In many settings, leaders are evaluated in contexts where complexities of production processes and conflicting pressures from interest groups pose challenges to performance evaluation. In education, school accountability systems assemble rich data and report both categorical rating and the underlying student pass rates that determine them, permitting direct investigation of how different information affects labor market outcomes of school leaders. Applying regression discontinuity methods that by design hold effectiveness constant, we find sizable positive impacts on Texas elementary school principal retention and salaries for crossing the unacceptable-acceptable boundary but not for crossing higher ratings cutoffs. The apparent information breakdown that leads to the unequal treatment of equals at the lowest boundary could raise the distribution of principal quality through disproportionate departures of less effective school leaders. However, there is substantial overlap in principal value-added distributions across rating categories, and failure to cross the acceptable threshold does not lead to future improvements in school performance. Supplementary analysis suggests that the labor market penalty to leading a school that receives the lowest rating is confined to the current district, where the stigma of a low rating is likely to be greatest.