Area-yield insurance feasibility in the context of remote sensing analysis

• Ano de defesa: 2021
• Autor(a) principal: Horn, Roberto Mattes
• Orientador(a): Campos, Rogério Costa
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Pelotas
• Programa de Pós-Graduação: Programa de Pós-Graduação em Desenvolvimento Territorial e Sistemas Agroindustriais
• Departamento: Centro de Ciências Sócio-Organizacionais; Faculdade de Agronomia Eliseu Maciel
• País: Brasil
• Palavras-chave em Português: Cluster analysis; Yield prediction; Soybean; Spacetime yield correlation; Insurance utility optimization; Análise de agrupamentos; Estimação produtividade; Soja; Correlação de rendimento espaço-temporal; Otimização da utilidade do seguro
• Área do conhecimento CNPq: CNPQ::CIENCIAS SOCIAIS APLICADAS::ADMINISTRACAO
• Link de acesso: http://guaiaca.ufpel.edu.br/handle/prefix/8220

The conventional insurance was not effective on promoting massive insurance uptake in developing countries like Brazil. Its well-known operational problems and the lack of information on covariant risk prevent the insurers to remove risk from the system and so operate with risk independent portfolios. Since farmers do not find the expected utility in the conventional insurance, less than 15% of the crop area is currently insured in the country. We evaluated the synergism between actuary and remote sensing tools to outline an alternative insurance product based on area-yield concept. Empirical analyses were performed over 10 years of soybean growth inside a continuous region encompassing 54 counties at the North of Rio Grande do Sul state in Brazil. We evaluate the feasibility of insurance companies to offer a soybean crop area-yield insurance as the actuarial/operational flow is supported by yield correlation assessment from remote sensing data analysis. Results suggest the area-yield based insurance find a new attractive operational applying context as remote sensing provides yield inputs in high spatial and temporal resolutions. It was showed the systemic risk can be successfully managed through the space-time clustering of the estimated yields from soybean areas and it is a better strategy than the structural correlation from the less meaningful political county limits. The actuarial simulations of producer utility and premium loading showed that using per cluster correlation structure to set insurance parameters outperforms both using per county correlations structure and using no insurance. In contrast to county political limits, the extracted utility rose 39% with clusters correlation structure.