We all observe that climate risks are bad for agriculture. However, we do not know whether climate risks can create complementary effect or substitution effect among different agricultural sectors. Some plants can sustain climate risks. For example, potatoes and peanuts are more resistant in drought weather, whereas vegetables are more fragile in flood. Some countries suffer from climate disasters more often. How do farmers response to climate shocks to adjust their agricultural production? How do countries and regions in different climate zones react to and trade during frequent climate disasters?
This paper is analytical, setting a simple empirical model to answer these questions. Building on the features of climate change and agricultural inputs with sectoral differences, we establish some prepositions to diagnose the climate-agriculture-trade match. We frame climate risks and trade in agriculture with regional heterogeneity, sectoral variety, and partner country. The climate-agriculture-trade match creates diverse results, especially when we consider different climate risks with different agricultural products. Agricultural policy is recommended with climate related trade policy.
The empirical model we extend has a number of ingredients. The first being that it has various climate risks, instead of a simple temperature. We split these “risks” and employ traditional trade models to describe them. The second is agricultural inputs; land, labor and water are the central drivers for trade in agri-food.
What can we obtain? First, we reaffirm that climate risks are bad for agriculture, while trade in agriculture and food can offset negative effects by climate risks. It is more precise to get the type of climate risks related to trade patterns in agri-food. Second, we specify climate risks in different regions and trade in such regions. The droughts have a significant effect on the international trade in agri-food in the north, while sunshine duration presents opposite effects on the agri-food trade of the north and south.
We analyse effects of different climate risks on trade in agri-food from the perspective of sectoral heterogeneity, regional differentiation, and partner countries. By doing this, we can observe agri-food sector related climate risk and provide a specific policy to release negative damage. Learning by trading can be obtained from partner countries.
The contribution of the study is that we identify climate risks with agri-food trade at a provincial and sectional level in China. Considering the complex weather differences among different provinces in China, identifying specific climate risk with agricultural trade can benefit China’s agricultural trade policy and agricultural production policy. Our empirical model is stylized in its climate-agriculture-trade structure. More generally, we find multifaceted effects of climate risk on China’s agricultural trade and its underlying mechanisms. Climate risk significantly influences China’s agricultural trade. Specifically, droughts and floods reduce international trade in provincial agricultural products in China.
The structure of this paper is organized as follows. In the next section we provide a review of the relevant literature. Then, we describe our research methodology and discuss the properties of the dataset. Subsequently, we report and interpret our estimation results. In section 5, we provide a diverse discussion for different agricultural sectors and in different regions. The final section summarizes and concludes.
Climate change has a clear effect on agriculture production. Fankhauser (2017) argues that humans have different adaptation actions responding to climate change. Malpede and Percoco (2024) prove that global warming can affect Human Development Index. To be more precise, Calzadilla et al. (2013) and Ignaciuk and Mason-D’Croz (2014) prove that climate change impacts global agricultural production. Ahvo et al. (2023) prove that climate change in fertilizer shocks leads to decrease of crop yields. Xie et al. (2020) argue that climate change has different effects on different crops production in China. Dumortier et al. (2021) report that climate change expands wheat production, while rice production expands due to CO2 fertilization. Burke and Emerick (2016) find that extreme heat decreases productivity of corn and soy in the USA. Wimmer et al. (2024) argue that both expected and actual weather determines crop production. Clearly, climate change has a significant effect on agricultural production, and the difference of the effects relies on the sectoral level.
The mechanism of climate’s effect can extend to unemployment rate, since the climate causes the output of agricultural production. For example, Bareille and Chakir (2023) used the repeat-Ricardian method to find that warmer summers can benefit French agriculture. Later, Bareille and Chakir (2024) also found that farmers can adjust agriculture production in response to climate change. Liu and Lin (2023) argue that global warming increases the unemployment rate via the inflation rate, agricultural production and urbanization. In terms of theoretical methodology, the cost of climate change on agriculture is often underestimated when using the supply-side approach, due to neglect of imperfect substitution of crops (Gouel, 2025).
