An approach to model-based small area estimation under covariate measurement errors is presented. Using a min-max approach, we proof that regularized regression coefficient estimation is equivalent to robust optimization under additive noise. Applying this equivalence, the Fay-Herriot model is...