Sowing Uncertainty: Assessing the Impact of Economic Policy Uncertainty on Agricultural Land Conversion in China

Abstract

This study examines the impact of economic policy uncertainty (EPU) on agricultural land conversion. Using a newspaper-based index of EPU and a comprehensive panel dataset covering 270 prefecture-level cities in China, we estimate a city fixed effects model to explore this relationship. Our results indicate that a one-standard-deviation increase in EPU leads to a 22.2% increase in the conversion of agricultural land to urban residential, commercial, and industrial uses. This finding suggests that the surge in EPU triggered by the global financial crisis accounts for approximately 45% of the increase in agricultural land conversion. The adverse effect on agricultural land preservation mainly stems from intensified fiscal pressures and heightened demands on local governments to meet economic growth targets. To address potential endogeneity concerns, we employ the one-period lagged U.S. EPU index and its temporal variations as an instrument for China’s EPU, leveraging cross-country spillover effects. Our instrumental variable estimates confirm the validity of the land conversion effect and its underlying mechanisms. Furthermore, we find that the effects of EPU are particularly pronounced in cities located in non-eastern China and those that depend heavily on fixed asset investment for local economic development. Finally, our analysis of potential policy interventions to mitigate EPU-induced agricultural land loss suggests that strengthening market-oriented reforms and reducing province-level quotas on agricultural land conversion can effectively offset the impact of rising EPU.

1. Introduction

How human societies use land is central to addressing the sustainability challenges outlined in the 2030 Agenda for Sustainable Development, the Paris Climate Agreement, and the Convention on Biological Diversity [1]. Agricultural land conversion is defined as the transformation of agricultural land into urban areas for purposes such as commercial and industrial uses, representing the most significant land use change, and it triggers a range of social, economic, and environmental issues in developing countries, such as land degradation [2], greenhouse gas emissions and climate disasters [3,4], and the deterioration of livelihood stability and household welfare for land losers in peri-urban areas [5,6]. An informed discussion about how to address these issues to accommodate sustainable development requires a good understanding of the drivers of agricultural land conversion. This study aims to identify economic policy uncertainty (EPU) as an important factor for agricultural land conversion to help policymakers understand the trade-off between economic policy interventions and agricultural land conservation.

Notable uncertainty arising from frequent macroeconomic policy interventions—such as changes in fiscal, monetary, or trade policies—has become a defining characteristic in the aftermath of the global financial crisis. Numerous studies utilized a weighted policy uncertainty index constructed by newspaper coverage frequency and found that EPU depresses firms’ investments [7,8,9], reduces corporate innovation activities [10,11], and impedes their mergers and acquisitions [12]. Azqueta-Gavaldón et al. [13] also constructed a newspaper-based policy uncertainty index in EU countries and revealed that EPU has strong negative effects on consumption. However, less attention is paid to the spillover effects of EPU on land use changes. Particularly, the question of whether and how EPU affects agricultural land conversion in developing countries cannot be ignored, as frequent policy reforms and the large demand for land urbanization are common features in the process of economic development.

In this study, we fill this gap by empirically investigating the impact of EPU on agricultural land conversion and exploring the underlying mechanisms driving this effect in China, the world’s largest developing country. With China’s growth miracle, the scale of agricultural land conversion significantly increases. According to statistics by the Ministry of Natural Resources of China, the area of agricultural land converted to urban residential, commercial, and industrial areas rose 295.3% from 1998 to 2013. The pace of conversion notably quickened after the 2008 global financial crisis, a time of heightened EPU. We constructed a panel dataset of 270 prefecture-level cities from 2004 to 2017 to investigate the effect of EPU on agricultural land conversion for urban use. We found that EPU significantly increased the area of agricultural land converted for urban residential, commercial, and industrial uses. Specifically, a one-standard-deviation increase in EPU led to a marginal expansion of 22.2% in agricultural land converted into residential buildings, business centers, and industrial estates. For instance, during the 2008 global financial crisis, the annual geometric mean of EPU rose by roughly two standard deviations relative to 2007. The results suggested that this increase led to an average of about 45% more agricultural land conversion per city for urban development.

To address potential endogeneity concerns, we drew on the international spillover effects of the U.S. EPU and exploited the temporal variation in the one-period lagged U.S. EPU as an instrument variable (IV) for China’s EPU. The two-stage estimation results further confirmed the expansionary effect of EPU on agricultural land conversion. The finding remained valid even under the plausibly exogenous IV framework [14], which substantially relaxes the exclusion restriction of the IV. The finding also remained robust to a series of robustness checks.

Moreover, we show that this effect was mainly driven by increased pressure on prefecture-level governments’ fiscal spending and their management of economic growth targets. On the one hand, EPU reduced local tax revenue while raising public expenditures, prompting local governments to seek additional sources of fiscal income. On the other hand, EPU hampered local economic growth, further intensifying the pressure on local governments to meet their growth targets. Confronted with these challenges, local officials were forced to convert more agricultural land to generate off-budget revenue and stimulate economic growth by attracting fixed asset investment through land incentives.

We further investigated city-level heterogeneity and found that the land conversion effect driven by EPU was particularly pronounced in cities heavily reliant on fixed asset investment for local economic development and are located in non-eastern China. By contrast, dependence on land finance did not significantly influence the heterogeneity of agricultural land conversion. Additionally, our analysis of relevant policy moderating effects suggests that stronger marketization reforms and stricter agricultural land quotas imposed by provincial governments can help to mitigate agricultural land loss caused by EPU.

The remainder of this paper is structured as follows. Section 2 reviews key literature and develops theoretical hypotheses. Section 3 describes the data and outlines the estimation methods. Section 4 evaluates the impact of EPU on agricultural land conversion. Section 5 explores the potential mechanisms. Section 6 discusses heterogeneous effects and potential policies to mitigate the adverse effects of EPU. Section 7 concludes the paper.

2. Literature and Hypotheses

2.1. Literature Review

EPU refers to the uncertainty that arises from the unpredictability or ambiguity surrounding government policies, particularly in fiscal, monetary, trade, and economic regulatory areas. This uncertainty often leads to caution in decision-making processes across various sectors, with significant economic consequences.

A substantial body of literature highlights the detrimental effects of EPU on economic variables such as output, employment, and financial markets. Utilizing a comprehensive index to quantify EPU in the United States, higher uncertainty is often associated with increased stock price volatility, reduced firm investment, and declines in output and employment, particularly in sectors that are highly sensitive to policy changes [15]. Moreover, EPU is counter-cyclical, meaning that increased uncertainty is typically followed by a decline in real output in G7 countries [16]. Similarly, EPU shocks observed in China lead to declines in equity returns, employment, and output, indicating that the heightened policy uncertainty negatively impacts broad economic performance [17].

EPU is particularly influential in shaping corporate investment decisions. Firms, especially those that are sensitive to government policies, tend to postpone or reduce investments during periods of heightened uncertainty. A doubling of EPU is associated with an 8.7% decrease in quarterly investment rates for American firms [7]. The effect is especially stronger for firms with irreversible investments, which is consistent with the real option theory, where the uncertainty surrounding policy decisions increases the value of waiting, thus delaying investments until clearer information emerges. In China, EPU resulting from structural reforms in 2015 significantly reduced corporate investment and net debt issuance for private firms [9]. Heightened policy uncertainty also discourages mergers and acquisitions (M&A) activity [12].

In addition to investment and M&A activity, EPU has a significant impact on corporate innovation. Several studies suggest that heightened policy uncertainty can hinder a company’s ability and willingness to invest in new technologies and innovation initiatives, particularly in firms that face financial constraints or rely on external financing. Rising EPU increases firms’ cost of capital, which in turn discourages innovation, especially in industries that require substantial financial resources [10]. Furthermore, higher EPU leads to reduced green innovation, whereas government environmental subsidies could mitigate the negative effects [11]. EPU can also moderate the relationship between corporate financialization and innovation investment [18].

