This study investigates the dynamic, bidirectional relationship between climate policy uncertainty (CPU) and agricultural investment (AGRI) in China. Employing a bootstrap rolling-window Granger causality test on monthly data from January 2012 to December 2024, we find a significant and time-varying interaction between the two variables. The impact of CPU on agricultural investment is predominantly negative, particularly during periods of policy ambiguity or unclear implementation. Conversely, this negative effect can be mitigated or even reversed when strong government support for agricultural modernization provides a stable and predictable environment. Conversely, agricultural investment also influences climate policy uncertainty, with the effects varying over time. The study underscores the complex, time-dependent interactions between climate policies and agricultural investments, highlighting the importance of transparent, stable climate policies to foster agricultural development and resilience.
agriculture investmentbootstrap rolling-windowclimate policy uncertaintygranger causalitytime-dependent interrelationshipThe author(s) declared that financial support was received for this work and/or its publication. This research was funded by the Chinese Academy of Engineering Strategic Research and Consulting Program, grant numbers 2022-XBZD-03 and 2024-XZ-15.section-at-acceptanceAgricultural and Food EconomicsIntroduction
In the global effort to combat climate change, climate policy uncertainty (CPU) is emerging as a key barrier to long-term sustainable investment (Moss et al., 2010). This challenge is particularly acute in the agricultural sector (Rogers, 2016), which is highly sensitive to the policy environment due to its long investment cycles and high sunk costs (Ju et al., 2020). Such uncertainty is especially consequential given the sector’s dual mission of ensuring food security and promoting climate adaptation (Godfray et al., 2010). In China, the world’s largest agricultural economy, ambitious climate goals coexist with this unpredictability. Variability in policy pathways, implementation strength, and future direction creates significant uncertainty for agricultural investors, potentially undermining the country’s sustainable transition.
However, focusing on the specific source of climate policy uncertainty, its impact mechanism on the agricultural sector remains underexplored. Previous studies have predominantly concentrated on less-developed, agriculture-based economies, and have often failed to adequately consider the unique political-economic system of China and the interaction between its investment and policy (Kotu et al., 2023; Mukashov et al., 2024). More importantly, few studies have investigated whether agricultural investment behavior, in turn, influences the adjustment of climate policy, thereby forming an endogenous interactive relationship (Zhang, 2025). We found that past research has studied the one way relationship between CPU and AGRI through static methods (Dai, 2025; Karagöl and Şahin, 2025; Lai et al., 2025), but the one way relationship research often ignores the dynamic relationship of time, which hinders the analysis of the relationship between CPU and AGRI and the direction of influence. In the modern era of climate change and faster and faster economic development, more research is needed to analyze the potential interrelationship between the two to deal with the problem of sustainable agricultural development, especially in a large food-consuming country like China. Therefore, the core questions of this paper are: Within China’s specific institutional context, how does climate policy uncertainty affect agricultural investment decisions? And is there a feedback mechanism between the two? Answering these questions has important theoretical and practical significance for understanding China’s agricultural transformation and development path and optimizing climate governance.
The potential merits of this research can be outlined as follows. First, while existing literature predominantly emphasizes the impact of climate policy on agricultural investment, relatively little attention has been paid to the reverse effect of agricultural investment on policy formulation. Given the long-term nature of agricultural investment cycles, which may shape policymakers’ decisions, this study highlights the feedback effects of agricultural investment behavior on climate policy adjustments, thereby offering a novel perspective for climate policy design. Secondly, the previous literature mainly relied on linear regression or static time series methods to assess the impact of policies on investment. Although these methods are valuable, they usually assume parameter constancy and one-way causality, so it is difficult to solve the endogenous and dynamic feedback loops between variables. In order to overcome these limitations, this study used a VAR framework to examine the multidimensional Granger causal relationship. By using the Bootstrap rolling window causal relationship test, we can resolve the endogenicity between climate policy uncertainty and agricultural investment, while also capturing the structural changes and time-varying characteristics of their interactions. Third, previous research has to some extent overlooked the possibility that the interaction between climate policy uncertainty and agricultural investment is subject to continuous external shocks. This study applies four tests of parameter stability to demonstrate that conventional full-sample estimates may be unreliable. By using sub-sample techniques that reduce estimation errors, this study uncovers the time-varying transmission mechanisms between climate policy uncertainty and agricultural investment.
The structure of this paper is as follows: Section 2 presents a review of the existing literature. Section 3 outlines the empirical methodology used in our models. Section 4 describes the data employed in the analysis. Section 5 provides a detailed discussion of the empirical results. Section 6 concludes the paper and offers policy implications. Finally, Section 7 highlights the research gaps and future directions for this study.
Theoretical framework and literature reviewConceptual framework and transmission mechanisms
To elucidate the complex, time-varying interaction between climate policy uncertainty and agricultural investment, this study constructs a dynamic theoretical framework incorporating China’s institutional context (see Figure 1). We posit that the impact in both directions is not constant but contingent on the macroeconomic and policy environment, manifesting through opposing theoretical channels.
The conceptual figure.
Diagram illustrating the relationship between Climate Policy Uncertainty (CPU) and Agricultural Investment (AGRI). The upper section shows negative effects via real options and risk premium and positive effects via strategic preemption signals. The lower section indicates negative effects through policy lock-in and investment stability, with positive effects through fiscal and resource competition.Transmission from CPU to AGRI
The impact of CPU on agricultural investment is determined by the net effect of two countervailing mechanisms:
The Inhibitory Channel. This mechanism is theoretically grounded in Real Options Theory, investment decisions are analogous to purchasing financial call options (McDonald and Siegel, 1986). Agricultural investments are characterized by long cycles and high sunk costs. Agricultural assets often possess high asset specificity and significant sunk costs, making liquidation difficult if policy conditions deteriorate. When climate policy signals become ambiguous, the option value of waiting increases (Abel, 1983). Rational agricultural producers, typically exhibiting risk aversion, will calculate that the cost of a potential policy misstep exceeds the opportunity cost of delay (Binswanger, 1980). Consequently, they postpone capital commitment to avoid the risk, leading to a temporary contraction in investment volume. Concerning the industry environment, the agricultural sector is highly dependent on government subsidies, regulations, and support mechanisms (Mittenzwei et al., 2017). Therefore, any fluctuation in climate policies, such as carbon taxes, subsidy adjustments, or environmental regulations, directly impacts the long-term expected returns and risk assessments of investments, compelling agricultural investors to adopt more conservative financial strategies (Autio et al., 2021; Sun and Baker, 2021). China’s agricultural sector has long been subject to strong government intervention and is characterized by significant regional development disparities (Dai, 2025; Wang et al., 2025).
The Catalyst Channel. Conversely, within the context of China’s state-guided economic transition, specific policy uncertainties can paradoxically stimulate investment. This phenomenon is often driven by the anticipation of stricter future regulations (e.g., the Dual Carbon goals). When policy direction is clear, but the timing or intensity is uncertain, forward-looking firms do not perceive ambiguity solely as a risk. Instead, they interpreted it as a signal of upcoming regulatory reforms. In this case, uncertainty creates an opportunity. Investors speed up the pace of investment, the purpose is to ensure the first-mover advantage of technology adoption before standards are tightened (Kulatilaka and Perotti, 1998). At the same time, ensure that investment activities are consistent with the national strategy in order to obtain political legitimacy and subsidies. And establish a competitive moat against the laggards. Therefore, uncertainty acts as a catalyst, stimulating expected investment to ensure survival in a changing policy environment and obtain government support (Guan et al., 2021).
Transmission from AGRI to CPU
Similarly, the feedback loop from agricultural investment back to policy uncertainty is bidirectional, operating through stabilization and provocation mechanisms:
The Stabilization Channel. Large scale agricultural investments can actively dampen policy volatility. This operates primarily through a signaling mechanism(Steinbuks and Hertel, 2013). Substantial sunk costs in climate-adaptive agriculture demonstrate the industry’s strong commitment and alignment with national ecological strategies. This alignment reduces information asymmetry between regulators and producers, enhancing government confidence and policy credibility. Furthermore, robust investment strengthens the agricultural supply chain, ensuring food security which is a core priority for the Chinese government(Tian et al., 2015). Especially in the context of increasingly frequent extreme climate events and external shocks (such as the COVID-19 pandemic), maintaining the stable adaptability of agricultural investment and production has become a key consideration influencing the priorities of climate policy (Okolie and Ogundeji, 2022; Nair, 2020). When the sector demonstrates resilience and self-sufficiency through green investment, it reduces the urgency for the government to implement sudden, disruptive interventions or corrective shocks. Consequently, high levels of stable investment foster a “policy lock-in” effect, where the regulatory environment becomes more predictable and CPU declines.
