By reducing information asymmetries across the hierarchy, the digitisation of government services presents an opportunity for centralised management of frontline staff. In particular, high-frequency granular data can enable senior government officials to hold poorly performing members of the service delivery chain to account. To be effective, however, centralised management must translate large volumes of data into appropriate management actions. This paper studies this tension by evaluating a large-scale centralised accountability approach to managing education carried out at scale in Punjab, Pakistan.

We find that a system that automatically identified poorly performing schools and jurisdictions for the attention of central management had no appreciable impact on the trajectory of school outcomes across any area of its focus. We contrast this result with the significant impact that frontline managers (head teachers) can have on school outcomes across the same areas and the potential for using centralised information systems to optimise the allocation of managerial talent across the public sector.

JEL CODES: D73, H11, H83