Person:

Hanna, Rema

Existing learning models attribute failures to learn to a lack of data. We model a different barrier. Given the large number of dimensions one could focus on when using a technology, people may fail to learn because they failed to notice important features of the data they possess. We conduct a field experiment with seaweed farmers to test a model of “learning through noticing”. We find evidence of a failure to notice: On some dimensions, farmers do not even know the value of their own input. Interestingly, trials show that these dimensions are the ones that farmers fail to optimize. Furthermore, consistent with the model, we find that simply having access to the experimental data does not induce learning. Instead, farmers change behavior only when presented with summaries that highlight the overlooked dimensions. We also draw out the implications of learning through noticing for technology adoption, agricultural extension, and the meaning of human capital.

We use a unique data-set from Indonesia on what individuals know about the income distribution in their village to test theories such as Jackson and Rogers (2007) that link information aggregation in networks to the structure of the network. The observed patterns are consistent with a basic diffusion model: more central individuals are better informed, and individuals are able to better evaluate the poverty status of those to whom they are more socially proximate. To understand what the theory predicts for cross-village patterns, we estimate a simple diffusion model using within-village variation, simulate network-level diffusion under this model for the over 600 different networks in our data, and use this simulated data to gauge what the simple diffusion model predicts for the cross-village relationship between information diffusion and network characteristics (e.g. clustering, density). The coefficients in these simulated regressions are generally consistent with relationships suggested in previous theoretical work, even though in our setting formal analytical predictions have not been derived. We then show that the qualitative predictions from the simulated model largely match the actual data in the sense that we obtain similar results both when the dependent variable is an empirical measure of the accuracy of a village’s aggregate information and when it is the simulation outcome. Finally, we consider a real-world application to community based targeting, where villagers chose which households should receive an anti-poverty program, and show that networks with better diffusive properties (as predicted by our model) differentially benefit from community based targeting policies.

In this paper, we provide a new framework for analyzing corruption in public bureaucracies. The standard way to model corruption is as an example of moral hazard, which then leads to a focus on better monitoring and stricter penalties with the eradication of corruption as the final goal. We propose an alternative approach which emphasizes why corruption arises in the first place. Corruption is modeled as a consequence of the interaction between the underlying task being performed by bureaucrat, the bureaucrat's private incentives and what the principal can observe and control. This allows us to study not just corruption but also other distortions that arise simultaneously with corruption, such as red-tape and ultimately, the quality and efficiency of the public services provided, and how these outcomes vary depending on the specific features of this task. We then review the growing empirical literature on corruption through this perspective and provide guidance for future empirical research.

We use a unique data-set from Indonesia on what individuals know about the income distribution in their village to test theories such as Jackson and Rogers (2007) that link information aggregation in networks to the structure of the network. The observed patterns are consistent with a basic diffusion model: more central individuals are better informed and individuals are able to better evaluate the poverty status of those to whom they are more socially proximate. To understand what the theory predicts for cross-village patterns, we estimate a simple diffusion model using within-village variation, simulate network-level diffusion under this model for the over 600 different networks in our data, and use this simulated data to gauge what the simple diffusion model predicts for the cross-village relationship between information diffusion and network characteristics (e.g. clustering, density). The coefficients in these simulated regressions are generally consistent with relationships suggested in previous theoretical work, even though in our setting formal analytical predictions have not been derived. We then show that the qualitative predictions from the simulated model largely match the actual data in the sense that we obtain similar results both when the dependent variable is an empirical measure of the accuracy of a village’s aggregate information and when it is the simulation outcome. Finally, we consider a real-world application to community based targeting, where villagers chose which households should receive an anti-poverty program, and show that networks with better diffusive properties (as predicted by our model) differentially benefit from community based targeting policies.

In this paper, we demonstrate that university students who cheat on a simple task in a laboratory setting are more likely to state a preference for entering public service. Importantly, we also show that cheating on this task is predictive of corrupt behavior by real government workers, implying that this measure captures a meaningful propensity towards corruption. Students who demonstrate lower levels of prosocial preferences in the laboratory games are also more likely to prefer to enter the government, while outcomes on explicit, two-player games to measure cheating and attitudinal measures of corruption do not systematically predict job preferences. We find that a screening process that chooses the highest ability applicants would not alter the average propensity for corruption among the applicant pool. Our findings imply that differential selection into government may contribute, in part, to corruption. They also emphasize that screening characteristics other than ability may be useful in reducing corruption, but caution that more explicit measures may offer little predictive power.