The Provocation Channel. However, investment behavior can also act as a source of uncertainty, particularly during phases of rapid expansion or economic constraint. In the early stages of green transformation, a disorderly surge in investment may trigger market fears of greenwashing or bubbles. This compels policymakers to frequently adjust standards, audit subsidies, or tighten verification processes, thereby increasing regulatory uncertainty(Yang et al., 2024). During economic downturns, aggressive agricultural investment competes with other critical sectors for limited fiscal resources. This crowding-out effect forces policymakers into difficult trade-offs regarding subsidy allocation and environmental enforcement intensity(Bai et al., 2019). The oscillation of government support in response to fiscal pressure heightens the volatility of climate policy signals, thereby exacerbating CPU.
Literature review
Having established the theoretical framework, we review the relevant empirical literature, focusing on the measurement of climate policy uncertainty and its documented economic impacts. Climate policy is a crucial instrument for achieving sustainable development, its effectiveness is often undermined by inherent instability (Iglesias et al., 2011). In China, although long-term climate goals are clearly articulated (Li et al., 2024), implementation is frequently inconsistent, driven by short-term economic and political agendas as well as a disconnect between international commitments and domestic action (Chmutina et al., 2012; Wu, 2023; Xiang and Gevelt, 2025).
A key issue in this series of studies is how to accurately measure climate policy uncertainty. The quantification of policy uncertainty builds upon the pioneering Economic Policy Uncertainty (EPU) index developed by Baker et al. (2016). Subsequent research has adapted this text-based methodology to specific domains, with Palikhe et al. (2024), for example, validating its effectiveness in capturing major environmental policy events. More recently, scholars have constructed CPU indices specifically for China via two main approaches: leveraging novel data sources such as social media and applying advanced algorithms like deep learning to analyze official media texts (Lee and Cho, 2023; Ma et al., 2023). Collectively, these pioneering studies affirm the feasibility of quantifying China’s CPU and provide the methodological foundation for this paper’s investigation into its economic consequences.
The empirical nexus between policy uncertainty and investment has been extensively debated. The mainstream view supports the depression effect, finding that macroeconomic uncertainty significantly inhibits corporate investment activities (Bloom et al., 2007; Gulen and Ion, 2016). Specific to the agricultural sector, agriculture is not only a source of greenhouse gas emissions but also a vulnerable sector affected by climate change (Calvin et al., 2016). Research has found that farmers generally adopt conservative strategies amidst ambiguity due to the sector’s vulnerability (Mittenzwei et al., 2017). However, existing literature presents two notable limitations. First, most studies assume a static, unidirectional impact from policy to investment, largely overlooking the potential reverse causality where agricultural investment behavior might influence policy formulation (Zhang, 2025). Second, traditional full-sample econometric models fail to capture structural breaks. As noted by Balcilar et al. (2010), the relationship between economic variables is often time-varying. Ignoring these temporal shifts may mask the coexistence of positive and negative shocks identified in our theoretical framework.
In summary, this study systematically investigates the dynamic causal relationship between climate policy uncertainty and agricultural investment within China’s unique institutional context. Employing the China Climate Policy Uncertainty index from Ma et al. (2023), we use a bootstrap rolling-window causality test to examine the time-varying nature of this linkage. This research aims to contribute novel micro-level evidence on the economic consequences of climate policy, providing an empirical foundation for policymakers to formulate more adaptive and effective strategies amidst prevailing uncertainty.
MethodologyFull-sample technique
The Granger causality test statistic follows the standard asymptotic distribution, which is a prerequisite for obtaining accurate results in traditional vector autoregression models. Any deviation from this distribution can lead to unreliable results (Diks and Panchenko, 2006). To overcome this limitation, our study employs the bootstrap (RB) method, which is particularly effective for tests based on standard asymptotic distributions (Shukur and Mantalos, 2004). At the same time, Monte Carlo simulations may not sufficiently enhance the Wald test in small sample studies. However, the likelihood ratio (LR) test proposed by Shukur and Mantalos offers a more accurate adjustment for simulation power (Shukur and Mantalos, 2000). Consequently, we use the RB-augmented modified LR statistic to analyze the relationship between CPU and AGRI, ensuring a thorough investigation of variables that do not follow normal distributions. To further account for macroeconomic uncertainties and sectoral price fluctuations, we introduce EPU and ANFPPI as control variables. To demonstrate this approach, we present the multivariate VAR(p) system below, which incorporates the RB-based modified-LR causality test. See Equation 1 for the initial form.
Xt=β0+β1Xt−1+……βpXt−p+εt,t=1,2,……,T
We used the Schwarz Information Criterion (SIC) to identify the optimal lag order p The variable X is given by Xt = (CPUt + AGRI2t)′, from which Equation 2 can be derived.
εt=(ε1t,ε2t) is a white-noise process; βij(L)=∑k=1pβij,kLk,i,j=1,2. L are the lag operators; then,LkXt=Xt−k. We can establish the null hypothesis: there is no significant Granger causality between CPU and AGRI, which is expressed as β12,K=0,k=1,2,……,p. Similarly, the reverse hypothesis is,β21,K=0,k=1,2,……,p.
Stability test of parameters
Causality tests conducted on the full sample typically assume that the parameters of the Vector Autoregressive (VAR) model remain constant over time (Yuan et al., 2022). However, in practice, these parameters often exhibit instability, which can lead to inaccurate results. To address this issue, the Sup-F, Ave-F, and Exp-F tests are commonly employed, as they effectively mitigate the problem of parameter instability (Andrews, 2003). Specifically, the Sup-F test helps identify structural changes within the model, while the Ave-F and Exp-F tests assess the stability of the parameters. Additionally, the Lc test allows for the detection of whether the parameters follow a random walk process (Nyblom, 1989). The presence of time-varying parameters can cause instability in the relationship between CPU and AGRI, potentially resulting in erroneous conclusions when applied to the full sample. To overcome this limitation, the present study utilizes a subsample technique to more accurately analyze the dynamic relationship between CPU and AGRI.
Subsample technique
The rolling-window test involves segmenting a time series into overlapping subsamples of a fixed size to analyze its dynamic properties (Balcilar et al., 2010). Choosing the window size is critical, as it involves a trade-off: larger windows increase estimator precision but can obscure structural breaks, whereas smaller windows excel at detecting breaks but reduce precision. This issue is often managed by setting a minimum window size, particularly in datasets with known parameter instability (Pesaran and Timmermann, 2005). In this study, we assume the time series length is denoted as 𝑇 and the rolling-window width as 𝑟, with each segment represented as 𝑟, 𝑟+1, …, T, leading to T − r + 1 time series. Then apply the RB-based revised-LR method to determine the Granger causality for each subsample, with the LR statistics and p-values organized chronologically based on the subsample technique. The average of all bootstrap estimations, Nb−1∑k=1pβ̂21,ak∗ and Nb−1∑k=1pβ̂21,k∗, represent the influence of CPU on AGRI and the influence of AGRI on CPU. Nb represent the frequency of repeated bootstraps, while β̂21,k∗ and β̂21,k∗ are parameters of Equation 2. The confidence interval used in this section is 90%, with the upper and lower bounds corresponding to the 95th and 5th percentiles.
We chose the bootstrap subsample rolling-window Granger causality test for its effectiveness with small or structurally unstable datasets. Our CPU and AGRI series exhibit such time-varying properties, necessitating a dynamic analysis rather than assuming constant parameters. Alternative methods are less suitable: the full-sample Granger test is biased by structural changes, panel data regression misses time-varying causality, and the NARDL model cannot precisely identify subsample shocks (Su et al., 2021). In contrast, the rolling-window approach is well-suited to our objectives, offering a robust tool to detect significant shifts driven by policy or market transitions (Du and Chen, 2025).
