In this note we deal with a panel data model with fixed effects. We show that a Bayesian procedure based on the reference prior suffers from the incidental parameter problem, as also happens with the Jeffrey’s prior [Econom. Lett. 82 (2004) 135–138]. Using an alternative prior distribution we present a solution to the problem.
"Does reference prior alleviate the incidental parameter problem?." Braz. J. Probab. Stat. 24 (3) 509 - 512, November 2010. https://doi.org/10.1214/09-BJPS108