Dataset on HYV Adoption, Climate Stress, and Agricultural Income Across 33 Districts of Maharashtra, India, 2014-15 to 2023-24

This dataset is a balanced district-year panel covering 33 districts of Maharashtra, India, over ten agricultural years from 2014-15 to 2023-24 (330 observations, 30 variables), compiled to study the relationship between high-yielding variety (HYV) seed adoption, climate stress, and agricultural income. Maharashtra was selected because its agro-climatic heterogeneity, spanning the humid Konkan coast, the semi-arid Deccan plateau, the drought-prone Marathwada belt, and the cotton-soybean Vidarbha region, produces the climate stress variation necessary for threshold identification, while its persistently high incidence of farmer distress gives the question high policy stakes. The dataset integrates variables from seven public secondary sources: the Directorate of Economics and Statistics (DES) Maharashtra, the India Meteorological Department (IMD), the Indian Council of Agricultural Research (ICAR), the National Bank for Agriculture and Rural Development (NABARD), Maharashtra Agricultural Produce Market Committees (APMC), the Pradhan Mantri Gram Sadak Yojana (PMGSY) road database, and three rounds of the Agriculture Census of India (2010-11, 2015-16, and 2021-22). Variables cover six domains: (1) agricultural income, measured as Gross District Value Added from Agriculture and Allied Sectors at constant 2011-12 prices; (2) seed system quality, proxied by HYV cultivation area in thousand hectares; (3) climate stress, including Maximum Consecutive Dry Spell in days (the primary threshold variable, computed from IMD daily records), standardised rainfall anomaly, and standardised temperature anomaly; (4) input intensification, including pesticide and NPK fertiliser consumption, irrigation coverage, and variable input cost per hectare; (5) institutional access, including KVK training intensity, extension worker density, PACS societies, APMC market yards, and rural road density; and (6) farm structure, including average holding size, small farmer share, cropping intensity, the Herfindahl crop concentration index, and rainfed area share. Key construction decisions: Thane and Palghar districts are merged into a single unit reflecting the pre-2014 administrative boundary; Mumbai City and Mumbai Suburban are excluded due to negligible agricultural activity; Agriculture Census variables are linearly interpolated to annual frequency between census rounds; and MCDS is computed as the longest consecutive run of days with rainfall below 2.5 mm within each agricultural year. The dataset supports the companion research article published in Land Use Policy (Manuscript No. LUP-D-26-02373) and accompanies four R replication scripts covering Hansen Panel Threshold Regression, the Spatial Durbin Model, and Canay panel quantile regression. A full variable dictionary and README with step-by-step replication instructions are included in the deposit.