Data select and testData select
This paper investigates the relationship between climate policy uncertainty and agricultural investment in China using monthly data from January 2012 to December 2024. Climate policy uncertainty is measured by the China Climate Policy Uncertainty index from the study of Ma et al. (2023). Unlike traditional indices based on simple keyword counting, this index is constructed using a MacBERT-based deep learning algorithm combined with rigorous manual auditing, ensuring high semantic accuracy. The data source covers 1.75 million articles from six authoritative mainstream newspapers in China, including the People’s Daily and GuangMing Daily, spanning from 2000 to 2024. The construction process involved training the model on a manually labeled dataset, achieving a classification accuracy of 97.83% relative to human auditing. The monthly index was then calculated based on the standardized frequency of news items related to climate policy uncertainty, following the methodology of Baker et al. (2016). This index has been widely validated and provides a reliable proxy for capturing the dynamics of China’s climate policy environment. Agricultural investment is measured by the cumulative growth of fixed asset investment in agriculture, sourced from the National Bureau of Statistics (NBS) of China. To isolate the specific impact of climate policy uncertainty and ensure the robustness of our results, we introduce two control variables. First, to control for broader macroeconomic shocks, we employ the Economic Policy Uncertainty (EPU) index developed by Baker et al. (2016). Second, to account for price fluctuations in the downstream sector, we include the ex-factory price index of the agricultural and non-staple food processing industry (ANFPPI), which is also sourced from the National Bureau of Statistics of China. Table 1 presents the definitions and sources of the data employed in this study.
The definition of variables.
Variables
Definition
Source
CPU
China’s climate policy uncertainty index
Figsharehttps://figshare.com/
AGRI
Growth in accumulated value of agricultural fixed-asset investment
National Bureau of Statistics of Chinahttps://www.stats.gov.cn/sj/
EPU
China’s economic policy uncertainty index
Figsharehttps://figshare.com/
ANFPPI
The ex-factory price index of the agricultural and non-staple food processing industry
National Bureau of Statistics of Chinahttps://www.stats.gov.cn/sj/
The selection of 2012 as the starting year for the sample period is based on two key considerations: First, it is based on data availability and the consistency of statistical standards. The NBS of China only began to separately publish the cumulative growth rate of fixed asset investment for agriculture (specifically, the crop cultivation industry) from 2012 onwards. Prior data represented a combined total for agriculture, forestry, animal husbandry, and fishery. Furthermore, after 2011, the NBS adjusted its statistical caliber for fixed asset investment. Therefore, 2012 was chosen as the starting point to ensure the homogeneity of the data series and the reliability of the results. Second, 2012 was a key turning point for China’s economy and policy. On the political and strategic level, the 18th National Congress of the Communist Party of China (CPC) in 2012 incorporated Ecological Civilization Construction into the country’s overall layout for the first time, marking a fundamental shift that elevated climate and environmental issues to a core national strategic level (Wu et al., 2021). On the agricultural development level, 2012 was also a significant period of adjustment for the structure of agricultural investment and the direction of policy support. The allocation of agricultural subsidies has transitioned from general grain growing support to sectors driven by technology and productivity, including seed industry innovation and facility modernization (Li et al., 2021).
Data trend analysis
By examining the developmental trends of CPU and agricultural investment, we can further analyze the interrelationship between them. According to Figure 2, the CPU index exhibits a general upward trend amidst fluctuations, and its volatility shows distinct cyclical and event-driven characteristics. The index consistently rises in the first quarter, likely influenced by the national “Two Sessions” in March, which sets the annual policy direction (Shi et al., 2020). Similarly, significant spikes occurred around the CPC National Congresses in 2012 and 2017, which define the policy agenda for the subsequent five years. Furthermore, major domestic and international events have also driven sharp fluctuations in the index. For example, the signing of the Paris Agreement in 2015 prompted intensive adjustments to domestic climate policy (Richards et al., 2018). In late 2019, at the beginning of the COVID-19 outbreak, a shift in policy focus caused uncertainty to drop sharply to a temporary low. Subsequently, China’s proposal of the dual carbon goals in 2020 led to exploration of new policy instruments like carbon taxes and carbon trading, causing the CPU index to rise sharply (Xu et al., 2021). However, the index fell back in 2022 as the focus returned to pandemic control due to virus variants. In 2023 and 2024, as the pandemic’s influence waned, a confluence of economic pressures and extreme weather events prompted policy oscillations, causing the CPU index to remain at a persistently high level.
The trends of CPU and AGRI. The solid line depicts the CPU trend, shown on the left axis; the dashed line illustrates the AGRI trend, shown on the right axis.
Line graph showing CPU and AGRI indices from 2012 to 2024. Key events like the 18th CPC National Congress, Paris Agreement, COVID-19, China-U.S. trade dispute, and carbon goals are marked. CPU fluctuates more than AGRI, with noticeable peaks and troughs around major events.
In contrast to the CPU index, agricultural investment (AGRI) has followed a general downward trajectory. Propelled by favorable government policies and budget increases, investment expanded rapidly from 2012 to its apex in 2015. However, a continuous decline began in 2016, driven by diminishing profit expectations, rising production costs, and a broader shift toward conservative corporate strategies. This capital reticence was compounded by frequent natural disasters and disappointing returns on investment (Cassar et al., 2017). The downturn was exacerbated by the 2018 Sino-US trade dispute, which raised input costs and impeded exports (Fajgelbaum and Khandelwal, 2022). Subsequently, the COVID-19 pandemic triggered a sharp fall in late 2019, with investment remaining negative throughout 2020. A modest, stimulus-spurred recovery in 2021 proved short-lived (Pan et al., 2020). After 2022, investment growth decelerated again due to waning policy dividends, a global economic slowdown, and more frequent extreme weather. Underlying structural issues may also contribute, such as poorly targeted subsidies and lagging mechanization creating diseconomies of scale amid rising labor costs (Jayne and Rashid, 2013).
In summary, there are significant differences in the trends of CPU and agricultural investment, suggesting that a complex dynamic relationship may exist between them. Traditional full-sample Granger causality tests may struggle to accurately capture this time-varying interaction. Therefore, this paper will subsequently employ subsample techniques with the aim of more precisely identifying the evolving relationship between the two.
Descriptive statistics
The Table 2 are the descriptive statistics for CPU and AGRI. The maximum value of CPU is 4.122, which is 1.66 times its mean, while the minimum value is 1.117. The standard deviation indicates that the CPU data exhibits significant fluctuation, suggesting that climate policy uncertainty varies noticeably across different time periods. The maximum value of AGRI is 0.799, which is 4.1 times its mean, and its minimum value is −0.431, indicating that agricultural fixed investment experienced a contraction in certain periods. Regarding the control variables, the log-transformed Economic Policy Uncertainty (EPU) shows a mean of 4.94 with a standard deviation of 0.19, while ANFPPI averages 101.1. In terms of skewness and kurtosis, CPU has a slightly longer right tail and a kurtosis of less than 3, indicating a platykurtic distribution. In contrast, AGRI is nearly symmetrical with data more concentrated around the mean, but its kurtosis is greater than 3, indicating the presence of some extreme values. The Jarque–Bera test results show that the p-values for CPU, AGRI and both control variables are greater than 0.05. This means we cannot reject the null hypothesis of the Jarque–Bera test, suggesting that the data for all series in the model are approximately normally distributed.
Descriptive statistics.
Statistical variables
CPU
AGRI
ANFPPI
EPU
Observations
156
156
156
156
Mean
2.479275
0.194109
101.1045
4.944127
Median
2.395530
0.193500
100.6000
4.933646
Maximum
4.121537
0.799000
108.7000
5.457541
Minimum
1.117342
−0.431000
95.00000
4.530339
Standard Deviation
0.588054
0.181929
3.033923
0.191857
Skewness
0.399716
−0.072396
0.332294
0.360707
Kurtosis
2.583365
3.765830
2.764714
2.853744
Jarque–Bera
5.282399*
3.948491
3.230737
3.521890
Probability
0.071276
0.138866
0.198817
0.171882
*, **, and *** denote significance levels of10, 5, and 1%, respectively.
Data test
This study uses the Dickey–Fuller test, Phillips–Perron test, and Kwiatkowski–Phillips–Schmidt–Shin (KPSS) test to evaluate the stationarity of CPU and AGRI following first-order differencing of both series. The results, presented in Table 3, show that both series are stationary. To investigate the Granger causality between the two variables, we use the VAR model on the full sample. We set the optimal lag order as 2 based on the SIC, with 1,000 bootstrap replications. The Granger causality results for the full sample between CPU and AGRI are displayed in Table 3. Based on the bootstrap p-values, we find that CPU is not a Granger cause of AGRI, but AGRI is a Granger cause of CPU. Previous studies often assume that time series data are free from structural changes and that a stable Granger causality relationship exists throughout the entire sample period (Balcilar et al., 2010). However, when structural changes are present in the data, the parameters of the VAR model may change over time, leading to an unstable Granger causality relationship between the variables under study. Therefore, full-sample tests that assume fixed parameters and constant Granger causality relationships across the sample period may yield inaccurate conclusions in such cases.