EPU is not confined to a single country and its effects often spill over across borders. Countries with strong bilateral trade relationships experience significant spillovers of policy uncertainty. These spillovers can manifest in various forms, including trade disruptions, investment delays, and shifts in financial market performance [19]. The U.S. EPU often transmits to BRIC countries, especially in the short term [20].

Recent literature in developed countries show that economic crises or fluctuations have profoundly reshaped land use patterns and policy orientations. For instance, in Athens, Greece, the economic recession slowed the expansion of new residential construction, while industrial land expansion became the dominant driver of land use change. The conversion of low-economic-value land, such as forests, to urban uses also increased markedly during this period [21]. Moreover, the 2008 global financial crisis served as a critical turning point in driving land policy reforms. The UK government promoted the conversion of office buildings into residential units by relaxing development rights, while the Netherlands streamlined planning procedures to facilitate the reuse of office space [22]. In Greece, fiscal imperatives during the crisis prompted the government to rapidly privatize and develop public land, resulting in an abrupt shift in land policy [23].

In other European countries such as Spain, Portugal, and Italy, spatial planning reforms in response to the pressures of crisis and austerity have centered on “streamlining administration and enhancing efficiency.” These reforms have simplified land use change procedures, weakened multilevel governance and coordination, and strengthened the entrepreneurial role of local governments [24]. In the Netherlands, local governments, facing fiscal pressures, shifted from proactive development to more flexible and incremental approaches to land management [25]. With respect to large-scale infrastructure financing, Dutch local governments have also bridged funding gaps through “developer obligations” negotiated with private developers [26].

Although previous studies have examined the economic impacts of EPU on firm behavior, financial markets, and macroeconomic outcomes, the influence of EPU on the conversion of agricultural land for urban development remains understudied. In particular, there is a notable lack of research on how local governments adjust their strategies for converting agricultural land in response to EPU. This research gap is especially significant in developing countries such as China, where fiscal decentralization, political centralization, and performance-based governance interact in complex ways. Moreover, while some studies have explored the effects of economic crises or fluctuations on land use in developed countries, there is still a scarcity of systematic analysis on how EPU shapes land use decisions within the institutional contexts of developing economies.

2.2. Hypothesis Development

From a political economy perspective, evidence on the economic impact of EPU suggests that local governments may adjust land use policies in response to EPU, thereby affecting the conversion of agricultural land. Particularly within China’s governance framework, the adverse economic effects triggered by EPU can exacerbate the dual pressures on local governments to achieve both fiscal sustainability and economic growth targets, which in turn alters land use patterns.

The pressures faced by local governments stem from the distinctive characteristics of China’s fiscal and political governance. As a major transitioning and developing country, China’s current governance system is characterized by the coexistence of political centralization and fiscal decentralization [27]. Since the Qin Dynasty in ancient times, successive Chinese dynasties have generally adopted a centralized system in which major decisions are made by the central government, with local governments responsible for implementation and subject to central assessment and incentives. In contemporary China, a strict hierarchical relationship between the central and local governments remains. The central government retains absolute authority over political and major economic decisions, and local governments are not only expected to maintain basic social stability but are also subject to explicit requirements regarding regional economic performance, mainly reflected in annual economic growth targets and fiscal sustainability [27,28]. For instance, in 2018, the central government set an annual economic growth target of 6.5% at the beginning of the year. This meant that, barring exceptional circumstances, local governments were required to ensure that the annual growth rate in their jurisdictions did not fall below this threshold, thereby intensifying the pressure on local authorities to promote economic growth. On the fiscal front, following the tax-sharing reform in 1994, local governments have become primarily responsible for financing public expenditures such as public services and infrastructure investment through their own fiscal revenues, leading to inherent pressures regarding the sustainability of local public finances.

To address these pressures, local governments have adopted a variety of strategies, such as actively attracting investment to expand the local tax base and promote investment-driven economic growth. They also utilize local government debt as a means to raise funds for infrastructure development. Notably, since the enactment of the 1998 Land Management Law, converting agricultural land and leasing it for urban development has become a key strategy for local governments in China to generate off-budget revenues and stimulate local economic growth [29,30]. Local governments can acquire farmland from rural residents at relatively low prices and then lease or sell it to developers at much higher prices, generating substantial revenue from land transactions. This revenue is often kept off the local budget and does not have to be shared with the central government, allowing local authorities to invest directly in infrastructure, attract investment, and fund economic development projects, thereby fueling rapid local economic growth. Moreover, to balance these fiscal and growth objectives, local governments typically structure their land lease portfolios to include a mix of high-priced commercial and residential land alongside lower-priced industrial land [31]. The first two categories are particularly effective in alleviating fiscal pressures—a strategy commonly referred to as the “land finance motive” [32]. By contrast, industrial land plays a crucial role in fostering investment-driven growth, a practice known as the “land investment motive” [33].

Although EPU is manifested through changes in both central and local government policies, the inherent complexity of the economic system often makes the origins and impacts of these policies both unanticipated. As a crucial node in the global supply chain and capital flows, China’s EPU is also shaped by macroeconomic uncertainty and policy changes in other major economies such as the United States [20]. In particular, international economic uncertainty also influences the land use policies of local governments in China through globalization. For instance, the 2011 downgrade of the U.S. sovereign credit rating significantly increased foreign capital inflows into Chinese regions, substantially alleviating fiscal and growth pressures on local governments, and thus potentially altering agricultural land conversion patterns to some extent.

However, due to the centralization of major economic decision making, local governments have limited capacity to influence such uncertainty, making it difficult to preemptively mitigate its adverse effects on local economies. As a result, local governments frequently face heightened pressures in achieving economic growth targets and maintaining fiscal sustainability, and must resort to additional ex post measures in response. For example, Dang, Fang, and He [34] found that, when confronted with fiscal pressures induced by EPU, local governments intensify tax collection efforts. Theoretically, since the conversion and leasing of agricultural land for urban development is a key policy tool for local governments, they have strong incentives to expand the conversion of agricultural land for commercial, residential, and industrial purposes to mitigate fiscal and economic growth pressures. Figure 1 summarizes these theoretical mechanisms, and we propose two testable hypotheses based on this framework.

Hypothesis 1. 

EPU leads to an expansion in the scale of agricultural land conversion for commercial, residential, and industrial uses.

Hypothesis 2. 

The agricultural land conversion induced by EPU stems mainly from increased pressures on local governments’ fiscal expenditure and the need to meet economic growth targets.

The impact of EPU on local governments’ land conversion may be shaped by two key institutional factors. The first is agricultural land conversion quotas policy, under which the Chinese central government imposes strict annual limits on the amount of agricultural land that can be converted into construction land in each province [35]. To promote regional economic balance, higher agricultural land conversion quotas have been allocated to inland cities in the west compared to coastal cities in the east. This policy constrains local governments’ flexibility in managing fiscal and economic growth pressures. Consequently, when agricultural land conversion quotas are lower, local governments may convert less agricultural land for urban use as a means of mitigating the effects of EPU.

The second factor is China’s ongoing transition from a planned economy to a market-oriented system, which has led to significant variations in the degree of marketization across regions. On the one hand, market-oriented reforms stimulate corporate investment, enhance firm productivity, and improve resource allocation efficiency by strengthening property rights protections and enhancing contract enforcement [36,37,38]. These mechanisms help to buffer the local economy against the negative effects of EPU. On the other hand, increased marketization reduces government intervention in the economy by standardizing local officials’ behaviors [39], thereby diminishing their discretionary control over agricultural land conversion. Both forces can help to mitigate the adverse impact of EPU on agricultural land conservation. As a result, the positive effect of EPU on land conversion is likely to be weaker in cities with lower land conversion quotas or higher levels of marketization. Based on this logic, we propose the following theoretical hypothesis:

Hypothesis 3. 

Lower agricultural land conversion quotas or greater marketization reform can mitigate the positive impact of EPU on agricultural land conversion.

