Silvia Castro

Why is peer support under supplied in organizations? This paper shows that helping is valuable but often invisible. When firms do not observe or recognize it, employees provide less of it. This paper uses a cluster-randomized field experiment in two banks. It shows that making peer support visible and rewarded leads to lasting productivity gains. 

How far can managerial strategy reach? This paper shows that its reach extends beyond the organizational chart. Structured group discussions among employees reduce stigma on menstrual health — a norm imposed by society, not management — and drive adoption of health technologies introduced by the firm.

See the coverage of this project in VoxDev, in The Decision Lab, or in LMU pressroom.

 Can a simple change in how managers run meetings shift the culture of their teams? Yes — and at scale. A randomized controlled trial across  4,300 employees at a global pharmaceutical firm shows that structured changes to manager-employee one-to-one meetings improve team psychological safety, innovation climate, and leadership quality.

See the coverage of the project at the MIT Sloan Review here. 

Can firms use the infrastructure they already have to shift norms they did not create? This project shows that they can, but only within limits. Daily audio broadcasts to over 4,000 workers in a Bangladeshi garment factory reduced menstrual stigma and taboos for both women and men. Norms shifted beyond the factory into workers' households.  Not all norms proved to be malleable to the same degree: communication taboos shifted, while norms rooted in beliefs about impurity resisted change, revealing the limits of managerial strategy.

Why does the same business idea get funded differently depending on who presents it? Not because of the idea, because of beliefs about who can execute it. In a study with 451 loan officers, evaluators rated the same business ideas equally, no matter the founder's gender. However, they penalised female entrepreneurs based on beliefs about the challenges they face. Team formation neutralizes this penalty. This shows that an organisational structure can act as a way to reduce bias.

What happens to critical information when supervisors lack the skills to receive it? This paper shows that it stays at the bottom of the organizational chart. A cluster-randomized field experiment in a large commercial bank shows that training supervisors in supportive leadership improves supervisory quality and team psychological safety. It also changes not how much employees speak up, but what they say. More critical, decision-relevant feedback flows upward in the organization.

Firms are adopting generative AI to improve individual productivity, but what happens to the social fabric of organizations in the process? This project studies how peer cooperation, knowledge sharing, and mentoring change once agentic AI enters an organization.

When AI tools change how employees perform, do supervisors update how they assess potential? This project uses a field experiment in a large firm to study whether GenAI shifts supervisors' reliance from performance metrics and whether this affects identification of talent.

Can you change organizational culture if you only intervene at one level? Culture change is hard — especially when it requires coordinated effort across hierarchical levels. This project uses a large-scale RCT spanning the full hierarchy of a US pharmaceutical firm to study exactly this.

What does firm productivity leave on the table when workplaces are not designed with women in mind? Many large manufacturing firms rely on large female workforces. Yet, core workplace routines, supervisory practices, and organizational policies are misaligned with workers’ menstrual health needs. This project examines menstrual health as a strategic human capital issue using a large-scale field experiment across Bangladeshi manufacturing firms employing over 100,000 workers.  We test if more supportive workplace design improves employee well-being and, in turn, firm productivity.