The results of data test.
Method
Unit root tests
CPU
AGRI
Statistics
1%Level
Statistics
1%Level
ADF
−3.813***
−3.4765
−12.714***
−3.4731
PP
−63.818***
−3.4731
−13.621***
−3.4731
KPSS
0.138
0.7390
0.177
0.7390
Method
Parameter stability tests
CPU
AGRI
VAR (s) process
Statistics
p-value
Statistics
p-value
Statistics
p-value
Sup-F
58.413***
0.001
72.819***
0.000
34.356**
0.015
Ave-F
12.705***
0.000
18.012***
0.0000
18.566**
0.037
Exp-F
25.812***
0.005
32.748***
0.0000
13.292**
0.017
Lc
4.001***
0.005
Bootstrap full-sample method
H0: CPU is not the Granger cause of AGRI
H0: AGRI is not the Granger cause of CPU
Statistic
p-value
Statistic
p-value
0.765
0.675
5.2947*
0.074
Unrestricted cointegration rank test
Hypothesized
Statistic
5% Critical value
p-value
None (Trace test)
13.395
15.495
0.1011
None (Max-Eigenvalue test)
9.876
14.265
0.2202
*, **, and *** denote significance levels of10, 5, and 1%. The null hypothesis for KPSS is that the time series is stationary.
Within this research, we tackle this problem by assessing parameter stability and detecting structural changes. To check the temporal stability of parameters in the VAR model for CPU and AGRI, we employ the Sup-F, Mean-F, and Exp-F tests put forward by Andrews and Ploberger (1994). Moreover, we use the Lc test proposed by Nyblom and Hansen to evaluate the stability of all parameters in the VAR system (Hansen, 1992). Findings from these tests are presented in Table 3. The results reveal significant parameter instability in both the CPU and AGRI series. As the VAR(s) process model indicates that some of its parameters are unstable, a standard analysis could yield misleading conclusions. This necessitates the use of a model that accommodates time-varying parameters.
Following the stationarity tests, we further investigated the potential long-run equilibrium between the variables. We conducted the Johansen cointegration test to determine the appropriateness of a Vector Error Correction Model (VECM). The results indicate that we cannot reject the null hypothesis of no cointegration at the 5% significance level, with the Trace Test p-value at 0.1011 and the Max-Eigenvalue Test p-value at 0.2202. The absence of a cointegration relationship, likely disrupted by the frequent structural breaks identified above, confirms that a VECM is not applicable in this context. Consequently, this study proceeds with the VAR framework using first-differenced stationary series to accurately capture the short-term time-varying interactions.
Quantitative analyses and discussions
To tackle potential structural shifts, we use rolling-window estimation to analyze the Granger causality between CPU and AGRI. This method provides a more detailed approach than the full-sample causality test, as it captures the dynamic character of the relationship by considering time variations across multiple subsamples. Specifically, we apply the RB bootstrap-based modified-LR causality test in each rolling window to explore the Granger causality between CPU and AGRI. To capture the rapid structural shifts in climate policy while ensuring statistical validity, we set the fixed rolling-window width to 24 months. Although this window size is relatively compact, the RB bootstrap method employed in this study is specifically designed to perform robustly in small-sample contexts, correcting for the size and power distortions that typically affect standard asymptotic tests(Shukur and Mantalos, 2004). This is also a common setting in existing studies(Li et al., 2016; Su et al., 2021). The results from the rolling-window estimates for each subsample are presented in Figures 3–6.
Examining the null hypothesis that CPU is not a Granger cause of AGRI. The research counts p-values by using 1,000 bootstrap repetitions. The solid line means the bootstrap p-values, and the dashed line indicates that the p-value is 0.1.
Line graph showing fluctuations from 2014 to 2024. Values range from 0.0 to 1.0, with notable peaks in 2015, 2017, and 2020, and dips in 2018 and 2020-2021.
The coefficients of the influence from CPU to AGRI. The shadow represents the interval where CPU has significant Granger causality to AGRI.
Graph showing the sum of coefficients with upper and lower bounds from 2014 to 2024. The sum line fluctuates around zero, with notable peaks and troughs. Shaded areas indicate specific time periods.
Examining the null hypothesis that AGRI is not a Granger cause of CPU. The research counts p-values by using 1,000 bootstrap repetitions. The solid line means the bootstrap p-values, and the dashed line indicates that the p-value is 0.1.
Line graph showing data from 2014 to 2024. The values start high in 2014, fluctuate, and peak multiple times, especially in 2019 and 2021. The overall trend declines sharply after 2019 and remains low into 2024, with occasional minor peaks. A dashed horizontal line is present around 0.2.
The coefficients of the influence from AGRI to CPU. The shadow represents the interval where AGRI has significant Granger causality to CPU.
Line graph displaying the sum of coefficients from 2014 to 2024, with shaded areas indicating certain periods. The solid line shows the sum, while dashed lines represent the lower and upper bounds. The sum fluctuates around the zero line with noticeable peaks and troughs.Rolling-window estimation: the influence of CPU to AGRI
From Figure 3, it can be observed that during the periods January to November 2014, April to May 2015, April to December 2018, and October 2019, the null hypothesis that CPU is not a Granger cause of AGRI is rejected at the 10 percent significance level. From Figure 4, it is further evident that in January to November 2014 and October 2019, CPU exerted a significant negative effect on AGRI, whereas in April to May 2015 and April to December 2018, CPU had a significant positive effect on AGRI.
During the prolonged period from January to November 2014, CPU exerted a continuous significant negative impact on agricultural investment. This sustained dampening effect was likely driven by the intensifying pressure of environmental regulations coupled with ambiguity regarding implementation standards. The onset of this trend in the first quarter was likely driven by a convergence of domestic policy tightening and rising international pressure. In January, China’s Central No. 1 Document initiated this shift by demanding eco-friendly agriculture under the strictest environmental regulations. This policy hardening signaled a future of sharply higher compliance costs and regulatory risks. Faced with an unpredictable implementation timeline, scope, and severity, rational investors chose to wait and see, pausing or deferring new investments for clarity (Wang et al., 2021). This cautious sentiment was reinforced in March by the IPCC’s Fifth Assessment Report (Working Group II), which underscored the grave agricultural risks from climate change (Intergovernmental Panel on Climate Change (IPCC), 2014) and intensified global expectations for stricter climate action (Myers et al., 2014). Consequently, the ambiguity regarding the future policy trajectory stalled investment activity at the beginning of the year.
In April and May 2015, CPU had a positive impact on AGRI, indicating that under these specific conditions, CPU actually promoted agricultural investment. We attribute this unusual effect to the concurrent release of the National Plan for Sustainable Agricultural Development (2015–2030), which reframed market perceptions of uncertainty. Issued by multiple government departments, the plan not only designated agricultural adaptation to climate change as a priority but also launched specific major projects and pledged central government funds to guide private capital investment (Chen and Gong, 2021; Chen and Li, 2024). This effectively transformed what was seen as policy ambiguity into a clear structural opportunity aligned with a national strategic vision. In a state-influenced economy like China’s, such top-level design significantly boosts investor confidence in long-term sectors like agricultural modernization (Drobetz et al., 2018). This finding suggests that when uncertainty is coupled with a credible industrial strategy, it can paradoxically act as a catalyst, accelerating capital allocation toward strategic goals (Erdoğan et al., 2025).