3. Data and Methods

In this section, we first describe the data and variables used for empirical analyses. We then provide suggestive evidence on the positive relationship between EPU and agricultural land conversion. Finally, we outline the empirical strategy used to investigate the impact of EPU on agricultural land conversion, its underlying mechanisms, and potential policy options to moderate the land conversion effect.

3.1. Data and Variables

To examine the impact of EPU on agricultural land conversion by local governments, we focused on 270 prefecture-level cities in mainland China, using data from 2004 to 2017. (Due to severe missing data and changes in administrative zoning, we excluded Bijie, Tongren, Haidong, Zhanzhou, Sansha, Lüliang, Bayan Nur, Ulanqab, Laibin, Chongzuo, Pu’er, Lincang, Dingxi, Longnan, Guyuan, Zhongwei, Turpan, Hami, and all cities in Tibet from our analysis). This sample selection served three key purposes. First, it avoided structural disruptions caused by major events before 2001, such as the tax-sharing reform, the implementation of the Land Administration Law, and China’s accession to the WTO—factors that likely had significant effects on local governments’ agricultural land conversion incentives. Second, it allowed us to capture nearly an entire business cycle in China, encompassing both the rapid growth period from 2004 to 2007 and the subsequent downturn from 2008 to 2017. Lastly, it addressed the data availability issue, as official yearbooks containing city-level agricultural land conversion statistics for residential, industrial, and commercial uses were not publicly available before 2004 or after 2017.

The dependent variable in our analysis was the scale of land conversion for urban residential, industrial, and commercial uses. Following a previous study [40], we measured this variable using the area of newly added construction land leased to developers. The raw data were sourced from the China Land and Resources Statistical Yearbook (2005–2018), published by the National Bureau of Statistics, the primary official statistics agency in China. Figure 2 illustrates the spatial distribution of the average agricultural land conversion area from 2004 to 2017. High conversion areas are concentrated in the eastern and coastal regions, particularly in the Yangtze River Delta, while the western regions exhibit lower conversion areas. This pattern likely reflects rapid urbanization and economic growth, resulting in substantial agricultural loss in densely populated and industrialized areas.

To measure EPU in China, we employed the China Economic Policy Uncertainty Index developed by Huang and Luk [17]. This index builds on the pioneering work of Baker, Bloom, and Davis [15], which measures policy uncertainty based on the frequency of uncertainty- and policy-related keywords in the South China Morning Post (SCMP), Hong Kong’s leading English-language newspaper. While this approach effectively captures the time-varying nature of policy uncertainty, it has certain limitations regarding news sources. As Davis, Liu, and Sheng [41] noted, “The SCMP-based index reflects more of an outsider perspective on aspects of China’s economic policy uncertainty that intersect with foreign editors’ interests and concerns.” To address this issue, Huang and Luk [17] instead relied on text analysis from the top ten newspapers in mainland China, ensuring a more locally representative measure of economic policy uncertainty in China. (The ten newspapers include Beijing Youth Daily, Guangzhou Daily, Jiefang Daily, People’s Daily Overseas Edition, Shanghai Morning Post, Southern Metropolis Daily, The Beijing News, Today Evening Post, Wen Hui Daily, and Yangcheng Evening News).

Figure 3 illustrates the significant fluctuations in China’s EPU from 2004 to 2017. The uncertainty remained relatively stable before 2008 but spiked sharply after the Lehman Brothers bankruptcy, reaching peaks around major global and domestic events, such as the U.S. sovereign credit rating downgrade and changes in the RMB fixing mechanism. Furthermore, to align with other annual variables, we followed Gulen and Ion [7] to convert the original monthly index into an annual measure using the geometric average method. Additionally, in the robustness analysis, we used the daily policy uncertainty index developed by Huang and Luk [17], which is based on 114 general interest daily newspapers in mainland China, as an alternative data source. We also aggregated the monthly index into an annual scale by calculating the arithmetic average.

We included three types of control variables in our analysis. First, cities’ demographic and economic endowments play a crucial role in local governments’ land use decisions. Following previous studies on agricultural land conversion, we incorporated several key variables in this category: GDP per capita, the share of secondary industry in GDP, population density, and foreign direct investment. These variables capture regional economic development, industrial structure, urbanization, and openness. The relevant socioeconomic data were sourced from the China City Statistical Yearbook.

Second, local government land use decisions are significantly influenced by the promotion incentives of local officials. To account for this factor, we controlled for individual characteristics of local officials, specifically the age and tenure of city party secretaries. These variables were manually compiled from officials’ resumes, following the methodology of Zhang and Gao [42]. The raw data were obtained from People’s Daily Online and Baidu Encyclopedia.

Third, in line with Gulen and Ion [7], we controlled for macroeconomic conditions by substituting the time fixed effects with three national representative expectation indicators: the leading economic index, the entrepreneur confidence index, and the investor sentiment index. (On the one hand, the overall EPU index lacks cross-sectional variation, so including time fixed effects would absorb its explanatory power. On the other hand, since policy uncertainty is counter-cyclical, controlling for macroeconomic cycles is necessary to mitigate potential endogeneity issues). The first two indicators were sourced from the MacroChina database, while the investor sentiment index came from Cheema, Man, and Szulczyk [43].

Moreover, the instrumental variables, mechanism variables, and policy variables will be discussed later.

Table 1. Summary statistics of main variables.

Variable	Obs.	Mean	SD	Min	Max
A. Agricultural land conversion
Area of newly added construction land (hectares)	3631	460.27	482.82	0.03	5788.56
B. Economic policy uncertainty
Annual economic policy uncertainty Index	3780	0.00	1.00	$-$1.65	1.23
C. City socioeconomic characteristics
GDP per capita (CNY Thousand)	3780	36.16	27.86	2.82	215.49
Population density (Person/km. sq)	3780	429.44	311.26	4.70	2661.540
Share of secondary industry (%)	3780	49.02	10.67	9.00	90.97
Foreign direct investment (USD Million)	3667	580.25	1166.13	0.02	14,004.53
D. City leader characteristics
Age of city party secretary	3737	53.64	3.54	40.00	62.00
Tenure of city party secretary	3743	2.18	1.70	0.00	10.00
E. National macroeconomic conditions
Leading economic index	3780	101.04	1.50	98.52	103.33
Entrepreneur confidence index	3780	124.93	8.92	110.22	141.93
Investor sentiment index	3780	0.02	0.57	$-$1.05	1.26
F. Instrumental variable
U.S. economic policy uncertainty (lagged)	3780	0.00	1.00	$-$1.36	1.72
G. Mechanism variables
Fiscal expenditure pressure	3780	9.49	0.99	5.97	13.04
Economic growth pressure	3679	2.17	2.87	$-$10.72	21.61
H. Policy variables
Provincial land conversion quotas (ln. hectares)	3776	8.31	0.81	3.01	10.45
Provincial marketization index	3780	6.93	1.69	2.33	11.80

Notes: The natural logarithms of prefecture-level new construction land area, GDP per capita, population density, and foreign direct investment are used in regressions.

presents the summary statistics for these main variables.

3.2. Suggestive Evidence

Before conducting the formal analysis, we provide descriptive evidence to contextualize the following empirical findings. Figure 4 illustrates the annual trends of the EPU index and the average agricultural land conversion for residential, commercial, and industrial uses across Chinese prefecture-level cities from 2004 to 2017. The figure reveals a notable degree of co-movement, particularly during periods of heightened EPU. From 2004 to 2008, both indicators exhibit a sharp upward trend, with the policy uncertainty spiking in 2008, coinciding with a surge in agricultural land conversion. Between 2009 and 2013, both series remain at elevated levels, fluctuating in response to major economic events. This synchronization suggests that EPU may have influenced prefectures’ land use decisions, potentially accelerating land conversion during uncertain periods.

