Strengthening State Capacity Through Evidence-Based Performance Diagnostics: Experimental Evidence from Ghana’s Civil Service
How to cite
Bancalari, Antonella; Rogger, Dan; Augsburg, Britta; Opuni, Kpadam; Jones, Maria Ruth; Vakis, Renos. 2025. "Strengthening State Capacity Through Evidence-Based Performance Diagnostics: Experimental Evidence from Ghana’s Civil Service." AEA RCT Registry. AEARCTR-0017267. https://doi.org/10.1257/rct.17267-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
We study how to improve performance assessments and the inner workings of hierarchies in bureaucracies. We conduct a field experiment with bureaucrats in all ministries staffed by the Ghanaian Civil Service. We randomly allocate directorates an additional module on managerial practices as part of the business-as-usual training in the annual performance review processes of the civil service. Additionally, among the treated, we randomize whether the team performance assessment, after the training, is led by managers or by workers, nudging managers on the importance of enabling worker voice.
Intervention
The link between state capacity and economic prosperity is now widely established (Khan et al., 2015, 2019; Finan et al., 2017; Bertrand et al., 2019; Besley et al., 2022). At a micro level, there is growing evidence that the administration’s capacity to complete bureaucratic tasks is a core feature of an effective state (Rasul et al., 2021). In Ghana, where about one-third of infrastructure projects are never completed (Williams, 2017) and government spending accounts for 25% of GDP (World Bank, 2024)—such administrative inefficiencies have significant macroeconomic consequences: a 10-percentage-point increase in task completion implies substantial gains in economic growth (Rasul et al., 2021). Similar patterns in other middle-income contexts show that unfinished public investment projects reduce welfare (Bancalari, 2024).
High-quality administrative activity is a key feature of bureaucratic capacity. For example, managerial quality is increasingly recognised as a key driver of bureaucratic task completion and broader public sector performance (Bloom et al., 2016; Rasul and Rogger, 2018). Wide variation in management practices across Ghana’s civil service, particularly poor practices in setting clear targets, monitoring progress and rewarding good performance, suggests significant room for reform (Rasul et al., 2021).
A key question for governments is how to improve poor management and foster a culture of evidence-based, continuous improvement within public administrations. In the private sector, particularly among manufacturing and agricultural firms in low- and middle-income countries, business training programmes have shown promise (Bloom et al., 2012, de Mel et al., 2014; Iacovone et al., 2022). However, the evidence is mixed when it comes to sustainable public sector improvements (McKenzie and Woodruff, 2013; Campos et al., 2017). The evidence gap is even greater in the public sector, where little is known about how to build and sustain effective managerial practices within the state apparatus.
The intervention proposed in this project embeds managerial training --grounded in international research evidence-- into Ghana’s civil service performance diagnostics. While Ghana’s civil service operates an institutionalized appraisal system, including Annual Performance Reports (APRs) and Chief Directors’ Performance Assessments (CDPA), qualitative evidence --including reflections shared by senior staff of the Office of the Head of the Civil Service-- highlights its limitations. These diagnostic tools are widely perceived as perfunctory, compliance-oriented exercise, with limited engagement in systematic analysis of what worked, what failed, and why. As a result, the full potential of these tools to support learning and continuous improvement remains untapped.
Within the Ghanaian Civil Service, we propose testing innovative approaches to embedding research-based insights on bureaucratic performance directly into the core diagnostic process of the government, and thus into the reform of administration and the policymaking machinery. The intervention seeks to shift processes from compliance-driven reporting to tools that identify and address the root causes of performance gaps.
We are uniquely positioned to study how best to promote the uptake of effective managerial practices and foster a culture of evidence-based policymaking, thanks to our longstanding collaboration with the Office of the Head of the Civil Service (OHCS) in Ghana. This partnership builds on more than a decade of engagement with the analytical unit at the centre of Ghana’s public sector reform efforts, providing us with rare access to the institutional processes and decision-making structures that shape civil service performance. Crucially, this proposal responds directly to a formal request from OHCS—and the wider civil service—for evidence on how to track and assess performance more effectively, integrate feedback into planning, and support continuous improvement. By rigorously evaluating reforms that have been explicitly demanded by our government counterparts, we ensure both the policy relevance of our research and strong potential for sustained uptake of findings at the highest levels of the Ghanaian state.
The main intervention (T1) tests a module designed to improve how organizational directorates within ministries conduct performance diagnostics and to ensure that these diagnostics feed into meaningful actions and improve service delivery. At its core is a training and a system of regular feedback, grounded in evidence from organizational and personnel economics (Bloom and Van Reenen, 2007; Bloom et al., 2012; Bloom et al., 2015; Rasul and Rogger, 2018; Rasul et al., 2021).
The second intervention (T2) studies the inner workings of hierarchies. In the top-down group, we highlight the importance of leadership and we ask seniors to lead discussions. In the bottom-up group, we nudge seniors to move away from a hierarchical approach, by highlighting evidence on the benefits of worker voice for organizational performance, and ask workers to chair discussions on team performance assessments.
Experimental design
We are joining forces with the Office of the Head of the Civil Service (OHCS), who conducts the Annual Performance Review among all ministries of the civil service.
The first experimental arm (T1) tests an intervention designed to improve how organizational directorates within ministries conduct performance diagnostics and to ensure that these diagnostics feed into meaningful actions and improved service delivery. At its core is a training module grounded in evidence from organizational and personnel economics (Bloom and Van Reenen, 2007; Bloom et al., 2012; Bloom et al., 2015; Rasul and Rogger, 2018; Rasul et al., 2021). This training combines a theoretical introduction to diagnostic tools with descriptive evidence on the status of the civil service in terms of the managerial scoring system adapted from the World Management Survey for bureaucrats. The control group conducts training business-as-usual, while the treatment group receives this additional training module.
The second treatment arm (T2) experiments with engagement modalities within hierarchies in teams. The OHCS convenes discussion groups, including the senior and juniors of directorates, after the training. Participants reflect on the previous year with their teams and assess multiple aspects of their teams de facto administrative performance and processes. We randomly allocate whether the discussion is chaired by the manager or by a worker. In the top-down group, we highlight the importance of effective leadership, while in the bottom-up group we nudge seniors to move away from a hierarchical approach by highlighting evidence on the benefits of worker voice for organizational performance. We expect the top-down engagement to be more effective if we need senior buy-in. On the contrary, the bottom-up could be more effective if junior-led diagnosis improves information flows, coordination and cooperation [Freeman 1980, Freeman and Lazear 1995]. We randomise T2 among a sub-set of those allocated to T1.
The randomisation unit is the directorate within ministries. This is appropriate because (i) each directorate consolidates inputs from its constituent units into a single diagnostic report, submitted to the OHCS, and (ii) the intervention targets internal diagnostics and planning, which involve substantial interaction within directorates but limited interaction across them, minimizing spillovers. To preserve internal coherence within these units, all units within a given directorate are therefore assigned to the same treatment arm, receiving identical training content and support in the structure of their discussions.
OHCS staff acts as facilitators to ensure the intervention follows research protocols.
Design details
Cited by
No citation data available for this registration yet.