Machine learning (ML) is a collection of data-driven techniques that originate in statistics, designed for prediction and pattern detection. ML works by learning from data and iteratively improving model performance. It differs from traditional econometric methods as its primary goal is testing theories, often through complex, flexible models, rather than explaining relationships between inputs and outputs via interpretable parameters like coefficients. Applications of ML in applied economics, including food economics and agribusiness, have gained importance following seminal contributions such as Athey (2019), and Graskemper et al. (2021, 2022). This article provides a brief overview of how ML methods are being used in applied economics, with a focus on recent examples in agribusiness, food, and resource markets. It also highlights areas where ML and traditional econometric approaches can be used in combination, as well as areas for continued research advancement.
The main tasks of ML include: (1) prediction with supervised ML, (2) pattern recognition with unsupervised ML, (3) reinforcement learning, (4) model variable selection with feature engineering, and (5) generative functions (Yu and Maruejols, 2023). Although each of these has applications in economics, supervised learning has received particular attention to date.
Supervised learning methods, which focus on pure prediction tasks, are particularly well-suited for exploratory research questions and emerging areas of inquiry. For example, ML has been used to identify predictors of food purchases from online influencers (Zhong and Yu, 2025), or to classify bioenergy villages based on communal characteristics (Maruejols et al., 2025). Instead of estimating coefficients, which is the primary outcome of classical econometric models, ML practitioners use variable importance metrics to interpret ML results. This aspect of ML is useful when examining predictors of pressing global issues, such as food insecurity (Zhao et al., 2025). Furthermore, predictive algorithms that move beyond point forecasting and provide density predictions hold great interest for capturing uncertainty and variability, which is especially relevant in agricultural economics. For instance, Xiong et al. (2025) applied quantile methods to predict corn yields. In short, supervised learning offers a flexible tool for improving prediction, in contexts where causal identification is not the primary goal.
Despite these advancements, concerns about ML exist due to its lack of transparency and causal interpretability. Historically, the need for new tools like ML to answer questions in agribusiness and applied economics has been met with resistance when the econometric toolkit already includes well-established, theory-driven methods. However, ML and traditional methods can serve as complementary, rather than competing, approaches. These complementarities can be seen throughout the articles of this special issue.
One promising area of complementarity lies in data preprocessing. ML can uncover latent patterns, such as identifying clusters of homogeneous observations via unsupervised learning, which can then be analyzed further using traditional econometric techniques (e.g., Zhang et al., 2025) or supervised ML models (e.g., Wang et al., 2024). In addition, there is scope for deeper integration of ML with classical methods by combining MLâs predictive power with the interpretability and structure of econometric models. A good example of this is the work by Li and Yu (2025), who use lasso regression to improve the detection of attribute non-attendance in choice experiments, enhancing willingness to pay estimates. Integrating ML in this way helps refine models, creating an âaugmentedâ version of classical models that improves parts of the methodology prone to human bias or limited by predictive capacity. Another example is provided by Njiru et al. (2025), who apply reinforcement learning (Q-learning) to improve modeling of strategic bidding behavior within an agent-based land market model, illustrating how ML can enhance simulation-based approaches in economics.
Although ML is typically associated with large datasets, economists often work with relatively small samples. This constraint of access to large-scale or high-dimensional data can be viewed as a limitation in using ML for applied economics and agribusiness questions. However, several studies featured in this special issue demonstrate the merit of ML methods for modestly sized datasets. Still, there remains substantial opportunity for expanding relatively small economic survey and take advantage of MLâs edge in analyzing larger datasets. Maruejols et al. (2025) combine remote sensing data with economic survey data demonstrating that the application of ML models to integrated datasets is a valuable application for continued research.
The potential of ML is also demonstrated across various contexts with implications for agribusiness and consumer food behavior. Cui et al. (2025) highlight the value of ML in agribusiness operations, especially when combined with technologies like robotics and remote sensing. This integration is particularly relevant for sectors facing labor shortages, such as specialty crops. On the consumer side, ML also holds promise for personalized nutrition and well-being. For example, Adhikari and McFadden (2025) explore how ML-based tools can support consumers in making healthier, more informed food choices.
Among other fast-growing areas of interest â though not covered in this special issue â is causal machine learning, which adapts ML techniques to estimate cause-and-effect relationships from observational data. Examples include causal forests (Stetter et al., 2022) and double machine learning (Hoeschle et al., 2025a). Combining behavioral economics or human intelligence (e.g., Bao and Yu, 2019) and artificial intelligence or ML (especially Reinforcement Learning) is another emerging frontier. Finally, the emergence of large language models (LLMs) enables the analysis of text data at scale, a resource that has traditionally been underutilized in economics (Hoeschle et al., 2025b). In addition to analyzing textual data, LLMs can be used to generate text, offering new ways to model communication, simulate behavior, or support decision-making in both academic and applied contexts.
Overall, the implications associated with machine learning applications in agribusiness and applied economics are vast. For academic researchers, ML does not replace econometric methods but adds useful tools, particularly in contexts where flexibility or predictive accuracy can improve model performance or support data-driven decision-making. ML provides analysts with opportunities to enhance existing approaches or to design entirely new ways to answer questions, offering significant value to industry and consumers alike.