Climate change, international trade and agriculture have close relationships. Reilly et al. (1994) and Gouel and Laborde (2018) also argue that international trade is a moderating role used to make the effect of climate change less severe. Similarly, Brenton et al. (2022) found that climate turbulence leads to food insecurity while trade in agriculture can reduce the insecurity. Likewise, Reilly and Hohmann (1993) show that the negative consequences from climate change can be released by interregional adjustment of agricultural production. Khan et al. (2019) criticize that climate change causes exports in agriculture to decline. In the context of country specific, Ahmed et al. (2012) find that Tanzania’s trading opportunity is enhanced when its trading partners experience dry conditions and reduce agriculture production. Kashem et al. (2024) find that environmental pollution decreases Australian agricultural export. Some existing literature investigates climate change at a country level. Yu et al. (2020) found that precipitation and temperature in Kazakhstan increases exports in wheat and rice and imports in maize. For instance, Dall’erba et al. (2021) argue that crop growers rely on local and trade partner’s weather. International trade can play a role on the effect of climate change on agriculture, because international trade can increase productivity loss from climate change (Huang et al., 2011). Dingel et al. (2023) reported that highly productive countries benefit bigger gains from trade because of spatial correlation of cereal production. Prior literature concludes that international trade can benefit agriculture loss caused by climate change.
Literature extends trade and agriculture to sustainable development. Drabo (2017) shows that the export in primary agricultural products increases greenhouse gas emissions. Porfirio et al. (2018) argue that CO2 emissions make agricultural trade agglomerate in few markets. Moreover, Candau et al. (2022) found that agricultural exports are less sensitive to water conditions if the exports are from vulnerable countries. Foong et al. (2022) find that certain countries emissions are largely dependent on consumption patterns and emission intensities of agricultural products, relative to their trading partners. Costinot et al. (2016) show that climate change causes global GDP to reduce by 26% with the consideration of production and trade pattern. Janssens et al. (2020) argue that climate change has caused an increase of 55 million undernourished people. Pothen and Hübler (2018) conclude that the reduction of non-tariff barriers contributes to larger welfare gains in Europe, while non-tariff barriers removal can reduce global CO2 emission. Therefore, existing studies already summarize the effect of trade in agriculture on sustainable development.
In terms of agri-food, climate change also plays an important role. Bozzola et al. (2023) show that climate change determines trade patterns in agri-food trade. Lobell et al. (2011) claim that the negative effect of climate change on food is larger than the positive effect of technology and carbon dioxide fertilization. Mosnier et al. (2014) report that production and trade can reduce the damage of food availability by climate change.
In the context of China, Zhao et al. (2021) argue that China should keep sustainable international trade to ensure China’s food consumption restricted by negative environmental impacts. Wu et al. (2021) prove that 1.5°C global warming will dampen China’s rice production through trade. Porfirio et al. (2018) argue China’s export increase is due to the improvement of agricultural productivity under high carbon emissions. China’s sectoral virtual water scarcity risk can transmit through international trade to distant countries (Zhao et al., 2019). Liu et al. (2025) find that reducing domestic trade cost is more conductive to increase food production than reducing import trade costs in China. Wang et al. (2024) argue that agricultural trade can reduce carbon emissions by scale, technological and structural effects. Kuai et al. (2024) find virtual water trade can lead to agricultural trade distortion in seven key provinces in China. The existing literature on China’s agricultural trade and climate mainly focus on the specific agricultural product and general climate risk. Our paper focuses more on various climate risks in different regions of China. However, we can also observe that existing literature pays close attention to the relationship between water usage and agricultural trade in China. This also aids our study to consider agricultural water consumption in our empirical study.
In summary, the existing literature has well documented the relationship between climate and agricultural production as well as agricultural trade. Trade in agriculture can also affect emissions. However, the prior studies have less focus on the specific type of climate risk and agricultural trade. In contrast to the existing literature, our paper focuses on the specific climate risk, including flood, drought and sunshine duration. Climate’s effect on agricultural production is clear and neat, but it is vague in China. In our paper, China’s diverse climate environment provides a rich database for us to diagnose different climate channels on trade in agriculture. To be more specific, our paper investigates the role of specific climate risks on agri-food trade in China.
