Speaker
Description
Keywords: Crop Farm Efficiency, FADN, Climate Extremes, Panel Data
JEL codes: Q12, Q54, D22
Introduction & Objective:
High land and farm productivity is key for a sustainable future under rising food demands, transitions towards circular economies, and limited natural resources, i.e. production factors. Effectively supporting and maintaining productivity requires accurate measures of sector, land, and farm productivity (Zelenyuk, 2023), for instance, as a base for identifying potential threats to productivity. Typical measures for productivity rely on its relative form, productive efficiency, i.e., the difference between actual and potential output. Under climate change, altered agro-climatic conditions and more frequent climate extremes (IPCC, 2022) lower the potential output and increase production risks, challenging productivity levels and their measurement (Boix-Fayos and Vente, 2023). For instance, farms may experience yield losses due to waterlogging (Chambers and Pieralli, 2020) or extreme drought events (Schmitt et al., 2022). This would be observed as productive inefficiency clearly attributable to climate extremes but not to management in the short run (Chambers and Pieralli, 2020).
The farm efficiency literature rarely explicitly accounts for climate extremes and agro-climatic conditions while measuring managerial-induced inefficiencies (Silva et al., 2020; Quiédeville et al., 2022). Not clearly disentangling inefficiencies by source, such as weather- and farm management-related, bears the risk of biased inefficiency measures and inflating policies incentivizing adaptation to climate change. We aim to close this gap by investigating the extent of managerial and weather-related farm-level inefficiencies' contribution to overall farm-level inefficiency. We answer this question for a sample of German specialized crop farms for 2004-2020. By this quantification, we expect to contribute to farm efficiency and productivity analysis and climate change economics literature.
Methodology:
To address our research question, we build our analysis on the four-component stochastic frontier model by Lien, Kumbhakar and Alem (2018). This allows us to quantify farm inefficiency and disentangling by managerial and weather-induced inefficiency while accounting for farm heterogeneity and technological progress.
We utilize an unbalanced panel of German crop farms from the E.U.'s Farm Accountancy Data Network (FADN) covering 2004 to 2020. To answer to what extent managerial and weather-related inefficiencies contribute to overall farm inefficiency, we model persistent and transient inefficiency over time. Transient inefficiency is modelled as a function of inefficiency determinants not under the control of the farm, i.e. weather realizations, including climate extremes and policy effects, for instance, from E.U.'s Common Agricultural Policy (Minviel and Latruffe, 2017). A farm's persistent and residual transient inefficiency is then attributed to managerial inefficiency.
To assign weather realizations to FADN farms' observations, we have computed probabilistic farm locations at NUTS-3 (Hartig, Seifert and Hüttel, 2023). We assign agro-climatic information (e.g., long-run precipitation, temperature, drought risk measures) and seasonal weather realizations (e.g. seasonal precipitation, evapotranspiration) provided by the German National Meteorological Service. Due to the abundancy of weather information, we reduce the complete set of available agro-climatic and weather variables. We build upon Li and Ortiz‐Bobea (2022) and consider machine-learning approaches for model selection (Cui et al., 2024). Our next steps include the identification of climate extremes relevant to German crop farms. We will consider, among others, plant-specific killing degree days, droughts, and waterlogging (Schmitt et al., 2022) for differrent seasons.
Results:
We expect lower estimates of managerial inefficiency than prior findings due to the captured productivity losses attributable to climate extremes (Wimmer et al., 2023). We further plan to highlight the importance of climate extremes in agricultural production by calculating farm losses from crop farming due to climate extremes. Using a monetary output variable allows us to calculate farm losses and compare the results to existing estimates (Schmitt et al., 2022).
In our preliminary model specification, we rely on a single-output translog specification. We use total crop output in Euros at 2015 prices as our output variable. We include labour, land, capital, and materials as production inputs measured by total labour hours, total land in hectares, and farm capital and fertilizer cost in Euros (Addo and Salhofer, 2022; Wimmer et al., 2023).
The first results indicate the highest mean elasticity for land (0.544), whereas elasticities with respect to capital, labour, and materials range between 0.1 and 0.2. Estimates suggest very small technical progress of 0.1% on average in all years. We find an overall technical efficiency of 75.2%, comprising both persistent technical efficiency (84.8%) and transient technical efficiency (88.3%).
We note that these are preliminary results. In the next steps, we will investigate the extent of managerial and weather-related farm-level inefficiencies' contribution to overall farm-level inefficiency.
Discussion & Conclusion:
With this paper, we expect to discuss to what extent managerial and weather-related inefficiencies contribute to crop farm inefficiency. We aim to provide reliable quantifications of managerial farm-level inefficiencies for German crop farms. Such reliable inefficiency quantifications and their unbiased attribution to management and climate extremes may support targeted, evidence-based policy for sustaining productivity in the agri-food sector under climate change with more frequent climate extremes. Contrary to the existing literature, we anticipate finding lower levels of managerial inefficiency and higher levels of weather-related inefficiency. Policymakers may, therefore, increase the efforts within the E.U.'s Common Agricultural Policy to support farm-level climate change adaptation and mitigation (Pe'er and Lakner, 2020).
As with any empirical study, our results will be limited due to data availability and the FADN sample. Farms included in the FADN are selected based on a sampling plan aiming to represent a region's population with respect to the type of farming, economic size, and region (Kempen et al., 2011). Selecting farms above a minimum size threshold might lead to an underrepresentation of agricultural activity in some areas. Also, voluntary participation may add to potential sample selection issues, limiting statistical inference. The single-output specification is another limitation (Lien, Kumbhakar and Alem, 2018). We consider an extension to a multi-output setting (Badunenko et al., 2021) and include mixed and livestock farms with branch-specific inefficiencies.
References
Please refer to the comments.