This inhibitory effect persisted through the latter half of the year, albeit with a diminishing magnitude. We posit that the sustained negative effect was primarily driven by a combination of external supply chain shocks and domestic implementation uncertainty. On one front, the severe 2014 drought in Brazil laid bare the vulnerabilities of China’s agricultural supply chain (de Melo et al., 2022). This external shock fueled market speculation about potential policy shifts toward grain self-sufficiency, prompting a continued cautious posture among investors (Sautner et al., 2023). On another front, the newly promulgated National Plan for Coping with Climate Change (2014–2020), while outlining macro-level commitments, failed to provide concurrent details on subsidies or an implementation roadmap. This approach, characterized by strategic clarity but tactical ambiguity, generated significant implementation-level uncertainty, thereby continuing to impede agricultural projects with long-term capital expenditure requirements (Zhang et al., 2024; Soko et al., 2023). Importantly, the gradual attenuation of this negative impact toward the end of the period was likely counterbalanced by positive governmental signals. China’s announcement at the UN Climate Summit to fund the China South–South Climate Cooperation Fund with 20 billion RMB sent a strong macroeconomic signal, which anchored long-term expectations and mitigated the adverse shock of uncertainty at the margin (Özkaragöz Doğan et al., 2021).
Crucially, a new significant positive effect emerged from April to December 2018. This finding is particularly insightful after controlling for general economic policy uncertainty. While 2018 is often associated with the Sino-US trade war, our specific climate-policy nexus reveals a supportive internal drive. The year 2018 was the inaugural year of China’s Rural Revitalization Strategy and a pivotal year for the Three-Year Action Plan for Winning the BlueSky War. Rigorous environmental enforcement, rather than suppressing investment, likely forced a wave of green upgrading investment to meet new compliance standards. Furthermore, the Strategic Plan for Rural Revitalization (2018–2022) released in September provided strong political assurance and fiscal support for green agriculture. This suggests that in the face of external trade uncertainties, strong domestic climate and agricultural policies effectively anchored investor confidence and stimulated the upgrading of the agricultural industry.
In October 2019 CPU’s negative impact on AGRI stemmed from domestic climate shocks, trade disruptions, and international policy stalemates. First, a severe drought in the Yangtze River basin threatened output, sparking anxiety over future water management policies (Feng et al., 2024). Second, the US-China Phase One trade deal mandated heavy agricultural imports, increasing competitive pressure and suppressing local investment appetite (He et al., 2019; Muhammad et al., 2022). Third, the COP25 stalemate regarding carbon markets obscured long-term pricing pathways, discouraging mitigation-related investments (Newell and Taylor, 2020; Wongpiyabovorn et al., 2023). However, the Central Economic Work Conference in December mitigated these uncertainties by prioritizing agricultural supply and structural reform. This declaration acted as a signal that the government would prioritize domestic agriculture amidst a worsening external climate, offering investors the expectation of a policy floor and easing concerns over a sudden policy tightening.
Rolling-window estimation: the influence of AGRI to CPU
The examination of the effect of AGRI on CPU is presented in Figures 5, 6. It can be observed that the null hypothesis that AGRI is not a Granger cause of CPU is rejected at the 10 percent significance level during the periods of March 2016 to January 2017, February 2020 to January 2021, March to December 2021, intermittent periods in 2023, and January to December 2024. Specifically, from March 2016 to January 2017, AGRI exerted a significant positive effect on CPU. During the distinct periods of February 2020 to January 2021 and March to December 2021, AGRI consistently exerted a significant negative effect on CPU. Entering the post-pandemic phase in 2023, the relationship exhibited structural volatility, alternating between positive (e.g., February–April, July–October) and negative (e.g., May–June, November–December) effects. Finally, a consistent positive effect was established throughout January to December 2024.
From March 2016 to January 2017, the test results show an anomalous finding: agricultural investment (AGRI) had a significant positive impact on climate policy uncertainty (CPU). We believe the root of this phenomenon lies in the wave of green agricultural transformation driven by the 13th Five-Year Plan (2016–2020) and the potential uncertainty it triggered. The year 2016 was the inaugural year of China’s 13th Five-Year Plan, in which green development was established as the core theme. Under the guidance of this national top-level design, institutions such as the State Council, the National Development and Reform Commission (NDRC), and the Ministry of Agriculture intensively issued a series of supporting plans, all pointing to a clear objective: advancing agricultural modernization toward resource conservation and environmental friendliness (Liu et al., 2022). Driven by strong policy signals and predictable financial support, a large amount of capital rapidly flowed into new fields that are highly dependent on policy subsidies and industry standards, such as agricultural environmental protection, green technology, and organic farming (Lu et al., 2024). However, as the scale of investment expanded rapidly in a short period, the market began to worry that subsidies might be phased out earlier than expected or that regulatory standards could be suddenly tightened. This expectation of policy adjustments, induced by the investment overheating, was a key factor in keeping the CPU index high (Zhang and Zhao, 2024). This indicates that in the initial stages of a policy-driven industrial transformation, rapid investment growth can temporarily exacerbate policy uncertainty by outpacing the development of institutional frameworks.
Across the periods of February 2020–January 2021, March–December 2021, AGRI consistently exerted a negative influence on CPU. In 2020, the onset of the COVID-19 pandemic elevated food security to an unparalleled strategic priority, compelling the government to minimize policy adjustments that could disrupt agricultural production, thus reducing climate policy uncertainty (Chan et al., 2022). This was reinforced in September when China formally committed to its dual carbon goals of peaking emissions by 2030 and achieving carbon neutrality by 2060. This ambitious top-level design delineated a clear, long-term green transition pathway for all sectors, agriculture included. The goals necessitated a high degree of policy continuity and stability, which systemically lowered policy uncertainty (Cowan et al., 2013; Deng et al., 2024). The strategic objectives of 2020 were translated into concrete action plans in 2021 with the release of the 14th Five-Year Plan and the National Plan for Green Agricultural Development. During this phase, substantial agricultural investment, especially in R&D for green and climate-friendly technologies, produced a critical “policy path lock-in” effect. These long-term capital commitments signaled the industry’s buy-in to the established policy direction, which in turn bolstered the credibility of the government’s emissions reduction commitments and diminished the probability of policy vacillation (Held et al., 2009; Zhang et al., 2024). Recent extreme weather events served as critical tests of China’s agricultural adaptation. The severity of the 2021 Henan torrential rain have caused an important threat to Henan, China’s main grain-producing region. But long-term investments in proactive measures like water conservancy infrastructure successfully mitigated crop damage (Liu et al., 2023). This real-world validation of existing policies reinforced decision-makers’ commitment to the current framework.
The period from 2023 to 2024 reveals a shift toward complexity and eventual reversal. The changes mainly occurred in 2023. From May to June and from November to December of this year, AGRI had a short-term negative impact on the CPU. Throughout 2023, the impact of AGRI on CPU exhibited marked volatility, alternating between positive and negative effects. As the pandemic waned, economic recovery took priority, though the administration sustained its focus on agricultural development and the green transition (Mi and Wang, 2024). In early 2023 (February–April), a positive shock stemmed from the No. 1 Central Document and its “New Round of 100 Billion Jin Grain Capacity Enhancement Action.” While this target drove investment in capacity expansion, it raised market concerns about conflicts with environmental regulations. Fears that enforcement might return to curb extensive growth caused this investment surge to paradoxically heighten policy uncertainty. The negative shock in May and June 2023 correlates with the harvest-disrupting rain events in the Huang-Huai-Hai region (Huang et al., 2025). This threat to food security prompted urgent government directives for emergency harvesting. Investment aligned with this clear political priority, eliminating ambiguity and reducing policy uncertainty. The positive effect resurfaced in Q3 (July–October), driven by speculative investment anticipating the China Certified Emission Reduction (CCER) restart (Chi et al., 2025). Capital flowed into agricultural carbon sinks before methodologies were standardized. This mismatch between preemptive investment and pending regulatory details exacerbated policy unpredictability. Finally, the negative effect returned in November–December due to the 1 trillion RMB special sovereign bond issuance. By allocating funds to high-standard farmland and disaster recovery, this fiscal injection signaled stable policy support, effectively anchoring investor expectations.
The turning point was arguably January 2024, when the UN’s World Economic Situation and Prospects 2024 report highlighted immense challenges to global growth from escalating geopolitical conflicts, sluggish trade, and climate disasters. Amidst a dual climate of recessionary fears and risk aversion to extreme weather, the ability of agricultural investment to reduce policy uncertainty was severely undermined. To avert a deep economic downturn, environmental regulations were relaxed, and the role of agricultural investment was fundamentally altered. In times of ample policy resources, investment is encouraged (Akan, 2025). Yet, in an economic downturn where policy space is limited, new large-scale investments begin to compete for the government’s finite fiscal, land, and environmental capacity. This forces policymakers into a difficult trade-off between growth and green. This underlying tension explains why China, for the foreseeable future, continues to emphasize the need to manage the relationship between development and security effectively.