To investigate the impact of climate risk on the total trade of agricultural products in China, we construct an empirical model to study the relationship between climate risk and China’s agri-food trade. The agri-food products include all agriculture, vegetable and food related commodities. Our empirical methodology is based on Gouel (2025). The empirical model is constructed based on two types of independent variables, including climate related variables and production input related variables. To examine the effect of climate risk on China’s agri-food trade, considering the facts of a significant number of zero values in the sample data, the Ordinary Least Squares (OLS) method cannot effectively address the endogeneity issue. Therefore, the Poisson Pseudo Maximum Likelihood Estimation with High Dimensional Fixed Effects (PPMLHDFE) can effectively correct the issue of zero values in the dependent variable. Therefore, this study employs the PPMLHDFE model. The regression model is shown in equation (1):
This paper takes agricultural trade, yijt, as the dependent variable. Given that the total import and export value of agricultural products between each province and its trading partners is the most direct indicator of agricultural trade, we use the total import and export value of agricultural products between each province and its trading partners to measure agricultural trade.
Climate risks can be categorized into physical risks, institutional risks, production risks, etc. Temperature is greatly influenced by geographical location, and the manifestation of climate warming is shown by the increase in local temperature relative to the historical average temperature of that area. Therefore, existing literature uses temperature anomalies (or standardized temperatures) to measure the climate risk. Different from existing literature, this paper determines a series of indicators to describe climate risks by reviewing the classification and descriptions of climate risks in relevant literature.
The core explanatory variables for climate risks include the proportion of area affected by drought (Mu & Khan, 2009; Zhou et al., 2017), the proportion of area affected by flood, the area of effective irrigation and the cumulative amount of sunshine. Other explanatory variables include: the number of urban employees in agriculture, forestry, animal husbandry, and fishery, the total mechanical power per Mu of land in agriculture, the expenditure on agriculture, forestry, and water affairs by local finance, and the water usage in agriculture (Zhao et al., 2019; Mu & Khan, 2009).
We construct a province-country-year panel data. The scope of the traded commodities in our model covers agriculture and food products. Our dataset includes trade sections of animal, vegetable, microbial fats, foodstuffs, beverages, alcohol and vinegar, tobacco, wool and cotton. This paper selects the period from 2015 to 2019 as the sample interval. It focuses on the 31 provinces and municipalities of China, as well as the countries engaged in agriculture and food trade with them; we get a total of 29,951 sample observations. The selection of provinces and municipalities excludes the two Special Administrative Regions of Hong Kong and Macau, as well as Taiwan, China. The choice of years is due to the availability of many explanatory variables starting from 2015 and ending in 2019.
Drawing on the characteristics of agriculture and food products and summarizing previous research, the dependent variable, which involves the total import and export trade of agriculture, is sourced from the General Administration of Customs of the People’s Republic of China. The core explanatory variables and other related data information come from the National Bureau of Statistics. Additionally, the control variables included in this study are primarily obtained from the World Bank database.
Based on the regression results from the PPMLHDFE as presented in Table 1, from the perspective of climate risk, the impact of drought disaster occurrence (DRO) on China’s agri-food trade is statistically significant and shows a negative sign. It indicates that the more area affected by drought disasters, the lower the trade volume of agricultural products in China. The area affected by drought disasters, serving as an important proxy variable for climate risk, suggests that higher climate risk correlates with lower trade volumes of agri-food products in China. Higher climate risk due to greater drought disaster areas leads to lower agricultural output (Mu & Khan, 2009; Zhou et al., 2017). Consequently, it causes lower trade volumes in agri-food products. There are two possible reasons for such result. The first reason is that lower agricultural output decreases exports, as stated in existing literature (Chen & Gong, 2021; Zhang et al., 2015). Traditional trade theory clearly explains that when export country has less exportable products available, the export value will decrease due to the price increase. The second reason is that higher drought disaster will cause less demand for intermediate inputs in agricultural production. Our finding can also respond to exiting literature. For example, Chen and Gong (2021) also find that extreme heat will reduce China’s input utilization in agriculture. The combined two effects contribute to the trade drop due to drought disasters.
