Research

I study how compliance responds dynamically to temporary, highly visible policy interventions. I develop a coordination framework in which visibility raises contemporaneous compliance by activating marginal adopters, while postintervention persistence depends on belief updating within local social networks. As visibility fades, compliance attenuates at a rate governed by social cohesion. I test the framework using a randomised field experiment in dense urban neighbourhoods. Compliance rises sharply during the intervention but decays afterward, with significantly slower decay in more cohesive communities. Welfare counterfactuals show that short, salient interventions can generate substantial gains when targeted to cohesive networks.

Public service delivery in low-capacity environments often operates through frontline institutions subject to binding organisational constraints. This paper studies the elasticity of public service utilisation with respect to frontline capacity, and how this elasticity depends on whether capacity expansions introduce an adaptive frontline rather than a static increase in staffing. I analyse a randomised assignment of newly trained health mobilisers to administrative units during a large-scale vaccination campaign in Lahore, Pakistan. The intervention expands frontline capacity by introducing a mobile, locally embedded layer that augments existing facilities. Assignment increases the aggregate pool of frontline capacity and alters implementation conditions at the facility level. However, realised deployment across facilities is incomplete and selectively implemented, reflecting endogenous patterns of frontline allocation. I show that utilisation increases sharply in settings where adaptive frontline capacity is realised, implying positive elasticity of service utilisation with respect to realised, adaptively allocated frontline capacity. Instrumental-variables estimates indicate economically large increases in uptake among facilities whose deployment status is induced by assignment. The results demonstrate that capacity expansions are most effective when they introduce adaptive frontlines rather than when they operate as static additions to existing delivery systems.

Large natural disasters impose severe disruptions on firms, particularly in environments where formal insurance coverage and post-disaster transfers are incomplete. This paper documents how firms reorganise production following the 2005 Pakistan earthquake and asks whether such adaptation improves long-run efficiency or instead generates persistent distortions. Using a difference-in-differences design on a nationally representative panel of manufacturing firms, I document a non-monotonic pattern of adjustment across the production structure. Firms with intermediate capital–labour ratios experience the largest and most persistent declines in sales, while both labour-intensive and highly capital-intensive firms display greater resilience. Firms respond through intensive short-run adjustment: extending operating hours, reallocating labour towards skilled workers, increasing reliance on privately provisioned backup inputs, and diversifying markets. These responses stabilise output but do not induce innovation, indicating a reorganisation of production towards short-run stabilisation rather than long-run productivity growth. To interpret these patterns, the paper develops a model of adaptive misallocation in which firms facing incomplete insurance over-invest in self-protective inputs. These privately optimal responses divert resources away from productivity-enhancing investment and generate persistent distortions in firm behaviour. The results show that disasters reshape production not only through capital destruction, but through expectation-driven misallocation under incomplete insurance.