government subsidies on economic policy uncertainty affecting enterprise technology innovation, this paper introduces the interaction term between government subsidy and economic policy uncertainty in the equation and uses the cross-product method for estimation. The model results are shown in columns (1) and (2) of

Table 14. The results of moderating effect of government subsidies.

Variables	(1)		(2)
	lnR&Di,t	t-Value	lnPati,t	t-Value
epu	0.665 ***	4.41	0.492 ***	4.74
sub	−0.005 **	−2.28	−0.011 ***	−6.80
epu∗sub	0.001 **	2.29	0.002 ***	6.85
top1	0.102	0.25	0.038	0.14
growth	−0.063 *	−1.75	−0.070 **	−2.50
mino	3.052 ***	3.42	2.007 ***	3.78
tangibility	−1.560	−1.31	0.369	0.61
cashflow	0.003	0.41	−0.015 ***	−3.57
tq	−0.028	−0.69	−0.080 ***	−3.66
age	0.021	0.21	0.275 ***	5.67
board	−0.194	−0.58	0.359 *	1.93
roa	3.745 *	1.72	2.879 **	2.26
roe	0.475	0.44	0.239	0.39
cfo	1.950 *	1.75	2.548 ***	4.01
Constant	14.310 ***	8.52	−3.767 ***	−4.09
n	1477		1506
R-squared	0.163		0.235
Industry	YES		YES
Year	YES		YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

. The regression results show that the epu∗sub coefficients of 0.001 and 0.002 in the regression analysis of R&D input and patent output of agricultural enterprises pass the significance test, respectively. This indicates that government subsidies strengthen the role of economic policy uncertainty in promoting technological innovation in agribusiness, and hypothesis H4 is verified.

Possible reasons for this result are: first, government subsidies provide direct financial support to agribusinesses, such as R&D subsidies and innovation funds. Second, agribusiness innovation products have the characteristics of public goods [59,60], and the stimulation of agribusinesses by government subsidies exerts a market failure correction effect. Agribusinesses that receive government subsidies are equivalent to receiving innovation certification provided by the government, which will enhance agribusinesses’ motivation to invest in R&D and thus become more conducive to agribusiness technology innovation.

5.3. Identifying the Effects of Environmental Uncertainty

Distinct from the impact of macro-environmental uncertainty on firms’ R&D investment, existing studies [54] have shown that micro-environmental uncertainty may lead to a decrease in firms’ R&D investment. Scholars have identified environmental uncertainty as an important factor affecting firms’ technological innovation [49,51] and used fluctuations in firms’ operating income to measure environmental uncertainty, demonstrating that this uncertainty also has an important impact on firm innovation in recent years. In order to remove the uncertainty of macroeconomic impact on agricultural firms and focus on the impact of micro-environmental uncertainty on technological innovation in agricultural firms, this paper introduces environmental uncertainty in the regression model.

Academics [55,56] usually measure environmental uncertainty in terms of fluctuations in sales revenue; to exclude the effect of industry, this paper uses the standard deviation of sales revenue over the past five years with industry-adjusted values to measure firms’ environmental uncertainty [61,62,63]. The regression results after introducing environmental uncertainty are shown in columns (1) and (2) of

Table 15. The results of identifying the impact of environmental uncertainty.

Variables	(1)	(2)	(3)	(4)
	lnR&Di,t	lnPati,t	lnR&Di,t+1	lnPati,t+1
epu	0.742 ***	0.446 ***	0.943 ***	0.367 ***
	(4.68)	(4.13)	(3.62)	(3.05)
eu	−0.320 ***	−0.229 ***	−0.341 ***	−0.225 ***
	(−3.76)	(−6.28)	(−2.77)	(−5.20)
top1	−0.188	−0.547 *	−0.660	−0.865 **
	(−0.40)	(−1.82)	(−1.07)	(−2.38)
growth	0.070 **	0.024 **	0.082 **	0.024
	(2.55)	(2.02)	(2.44)	(1.41)
mino	2.705 ***	1.891 ***	3.257 **	1.568 **
	(3.01)	(3.71)	(2.42)	(2.44)
tangibility	−1.585	0.682	−0.569	1.182 *
	(−1.38)	(1.19)	(−0.37)	(1.77)
cashflow	0.001	−0.020 ***	−0.001	−0.020 ***
	(0.12)	(−4.09)	(−0.07)	(−3.29)
tq	−0.034	−0.063 ***	−0.093	−0.086 **
	(−0.79)	(−2.89)	(−1.48)	(−2.53)
age	−0.105	0.287 ***	−0.120	0.255 ***
	(−0.75)	(4.35)	(−0.59)	(3.13)
board	−0.340	0.024	0.103	−0.124
	(−0.93)	(0.12)	(0.19)	(−0.54)
roa	5.136 **	3.450 ***	11.959 ***	3.228 **
	(2.15)	(2.70)	(4.31)	(1.97)
roe	−0.184	−0.016	−1.818	0.580
	(−0.17)	(−0.03)	(−1.17)	(0.72)
cfo	1.784	2.729 ***	0.890	3.730 ***
	(1.31)	(3.93)	(0.41)	(4.33)
Constant	15.048 ***	−2.703 ***	11.994 ***	−1.983 *
	(8.95)	(−2.91)	(5.14)	(−1.75)
Observations	1266	1288	855	864
R-squared	0.178	0.272	0.182	0.285
Industry	YES	YES	YES	YES
Year	YES	YES	YES	YES

Note: Data in parentheses are t-values; *** indicates p < 0.01; ** indicates p < 0.05; * indicates p < 0.1.