Baseline estimation for effect of climate risk on China’s agricultural trade
| (1) | (2) | |
|---|---|---|
| Index | Total trade | |
| DRO | Drought-affected area | −0.0800** |
| (0.0403) | ||
| FLO | Flood-affected area | −0.3100*** |
| (0.0277) | ||
| SH | Cumulative sunshine duration | −0.0440 |
| (0.1716) | ||
| EPU | Employment in agriculture, forestry, animal husbandry, and fishery in urban areas | 0.2907*** |
| (0.1044) | ||
| PU | Pesticide usage | 0.6976*** |
| (0.1042) | ||
| TAC | Total power of agricultural machinery/Total sown area | 2.1882*** |
| (0.1775) | ||
| LAF | Local financial expenditure on agriculture, forestry, and water affairs | 1.3497*** |
| (0.2891) | ||
| TWC | Agricultural water consumption | −0.2335 |
| (0.2291) | ||
| IAP | Proportion of irrigated land in total agricultural land | 0.2363 |
| (0.5894) | ||
| AAP | Proportion of agricultural land in total land area | 1.4999 |
| (3.0269) | ||
| MIP | Proportion of agricultural raw material imports in total commodity imports | −0.0427 |
| (0.3603) | ||
| Constant | 1.6692 | |
| (11.3001) | ||
| Year effect | Y | |
| Country effect | Y | |
| No. of Obs. | 4,140 |
Note: Standard errors in parentheses
p<0.01,
p<0.05,
p<0.1.
All variables are in log values.
The impact of flood disaster occurrence (FLO) on China’s agri-food trade is also statistically significant and shows a negative sign. It indicates that the higher the accumulated flood disaster areas, the lower the trade volume of agricultural products between the province in China and partner countries. Flood disaster areas, as another important proxy variable for climate risk, also indicate that higher climate risk correlates with lower trade volumes of agri-food products in China. The same mechanism as DRO can explain the negative relationship between flood disaster and agricultural trade.
The cumulative sunshine duration (SH) does not significantly impact China’s agricultural trade, showing that it does not affect the trade of agri-food products. As we can observe, the sunshine duration can hardly affect agricultural trade between China’s provinces and partner countries.
The impact of climate risk on crop production is multifaceted. The current global trend of rising temperatures and other climate risks has a negative effect on China’s agri-food trade. Global climate risks will exacerbate the production and trade environment for agricultural products, reducing China’s trade in these commodities. Climate changes, such as variations in precipitation and temperature, lead to increased climate risk, which directly affects agricultural economic output rather than indirectly through impacts on inputs. It also indirectly affects China’s international trade in agricultural products by increasing the severity of natural disasters.
From the perspective of agricultural production inputs, employment in agriculture, forestry, animal husbandry, and fishery in urban areas (EPU) is an indirect indicator of agricultural labor input. This significantly and positively impacts China’s agricultural trade. It means that the more employment in these sectors, the greater the trade volume of agricultural products in China. Pesticide usage (PU) is statistically significant and shows a positive sign. It means the higher pesticide usage in agricultural production, the higher trade in agriculture between provinces in China and partner countries.
The total power of agricultural machinery used per unit of sown area (TAC) and local financial expenditure on agriculture, forestry, and water affairs (LAF) also significantly and positively impact China’s agricultural trade. The enhancement of total agricultural machinery power generally shows a significant promotional effect. While other factors depend on the type of agricultural product and demonstrate differentiated characteristics. Other variables related to agricultural input, such as agricultural water consumption, the proportion of irrigated land to total agricultural land, the proportion of agricultural land to total land area, and the proportion of agricultural raw material imports to total commodity imports, do not significantly impact China’s agricultural trade; their changes have not produced a significant effect on China’s trade in agricultural products. Overall, climate change brought about by climate risk plays a significant role in China’s trade of agricultural products.
To investigate whether the climate risks have different effects on different types of agri-food trade, we split agri-food trade into different sectors. Since different agricultural products have different response from climate change, we can observe the consequences of such differences.
