Ideas Generation in Hierarchical Bureaucracies: Evidence from a Field Experiment and Qualitative Data
joint with Margherita Fornasari (World Bank), Daniel Rogger (World Bank)
and Martin J.Williams (Michigan)
November 2025
Abstract We study ideas generation in bureaucracies, combining qualitative and quantitative evidence. We do so at-scale, using data from bureaucrats in all ministries staffed by the Ghanaian Civil Service. Our qualitative evidence shows these organizations have strong hierarchical cultures, where juniors feel unable to raise innovative ideas, and the norm is to cite Civil Service-wide resourcing issues as the main bottleneck to raising productivity. Within the confines of these hierarchical cultures, our quantitative evidence comes from a field experiment training bureaucrats how to identify simple innovations in work processes and raise these ideas with colleagues. We implemented training at the individual level, and at the division-level to bureaucrats working together day-to-day. Our key finding is that individual trainings were more effective in shifting perceived climates towards innovation, leading treated individuals to be more likely to raise and share new ideas with colleagues. This ultimately improved administrative processes and public service delivery. Division-level training was ineffective on all margins. We document the key reason for this was that divisions failed to take on board core learnings from the intervention when it was delivered collectively. Rather, division-level training led teams to suggest ideas reflecting pre-existing workplace norms, falling back on unrealistically aiming for resource intensive Civil Service-wide change. Our findings highlight the impact of interventions promoting innovation in organizations can depend critically on pre-existing hierarchical workplace cultures.
Legacies of Conflict: Self-efficacy and the Formation of Conditional Trust
joint with Niklas Buehren (World Bank), Markus Goldstein (World Bank) and Andrea Smurra (UCL)
October 2025
Abstract Exposure to armed conflict in early life is a traumatic experience, affecting 400 million children worldwide. We combine theory, measurement and evidence to study how psychological legacies of conflict mediate the relationship between exposure to conflict and the long-term formation of trust preferences. Our analysis is based on a sample of 3900 women born during the Sierra Leonean civil war and surveyed 14 years later. We first develop a framework describing the link between exposure to conflict and trust. This makes precise what individuals have in mind when expressing conditional trust in others, and establishes the role of self-efficacy in linking conflict and trust. Our empirical analysis then shows that exposure to conflict significantly increases self-efficacy, and through this channel, conflict leads conditional trust to rise and for outright trust of others to fall, relative to those never exposed to conflict. To further microfound how exposure to conflict translates into psychological legacies, we construct a granular typology of experiences of conflict, combining information on age of exposure to conflict and recall of victimization. We use this to show how direct exposure, memories and trauma, and narratives of conflict from others each distinctively shape self-efficacy. Finally, we show how our model can help reconcile heterogeneous findings across conflict scenarios, and suggest avenues for future work on the more general role of psychological legacies from traumatic shocks early in life on the long-term formation of economic preferences.
Impacts of Pro-poor Transfers on Perceptions, Mindsets, Political Views and Participation of the Rich and Poor
joint with Adnan.Q.Khan (LSE), Nicolas Cerkez (UCL) and Anam Schoaib (CERP)
September 2025
Abstract Does exposure to big push pro-poor interventions impact the political participation of the rich and poor, and is this driven by shifts in perceptions, mindsets and policy views? We study the issue using a field experiment tracking 15,000 households over four years in rural Pakistan. Villages are randomly assigned to receive an intervention where the poor are either offered a one-time asset transfer of value $620 or an equivalent valued one-time unconditional cash transfer. In treated villages we randomize which of the poor receive the transfer, and then track treated poor, not treated poor and not poor households at two and four years post-intervention. The transfers cause large and persistent economic gains to the treated poor, and reduce village-level consumption inequality. Despite these measurable changes, we document a wedge between economic reality and household's perceptions of their economic standing and village inequality. Two years post-intervention, all households, irrespective of their beneficiary status, more strongly view the rich as deserving, hold stronger pro-market mindsets and become more trusting of neighbors, while redistributive preferences remain unchanged. Finally, political participation of households rises in local elections. For all households, pro-market mindsets mediate this outcome, but for the treated poor, political participation is also mediated by their improved economic standing. Results four years post-intervention highlight how exposure to economically effective big push pro-poor interventions is however unlikely to persistently shift perceptions, mindsets and policy preferences of the rich and poor, even as economic gains to the treated poor persist.
