Loss ratio dynamics

Abstract Many studies of the insurance profit cycle use industry‐level annual data and focus on the existence of an AR(2) process. We take a different approach by adopting the idea of possible hard and soft markets, but they are not necessarily cyclical in the classic sense. In addition to aggregated data, we use quarterly firm‐level data to examine loss ratio behavior over time. This approach allows one to assess the firm‐level heterogeneity in the insurance market. We further use a Markov switching model to assess the heterogeneity of response to economic variables. Using a K‐means cluster approach, we examine the different clusters of firms and their different behavior over 2001q1−2020q4.