The definitions of agriculture and food products fully follow HS Nomenclature. We first examine the definitions of HS products according to HS Nomenclature. The HS codes which are closely related to agricultural productions are selected. Hence, animal and vegetable related products are selected. It is common sense that food products are closely related to agricultural production. However, mineral products are excluded since they are less likely to be affected by climate. Though textiles and textile articles are typically manufacturing products, we still select animal hair and cotton in our study scope. We can observe that animal hair, such as wool, is closely related to animal breeding. We can observe that cotton plants highly rely on weather and climate factors. Based on the reasons and analysis above, we categorize agricultural products into five major sectors based on the WTO Agreement on Agriculture and China’s customs statistical directory. The first category includes live animals and animal products (HS Chapters 01–05); the second category comprises plant products (HS Chapters 06–14); the third category encompasses animal and vegetable fats and oils and their cleavage products, refined edible fats and oils, and waxes of animal or vegetable origin (HS Chapter 15); the fourth category consists of foodstuffs, beverages, alcohol and vinegar, tobacco and tobacco substitutes (HS Chapters 16–24); the fifth category includes wool and its products, as well as cotton (HS Chapters 51–53).
Table 2 presents the regression results of agricultural trade across the five categories. Among columns (1)–(5), DRO is statistically significant and shows a negative sign for category IV in column (4) only. It implies that drought only affects foodstuffs, beverages, alcohol and vinegar related products trade at provincial level. The rest of the agricultural products trade are less affected by drought at the provincial level.
Effect of climate risk on China’s agricultural trade at sectoral level
| (1) | (2) | (3) | (4) | (5) | |
|---|---|---|---|---|---|
| Category I | Category II | Category III | Category IV | Category V | |
| DRO | −0.0786 | −0.0697 | −0.2226 | −0.0942** | −0.0706 |
| (0.0587) | (0.0549) | (0.0000) | (0.0466) | (0.0806) | |
| FLO | −0.3907*** | −0.3443*** | −0.2887 | −0.2479*** | −0.2940*** |
| (0.0550) | (0.0350) | (0.0000) | (0.0267) | (0.0433) | |
| SH | −0.0479 | 0.2504 | −1.1799 | −0.1964* | −0.6297** |
| (0.3049) | (0.2813) | (0.0000) | (0.1136) | (0.2789) | |
| EPU | 0.7497*** | 0.3677*** | 0.6657 | 0.1412 | −0.1530 |
| (0.2036) | (0.1236) | (0.0000) | (0.1003) | (0.1544) | |
| PU | 2.2594*** | 0.7417*** | 0.1826 | 0.7024*** | 0.1379 |
| (0.2693) | (0.1423) | (0.0000) | (0.0938) | (0.1943) | |
| TAC | 3.2962*** | 2.1011*** | 3.2033 | 1.8905*** | 2.3177*** |
| (0.3530) | (0.2199) | (0.0000) | (0.2110) | (0.3285) | |
| LAF | 0.4537 | 1.3033*** | 4.3430 | 0.6510** | 2.1729*** |
| (0.3912) | (0.4634) | (0.0000) | (0.2642) | (0.6257) | |
| TWC | −1.3146*** | −0.6922** | 0.1442 | −0.2971 | 1.4083*** |
| (0.3811) | (0.3502) | (0.0000) | (0.2032) | (0.3025) | |
| IAP | −0.1920 | 1.1049 | 0.2499 | 0.1314 | 1.2283 |
| (0.9428) | (1.2091) | (0.0000) | (0.6457) | (1.6215) | |
| AAP | 5.3175 | 3.2108 | 0.9318 | 0.5081 | 2.7678 |
| (3.7666) | (4.2910) | (0.0000) | (2.2753) | (5.3678) | |
| MIP | 0.4295 | 0.2087 | −0.6172 | 0.1137 | 0.0506 |
| (0.5357) | (0.6011) | (0.0000) | (0.3808) | (0.5592) | |
| Constant | −3.3414 | −10.1048 | −13.2491 | 11.8166 | −14.1921 |
| (15.1269) | (17.1195) | (0.0000) | (9.0853) | (21.6627) | |
| Year effect | Y | Y | Y | Y | Y |
| Country effect | Y | Y | Y | Y | Y |
| No. of Obs. | 4,073 | 4,140 | 3,834 | 4,140 | 4,140 |
Note: Standard errors in parentheses
p<0.01,
p<0.05,
p<0.1
In Table 2, FLO is statistically significant and shows negative signs for category I, II, IV and V. The different sign falls into category III. It implies that flood has little impact on animal and vegetable related products. It is not difficult to conclude that production conditions for animal and vegetable related products are usually far away from big rivers or lakes, and flood can hardly damage or create obstacles for the production and trade.