However, some divergences between the two series are observed in certain years, particularly after 2014, indicating that other factors, such as macroeconomic fluctuations, may have played an independent role in shaping agricultural land conversion dynamics. These observations underscore the need for a more rigorous empirical analysis to disentangle the effect of EPU on agricultural land conversion from broader confounding factors.

3.3. Empirical Strategy

3.3.1. The Impact of Economic Policy Uncertainty on Agricultural Land Conversion

OLS estimation. To assess the impact of EPU on agricultural land conversion and empirically test Hypothesis 1, we adopted a city fixed effects model following prior studies on policy uncertainty, such as Gulen and Ion [7]:

$${ALC}_{ct}={\beta }_0+{\beta }_1{EPU}_{ct}+{\beta }_2{\mathit{X}}_{\mathit{c}\mathit{t}}+{\mu }_c+{ϵ}_{ct},$$

(1)

where the dependent variable ${ALC}_{ct}$ represents the natural logarithm of the agricultural land area converted for urban residential, industrial, and commercial purposes in city $c$ during year $t$. The term ${\mu }_c$ captures city fixed effects, accounting for time-invariant characteristics specific to each prefecture-level city that may affect agricultural land conversion. The vector ${\mathit{X}}_{\mathit{c}\mathit{t}}$ includes a set of control variables, such as the city’s socioeconomic conditions, the personal attributes of local political leaders, and national macroeconomic conditions, as discussed in the Section 3.1. ${ϵ}_{ct}$ indicates the random error term.

The key explanatory variable of interest is ${EPU}_{ct}$, which represents the average EPU in city $c$ during year $t$, measured as the annual geometric average of the monthly China Economic Policy Uncertainty Index from Huang and Luk [17]. The parameter of interest ${\beta }_1$ captures the average effect of EPU on agricultural land conversion at the prefecture-level city. According to Hypothesis 1, we expected ${\beta }_1$ to be significantly positive. Notably, using a national-level EPU measure implies uniform exposure of all cities to EPU, embodying an intention-to-treat framework that facilitates policy analysis.

However, estimating the causal effect of EPU on agricultural land conversion using Equation (1) may be subject to two primary endogeneity concerns. First, the effect of EPU may be confounded by broader macroeconomic and political uncertainty. Major macroeconomic or political events often contribute to increased policy uncertainty, meaning that local governments facing EPU are likely also experiencing wider macroeconomic and political uncertainties.

Second, there is the potential for measurement errors in EPU, as this newspaper-based policy uncertainty index is influenced by the choice of newspaper sources and the selection of keywords related to macroeconomic policy. These concerns may bias our OLS estimates.

Instrumental variable estimation. The standard approach in the literature for addressing endogeneity concerns is the use of IVs. Therefore, inspired by prior studies [34], we employed the one-period lagged U.S. EPU as an IV for China’s EPU to reinforce the validity of baseline estimates in Equation (1). Specifically, we implemented a two-stage estimation, with Equation (1) as the second stage and the first stage specified as follows:

$${EPU}_{ct}={\alpha }_0+{\alpha }_1{USPU}_{t-1}+{\alpha }_2{\mathit{X}}_{\mathit{c}\mathit{t}}+{\mu }_c+{\nu }_{ct}.$$

(2)

where ${USPU}_{t-1}$ represents the average U.S. EPU during year $t-1$, measured as the geometric average of the monthly U.S. Economic Policy Uncertainty Index from Baker, Bloom, and Davis [15].

The rationale behind our instrument is twofold. First, as the world’s largest economy, fluctuations in U.S. macroeconomic policies exert significant and asymmetric spillover effects on business cycle fluctuations and EPU in other economies, including China [20,44]. Second, the impact of the U.S. EPU on agricultural land conversion by Chinese local governments is largely mediated through domestic EPU in China. This is due to the high centralization of foreign policy by the Chinese central government and the country’s top-down approach to macroeconomic management.

To test the plausibility of the IV, we first assessed its strength using the Kleibergen–Paap rank F-statistic from the first-stage regression. Following Staiger and Stock [45], an F-statistic above 10 indicates a sufficiently strong instrument. Then, we used Conley, Hansen, and Rossi’s [14] framework to test the exclusion restriction assumption of the IV that U.S. economic policy uncertainty affects China’s agricultural land conversion only through China’s own EPU. Under this method, the true effect of EPU on agricultural land conversion is assumed to be estimated by the equation:

$${ALC}_{ct}={\gamma }_0+{\gamma }_1{EPU}_{ct}+\delta {USPU}_{t-1}+{\gamma }_3{\mathit{X}}_{\mathit{c}\mathit{t}}+{\mu }_c+{\eta }_{ct},$$

(3)

where $\delta$ represents the direct effect of the U.S. EPU on agricultural land conversion in Chinese cities, which is assumed to be zero under the standard IV framework. However, under the framework of plausibly exogenous instruments [14], $\delta$ is typically assumed to be non-zero. By re-estimating the two-stage model under this prior assumption, the framework indirectly addresses the challenge that the exclusion restriction cannot be directly tested due to the inherent complexity of the error term. Specifically, we applied the Union of Confidence Intervals (UCI) method of Conley, Hansen, and Rossi [14] to construct 95% confidence intervals for the average effect of EPU under different prior interval assumptions for $\delta$. The inference results indicate how severe a violation of the IV exclusion would need to be for the estimated effect of EPU on agricultural land conversion to remain positive, thereby further assessing the plausibility of the instrument.

Robustness checks and heterogeneity analyses. We conducted several robustness checks to validate the effect of EPU on agricultural land conversion. First, we controlled for general macroeconomic uncertainty and political uncertainty. This approach allowed us to isolate the specific effects of policy-related uncertainty from broader macroeconomic and political fluctuations. Second, we used alternative measures of EPU for robustness. Specifically, we employed Huang and Luk’s [17] daily policy uncertainty index, constructed from 114 general interest daily newspapers in mainland China, as an alternative data source. We also included a dummy variable for the global financial crisis, set to 1 for years after 2008 and 0 otherwise, as a proxy for heightened EPU.

Further robustness checks included winsorizing the key variables to mitigate the influence of outliers in both agricultural land conversion and EPU and detrending the policy uncertainty index to address potential spurious correlation. Finally, we applied alternative standard error adjustments, including heteroskedasticity-robust and bootstrap standard errors.

To explore heterogeneous effects, we performed subsample analyses based on cities’ geographic locations and differing levels of dependence on land finance and fixed asset investment.

3.3.2. Mechanisms and Policy Discussion

Mechanism analyses. To illuminate the underlying mechanisms linking EPU to agricultural land conversion, we empirically tested Hypothesis 2 by estimating the following model:

$${Pressure}_{ct}=\alpha +{\theta MPU}_t+\gamma {\mathit{X}}_{\mathit{c}\mathit{t}}+{\mu }_c+{\nu }_{ct},$$

(4)

where ${Pressure}_{ct}$ represents a potential mechanism variable, such as fiscal pressure or economic growth target pressure faced by prefecture-level local governments. The key parameter of interest $\theta$ captures the degree to which local governments’ fiscal or economic growth pressures respond to EPU. According to Hypothesis 2, we expected $\theta$ to be significantly positive. To strengthen our estimation in Equation (4), we also employed the one-period lagged U.S. EPU as an instrumental variable for China’s economic policy uncertainty and applied a two-stage least squares estimation.

Policy analyses. We examined the potential policies that mitigate the effect of EPU on agricultural land conversion by estimating a city fixed effect model with interaction terms:

$${ALC}_{ct}=\alpha +{\delta MPU}_t+\varphi {MPU}_t\times {Policy}_{ct}+\gamma {\mathit{X}}_{\mathit{c}\mathit{t}}+{\mu }_c+{\eta }_{ct},$$

(5)

where ${Policy}_{ct}$ represents potential policy measures implemented by upper-level governments to regulate agricultural land conversion driven by EPU in prefecture-level city $c$ during year $t$, including provincial land quotas and marketization policies. The parameter of our interest $\varphi$ captures the potential effectiveness of policy interventions in regulating such land conversion, helping us empirically test Hypothesis 3.

