Texas dryland upland cotton yields have historically exhibited greater variation and more distributional irregularities than the yields of other crops, raising concerns that conventional parametric distribution models may generate biased or otherwise inaccurate crop insurance premium rate estimates. Here, we formulate and estimate regime-switching models for Texas dryland cotton yields in which the distribution of yield is conditioned on local drought conditions. Our results indicate that drought-conditioned regime-switching models provide a better fit to Texas county-level dryland cotton yields than conventional parametric distribution models. They do not, however, generate significantly different Group Risk Plan crop insurance premium rate estimates.