, and the results indicate that economic policy uncertainty and environmental uncertainty both have an effect on agribusiness innovation. Columns (3) and (4) show that economic policy uncertainty and environmental uncertainty do not change significantly in their effects on technological innovation in the next period, indicating that economic policy uncertainty has an incremental explanatory effect on technological innovation in agribusiness.

5.4. Discussion

The Global Economic Policy Uncertainty Index (GEPI) compiled by Prof. Baker’s team shows that the GEPI is near its highest ever at this stage. At present, the world economy is in a period of deep adjustment, and after two major economic crises, the “global financial crisis” and the “global COVID-19”, the “pandemic outbreak in 2020” and “widespread vaccination in 2021” are coming. Geopolitical risks and other changes in the external environment will also increase economic policy uncertainty, such as the Brexit referendum, the Russia-Ukraine war, the election of Donald Trump, and the nuclear and ICBM tests conducted by North Korea. Both macroeconomic and geopolitical developments will have an impact on innovation and uncertainty in China.

The banking crisis is likely to cause firms to reduce investment while increasing global economic policy uncertainty. First, the banking crisis will discourage firms from investing in innovation. Theoretically, the banking crisis may restrict agribusiness financing, as evidenced by the increased difficulty of financing, the decline in the scale of financing, and the increase in the cost of financing, so that agribusinesses will reduce their innovative investments. Taking the transmission mechanism of government guarantees as an example, which play a central role in the stability and functioning of the banking sector [64,65]. Economic crises can affect the strength of government guarantees, which in turn affects banks through their tail risk-taking, endogenous risk-taking, investor supervision, and government supervision [66], among other four channels that affect banks’ performance. When banks’ profitability is affected by a crisis, there will be earnings volatility, while investors and regulators may prefer smoother earnings [67,68], so the banking crisis may cause companies to reduce their investments. Second, the Global Financial Crisis (2007–2009), the Eurozone sovereign-debt Crisis, and the COVID-19 pandemic contributed to increased economic policy uncertainty. Dantas et al. [69] analyze the time series of the economic policy uncertainty (EPU) index for the euro area and find that there is a significant movement in the EPU index before and after the creation of the euro area. This indirectly confirms that large global economic events will affect the economic policy uncertainty index.

The monetary expansion will increase global economic policy uncertainty and has the potential to promote innovation in Chinese agribusiness in the short term. First, in terms of monetary policy, in response to the Subprime and COVID–19 crises, central banks of developed economies, including US Fed, ECB, Bank of England and Bank of Japan, implemented a series of quantitative easing policies [70]. It is worth noting that from the Subprime to the COVID-19 crisis, the policy response transitioned from a unilateral response by the Federal Reserve to a diversified response by central banks, especially the Bank of Canada, which saw significant balance sheet expansion in 2020. As widely documented, there is a strong correlation between financial conditions in the U.S. and international markets [71], so a Fed monetary expansion would increase the global economic policy uncertainty index. Second, in theory, the implementation of quantitative easing monetary policy in advanced economies will lead to a significant depreciation of their currencies and lower interest rates, and capital inflows to emerging economies such as China will increase the supply of funds and lower interest rates in emerging economies. In the short term, the dual favorable policies of increased capital supply and lower capital costs will likely prompt Chinese agribusinesses to increase their investment in innovation. However, in the long run, quantitative easing monetary policy will sow the potential for global inflation and worsen the economic situation of the relevant trading body. Dedola et al. [72] find that quantitative easing measures have a large and lasting impact on exchange rates. Quantitative easing announcements lead to an increase in relative balance sheets of about 20% over the sample period, which in turn leads to a sustained depreciation of exchange rates of about 7%. One of the most direct manifestations of quantitative easing is that a significant depreciation of the domestic currency benefits the domestic export sector but conversely leads to an appreciation of the currency of the economy in the relevant economies, thus being highly detrimental to export-oriented emerging economies in a financial whirlwind.