4. Main Results

In this section, we first report the OLS and IV estimation results on the impact of EPU on agricultural land conversion. We then present the results of robustness checks.

4.1. OLS Estimation

As a baseline investigation into the impact of EPU on agricultural land conversion, we estimated the specification in Equation (1) using ordinary least squares (OLS). The results are presented in

Table 2. The impact of economic policy uncertainty on agricultural land conversion.

	(1)	(2)	(3)
Dependent Variable:	Agricultural Land Conversion
	Without Policy Uncertainty	Only Policy Uncertainty	Baseline
Economic policy uncertainty		0.5749 ***	0.2220 ***
		(0.03)	(0.03)
GDP per capita	1.3113 ***		1.0218 ***
	(0.07)		(0.07)
Population density	0.0930		0.0641
	(0.09)		(0.09)
Share of secondary industry	0.0366 ***		0.0338 ***
	(0.01)		(0.01)
Foreign direct investment	0.0590		0.0427
	(0.03)		(0.03)
Age of city party secretary	$-$0.0001		$-$0.0000
	(0.01)		(0.01)
Tenure of city party secretary	0.0087		0.0183
	(0.02)		(0.02)
Tenure square of city party secretary	$-$0.0014		$-$0.0034
	(0.00)		(0.00)
National leading economic index	0.0693 ***		0.0455 **
	(0.02)		(0.02)
National entrepreneur confidence index	$-$0.0023		0.0018
	(0.00)		(0.00)
National investor sentiment index	0.0768 **		0.2404 ***
	(0.03)		(0.04)
City fixed effects	Yes	Yes	Yes
Adjusted R-squared	0.625	0.523	0.631
Observations	3492	3631	3492

Notes: Robust standard errors in parentheses are clustered at the city level. All regressions employ OLS. Variable definitions and summary statistics are described in Section 3.1. ** and *** denote statistical significance at the 5% and 1% levels, respectively.

.

To connect with the literature and emphasize the role of EPU, Column (1) first reports estimates without the policy uncertainty, which are largely align with previous empirical findings. Regarding city-level socioeconomic characteristics, higher GDP per capita is statistically significantly associated with increased agricultural land conversion [46]. A higher population density can drive more agricultural land conversion, although this relationship is not statistically significant, which can account for the effect of urbanization and population growth on land use changes [47]. Cities with a higher secondary industry share in GDP tend to convert more agricultural land for urban construction, highlighting the role of land demand for industrial development [48]. Additionally, an increase in foreign direct investment is positively associated with agricultural land conversion for urban uses, likely reflecting the impact of economic globalization on land use transitions [49].

Regarding individual characteristics of city leaders, agricultural land conversion appears to decline with the age of the city party secretary but follows an inverted U-shaped relationship with tenure length. However, these effects are not statistically significant. These leadership traits may capture the impact of career incentives on urban spatial expansion [50]. Furthermore, the coefficients on the national leading economic index and the investor sentiment index are both significantly positive, suggesting that fluctuations in the real economy and financial markets affect agricultural land conversion. By contrast, the national entrepreneur confidence index does not show a statistically significant coefficient.

Next, Column (2) reports the estimation results from the city fixed effects regression, incorporating only the EPU index. The estimated coefficient on EPU is significantly positive at the 1% level, indicating a strong positive relationship between EPU and agricultural land conversion. Furthermore, Column (3) follows the specification in Equation (1), adding controls for city-level socioeconomic characteristics and the demographic background of city leaders. The estimated coefficient on EPU remains positive and statistically significant at the 1% level, suggesting that heightened policy uncertainty contributes to the expansion of agricultural land conversion for commercial, residential, and industrial purposes. These findings provide empirical support for Hypothesis 1.

The average effect of EPU on agricultural land conversion is ecologically significant. As shown in Column (3), the magnitude of the coefficient on the policy uncertainty (0.2220) indicates that a one-standard-deviation increase in EPU can result in an average increase of 22.2% in agricultural land conversion for residential, commercial, and industrial uses at the prefecture-level city. For instance, during the 2008 global financial crisis, the annual geometric average of EPU increased by approximately two standard deviations compared to 2007. According to the coefficient in Column (3) of Table 2, this surge led to an average conversion of 204.36 hectares of agricultural land per city for urban development. These findings underscore EPU as a significant driver of agricultural land conversion, shaping land use patterns amid economic policy volatility.

4.2. Instrumental Variable Analysis

An effective approach to addressing concerns about omitted variable bias and measurement errors in the baseline estimation is to use instrumental variable analysis. As discussed in Section 3.3.1, one-period lagged EPU in the United States is a suitable instrumental variable for EPU in China. Accordingly, we used the U.S. Economic Policy Uncertainty Index from Baker, Bloom, and Davis [15] to quantify EPU in the United States, and we implemented a two-stage least squares regression to analyze agricultural land conversion at the prefecture level.

Table 3. Instrumental variable analysis.

	(1)	(2)
Dependent Variable:	Economic Policy Uncertainty in China	Agricultural Land Conversion
Economic policy uncertainty in China		6.0263 ***
		(1.10)
U.S. economic policy uncertainty (lagged)	0.0283 ***
	(0.01)
City fixed effects	Yes	Yes
Observations	3624	3493
First-stage F-statistic		30.89

Notes: Robust standard errors in parentheses are clustered at the city level. Column (1) employs OLS, while Column (2) uses 2SLS estimation. The first-stage F-statistic correspond to the Kleibergen–Paap rank F-statistic. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

presents the results using this instrumental variable. Column (1) reports the first-stage regression result on EPU in China, which shows that the coefficient on the U.S. EPU is positive and statistically significant at the 1% level. The magnitude of this coefficient indicates that a one-standard deviation increase in the U.S. EPU in the previous year corresponds to a 0.0283 standard-deviation increase in China’s EPU. The first-stage estimation provides evidence on a significant positive spillover effect from the U.S. EPU to China’s, consistent with prior literature [20].

Column (2) presents the second-stage estimation result, where the one-period lagged U.S. EPU is used as an instrument. The first-stage F-statistic exceeds 10, surpassing Staiger and Stock’s [45] rule-of-thumb threshold, confirming the instrument’s strength. Notably, the coefficient on China’s EPU remains positive and significant at the 1% level, with a larger magnitude compared to the baseline estimate in Table 2. This increase may be attributable to adjustments for omitted variable bias and measurement errors. These findings underscore the significant and positive impact of EPU on agricultural land conversion and further validate the findings from our OLS estimation results.

The primary threat to the validity of the IV estimates is that the U.S. EPU may directly affect agricultural land conversion in Chinese prefecture-level cities, violating the perfect exclusion restriction and undermining the IV strategy. As discussed in Section 3.3.1, we applied the UCI method of Conley, Hansen, and Rossi [14] to infer 95% confidence intervals for the average effect of EPU under different prior interval assumptions for $\delta$. Specifically, we assumed $\delta$ lies within a symmetric interval centered at zero and bounded by the absolute value of the reduced-form effect in Column (2) of

Table A1. Reduced-formed effect of U.S. economic policy uncertainty on agricultural land conversion.

	(1)	(2)
Dependent Variable:	Agricultural Land Conversion
U.S. economic policy uncertainty (lagged)	0.5054 ***	0.2806 ***
	(0.02)	(0.02)
City and national characteristics	No	Yes
City fixed effects	Yes	Yes
Adjusted R-squared	0.491	0.650
Observations	3631	3492

Notes: All models are estimated by OLS. City and national characteristics are summarized in Table 1. Robust standard errors in parentheses are clustered at the city level. *** denote statistical significance at the 1% level.

. The inference results in Figure 5 reveal that the bounds for the second-stage estimates exclude zero and remain positive as long as the absolute value of the U.S. EPU’s direct effect is smaller than about 23% of the area of agricultural land conversion for prefecture-level cities. These thresholds correspond to approximately 82% of the respective reduced-form effects and suggest that the baseline IV estimates in Table 3 are robust to a substantial degree of instrument exclusion, providing indirect support to the validity of our IV strategy.