Firm Responses to Uncertainty and Implications for Workers: Experimental Evidence from Uganda During the Pandemic
joint with Livia Alfonsi (HBS), Vittorio Bassi (USC) and Elena Spadini (UCL)
July 2025
Abstract The Covid-19 pandemic represents one of the most rapid and severe shocks to hit global labor markets. We study how firms reacted to the heightened uncertainty and consequences for workers, in a developing country economy: Uganda. Our analysis is based on a panel of firms and workers, tracked from 2012 to 2022, including high frequency surveys during the pandemic. We find that firms' common response to the heightened uncertainty of the pandemic was to immediately lay off the highest earning workers, that is, the most experienced or skilled employees. We then study the differential impacts of such firm survival strategies on workers' labor market dynamics across the skills distribution, exploiting the fact that we randomly assigned individuals to the offer of vocational training in 2013. We find that high skill trained workers, who enjoyed better labor market outcomes pre-pandemic, were more likely to be laid off early in the pandemic given firm's survival strategies. However, trained workers recovered from this job loss and were resilient to the shock. Cumulatively over the pandemic, trained workers spend 61% more time employed than untrained controls, and earn 17% more. Hence, the returns to training survive the pandemic. The mechanisms driving the resilience of trained workers are the certifiability of their skills and their greater accumulation of sector-specific experience, both of which enable them to switch employers within the same sector during the crisis. Our findings provide new insights on firm responses to fast-moving aggregate shocks in low-income settings, consequences for workers' labor market trajectories, and drivers of the returns to training in good times and bad.
The Anatomy of a Public Health Crisis: Household Responses Over the Course of the Zika Epidemic in Brazil
joint with Ildo Lautharte (World Bank)
June 2025
Abstract The global frequency and complexity of viral outbreaks is increasing. In 2015, Brazil experienced an epidemic caused by the Zika virus. This represents the first known association between a flavivirus -- carried by the Aedes Aegypti mosquito -- and congenital disease, marking a `new chapter in the history of medicine' [Brito 2015]. We use hundreds of millions of administrative records to document household responses to the first public health alert linking the Zika virus to the risk of congenital disease for those in utero. We study the two margins of household behavior emphasized by economic epidemiological models of disease diffusion: risk avoidance behaviors (avoiding pregnancy), and risk mitigation behaviors during pregnancy (ultrasounds and abortions). On risk avoidance, we find a 7% reduction in pregnancies post-alert, a response triggered immediately after the alert, and driven by higher SES women. On risk mitigation during pregnancy, we find a 9% increase in the use of ultrasounds in the first trimester of pregnancy, and abortions rise by 5%, being concentrated among late term abortions. We document that post-alert all groups -- irrespective of race, marital status, education, age, and socioeconomic status -- were able to reduce risks during pregnancy, in line with preventative measures not being costly. Combining the evidence on avoidance and mitigation behaviors rules out that higher educated mothers responded more in terms of delayed pregnancies because they were better informed of the risks from Zika. Rather, the combined evidence suggests more educated households might be able to delay conception because they face lower costs of altering their fertility timing from their plan. We conclude by discussing consequent impacts on birth outcomes, and the extent to which our findings extend to household responses to public health alerts on other emerging viral threats.
Revisions requested, Journal of Health Economics.
Majority Perceptions of Minority Groups: Economic Inequalities, Their Causes, and Policy Solutions
joint with Lucinda Platt (LSE) and Pratyush Tiwari (UCL and IFS)
March 2025
Abstract How does the majority population view the societal contributions, economic outcomes and opportunities available to specific ethnic minority groups, root causes of ethnic disadvantage, and policy solutions to address them? We answer these questions in the UK context, a multi-ethnic society where some minorities outperform the majority in economic outcomes, while others underperform. We use an online survey fielded to 3200 White British individuals, into which we embed a survey experiment that presents respondents with narratives about the economic success or disadvantage of specific minority groups. The experiment was purposefully implemented in the run up to the 2024 UK General Election, that saw the rise of populist anti-immigration parties. We find that even in such charged times, light-touch narratives can correct majority misperceptions of the economic outcomes of specific minorities, and shift views on policies to address ethnic inequalities. Views on the opportunities available to minorities and root causes of disadvantage, such as luck or effort, are harder to shift irrespective of the minority outgroup. By considering perceptions towards their ingroup, we document that narratives about the economic success of minorities can shift majority perceptions in ways consistent with zero sum thinking. Given strong political differences in perceptions of minorities, we examine heterogeneous responses to the narratives by political leaning. Narrative treatments can shift perceptions, including those of right-leaning individuals, with zero sum mindsets being independent of political leaning. We conclude by examining how perceptions across domains shape the reasoning behind support for policies targeted to specific minorities to address ethnic inequalities.