Geopolitical uncertainty will make Chinese agribusiness innovation investments more attractive. Rising political uncertainty appears to have been a global phenomenon in recent years [28]. The 2016 U.S. election, the Brexit vote, and the Russia-Ukraine conflict have exacerbated political risks in their countries. In terms of uncertainty, Born et al. [73] find that shocks to economic policy uncertainty and growth expectations explain almost the entire output loss in the quarter following the Brexit vote. Overall, uncertainty and growth expectations account for about half of the total economic costs of the Brexit vote. At the level of impacts on firms, Campello et al. [28] focus on the international transmission of uncertainty and find that the 2016 Brexit vote affected employment, investment, and R&D at U.S. firms. The findings suggest that foreign-generated political uncertainty will propagate across international borders, thereby affecting domestic capital formation and labor allocation. For local Chinese agribusinesses, theoretically, on the one hand, political risk will increase uncertainty about economic policies in countries involved in geopolitical conflicts, compared to China, where politics are more stable, political risk is less, and market prospects are better, so Chinese stocks become more attractive and will promote technological innovation in Chinese agribusinesses. On the other hand, geopolitical crises will affect the economies and agricultural production of other countries, which will be more favorable to China’s agricultural exports, and the market potential of Chinese agricultural products will become larger, thus driving Chinese agricultural enterprises to innovate technology, transform their achievements, and promote their application.

Overall, macroeconomic events have both positive and negative effects on Chinese agribusinesses innovation. We do not put these factors in our econometric model, mainly because we consider that these shocks are currently in the error term and may be able to maintain a kind of offset between each other, thus having little impact on our research findings. In addition, data related to macroevents and geopolitical changes are difficult to obtain, and there is a lack of such data in China at present, which is a limitation of this paper. We intend to refine it at a later stage and try to collect data in this area in the future for an in-depth study.

6. Conclusions and Policy Implications

6.1. Conclusions

Using the panel data of Chinese A-share listed agricultural enterprises from 2007–2021, we empirically examine the impact, heterogeneity, mechanisms, and the moderating effect of economic policy uncertainty on technological innovation in agricultural enterprises. We can draw the following conclusions in four aspects: First, increased economic policy uncertainty significantly promotes technological innovation in agricultural firms, and the findings remain valid after endogeneity treatment and robustness checks. Second, in terms of ownership nature, the “incentive” effect of economic policy uncertainty on technological innovation is more pronounced for state-owned agribusinesses than for non-state-owned agribusinesses. Economic policy uncertainty promotes technological innovation more significantly for basic agribusinesses than for manufacturing agribusinesses. In times of increased economic policy uncertainty, high adjustment cost agribusinesses may not invest enough in R&D due to financing constraints compared to low adjustment cost agribusinesses. However, the innovation outputs of high-adjustment-cost agribusinesses are more significant than those of low-adjustment-cost agribusinesses, which may be due to the return on the previous innovation investment. Third, the path analysis of the mediating mechanism suggests that when economic policy uncertainty is enhanced, the financial flexibility of agribusinesses tightens, which makes them shift their development strategies to internal innovation and improve their technological innovation; meanwhile, the managerial capacity will create a masking effect in the process of the impact of economic policy uncertainty on agribusinesses’ technological innovation when economic policy changes. Fourth, government subsidies reinforce the positive contribution of economic policy uncertainty to agribusiness technological innovation.

6.2. Recommendations

Based on the above findings, this paper proposes the following policy recommendations: for agricultural enterprises, they should raise their awareness of financial flexibility reserves and enhance their financing capacity in times of economic change. When economic policy uncertainty increases, agribusinesses should actively play a “preventive” role for their financial flexibility in response to the external environment and at the same time seize innovation opportunities to realize the “utilization” property of financial flexibility to ensure that agribusinesses have more options than independent innovation in the face of macroeconomic shocks.

In terms of economic policy, first, policy transparency should be improved as much as possible to reduce the impact of economic policy uncertainty on heterogeneous agricultural enterprises. Second, focus on the intensity and frequency of economic policy formulation and adjustment to reduce the impact on technological innovation of non-state enterprises, manufacturing enterprises, and low-adjustment-cost enterprises. Moreover, the government should give full play to the effect of subsidies to agribusiness when the macroeconomic environment is uncertain. On the one hand, when the government subsidizes and supports key industries in the agricultural field, it should consider various factors, such as R&D projects and the nature of the industry, so as to provide precise support to agricultural enterprises and support the innovation and development of leading agricultural industrialized enterprises. On the other hand, government subsidies should be used as a “accomplishing a great task with little effort by clever maneuvers” function to guide social capital to enter to support manufacturing agricultural enterprises. Finally, the flexibility of economic policies should be brought into play to promote the development of innovative agricultural enterprises. Different agricultural enterprises in different industries and of different natures have different perceptions of economic policy changes, so policy formulation and implementation should be tailored to local conditions and vary from enterprise to enterprise. While ensuring that state-owned agribusinesses enjoy domestic agricultural support policies, private agribusinesses should be encouraged by tax incentives, and foreign-funded agribusinesses should focus on improving border protection policies to stimulate the innovation drive of various agribusinesses. Manufacturing agribusinesses and agribusinesses with low R&D adjustment costs are less motivated to innovate due to uncertainty factors, so the government should guide the technological innovation activities of relevant agribusinesses when designing policies.