4.3. Robustness Checks

As discussed in Section 3.3.1, the results based on the baseline specification may suffer from endogeneity issues. Therefore, we provide a set of robustness checks in this section, particularly considering the additional control variables, using alternative measures of EPU, and other robustness checks.

4.3.1. Controlling for Macroeconomic and Political Uncertainty

The baseline effect of EPU may be influenced by broader macroeconomic or political uncertainties. In particular, macroeconomic uncertainties closely correlate with EPU. As shown in Figure 3, key events such as the 2008 bankruptcy of Lehman Brothers, the 2011 downgrade of the U.S. sovereign credit rating, and the 2015 adjustment of China’s RMB exchange rate mechanism coincide with notable peaks in China’s EPU. Economic uncertainties have also significantly affected China’s land supply and land use policies. The 2008 Lehman Brothers bankruptcy triggered a global financial crisis, prompting China’s “Four Trillion Yuan” stimulus package, significantly expanding land supply and intensifying local governments’ reliance on land finance through rapid infrastructure and real estate development. Subsequently, the 2011 U.S. credit rating downgrade increased global uncertainty and capital inflows into China, driving up land prices and prompting stricter land supply and real estate regulations to mitigate financial risks. Additionally, the 2015 RMB exchange rate adjustment created depreciation pressures and capital outflow risks, leading authorities to adjust land supply policies by emphasizing urban renewal and industrial land provisions to optimize land use structure and sustain economic stability.

To address this, we incorporated general macroeconomic uncertainty and provincial political uncertainty into the baseline specification. Specifically, we followed Jurado, Ludvigson, and Ng [51] to measure macroeconomic uncertainty using China’s macroeconomic and financial data. For provincial political uncertainty, we introduced a dummy variable that equaled 1 if a provincial party congress was held in a given year and 0 otherwise.

Table 4. Robustness checks: Controlling for macroeconomic and political uncertainty.

	(1)	(2)	(3)
Dependent Variable:	Agricultural Land Conversion
	Only economic uncertainty	Only political uncertainty	Both types of uncertainty
Economic policy uncertainty	0.7487 ***	0.2207 ***	0.7544 ***
	(0.06)	(0.03)	(0.06)
Macroeconomic uncertainty	$-$3.8086 ***		$-$3.8455 ***
	(0.32)		(0.33)
Provincial political uncertainty		0.0522 *	$-$0.0239
		(0.03)	(0.03)
City fixed effects	Yes	Yes	Yes
Adjusted R-squared	0.648	0.631	0.647
Observations	3492	3492	3492

Notes: Robust standard errors in parentheses are clustered at the city level. All regressions employ OLS and also include the covariates from Table 2. * and *** denote statistical significance at the 10% and 1% levels, respectively.

presents the estimation results accounting for macroeconomic and provincial political uncertainty. Consistent with the baseline findings, the estimated coefficients for EPU remain positive and highly significant at the 1% level, although their magnitudes range from 0.2207 to 0.7544 depending on the control group. Notably, unlike EPU, macroeconomic uncertainty exhibits negative and significant coefficients at the 1% level, suggesting that macroeconomic instability may deter agricultural land conversion, likely due to weakened long-term expectations of land supply and demand. Meanwhile, the coefficients for provincial political uncertainty indicate a relatively weaker impact on agricultural land conversion. These results confirm that even after accounting for macroeconomic and provincial political uncertainty, the positive effect of EPU on agricultural land conversion remains robust.

4.3.2. Alternative Measurements of Economic Policy Uncertainty

In the baseline estimation, we measured EPU in China using the annual geometric mean of Huang and Luk’s [17] monthly China Economic Policy Uncertainty Index, which was derived from the ten newspapers in mainland China. To test the robustness of our findings, this section introduces a number of other potential proxies for EPU. First, we modified the aggregate method by transitioning from a monthly index to an annual scale. Second, we used Huang and Luk’s [17] daily policy uncertainty index, constructed from 114 general interest daily newspapers in mainland China, as an alternative data source. Lastly, we introduced a year dummy variable that equaled 1 for years after 2007 and 0 otherwise, capturing heightened EPU following the global financial crisis, a widely used robustness proxy of EPU [34].

Table 5. Robustness checks: Alternative measurements of economic policy uncertainty.

	(1)	(2)	(3)	(4)
Dependent Variable:	Agricultural Land Conversion
	Arithmetic mean of monthly policy uncertainty	Geometric mean of daily policy uncertainty	Arithmetic mean of daily policy uncertainty	Global Financial Crisis
Economic policy uncertainty	0.2181 ***	0.1633 ***	0.1615 ***	0.6975 ***
	(0.03)	(0.03)	(0.03)	(0.10)
City fixed effects	Yes	Yes	Yes	Yes
Adjusted R-squared	0.631	0.630	0.630	0.633
Observations	3492	3492	3492	3492

Notes: Robust standard errors in parentheses are clustered at the city level. All regressions employ OLS and include the same covariates as in Table 2. *** denote statistical significance at the and 1% level.

presents the results using these alternative measures. Although the magnitudes of the coefficient on EPU vary depending on the measure used, the coefficients remain positive and statistically significant at the 1% level across all specifications. These results reinforce the baseline finding that higher EPU drives agricultural land conversion.

4.3.3. Other Robustness Checks

We performed additional robustness checks on the baseline effect of EPU, with the results reported in Appendix A. Specifically,

Table A2. Robustness checks: Addressing outliers in the baseline measurement of agricultural land conversion.

	(1)	(2)	(3)	(4)
Dependent Variable:	Winsorized Agricultural Land Conversion
Two-tailed winsorized at:	0.5% level		1% level
	OLS	IV	OLS	IV
Economic policy uncertainty	0.2101 ***	6.0487 ***	0.2031 ***	6.0351 ***
	(0.03)	(1.10)	(0.03)	(1.10)
City fixed effects	Yes	Yes	Yes	Yes
Observations	3492	3493	3492	3493
First-stage F-statistic		30.89		30.89

Notes: Robust standard errors in parentheses are clustered at the city level. The first-stage F-statistics correspond to the Kleibergen–Paap rank F-statistics. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

and

Table A3. Robustness checks: Addressing outliers in the baseline measurement of economic policy uncertainty.

	(1)	(2)	(3)	(4)
Dependent Variable:	Agricultural Land Conversion
Two-tailed winsorization at:	0.5% level		1% level
	OLS	IV	OLS	IV
Winsorized policy uncertainty	0.2220 ***	6.0263 ***	0.2220 ***	6.0263 ***
	(0.03)	(1.10)	(0.03)	(1.10)
City fixed effects	Yes	Yes	Yes	Yes
Observations	3492	3493	3492	3493
First-stage F-statistic		30.89		30.89

Notes: Robust standard errors in parentheses are clustered at the city level. The first-stage F-statistics correspond to the Kleibergen–Paap rank F-statistics. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

address potential outlier influence by winsorizing the baseline measures of agricultural land conversion and EPU at the 0.5% and 1% tails, respectively.

Table A4. Robustness checks: Addressing time trends in the baseline measurement of economic policy uncertainty.

	(1)	(2)	(3)	(4)
Dependent Variable:	Agricultural Land Conversion
Detrended using:	Linear regression		Hodrick-Prescott filter
	OLS	IV	OLS	IV
Detrended policy uncertainty	0.1386 ***	3.3718 ***	0.5890 ***	2.6510 ***
	(0.02)	(0.50)	(0.07)	(0.23)
City fixed effects	Yes	Yes	Yes	Yes
Observations	3492	3493	3492	3493
First-stage F-statistic		52.71		289.13

Notes: Robust standard errors in parentheses are clustered at the city level. The first-stage F-statistics correspond to the Kleibergen–Paap rank F-statistics. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

removes time trends in EPU using both linear detrending and the Hodrick-Prescott filter.

