The argument that pay comparisons systematically affect effort behavior has found weak support in the experimental literature. This paper uses laboratory experiments to test a possible explanation for these mixed findings, namely that marked reactions to pay comparisons only occur in environments where co-worker wages are a prominent reference standard for what constitutes a fair pay. Using a three-person gift-exchange game we study effort behavior in three treatment conditions: in a benchmark treatment co-worker wages are secret, while in two other treatments co-worker wages are publicly observed before employees choose effort. These ‘public wages’ treatments differ in how co-worker wages are determined: in one treatment the employer can arbitrarily choose the wage for the co-worker, while in the other treatment the level of the co-worker wage is mandated exogenously by the experimenter. This exogenous manipulation aims at prompting individuals to refer to co-worker wages as a suitable source of information about what is an appropriate (or acceptable) pay standard in an experimental firm. No such cue is instead present in the treatment where co-worker wages are chosen by the employer. We find that pay comparisons have a detrimental effect on effort behavior: for any given level of the own wage, effort is higher when employees cannot observe what co-workers earn than when co-worker wage are public. Crucially, the negative impact of pay comparisons on effort choices is only significant in the treatment where co-worker wages are mandated exogenously, while the same relative pay standings do not produce sufficiently strong reactions when co-worker wages are chosen by the employer.