Robustness test
Considering that the choice of window period may have an impact on the results of this article. In this article, the window period is adjusted to 18 months and 30 months, respectively, to verify the robustness of the results. The results show that after the adjustment window period, the main results of this study did not undergo significant structural changes.
As shown in the figures(See attached materials), the dynamic causal relationships between CPU and AGRI exhibit high consistency across different window widths, confirming the robustness of our main findings. The direction of causality remains consistent in key periods. Regardless of the window size, CPU exerts a significant negative impact on AGRI during 2014, and AGRI exerts a significant negative feedback on CPU during the 2020–2021 period. Furthermore, the distinctive structural break in 2024, where the impact of AGRI on CPU shifts to positive, is captured by all three window specifications. This indicates that the identified economic mechanisms are not artifacts of a specific window selection.
The temporal distribution of significant intervals follows a logical pattern. The results from the 18-month window appear more fragmented (e.g., the AGRI-to-CPU effect in 2023), reflecting the higher sensitivity of smaller windows to short-term fluctuations. Conversely, the 30-month window yields smoother and more continuous intervals (e.g., the consolidated negative period in 2021), as larger windows tend to filter out high-frequency noise. The benchmark 24-month window effectively balances the trade-off between sensitivity to structural changes and estimation stability. The core conclusion remains robust to alternative specifications.
Conclusions and policy recommendations
This paper systematically examines the bidirectional dynamic relationship between climate policy uncertainty (CPU) and agricultural investment (AGRI) in China. Through rolling-window causality estimation, it confirms that a bidirectional relationship exists between the two, which evolves with the macroeconomic and policy environment. The study finds that the impact of CPU on AGRI exhibits both positive and negative effects over time, but is predominantly characterized by a significant negative impact. A positive effect occurs when the intensive rollout of climate policies with clear implementation paths acts as a catalyst for agricultural investment. The negative impact manifests when insufficient policy enforcement and the expected damages from extreme climate events create negative shocks to agricultural investment. The study also finds that the impact of AGRI on CPU shows both positive and negative effects over time, but is mainly characterized by a significant negative impact. A positive effect arises when an expansion in agricultural investment, combined with shifts in international climate policy direction, leads to heightened expectations of a partial policy withdrawal, thereby increasing policy uncertainty. The negative impact is primarily due to the certainty and technological spillover effects from green investment in the agricultural sector, which enhance policy credibility and thus reduce climate policy uncertainty.
Based on the research conclusions above, we propose the following policy recommendations.
First, adopting a package of start-up strategies for climate policies to promote investment.our results show that important strategies with clear actionable goals and implementation rules can effectively stimulate investment. Therefore, future policy releases should be launched simultaneously with detailed implementation guidelines and plans to transform strategic intentions into investment opportunities.
Second, establish a green channel for investment projects across administrative departments to stabilize investment expectations and resist external shocks. We found that the ambiguity of policy implementation and the impact of foreign trade significantly suppressed investment. In order to alleviate this situation, a stable coordination mechanism between multiple departments should be established. We create green channels for pre-approved climate-adaptive agricultural project categories. By standardizing approval standards, these key projects are exempted from frequent regulatory fluctuations, and the government’s credibility is used to protect long-term capital from short-term policy oscillations.
Third, anchor financial incentives to specific Dual Carbon instruments. While financial support is crucial, it must be precise. Drawing on the finding that investment stability reduces policy uncertainty, policy should encourage long-term capital commitment. We recommend leveraging the emerging green finance system under China’s Dual Carbon goals. Specifically, policymakers should develop performance-based subsidies linked to the Green Agricultural Development. Furthermore, financial institutions should be encouraged to innovate products such as carbon-sink pledged loans or climate resilience insurance, which directly address the risk premium channel identified in our theoretical framework.
Research gaps and prospects
First, this research relies on comprehensive data at the national level, which may mask the huge spatial gaps inherent in China’s vast geography. As mentioned in the literature, agricultural vulnerability and policy implementation vary greatly from province to province. In future research, we will try to use hierarchical climate policy uncertainty data and classified investment data to explore how regional differences in resource endowments and local governance efficiency can alleviate the impact of climate change.
Second, we have found the delay and wait-and-see effects of policies on investment at the macro level, but we cannot directly analyze the activities at the corporate level. In future research, we will consider using micro-enterprise-level data to test and analyze specific impact paths such as risk aversion and real option valuation in this research.
Third, this research is based on China, a country characterized by a strong state-led governance model. The two-way influence mechanism discovered in this paper may have different functions in economies with different political structures. In the future, we hope to expand our research to other large economies that depend on agriculture (e.g., Brazil, India, and the European Union), and use comparative research to determine whether the findings in this article are universal or based on a special political system.
Furthermore, due to the lack of consistent monthly statistical data distinguishing between investment subjects, this study utilizes aggregate fixed asset investment in agriculture. We acknowledge that government and private investments may have distinct interaction mechanisms with climate policy uncertainty. Theoretically, a transmission chain may exist where “Government investment → CPU → Private investment.” Government investment often serves as a policy signal that may temporarily fluctuate policy uncertainty, subsequently influencing the risk aversion and decision-making of private capital. Future research should attempt to verify this specific transmission mechanism using micro-level firm data or annual data with broader coverage.
Data availability statement
The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author/s.
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that Generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsufs.2025.1735435/full#supplementary-material