As with any innovation, new opportunities come with new challenges. It is important to remain aware of the limitations of ML and data-driven methods, including issues related to interpretability, bias amplification, generalizability and resources usage. Applying ML effectively in economics requires a good understanding of how these methods work, how they differ from classical approaches, and where they can be meaningfully integrated. In this context, Bittmann (2025) offers a valuable bridge between the two cultures, providing a step-by-step breakdown of methods like lasso and causal ML from a traditional econometric perspective. This resource is especially helpful for students, teachers, and researchers with a classical statistics background.
References
Adhikari, S. and B.R. McFadden. 2025. Bridging taste and health: the role of machine learning in consumer food selection. International Food and Agribusiness Management Review 28(2): 440â455. https://doi.org/10.22434/ifamr1131
Athey, S. 2019. The Impact of machine learning on economics. In The economics of artificial intelligence. University of Chicago Press, Chicago, IL, pp. 507â552. https://doi.org/10.7208/chicago/9780226613475.003.0021
Bittmann, T. 2025. A practical guide from Ordinary Least Squares to causal machine learning. International Food and Agribusiness Management Review 28(2): 456â485. https://doi.org/10.22434/IFAMR.1087
Bao, T. and X. Yu 2019. Social norm and giving with indivisibility of money: an experiment of selfishness, equality and generosity. Journal of Institutional and Theoretical Economics 175(2): 272â290. https://doi.org/10.1628/jite-2019-0020
Cui, X., Z. Guan and D. Farnsworth. 2025. Advancing specialty crop management: a review of recent developments in robotics, remote sensing, and machine learning systems. International Food and Agribusiness Management Review 28(2): 422â439. https://doi.org/10.22434/ifamr1104
Graskemper V., X. Yu and J.-H. Feil. 2021. Farmer typology and implications for policy design â an unsupervised machine learning approach. Land Use Policy 103: 105328. https://doi.org/10.1016/j.landusepol.2021.105328
Graskemper V., X. Yu and J.-H. Feil. 2022. Values of farmers â evidence from Germany, Journal of Rural Studies 89: 13â24. https://doi.org/10.1016/j.jrurstud.2021.11.005
Hoeschle L., S. Liu and X. Yu. 2025a. âLet the poor talk about âpovertyâ: Revisiting poverty alleviation in rural China with machine learningâ. Public Policy and Poverty: in press. https://doi.org/10.1002/pop4.7000
Hoeschle, L., L. Maruejols and X. Yu. 2025b. The impact of energy justice on local economic outcomes: Evidence from the bioenergy village program in Germany. Energy Economics 145: 108432. https://doi.org/10.1016/j.eneco.2025.108432
Li, Y. and X. Yu. 2025. Attribute non-attendance in the choice experiment with machine Learning: WTP for organic apples in Germany. International Food and Agribusiness Management Review 28(2): 374â390. https://doi.org/10.22434/ifamr.1133
Maruejols, L., L. Höschle and X. Yu. 2025. Energy independence, rural sustainability and potential of bioenergy villages in Germany: machine learning perspectives. International Food and Agribusiness Management Review 28(2): 263â295. https://doi.org/10.22434/ifamr1132
Njiru, R.D.G., C. Dong, F. Appel and A. Balmann. 2025. Strategic bidding behaviour in agricultural land rental markets: reinforcement learning in an agent-based model. International Food and Agribusiness Management Review 28(2): 391â421. https://doi.org/10.22434/ifamr.1126
Stetter, C., P. Mennig and J. Sauer. 2022. Using machine learning to identify heterogeneous impacts of agri-environment schemes in the EU: a case study. European Review of Agricultural Economics 49 (4): 723â759. https://doi.org/10.1093/erae/jbab057
Wang, H., W. Huang, Z. Yu and J. Han. 2024. Prediction and pattern recognition of the large-scale poverty-returning risks: empirical evidence from machine learning. International Food and Agribusiness Management Review 28(2): 356â373. https://doi.org/10.22434/ifamr.1096
Xiong, T., M. Xia, G. Li, J. Li and W. Xia. 2025. Beyond point forecasting: Probability density forecasting of corn yield based on quantile regression forest. International Food and Agribusiness Management Review 28(2): 317â336. https://doi.org/10.22434/ifamr1134
Yu, X. and L. Maruejols. 2023. Prediction, pattern recognition and machine learning in agricultural economics. China Agricultural Economic Review 15(2): 375â378. https://doi.org/10.1108/caer-05-2023-307
Zhang, X., X. Yu, Y. Liu and Y. Xie. 2025. The cluster characteristics and influencing factors of Chinaâs agricultural product importing countries: an analysis using machine learning. International Food and Agribusiness Management Review 28(2): 337â355. https://doi.org/10.22434/ifamr1127
Zhao, L., M. Yang, S. Min and P. Qing. 2025. Prediction of household food insecurity in rural China: an application of machine learning methods. International Food and Agribusiness Management Review 28(2): 296â316. https://doi.org/10.22434/ifamr1137
Zhong, X. and X. Yu. 2025. Who buys food products from online influencers? predictions with machine learning. International Food and Agribusiness Management Review 28(2): 241â262. https://doi.org/10.22434/ifamr.1130
