A machine learning approach to assess implications of Climate Risk Factors on Agriculture: The Indian case

Abstract

The paper examines the direct and indirect implications of the risk factors relating to climate change on various parameters of agricultural production/productivity in India. India, being an emerging economy with considerable dependence on agriculture both for food security and employment generation, offers an important case study for understanding the macro-economic issues and designing the right set of policy approach for mitigating implications of climate change. Furthermore, unlike other central banks, the Reserve Bank of India (RBI) shares a close association with agriculture owing to the continued credit support to the sector as part of its priority sector lending policy, and because it is closely involved in addressing climate change. The paper focuses on the decade of 2010s, the recorded warmest decade till now and adopts a machine learning approach (sequential multivariate adaptive regression splines model) to assess the interaction between climate risk factors and agriculture. To the best of our knowledge, the modelling approach applied in this paper is unique, novel and of the first kind to assess the implications of climate change on agriculture. The results indicate that carbon dioxide (CO2) emission, precipitation, irrigation water used, and rainfall are the most prominent factors affecting different parameters of agricultural production. These factors are taken at the yearly aggregate level and their interactions among themselves, particularly CO2 emissions, affect productivity of foodgrains and oilseeds, providing a detailed insight about the recent decade of climate change in the context of Indian economy.