ReferencesAbelA. B. (1983). Optimal investment under uncertainty. Am. Econ. Rev.73, 228–233.AkanT. (2025). Forecasting the future of carbon emissions by business confidence. Appl. Energy382:125146. doi: 10.1016/j.apenergy.2024.125146AndrewsD. W. (2003). Tests for parameter instability and structural change with unknown change point: a corrigendum. Econometrica71, 395–397. doi: 10.1111/1468-0262.00405, 41447298AndrewsD. W.PlobergerW. (1994). Optimal tests when a nuisance parameter is present only under the alternative. Econometrica62, 1383–1414. doi: 10.2307/2951753AutioA.JohanssonT.MotarokiL.MinoiaP.PellikkaP. (2021). Constraints for adopting climate-smart agricultural practices among smallholder farmers in Southeast Kenya. Agric. Syst.194:103284. doi: 10.1016/j.agsy.2021.103284BaiJ.LuJ.LiS. (2019). Fiscal pressure, tax competition and environmental pollution. Environ. Resour. Econ.73, 431–447. doi: 10.1007/s10640-018-0269-1BakerS. R.BloomN.DavisS. J. (2016). Measuring economic policy uncertainty. Q. J. Econ.131, 1593–1636. doi: 10.1093/qje/qjw024BalcilarM.OzdemirZ. A.ArslanturkY. (2010). Economic growth and energy consumption causal nexus viewed through a bootstrap rolling window. Energy Econ.32, 1398–1410. doi: 10.1016/j.eneco.2010.05.015BinswangerH. P. (1980). Attitudes toward risk: experimental measurement in rural India. Am. J. Agric. Econ.62, 395–407. doi: 10.2307/1240194BloomN.BondS.Van ReenenJ. (2007). Uncertainty and investment dynamics. Rev. Econ. Stud.74, 391–415. doi: 10.1111/j.1467-937X.2007.00426.xCalvinK. V.BeachR.GurgelA.LabrietM.RodriguezA. M. L. (2016). Agriculture, forestry, and other land-use emissions in Latin America. Energy Econ.56, 615–624. doi: 10.1016/j.eneco.2015.03.020CassarA.HealyA.Von KesslerC. (2017). Trust, risk, and time preferences after a natural disaster: experimental evidence from Thailand. World Dev.94, 90–105. doi: 10.1016/j.worlddev.2016.12.042ChanF. K. S.ZhuY. G.WangJ.LiG.WangJ.ChenJ. (2022). China: protect home-grown solutions to food security. Nature610:448. doi: 10.1038/d41586-022-03289-0, 36258037ChenS.GongB. (2021). Response and adaptation of agriculture to climate change: evidence from China. J. Dev. Econ.148:102557. doi: 10.1016/j.jdeveco.2020.102557ChenP.LiJ. (2024). Sustainable agricultural management: how to achieve carbon neutrality in agriculture–evidence from China agricultural sustainable development plan. Sustain. Dev.32, 2846–2857. doi: 10.1002/sd.2821ChiY. Y.ZhangY. Y.LiuB. L.ZhangM. W.LiJ. L.LiG. Z. (2025). Assessing the combined effect of China's restarting of the CCER market: coupling of CGE and GAINS models. Adv. Clim. Chang. Res.16, 433–44. doi: 10.1016/j.accre.2025.01.007ChmutinaK.ZhuJ.RiffatS. (2012). An analysis of climate change policy-making and implementation in China. Int. J. Clim. Change Strateg. Manag.4, 138–151. doi: 10.1108/17568691211223123CowanL.KaineG.WrightV. (2013). The role of strategic and tactical flexibility in managing input variability on farms. Syst. Res. Behav. Sci.30, 470–494. doi: 10.1002/sres.2137DaiJ. (2025). Impact of climate policy uncertainty on agriculture development: multidimensional analysis from land use, food structure, and carbon emissions. Land Degrad. Dev.36, 4545–4561. doi: 10.1002/ldr.5652de MeloD. B.DolbethM.PaivaF. F.MolozziJ. (2022). Extreme drought scenario shapes different patterns of Chironomid coexistence in reservoirs in a semi-arid region. Sci. Total Environ.821:153053. doi: 10.1016/j.scitotenv.2022.153053, 35038537DengY.LiuJ. Y.XieW.LiuX.LvJ.ZhangR.. (2024). Impact of carbon pricing on mitigation potential in Chinese agriculture: a model-based multi-scenario analysis at provincial scale. Environ. Impact Assess. Rev.105:107409. doi: 10.1016/j.eiar.2023.107409DiksC.PanchenkoV. (2006). A new statistic and practical guidelines for nonparametric granger causality testing. J. Econ. Dyn. Control.30, 1647–1669. doi: 10.1016/j.jedc.2005.08.008.DrobetzW.El GhoulS.GuedhamiO.JanzenM. (2018). Policy uncertainty, investment, and the cost of capital. J. Financ. Stab.39, 28–45. doi: 10.1016/j.jfs.2018.08.005DuZ.ChenC. (2025). AI vs. ESG? Uncovering a bidirectional struggle in China’s sustainable finance. Sustainability17:4238. doi: 10.3390/su17094238ErdoğanS.KantarcıT.YıldırımD. Ç. (2025). Does economic policy uncertainty affect venture capital investments for OECD countries?Ventur. Cap.27, 203–223. doi: 10.1080/13691066.2023.2270162FajgelbaumP. D.KhandelwalA. K. (2022). The economic impacts of the US–China trade war. Annu. Rev. Econ.14, 205–228. doi: 10.1016/j.eap.2025.07.038FengS.GuX.GuanY.WangQ. J.ZhengY.WangL.. (2024). Anthropogenic exacerbation in dry-hot probability and consequential record-shattering droughts in the middle and lower reaches of the Yangtze River. J. Geophys. Res. Atmos.129:e2024JD041603. doi: 10.1029/2024JD041603GodfrayH. C. J.BeddingtonJ. R.CruteI. R.HaddadL.LawrenceD.MuirJ. F.. (2010). Food security: the challenge of feeding 9 billion people. Science327, 812–818. doi: 10.1126/science.1185383, 20110467GuanJ.XuH.HuoD.HuaY.WangY. (2021). Economic policy uncertainty and corporate innovation: evidence from China. Pac. Basin Financ. J.67:101542. doi: 10.1016/j.pacfin.2021.101542GulenH.IonM. (2016). Policy uncertainty and corporate investment. Rev. Financ. Stud.29, 523–564. doi: 10.1093/rfs/hhv050HansenB. E. (1992). Testing for parameter instability in linear models. J. Policy Model14, 517–533. doi: 10.1016/0161-8938(92)90019-9HeR.ZhuD.ChenX.CaoY.ChenY.WangX. (2019). How the trade barrier changes environmental costs of agricultural production: an implication derived from China's demand for soybean caused by the US-China trade war. J. Clean. Prod.227, 578–588. doi: 10.1016/j.jclepro.2019.04.192HeldH.KrieglerE.LessmannK.EdenhoferO. (2009). Efficient climate policies under technology and climate uncertainty. Energy Econ.31, S50–S61. doi: 10.1016/j.eneco.2008.12.012HuangH.FengA.ChaoQ.GuoY.LiY.ShiD.. (2025). Identification and formation mechanisms of harvest-disrupting rain events in winter wheat based on an intensity–size–duration model. J. Hydrol. Reg. Stud.61:102615. doi: 10.1016/j.ejrh.2025.102615IglesiasA.QuirogaS.DizA. (2011). Looking into the future of agriculture in a changing climate. Eur. Rev. Agric. Econ.38, 427–447. doi: 10.1093/erae/jbr037Intergovernmental Panel on Climate Change (IPCC). (2014). Climate Change 2014 – Impacts, Adaptation and Vulnerability: Part A: Global and Sectoral Aspects: Working Group II Contribution to the IPCC Fifth Assessment Report. Cambridge: Cambridge University Press. doi: 10.1017/CBO9781107415379JayneT. S.RashidS. (2013). Input subsidy programs in sub-Saharan Africa: a synthesis of recent evidence. Agric. Econ.44, 547–562. doi: 10.1111/agec.12073JuH.LiuQ.LiY.LongX.LiuZ.LinE. (2020). Multi-stakeholder efforts to adapt to climate change in China’s agricultural sector. Sustainability12:8076. doi: 10.3390/su12198076KaragölV.ŞahinA. (2025). The impact of climate policy uncertainty on agricultural investments. Turk. J. Agric. Econ., 31, 91–103. doi: 10.24181/tarekoder.1628954KotuB. H.MandaJ.MutungiC.FischerG.GasparA. (2023). Farmers' willingness to invest in mechanized maize shelling and potential financial benefits: evidence from Tanzania. Agribusiness39, 854–874. doi: 10.1002/agr.21801KulatilakaN.PerottiE. C. (1998). Strategic growth options. Manag. Sci.44, 1021–1031. 