We study how the gender composition of supervisor-worker dyads affects workers' outcomes. We use a fine-grained longitudinal personnel data on workers from an Italian insurance company over the period 2014-2021 and assign to each worker the gender of the direct supervisor. We implement an individual worker's fixed effect model, together with a dichotomous variable that captures pre- and post- Covid-19 period and time-varying individual characteristics. Our findings show that, although both male and female people managers evaluate similarly the performance of male and female workers, female supervisors grant lower amount of one-off bonus than male people managers to both male and female workers. Moreover, both male and female workers have a lower probability of receiving a promotion from employee of level VI to middle-managers when the people manager is a female than when is a male. When exploiting a heterogeneous analysis by gender, results confirm that the gender of the supervisors does not affect workers' performance assessments, while it negatively impacts the total amount of bonus of both male and female workers. We interpret these results either as evidence that female people managers are more severe to conform to a masculine gender stereotype associated with a leadership position, or as evidence of the fact that female people managers are at the head of marginal areas and offices, hence they receive less funds to provide bonus and promotions to workers they supervise.