We can also observe different signs for SH. Different from the estimation results in Table 1, SH is statistically significant and shows negative signs for category IV and V. It means foodstuffs, alcohol and vinegar, and cotton are more sensitive to sunshine hours. Hence, we can find that SH is statistically insignificant, but it is not for specific trade in agri-food sectors.
Therefore, it can be observed that different climate risks have varying degrees of impact on the trade of different types of agricultural products. Drought has a notably adverse effect on the trade of category IV, which includes food, alcohol, and tobacco; floods have a notably adverse effect on the trade of all agricultural products except for fats and oils. Sunshine duration has a notably adverse effect on the trade of food, alcohol, tobacco, and wool and cotton.
The effect of climate risks on trade in agri-food is diverse on agricultural production. Our estimations above prove it. Climate risks, such as global warming, have negative effect on China’s agricultural trade. Climate changes, such as precipitation drop, temperature jump, causes output of agricultural output directly. It results in the decrease of land productivity. As a result, it leads to the reallocation of production input among different crops. Hence, the complementary effect within food sector is significant. We can also observe such result in existing literature. For instance, climate change threatens food availability and trade can adjust to that (Mosnier et al., 2014). More generally, food security can not only rely on domestic production (Brenton et al., 2022).
Between different agricultural sectors, the substitution effect is outstanding. Therefore, we conclude that different climate risks cause complementary effect or substitution effect among different agricultural sectors.
China’s vast territory and the climatic differences between the north and south, along with the differential conditions in agricultural production, lead to heterogeneous impacts of climate risks on agri-food trade. Based on the line of the Qinling Mountains and the Huai River, we divide Chinese provinces into northern and southern groups. The reason to differentiate regions using the Qinling Mountains and the Huai River is that they are the border between subtropical climate and temperate climate. The southern group includes Anhui, Jiangxi, Hubei, Hunan, Guangxi, Shanghai, Jiangsu, Zhejiang, Fujian, Guangdong, Hainan, Chongqing, Sichuan, Guizhou, and Yunnan; all other provinces are considered as the northern group.
Columns (1) and (2) of Table 3 show the impact of climate risks on agricultural trade in the northern and southern regions. The area affected by drought (DRO). It is statistically significant and shows a positive sign for northern group. It means that more DRO can increase agri-food trade in northern China. However, DRO has no significant impact on agri-food trade in the southern region. The larger the drought-affected area in the north, the greater the agri-food trade. This is because the larger the drought-affected area in the north, the more agricultural imports, and thus, the greater the agri-food trade. The southern region, predominantly characterized by paddy fields and abundant water resources, can increase agricultural production by expanding the effective irrigation area, making the impact of drought on agricultural production less noticeable in comparison.
Effect of climate risk on China’s agricultural trade at regional level
| (1) | (2) | |
|---|---|---|
| North | South | |
| DRO | 0.1369** | −0.0461 |
| (0.0550) | (0.0598) | |
| FLO | −0.3112*** | −0.1737*** |
| (0.0686) | (0.0343) | |
| SH | 3.2629*** | −0.8778*** |
| (0.4260) | (0.1121) | |
| EPU | 0.5623*** | 1.4958*** |
| (0.1605) | (0.1478) | |
| PU | 0.4094*** | −1.3611*** |
| (0.1581) | (0.2342) | |
| TAC | 3.1275*** | 5.8725*** |
| (0.3559) | (0.4437) | |
| LAF | 2.5950*** | 3.3587*** |
| (0.5734) | (0.6042) | |
| TWC | −1.1197*** | −0.4607 |
| (0.3634) | (0.4342) | |
| IAP | 0.2316 | 0.2133 |
| (0.6247) | (0.9165) | |
| AAP | 1.3686 | 1.2494 |
| (2.3309) | (3.5280) | |
| MIP | 0.1365 | −0.2050 |
| (0.4566) | (0.4721) | |
| Constant | −36.6075*** | 2.0184 |
| (10.3392) | (13.5854) | |
| Year effect | Y | Y |
| Country effect | Y | Y |
| Observations | 2,332 | 1,808 |
Note: Standard errors in parentheses
p<0.01,
p<0.05,
p<0.1
The area affected by flood (FLO) is statistically significant and shows a negative sign for both northern and southern regions. It indicates that the larger the flood-affected area, the smaller the agricultural trade in both northern and southern regions. That is because the larger the flood-affected area, the lower the agricultural output, and the fewer agri-food exports. Consequently, it results in less agri-food trade. The cumulative sunshine hours (SH) significantly affect agri-food trade in both the northern and southern regions but has a significantly positive impact on northern agri-food trade and a significantly negative impact on southern agri-food trade. It indicates a substantial difference in the structure of agri-food trade between the north and south.