Table A5. Robustness checks: Alternative approaches to standard error adjustment.

	(1)	(2)	(3)	(4)
Dependent Variable:	Agricultural Land Conversion
SE adjusted by:	Heteroscedasticity-robust		Bootstrap
	OLS	IV	OLS	IV
Economic policy uncertainty	0.2220 ***	6.0263 ***	0.2220 ***	6.0263 ***
	(0.0316)	(1.4489)	(0.0313)	(1.3207)
City fixed effects	Yes	Yes	Yes	Yes
Observations	3492	3493	3493	3493

Notes: Robust standard errors in parentheses are obtained by different methods. Bootstrap standard errors are obtained from 50 replications, with the random seed set to 1. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

implements alternative standard error adjustments, including heteroskedasticity-robust and bootstrap standard errors. Across all specifications, the results consistently support the robustness of the baseline findings.

5. Mechanisms

After confirming the positive impact of EPU on agricultural land conversion, this section explores the underlying mechanisms driving this effect. As discussed in Section 2, local governments’ pressures related to fiscal expenditure and economic growth target management may be the key mechanisms driving the increased agricultural land conversion associated with EPU, as proposed in Hypothesis 2.

To empirically identify these mechanisms, we first quantified the fiscal pressure faced by prefecture-level governments. Fiscal pressure was measured as the natural logarithm of public budget expenditure. The fiscal data for prefecture-level governments were obtained from the China Urban Statistical Yearbook and the National Fiscal Statistics of Prefectures, Cities, and Counties. We also assessed economic growth target pressure by calculating the difference between local economic growth targets set at the beginning of the year and the realized national real GDP growth rate. Data on prefecture-level economic growth targets were manually collected from the Government Work Reports of these governments. Next, we estimated the average effect of EPU on these mechanism variables by following the specification in Equation (2) and performing two-stage least squares regressions, using the U.S. EPU from the previous year as an instrumental variable to enhance the validity of our results.

Table 6. Mechanism analysis: local pressures on fiscal expenditure and economic growth target management.

	(1)	(2)	(3)	(4)
Dependent Variable:	Fiscal Expenditure Pressure		Economic Growth Pressure
	OLS	IV	OLS	IV
Economic policy uncertainty	0.1131 ***	1.1492 ***	0.4375 ***	18.4078 ***
	(0.01)	(0.29)	(0.08)	(3.43)
City fixed effects	Yes	Yes	Yes	Yes
Observations	3624	3625	3529	3530
First-stage F statistic		12.70		30.26

Notes: Robust standard errors in parentheses are clustered at the city level. The first-stage F-statistics correspond to the Kleibergen–Paap rank F-statistics. All regressions include the same covariates as in Table 2. *** denote statistical significance at the 1% level.

presents the estimation results of the mechanism analyses. As shown in Columns (1) and (2), the coefficients on EPU are positive and statistically significant at the 1% level, indicating that EPU significantly increases fiscal expenditure pressure on prefecture-level governments, thereby empirically supporting fiscal expenditure pressure as an important mechanism. In Columns (3) and (4), the results also identify the economic growth pressure mechanism, as the coefficients of EPU in the regressions on prefecture-level economic growth pressure are positive and significant at the 1% level.

In summary, our findings suggest that the observed pattern in the data aligns most closely with the explanation that EPU heightens local governments’ pressures related to fiscal expenditure and economic growth target management, which in turn drives the conversion of agricultural land to urban uses as a way to alleviate these political and economic pressures.

6. Further Discussion

In this section, we further explore whether all prefecture-level cities are equally affected by EPU and examine potential policy measures to mitigate its impact on agricultural land conversion. Specifically, we first analyze the heterogeneous effects based on prefectures’ land finance reliance and fixed asset investment dependence, offering deeper insights into how different economic contexts shape these impacts. We then assess the moderating roles of agricultural land quotas and marketization policies, highlighting potential policy implications for provincial governments in preserving agricultural land.

6.1. Heterogeneity Analysis

The first source of heterogeneity may arise from prefecture-level governments’ reliance on land finance, as EPU increases fiscal expenditure pressure on local governments, prompting greater agricultural land conversion for urban use. However, the different impact of EPU on agricultural land conversion across different levels of land finance reliance remained ambiguous. On the one hand, higher land finance dependence may drive local governments to use land-based fiscal tools more frequently, amplifying the effect of EPU on agricultural land conversion (“Frequency effect”). On the other hand, pro-industrial land supply strategies often keep industrial land prices lower than those of commercial and residential land [52], giving the latter a comparative advantage in alleviating fiscal stress. Consequently, under fiscal pressure from EPU, governments with higher land finance dependence may adjust their land supply structure by substituting industrial land for commercial and residential land, potentially reducing overall agricultural land conversion (“Structure effect”).

To explore this heterogeneity, we followed Fan, Qiu, and Su [53] in defining land finance dependence as the ratio of land transfer fees to total revenue from land transfer fees and general budgetary revenue. The sample was then divided into high and low land finance reliance groups based on quantiles. We estimated the average impact of EPU on agricultural land conversion separately for each subgroup. Columns (1) and (2) of

Table 7. Heterogeneous effects by land finance, investment dependence, and geographical regions.

	(1)	(2)	(3)	(4)	(5)	(6)
Dependent Variable:	Agricultural Land Conversion
Grouped by:	Land finance dependence		Investment dependence		Geographical regions
	Low	High	Low	High	Eastern China	Non-eastern China
Economic policy uncertainty	0.1894 ***	0.1657 ***	0.0136	0.3046 ***	0.1502 *	0.2971 ***
	(0.05)	(0.05)	(0.06)	(0.08)	(0.06)	(0.04)
City fixed effects	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R-squared	0.621	0.644	0.634	0.653	0.574	0.641
Observations	1680	1770	1696	1775	1139	2353

Notes: Robust standard errors in parentheses are clustered at the city level. All regressions employ OLS and include the same covariates as in Table 2. * and *** denote statistical significance at the 10% and 1% levels, respectively.

present the results. As expected, the coefficients on EPU are positive and significant at the 1% level, with similar magnitudes across both groups. This suggests that the average effect of EPU on agricultural land conversion does not exhibit significant heterogeneity, as the frequency effect and structure effect offset each other.

The second source of heterogeneity may stem from local governments’ reliance on fixed asset investment for economic development, as economic policy uncertainty can dampen local economic growth. Chinese local governments often use land incentives to attract enterprises and stimulate fixed asset investment, thereby driving economic expansion [33]. Consequently, the extent to which agricultural land conversion contributes to economic growth likely depends on the importance of fixed asset investment in a region’s development. This suggests that the positive impact of EPU on agricultural land conversion may be more pronounced in prefecture-level cities with a high reliance on fixed asset investment, as reflected by its substantial share in local GDP.

To empirically test this heterogeneity, we divided the sample into high and low fixed asset investment dependence groups based on the quantiles of prefectures’ fixed asset investment share in GDP. The subsample estimation results are presented in Columns (3) and (4) of Table 7. As expected, the coefficient on EPU is statistically significant at the 1% level only in the high investment dependence group. Moreover, in this group, the magnitude of the EPU coefficient is 30 times larger than in the low investment dependence group, while all coefficients remain positive. This highlights a strong cross-sectional variation in the impact of EPU on agricultural land conversion, driven by differences in economic reliance on fixed asset investment.

The geographical location of a city may also influence its land conversion in response to EPU. This effect is particularly evident in the differences between cities in eastern and non-eastern regions of China. Compared to other regions, the eastern region has a more diversified economic structure. As a result, when confronted with fiscal and growth pressures arising from EPU, local governments in eastern China tend to rely more on non-land instruments. Moreover, the proportion of land that has already been developed is higher in eastern China, and the central government imposes stricter agricultural land conversion policies in this region. Even when local governments in the east respond to pressure by converting land for non-agricultural uses, the scale of such land conversion may be relatively limited. These factors may lead to cities in eastern China being less affected by EPU in terms of land conversion decisions than cities in other regions, even when facing EPU shocks of the same magnitude.