1 doi: 0.1287/mnsc.44.8.1021. doi: 10.1287/mnsc.44.8.1021, 19642375LaiY.ZhouJ.YaoL. (2025). Signal of Sustainable Agriculture!Nonlinear Impact of Climate Policy Uncertainty: New Evidence From China’s Agricultural Sector. Sustain. Dev. doi: 10.1002/sd.70167LeeK.ChoJ. (2023). Measuring Chinese climate uncertainty. Int. Rev. Econ. Finance88, 891–901. doi: 10.1016/j.iref.2023.07.004LiX. L.BalcilarM.GuptaR.ChangT. (2016). The causal relationship between economic policy uncertainty and stock returns in China and India: evidence from a bootstrap rolling window approach. Emerg. Mark. Finance Trade52, 674–689. doi: 10.1080/1540496X.2014.998564LiY.FanZ.JiangG.QuanZ. (2021). Addressing the differences in farmers’ willingness and behavior regarding developing green agriculture—a case study in Xichuan County, China. Land10:316. doi: 10.3390/land10030316LiX.PattbergP.WiderbergO. (2024). China’s climate governance from 2009 to 2019: motivations, instruments, actors, and geopolitics. Clim. Pol.24, 378–395. doi: 10.1080/14693062.2023.2260352LiuY.LiY.WangG.GaoG.ChenY. (2023). Quantifying multi-regional indirect economic losses: an assessment based on the 2021 rainstorm events in China. Front. Earth Sci.10:1057430. doi: 10.3389/feart.2022.1057430LiuK. D.ShiD.XiangW.ZhangW. (2022). How has the efficiency of China's green development evolved? An improved non-radial directional distance function measurement. Sci. Total Environ.815:152337. doi: 10.1016/j.scitotenv.2021.152337, 34958840LuF.MengJ.ChengB. (2024). How does improving agricultural mechanization affect the green development of agriculture? Evidence from China. J. Clean. Prod.472:143298. doi: 10.1016/j.jclepro.2024.143298MaY. R.LiuZ.MaD.ZhaiP.GuoK.ZhangD.. (2023). A news-based climate policy uncertainty index for China. Scientific Data10:881. doi: 10.1038/s41597-023-02817-5, 38065994McDonaldR.SiegelD. (1986). The value of waiting to invest. Q. J. Econ.101, 707–727. doi: 10.2307/1884175MiJ. J.WangY. (2024). How do industries recover through Covid-19 pandemic? An empirical investigation on Chinese online recruitment data. Appl. Econ. Lett.31, 1083–1089. doi: 10.1080/13504851.2023.2176430MittenzweiK.PerssonT.HöglindM.KværnøS. (2017). Combined effects of climate change and policy uncertainty on the agricultural sector in Norway. Agric. Syst.153, 118–126. doi: 10.1016/j.agsy.2017.01.016MossR. H.EdmondsJ. A.HibbardK. A.ManningM. R.RoseS. K.Van VuurenD. P.. (2010). The next generation of scenarios for climate change research and assessment. Nature463, 747–756. doi: 10.1038/nature08823, 20148028MuhammadA.SmithS. A.GrantJ. H. (2022). Can China meet its purchase obligations under the phase one trade agreement?Appl. Econ. Perspect. Policy44, 1393–1408. doi: 10.1002/aepp.13180MukashovA.ThomasT.ThurlowJ. (2024). Revisiting development strategy under climate uncertainty: case study of Malawi. Clim. Chang.177:91. doi: 10.1007/s10584-024-03733-2, 38827780MyersS. S.ZanobettiA.KloogI.HuybersP.LeakeyA. D.BloomA. J.. (2014). Increasing CO2 threatens human nutrition. Nature510, 139–142. doi: 10.1038/s41586-019-1602-8, 24805231NairS. (2020). Designing policy pilots under climate uncertainty: a conceptual framework for comparative analysis. J Comparative Policy Analysis22, 344–359. doi: 10.1080/13876988.2019.1695973, 41307611NewellP.TaylorO. (2020). Fiddling while the planet burns? COP25 in perspective. Globalizations17, 580–592. doi: 10.1080/14747731.2020.1726127NyblomJ. (1989). Testing for the constancy of parameters over time. J. Am. Stat. Assoc.84, 223–230. doi: 10.1080/01621459.1989.10478759OkolieC. C.OgundejiA. A. (2022). Effect of COVID-19 on agricultural production and food security: a scientometric analysis. Humanit. Soc. Sci. Commun.9. doi: 10.1057/s41599-022-01080-0Özkaragöz DoğanE.UygunZ.Akçomakİ. S. (2021). Can science diplomacy address the global climate change challenge?Environ. Policy Gov.31, 31–45. doi: 10.1002/eet.1911PalikheH.SchaurG.SimsC. (2024). Environmental policy uncertainty. J. Assoc. Environ. Resour. Econom.11, 1135–1163. doi: 10.1086/729529PanD.YangJ.ZhouG.KongF. (2020). The influence of COVID-19 on agricultural economy and emergency mitigation measures in China: a text mining analysis. PLoS One15:e0241167. doi: 10.1371/journal.pone.0241167, 33095814PesaranM. H.TimmermannA. (2005). Small sample properties of forecasts from autoregressive models under structural breaks. J. Econom.129, 183–217. doi: 10.1016/j.jeconom.2004.09.007RichardsM. B.WollenbergE.van VuurenD. (2018). National contributions to climate change mitigation from agriculture: allocating a global target. Clim. Pol.18, 1271–1285. doi: 10.1080/14693062.2018.1430018RogersS. (2016). Adaptation science and policy in China's agricultural sector. Wiley Interdiscip. Rev. Clim. Chang.7, 693–706. doi: 10.1002/wcc.414, 41452081SautnerZ.Van LentL.VilkovG.ZhangR. (2023). Pricing climate change exposure. Manag. Sci.69, 7540–7561. doi: 10.1287/mnsc.2023.4686ShiQ.ShiC.GuoF. (2020). Political blue sky: evidence from the local annual “two sessions” in China. Resour. Energy Econ.61:101165. doi: 10.1016/j.reseneeco.2020.101165ShukurG.MantalosP. (2000). A simple investigation of the granger-causality test in integrated-cointegrated VAR systems. J. Appl. Stat.27, 1021–1031. doi: 10.1080/02664760050173346ShukurG.MantalosP. (2004). Size and power of the RESET test as applied to systems of equations: a bootstrap approach. J. Mod. Appl. Stat. Methods3:10. doi: 10.22237/jmasm/1099267800SokoN. N.KaitibieS.RatnaN. N. (2023). Does institutional quality affect the impact of public agricultural spending on food security in sub-Saharan Africa and Asia?Glob. Food Secur.36:100668. doi: 10.1016/j.gfs.2022.100668SteinbuksJ.HertelT. W. (2013). Energy prices will play an important role in determining global land use in the twenty first century. Environ. Res. Lett.8:014014. doi: 10.1088/1748-9326/8/1/014014SuC. W.KhanK.UmarM.ZhangW. (2021). Does renewable energy redefine geopolitical risks?Energy Policy158:112566. doi: 10.1016/j.enpol.2021.112566SunB.BakerM. (2021). Towards an analytical governance framework within the policy dimension in China: the evolution of national climate policies since 1978. J. Environ. Plan. Manag.64, 202–223. doi: 10.1080/09640568.2020.1760800TianZ.LiangZ.SunL.ZhongH.QiuH.FischerG.. (2015). Agriculture under climate change in China: mitigate the risks by grasping the emerging opportunities. Hum. Ecol. Risk Assess. Int. J.21, 1259–1276. doi: 10.1080/10807039.2014.955392WangM.SongY.ZhangX. (2025). Climate risk and green total factor productivity in agriculture: the moderating role of climate policy uncertainty. Risk Anal.45, 981–995. doi: 10.1111/risa.17639, 39218805WangY.WangJ.WangX.LiQ. (2021). Does policy cognition affect livestock farmers' investment in manure recycling facilities? Evidence from China. Sci. Total Environ.795:148836. doi: 10.1016/j.scitotenv.2021.148836, 34237532WongpiyabovornO.PlastinaA.CrespiJ. M. (2023). Challenges to voluntary ag carbon markets. Appl. Econ. Perspect. Policy45, 1154–1167. doi: 10.1002/aepp.13254WuS. (2023). A systematic review of climate policies in China: evolution, effectiveness, and challenges. Environ. Impact Assess. Rev.99:107030. doi: 10.1016/j.eiar.2022.107030WuM.LiuY.XuZ.YanG.MaM.ZhouS.. (2021). Spatio-temporal dynamics of China’s ecological civilization progress after implementing national conservation strategy. J. Clean. Prod.285:124886. doi: 10.1016/j.jclepro.2020.124886XiangC. van GeveltT.2025China's global leadership aspirations and domestic support for climate policyEcol. Econ.227:108440 doi: 10.1016/j.ecolecon.2024.108440XuH.TangA.PittockJ. (2021). The dark side of ambition: side-effects of China’s climate policy. Environ. Res. Lett.16:111001. doi: 10.1088/1748-9326/ac30bfYangL.HuaH.ZhangJ. (2024). The impact of phasing out subsidy for financial performance of photovoltaic enterprises: evidence from “531 new policy” on China’s photovoltaic industry. Front. Energy Res.12:1486351. doi: 10.3389/fenrg.2024.1486351YuanX.SuC. W.UmarM.ShaoX.LobonŢO. R. (2022). The race to zero emissions: can renewable energy be the path to carbon neutrality?J. Environ. Manag.308:114648. doi: 10.1016/j.jenvman.2022.114648, 35149405ZhangX. (2025). Integrating policy design with agricultural emissions reduction in China: a multi-sector DSGE approach. Econ. Anal. Policy. 86, 2019–2048. doi: 10.1016/j.eap.2025.05.037ZhangZ.GaoQ.ShaoS.ZhangY.BaoY.ZhaoL. (2024). Carbon emission scenarios of China's construction industry using a system dynamics methodology–based on life cycle thinking. J. Clean. Prod.435:140457. doi: 10.1016/j.jclepro.2023.140457ZhangK. C.SafiA.KchouriB.BanerjeeA.WangL. (2024). The three greens: innovation, finance, and taxes—performance analysis and future implications. J. Innov. Knowl.9:100627. doi: 10.1016/j.jik.2024.100627ZhangY.ZhaoM. (2024). Environmental regulations or expected revenue: what plays a more important role in China’s green transition of agriculture?Agric. Econ.70, 425–435. doi: 10.17221/142/2024-AGRICECON
Edited by: Kai-Hua Wang, Qingdao University, China
Reviewed by: Tural Yusifzada, Innovation and Digital Development Agency, Azerbaijan
Qifeng Yang, Northeast Institute of Geography and Agroecology (CAS), China