The Role of Data-Driven Methodologies in Weather Index Insurance

There are several index insurance methodologies. Most of them rely on linear piece-wise methods. Recently, there has been studies promoting the potential data-driven methodologies in construction models due to their ability capture intricate non-linear structures. However, these types frameworks have mainly implemented high-income countries large amounts data and high-frequency requirements. This paper adapts a methodology based satellite-based climate indices explain flood risk agricultural losses Antioquia area (Colombia). We used records as proxy crop losses, while satellite comprises run-off, soil moisture, precipitation variables. analyse period between 3 June 2000 31 December 2021. logistic regression model reference point assess performance deep neural network. The results show that network performs better than traditional for available loss event selected metrics. Additionally, we obtained utility measure derive costs associated both parts involved including policyholder provider. When using networks, with lower majority range cut-off values. approach contributes future weather indexes region where decrease base would be expected, thus, resulting reduction costs.