We present an experimental test of a shirking model where monitoring intensity is endogenous and effort a continuous variable. Wage level, monitoring intensity and consequently the desired enforceable effort level are jointly determined by the maximization problem of the firm. As a result, monitoring and pay should be complements. In our experiment, between and within treatment variation is qualitatively in line with the normative predictions of the model under selfishness assumptions. Yet, we also find evidence for reciprocal behavior. The data analysis shows, however, that it does not pay for the employer to rely on the reciprocity of employees.