Thus, SH has different effects in northern and southern provinces on their agri-food trade. The impacts of climate risks on these regions are marked differently. Apart from floods, droughts have a significant impact on the international trade in agri-food in north, while SH presents opposite effects on the agri-food trade of the north and south.
To further analyze the mechanism by which climate risks affect China’s agri-food trade, we continue to explore whether the effect of climate risks differs when different agricultural trade partners engage in trade with various provinces in China. Traditional international trade theory suggests that the intensity of production factors varies across countries. Factor intensity determines the patterns of international trade. Therefore, based on the World Bank’s classification of countries, we categorize China’s provincial agri-food trade partners into high-income countries and other countries. The reason to select a high-income country is because a high-income country usually manages high technology in agriculture. The high technology in agriculture can offset some negative effect by climate risks.
Table 4 presents the effect of climate risks on agricultural trade between China’s provinces and high-income countries, as well as non-high-income countries. Column (1) of Table 4 shows that area affected by drought (DRO) is statistically insignificant on agri-food trade between Chinese provinces and high-income countries. However, the area affected by flood (FLO) and cumulative sunshine hours (SH) are statistically significant and show a negative sign. It means more severe the floods and the longer the sunshine hours, the smaller the trade value with high-income countries. Comparing to the results in column (1) of Table 4, column (2) indicates that the area affected by drought (DRO) becomes statistically significant and shows a negative sign. While DRO is statistically insignificant in column (1). The area affected by FLO is statistically significant and shows a negative impact on agri-food trade between Chinese provinces and non-high-income countries. This is the same as column (1).
Effect of climate risk on China’s agricultural trade at income level
| (1) | (2) | |
|---|---|---|
| High income | Others | |
| DRO | −0.0635 | −0.1069** |
| (0.0629) | (0.0513) | |
| FLO | −0.3676*** | −0.2648*** |
| (0.0449) | (0.0261) | |
| SH | −0.3086* | 0.1391 |
| (0.1634) | (0.2547) | |
| CNTR | - | - |
| Year effect | Y | Y |
| Country effect | Y | Y |
| Observations | 1,777 | 2,363 |
Note: Standard errors in parentheses
p<0.01,
p<0.05,
p<0.1
It suggests that the more severe the droughts and floods, the lower the trade volume between Chinese provinces and non-high-income countries. Therefore, we can conclude the effects of the area impacted by drought (DRO) on agri-food trade between China and high-income countries, as well as non-high-income countries, are different. Droughts are more likely to affect agri-food trade between China and non-high-income countries. It also proves that the technology advantage in high income countries can offset the negative effect of droughts.
Due to the high sensitivity of agri-food trade to meteorological conditions, climate change places agricultural ecosystems under severe stress. Agriculture, as the largest water consumer, accounts for more than seventy percent of global water usage. Existing literature also implies that water availability, as well as water trade, is crucial for agricultural trade (Candau et al., 2022; Kuai et al., 2024; Zhao et al., 2019). Climate risk alters precipitation patterns and threats the stability of the entire agricultural system. Concurrently, the rapid changes in aquatic ecosystems caused by climate change exert pressure on agricultural water resources, especially in low-lying coastal areas and river delta regions where land degradation is severe, effective irrigation areas change, and agricultural productivity is affected. Therefore, the mechanism examines whether drought affects China’s agricultural trade through water resources using the PPMLHDFE model. Then, we test the interaction effect between climate risks and agricultural water usage.