In Columns (5) and (6) of Table 7, we test this hypothesis regarding regional heterogeneity by dividing the sample into cities located in eastern China and those in non-eastern regions. The results show that although the EPU coefficients are positive for both groups, both the significance level and magnitude of the EPU coefficient for eastern cities are markedly smaller than those for non-eastern cities. In particular, the EPU effect for eastern cities is not statistically significant at the 5% level. These findings suggest that the impact of EPU on land conversion in eastern cities is relatively smaller, which was consistent with our theoretical expectations.

6.2. Policy Analysis

Given the critical importance of agricultural land conservation for the ecological environment, it is essential for policymakers to mitigate the impact of EPU on agricultural land loss. Therefore, in this section, we explore potential policy options. As discussed in Section 2, policies related to agricultural land conversion quotas and marketization reforms may help to reduce the effects of EPU on the scale of agricultural land conversion. According to Hypothesis 3, lower land conversion quotas or greater marketization reforms may lead to a smaller increase in agricultural land conversion for urban residential, industrial, and commercial uses induced by EPU.

To assess the effectiveness of these policies, we estimated their moderating effects based on the specification in Equation (3). Since exact data on agricultural land conversion quotas were unavailable, we used the natural logarithm of the provincial areas of agricultural land conversion approved by the State Council, sourced from the China Land and Resources Statistical Yearbook, to proxy provincial agricultural land quotas. This proxy was reasonable, as the State Council approves agricultural land conversion at the provincial level based on land conversion quotas stipulated by the Land Administration Law. To measure the degree of provincial marketization reforms, we used the China Marketization Index from the National Economic Research Institute in Beijing.

Figure 6 visually presents the estimated moderating effects of land conversion quotas and marketization reforms on the relationship between EPU and agricultural land conversion. Panel A examines the role of provincial land conversion quotas, showing that the coefficients of EPU in the regression on agricultural land conversion increase with provincial land quotas. This suggests that reducing agricultural land conversion quotas can effectively limit the expansion of agricultural land conversion driven by EPU. Panel B investigates the moderating effect of provincial marketization reforms. It reveals a negative marginal effect, indicating that greater marketization can reduce the impact of EPU on agricultural land conversion.

These findings provide empirical evidence for Hypothesis 3 and highlight the institutional mechanisms of land quotas and marketization reforms in shaping agricultural land conversion amid EPU. Furthermore, they offer important policy insights for the central government in China and other developing and transitioning economies facing similar challenges.

7. Conclusions

This study investigated the impact of EPU on the scale of agricultural land conversion. Using the data from Chinese 270 prefecture-level cities and the China Economic Policy Uncertainty Index from Huang and Luk [17], we provide strong evidence on the positive impact of EPU on the area of agricultural land conversion for urban uses. On average, a one-standard-deviation increase in EPU leads to a 22.2% increase in the area of agricultural land converted for industrial, commercial, and residential purposes.

Our mechanism analysis revealed that this land conversion effect is primarily driven by local governments’ pressures related to increased fiscal expenditure and challenges in managing economic growth targets. When considering city-level heterogeneity, the effect is particularly pronounced in cities located in non-eastern China and highly dependent on fixed asset investment for economic development. Furthermore, our policy analysis suggested that provincial policies related to agricultural land conversion quotas and marketization reforms may help to mitigate the effects of EPU on the scale of agricultural land conversion.

Unlike previous studies, these findings highlight how local governments respond to EPU through a political economy lens, uncovering a novel and significant driver for agricultural land conversion to urban uses. This study bridges and expands three streams of the existing literature. First, it contributes to the growing body of research on the adverse effects of overall EPU on domestic economies [7,9,15,16,54]. Unlike previous studies that focused on corporate investment, financing decisions, or national-level macroeconomic and financial conditions, our work specifically addresses the relatively understudied impact of policy uncertainty on local governments’ resource allocation in China [34,55]. By examining how local governments respond to EPU through agricultural land conversion, we offer a unique ecological perspective on the consequences of EPU in developing countries. Second, we contribute to the literature on the socioeconomic drivers of agricultural land conversion [56,57,58] by emphasizing the critical role of uncertainty. Unlike Tegene, Wiebe, and Kuhn [59], who focused on uncertainty in the private land market, we examine uncertainty within the macroeconomic policy environment. Additionally, we make a marginal contribution to the political economy of land use in China [29,50] by demonstrating how fiscal pressures on local governments and economic growth target management translate policy uncertainty into agricultural land conversion. Notably, we differentiate between land use changes driven by the political and economic pressures on local governments resulting from EPU and those induced by other institutional constraints within the Chinese context.

The findings in this study also carry significant policy implications for agricultural land conservation. First, in the face of increased agricultural land conversion driven by EPU, this study suggests that central governments concerned with agricultural land conservation should prioritize maintaining the stability and transparency of macroeconomic policies. Second, even though EPU may be inevitable, central governments can protect agricultural land by reducing provincial quotas on land conversion or promoting marketization reforms, such as minimizing administrative market interventions and improving property rights protection. However, the design of agricultural land quotas should be handled cautiously to avoid the side effect of resource misallocation [40]. Finally, when designing pilot projects for agricultural land conservation, central governments should pay particular attention to cities in the economic developed region and those with a high economic dependence on fixed asset investment.

Although this study focuses on China, its findings are also relevant to other economies. EPU is a global phenomenon that has produced varying negative impacts on the economies and public finances of many countries [13,15,17]. Over recent decades, the decentralization of public finance and land use planning has also become widespread [60,61], with numerous local governments employing land policies to achieve fiscal objectives. For example, local governments in Germany and the Netherlands have actively engaged in land development and leasing to increase fiscal revenues [62]. While local governments in other democratic systems differ from China’s politically centralized model and face less explicit external pressure for economic growth through promotion or assessment, they are nonetheless subject to internal growth pressures due to fiscal autonomy and the strong link between fiscal revenues and local economic performance. Moreover, party competition in democratic elections often intensifies these incentives. In pursuit of electoral gains through local economic growth, some parties use land use policy tools to stimulate local development. For instance, from 2003 to 2007, Spain’s right-wing ruling party converted large areas of agricultural land to urban use to promote economic growth and increase the urban housing supply [63].

Additionally, in most of these countries, central governments retain significant control over key macroeconomic policy institutions, such as central banks and ministries of finance, while local governments often lack effective mechanisms for communication and coordination with the central authorities. This makes them more vulnerable to unpredictable national economic policy shocks, which can further exacerbate fiscal and economic growth pressures. Although direct evidence is currently limited, it is theoretically plausible that EPU significantly increases agricultural land conversion in these countries by heightening the fiscal and economic growth pressures faced by local governments. Therefore, the analytical framework, findings, and policy implications of this study offer a valuable foundation for future research in other countries.

Due to limited data availability, this study faced some constraints that future research should address. First, as recent city-level data on newly added construction land have not been released, our dataset only covered up to 2017. Consequently, we were unable to empirically analyze the effects of recent major events such as the COVID-19 pandemic or recent land use policy changes in China on EPU-induced agricultural land conversion. With more timely data, future studies could extend the analysis period and potentially uncover dynamic changes or additional mechanisms, thus enriching this research area.

Second, the absence of long-term, nationally representative data on agricultural land conversion across cities means that studies, including ours, rely on proxy variables. Following Fu, Xu, and Zhang [40], we used the supply of newly added construction land as a measure of agricultural land conversion. However, there is often a time gap between land supply and actual land use in China. Although land use regulations such as the Regulations on Idle Land Disposal aim to limit this gap to within one year, some annual data discrepancies remain inevitable due to non-compliance. Future research could improve measurement approaches, allowing for more accurate effect estimation.