Moderate effects of pollution on health may exert an important influence on labor market decisions. We exploit exogenous variation in pollution due to the closure of a large refinery in Mexico City to understand how pollution impacts labor supply. The closure led to an 8 percent decline in pollution in the surrounding neighborhoods. We find that a one percent increase in sulfur dioxide results in a 0.61 percent decrease in the hours worked. The effects do not appear to be driven by labor demand shocks nor differential migration as a result of the closure in the areas located near the refinery.

To assess ways to achieve widespread health insurance coverage with financial solvency in developing countries, we designed a randomized experiment involving almost 6,000 households in Indonesia who are subject to a nationally mandated government health insurance program. We assessed several interventions that simple theory and prior evidence suggest could increase coverage and reduce adverse selection: substantial temporary price subsidies (which had to be activated within a limited time window and lasted for only a year), assisted registration, and information. Both temporary subsidies and assisted registration increased initial enrollment. Temporary subsidies attracted lower-cost enrollees, in part by eliminating the practice observed in the no subsidy group of strategically timing coverage for a few months during health emergencies. As a result, while subsidies were in effect, they increased coverage more than eightfold, at no higher unit cost; even after the subsidies ended, coverage remained twice as high, again at no higher unit cost. However, the most intensive (and effective) intervention – assisted registration and a full one-year subsidy – resulted in only a 30 percent initial enrollment rate, underscoring the challenges to achieving widespread coverage.

This paper investigates the impact of elite capture on the allocation of targeted government welfare programs in Indonesia, using both a high-stakes field experiment that varied the extent of elite influence and non-experimental data on a variety of existing government transfer programs. Conditional on their consumption level, there is little evidence that village elites and their relatives are more likely to receive aid programs than non-elites. Looking more closely, however, we find that this overall result masks a difference between different types of elites: those holding formal leadership positions are more likely to receive benefits, while informal leaders are actually less likely to. We show that capture by formal elites occurs during the distribution of benefits under the programs, and not during the processes when the beneficiary lists are determined by the central government. However, while elite capture exists, the welfare losses it creates appear quite small: since formal elites and their relatives are only 9 percent richer than non-elites, are at most about 8 percentage points more likely to receive benefits than non-elites, and represent at most 15 percent of the population, eliminating elite capture entirely would improve the welfare gains from these programs by less than one percent.

Existing learning models attribute failures to learn to a lack of data. We model a different barrier. Given the large number of dimensions one could focus on when using a technology, people may fail to learn because they failed to notice important features of the data they possess. We conduct a field experiment with seaweed farmers to test a model of "learning through noticing." We find evidence of a failure to notice: On some dimensions, farmers do not even know the value of their own input. Interestingly, trials show that these dimensions are the ones that farmers fail to optimize. Furthermore, consistent with the model, we find that simply having access to the experimental data does not induce learning. Instead, farmers change behavior only when presented with summaries that highlight the overlooked dimensions. We also draw out the implications of learning through noticing for technology adoption, agricultural extension, and the meaning of human capital.

This paper explores whether ordeal mechanisms can improve the targeting of aid programs to the poor ("self-targeting"). We first show that theoretically the impact of increasing ordeals is ambiguous: for example, time spent applying imposes a higher monetary cost on the rich, but may impose a higher utility cost on the poor. We examine these issues by conducting a 400-village field experiment with Indonesia’s Conditional Cash Transfer program (PKH), where eligibility is determined through an asset test. Specifically, we compare targeting outcomes from self-targeting, where villagers came to a central site to apply and take the asset test, against the status quo, an automatic enrollment system among a pool of potential candidates that the village pre-identified. Within self-targeting villages, we find that the poor are more likely to apply, even conditional on whether they would pass the asset test. Exploiting the experimental variation, we find that self-targeting leads to a much poorer group of beneficiaries than the status quo. Self-targeting also outperforms a universal asset-based automatic enrollment system that we construct using our survey data. However, while experimentally increasing the distance to the application site reduces the number of applicants, it does not differentially improve targeting. Estimating our model structurally, we show that there are large unobserved shocks in the decision to apply, and therefore increasing waiting times to 9 hours or more would be required to induce detectable additional selection. In short, ordeal mechanisms can induce self-selection, but marginally increasing the ordeal can impose additional costs on applicants without necessarily improving targeting.