Table 5 is the estimation results of interactive effect between climate risk and agricultural water consumption. The interaction term between the area affected by drought (DRO) and agricultural water usage (TWC) is statistically insignificant. It means that the area affected by drought (DRO) does not intensify or mitigate China’s agri-food trade through agricultural water usage. In contrast, the interaction term between the area affected by flood (FLO) and agricultural water usage (TWC) is statistically significant and shows a positive sign. It indicates that flood-affected areas reinforce China’s agri-food trade through agricultural water usage. This implies that flood-affected regions increase agricultural production and agri-food trade by reducing agricultural water usage. The interaction term between cumulative sunshine hours (SH) and agricultural water usage (TWC) is statistically significant and shows a positive sign. It means that cumulative sunshine hours reinforce China’s agri-food trade through agricultural water usage. It suggests that agricultural production areas adjust agricultural water usage based on the duration of sunlight to achieve optimal agricultural production output, thereby increasing agri-food trade.
Interactive effect of climate risk and agricultural water consumption
| (1) | (2) | (3) | |
|---|---|---|---|
| Total trade | Total trade | Total trade | |
| c. DRO#c. TWC | −0.0872 | ||
| (0.0584) | |||
| c.FLO#c.TWC | 0.1898** | ||
| (0.0742) | |||
| c.SH#c.TWC | 0.4494*** | ||
| (0.1129) | |||
| CNTR | - | - | - |
| Year effect | Y | Y | Y |
| Country effect | Y | Y | Y |
| Observations | 4,140 | 4,140 | 4,140 |
Note: Standard errors in parentheses
p<0.01,
p<0.05,
p<0.1
Thus, the effect of climate risks on agri-food trade can be intensified through water resources in production factors, aiming to achieve an optimal state of agricultural water usage.
This paper examines the impact of climate risk on China’s agri-food trade, delving into the mechanisms by which different climate risks at the provincial level affect trade in various types of agricultural products. The study provides an in-depth analysis of the multifaceted effects of climate risk on China’s agri-food trade and its underlying mechanisms. We conclude that climate risk significantly influences China’s agri-food trade. Specifically, droughts and floods suppress trade in provincial agricultural products in China. However, different types of agricultural products exhibit varying sensitivities to climate risk, which in turn affects trade differently. Droughts have a pronounced inhibitory effect on the trade of agricultural products such as food, beverages, and tobacco; floods do not impact the trade of animal and vegetable oils; and solar radiation hours only affect the trade of food, beverages, tobacco, and wool and cotton products. These differential impacts primarily operate through the agricultural water usage in different provinces of China. Agri-food imports in the northern regions increase following droughts; however, sunshine duration in the north favors agri-food trade, while the opposite is true for the south. Droughts significantly inhibit China’s agri-food trade with non-high-income countries but do not affect trade with high-income countries.
Based on the aforementioned research, this paper offers policy recommendations for policymakers to address climate risk, promote agri-food trade, and ensure food security. First, develop a diversified agricultural trade ecosystem, advocating for different agricultural production and trade between the north and south, and maintaining trade relations with various types of trade partners. Second, adopt different trade policies and response measures for different types of agricultural products facing different climate risks. Third, given that climate risk indirectly affects agricultural trade through natural disasters, it is essential to strengthen natural disaster risk management, including disseminating knowledge and skills for disaster prevention and reduction, enhancing farmers’ coping abilities, and establishing a robust risk monitoring and early warning system by the government to ensure the timeliness and effectiveness of disaster response. These measures collectively form a solid support for safeguarding national food security and promoting sustainable agricultural development.
The limitation of our study is that we are not able to incorporate partner countries features in our empirical model due to the availability of climate risks. By doing this, we need have similar climate risks that we measure at provincial levels in China. Unfortunately, similar measures of climate risks for partner countries are not available for us. Therefore, it will be valuable to include partner country’s climate risks in the model to constructure a more general framework. Besides, the direction for further research can concentrate on evaluating the effect of heterogenous climate-trade policy. It can improve the adaption of agricultural trade policy in different provinces in China.