Climate change and extreme heat effects on output in Vietnamese agriculture

Technical Summary: Climate Change and Extreme Heat Effects on Output in Vietnamese Agriculture

Problem Statement
Vietnamese agriculture has experienced significant growth in output volume over recent decades, accompanied by a parallel increase in the use of agrochemicals, specifically fertilizers and pesticides. However, rising global temperatures and the increasing frequency of extreme heat events pose a critical threat to this trajectory. While farmers often respond to heat stress by intensifying the use of agrochemicals, existing literature suggests that the efficiency of these inputs is temperature-dependent. Specifically, there are hypothesized threshold values (e.g., 30°C for major crops, 35°C for agrochemical application) beyond which the efficacy of fertilizers and pesticides declines, potentially leading to negative marginal returns. The central problem addressed is whether extreme heat in Vietnam has reached a point where it not only directly reduces agricultural output but also fundamentally alters the productivity of chemical inputs, rendering increased application counterproductive.

Methodology
The study employs a macroeconomic time-series analysis covering the period from 1991 to 2023. The data is primarily sourced from FAOSTAT, supplemented by the World Bank's Climate Change Knowledge Portal and Vietnam's National Statistics Office. The analysis utilizes a modified Cobb-Douglas production function (C-D-C model) within a Fully Modified Ordinary Least Squares (FMOLS) framework to address potential endogeneity and non-stationarity in the time-series data.

The model specification includes traditional input factors: land ($L$), labor ($W$), capital stock ($C$), fertilizers ($F$), and pesticides ($P$). Crucially, the model integrates climate variables, with a specific focus on an "extreme heat" index ($Heat$). This index is defined as the sum of the number of days with maximum temperatures exceeding 35°C and the number of tropical nights (temperatures remaining above 26°C).

To test the hypothesis that heat diminishes the efficiency of agrochemicals, the authors introduce interaction terms between the heat variable and the chemical inputs. Two primary specifications are estimated:

1. $Y_t = \alpha_0 + \alpha_1 L_t + \alpha_2 W_t + \alpha_3 C_t + \alpha_4 F_t + \alpha_5 P_t + \alpha_6 Heat_t + \alpha_7 (F_t \times Heat_t) + \epsilon_t$
2. $Y_t = \alpha_0 + \alpha_1 L_t + \alpha_2 W_t + \alpha_3 C_t + \alpha_4 F_t + \alpha_5 P_t + \alpha_6 Heat_t + \alpha_7 (P_t \times Heat_t) + \epsilon_t$

The study distinguishes between mean temperature and extreme heat, arguing that the latter better captures the nonlinear thresholds relevant to agrochemical efficiency.

Key Results
The estimation results, presented in Table 1, indicate that while land, labor, capital, and agrochemicals generally have positive effects on output, the interaction between extreme heat and chemical inputs is statistically significant and negative.

• Fertilizer Efficiency: In Model 3, the interaction term between fertilizer and heat ($\alpha_7$) is negative and significant (-0.3223). The marginal effect of fertilizer is calculated as $1.287 - 0.3223 \times \ln(Heat)$. The analysis reveals that when the heat index exceeds approximately 54.2 (days/nights), the marginal contribution of additional fertilizer turns negative. This suggests that in years with severe heatwaves, increasing fertilizer application may actually reduce total output or fail to provide economic benefits.
• Pesticide Efficiency: In Model 4, the interaction term between pesticides and heat is also negative (-0.081). The analysis indicates that the marginal effect of pesticides remains positive up to approximately 80 days of extreme heat. The data shows that only 2 years in the observed period reached a point of zero marginal effect, though the authors note this finding lacks direct evidence in the data and does not confirm the absence of years with negative marginal returns. However, the trend is downward, suggesting that efficiency is declining as heat exposure increases.
• Heat vs. Mean Temperature: The study finds that the extreme heat index is a more robust predictor of productivity losses than annual mean temperature. Models including mean temperature, precipitation, or humidity did not retain significance once the extreme heat index was included, supporting the argument that extreme events drive production risks more than average conditions.
• Historical Trend: Extrapolating the model parameters back to 1961 (Figure 8) indicates a long-term decline in the marginal efficiency of fertilizers, with the most pronounced negative effects occurring in recent decades as the frequency of extreme heat days has increased.

Significance and Claims
The paper claims three primary contributions to the literature on climate change and agricultural economics:

1. Non-Constant Elasticity: The study challenges the standard assumption in Cobb-Douglas frameworks that input elasticities are constant. It demonstrates that the output elasticity of chemical inputs in Vietnam is not fixed but varies systematically with heat exposure. As heat exposure increases, the proportional increase in output from additional fertilizer or pesticide use diminishes.
2. Superiority of Extreme Heat Metrics: The authors argue that annual mean temperatures are insufficient for modeling agricultural production risks in tropical contexts. They posit that the frequency, duration, and cumulative intensity of extreme weather events (captured by the heat index) are the critical drivers of production outcomes, particularly regarding input efficiency.
3. Policy Relevance for Input Management: The findings suggest that the common adaptation strategy of increasing agrochemical use in response to heat stress is increasingly ineffective and potentially harmful. The paper concludes that in periods of extreme heat, the marginal use of fertilizers may be economically and environmentally detrimental. Consequently, the study advocates for a shift in policy from subsidizing chemical quantities to improving the efficiency of their application through smart farming, precise weather forecasting, and soil-specific management strategies.

The authors maintain a modest tone regarding causality, noting that while the statistical evidence strongly supports a connection between extreme heat and declining agrochemical efficiency, the complex interplay of climate variables requires careful interpretation. The study serves as a macroeconomic indicator of these trends, highlighting the need for adaptation strategies that account for the nonlinear degradation of input productivity under extreme thermal conditions.