Session 4: Social Structure and Economic Development

China’s accession to the WTO created new economic opportunities in certain cities. A shift-share identification strategy shows that residents of adjacent rural areas migrated in and advanced economically. Longitudinal panel data on children reveals that their sons benefit, but counter-intuitively, daughters suffer worse mental and physical health, complete fewer years of schooling, and remain poor later in life. We explore why, and learn that hukou policy that restricts migrant children’s access to urban schools is a factor. Triple difference research designs reveal that migrant parents become discontinuously more likely to leave middle-school-aged daughters (but not sons) behind in rural areas– often without either parent present– exactly when and where hukou policy makes schooling more expensive. 69 million Chinese children are left behind in rural areas, and girls are harmed even when trade liberalization increases family income.

Internal labor migration is widespread in low- and middle-income countries despite substantial unemployment among local workers in destination labor markets. A common explanation is that firms hire migrants because migrants accept lower wages than locals. We test an alternative demand-side account in a field experiment with construction firms in Patna, India, in which firms receive real offers to hire either local or migrant workers. At the prevailing market wage, demand for migrant workers is 78.7% higher than demand for locals, and firms’ average willingness to pay is 12.7% higher for a migrant. To understand the preference, we experimentally vary migrants’ residence status. When firms cannot have workers reside on site, the demand premium falls to 56%, indicating that on-site residence affords firms greater control over migrant labor. Combining cross-cutting contract variations with incentivized belief elicitation, we decompose the premium into three channels. The largest is differential absenteeism and reneging risk, at 39.8%. The scope to extract unpaid hours, through both on-site residence and migrants’ generally longer hours, accounts for 29.3% and higher perceived effort per hour accounts for 30.9%. Firms’ incentivised beliefs are consistent with this decomposition. They expect migrants to work longer hours, miss fewer days, and have weaker recourse in disputes, with these beliefs strongest for on-site migrants. Workers themselves recognize these conditions. In incentivized belief surveys, both migrants and locals workers expect migrants to face more unpaid extra work, more wage theft, and weaker recourse than locals. In a real-stakes job-offer experiment, migrants forgo meaningful daily wages for insurance against wage theft, dispute mediation, and an outside-option guarantee. A matching exercise places experimental workers into firms from our sample, where migrants complete more contract days, are more likely to be asked for unpaid extra hours, and receive productivity ratings from firms more favorable than those given to locals. The firm- and worker-side results together suggest that migration and persistent local unemployment can coexist in informal labor markets because firms value employment relationships that grant them greater control over workers, not just labor input.

Mechanistic explanations—descriptions of a system through the causal interactions of its parts—play a key role in human cognition and scientific progress. Despite their importance, we lack systematic evidence on whether and how mechanistic explanations help lay decision-makers interpret information in complex economic environments. We evaluate the causal impact of including mechanistic explanations in an information intervention: public demonstrations of fertilizer use for smallholder tomato farmers in Eastern Uganda. In all demonstrations, extension officers showcased the impact of a recommended fertilizer recipe. In the treatment group, officers also explained the mechanisms underlying the recipe’s effects—introducing the language of macronutrients and the causal processes linking nutrients, soil features, and plant growth. We collected detailed data on beliefs and behaviors from 797 farmers in a lab-in-the-field experiment conducted at the demonstration site and followed up with them over two growing seasons. In the lab-in-the-field, treated farmers generalized more effectively—making better substitution and arbitrage decisions among fertilizers and achieving 9% higher simulated profits in an incentivized fertilizer-application task. At endline, treated farmers’ real fertilizer choices reflected improved nutrient timing and balance, and their yields were 14% higher.

Low-income workers face frequent shocks to their ability to work, including social obligations, illness, and volatile demand. We examine how this affects workers’ labor supply by testing whether labor supply is habit forming. Using a field experiment with casual urban laborers in India, we randomly provide treated workers with small financial incentives for attendance at labor stands for 7 weeks, leading to a 26% increase in labor supply. We then test for the persistence of impacts after the incentives are removed. First, we see a persistent 18% increase in labor supply over the following 2 months, resulting in a 10% increase in overall employment days. Second, treated workers exhibit a higher willingness to accept work contracts that are of longer duration and less flexible. Third, labor market disruptions deplete habit: shocks that temporarily pull workers out of the labor market instantly eliminate treatment effects on labor supply and work contract choice; in the absence of these shocks, we cannot reject that there is no decay in persistence over time. Fourth, we see no “fixed cost” changes in household time use, or learning among workers or employers—consistent with “true” state-dependence in labor supply. Rather, workers self-report an increase in automaticity—suggesting a change in their psychological default. Fifth, employers accurately predict treatment effects, and are willing to pay to hire workers who have been treated with our habit intervention. Our results suggest that in low income settings, intermittent employment and frequent shocks may inhibit workers from becoming habituated to regular work—with potential implications for absenteeism and barriers to structural transformation in developing countries.

The increasing integration of the cultural, social, and historical contexts of societies into empirical research has been limited by the availability of ethnographic data. We construct a new database of ethnographic data for groups within the African continent, using the Ethnographic Survey of Africa (ESA), a 50-volume series containing descriptions of ethnic groups published between 1950 and 1977. The database provides information on a wide range of economic, social, and cultural characteristics – approximately 1,800 variables in total – currently unavailable from quantitative ethnographic sources. It also provides information on groups observed at multiple levels – over 2,200 groups in all – including granular variation within typically defined ethnicities. This document outlines the construction of the database, provides a comprehensive overview of the data available, and validates the information against existing ethnographic data. We also provide applications that exemplify the insights that can be gained from the database relative to existing sources.

This paper examines the effect of limited access to grid electricity on inequality in South Africa. We document that exposure to scheduled outages is similar across neighborhoods. Nevertheless, the outages had a larger adverse effect on luminosity and mobile phone activity in poorer ones. Consistent with the use of independent generators to substitute for grid energy during outages, we find that the impact of outages is muted when the world prices coal and oil, which are used to fuel these generators, are low. Using measures of solar panel usage inferred from satellite images, we find that the adoption of this other alternative energy source increased more in higher-income neighborhoods than lower-income ones. Together, the results suggest that low public energy provision and low-income households’ limited ability to adopt alternative energy sources for independent generators contributed to inequality in South Africa.

Randomized Controlled Trials (RCTs) are the gold standard for evaluating the effects of interventions because they rely on simple assumptions. Their validity also depends on an implicit assumption: that the research process itself, including how participants are assigned, does not affect outcomes. In this paper, I challenge this assumption by showing that outcomes can depend on the subject's knowledge of the study, their treatment status, and the assignment mechanism. I design a field experiment in India around a soil testing program that exogenously varies how participants are informed of their assignment. Villages are randomized into two main arms: one where treatment status is determined by a public lottery, and another by a private computerized process. My design temporally separates assignment from treatment delivery, allowing me to isolate the causal effect of the assignment process itself. I find that estimated treatment effects differ across assignment methods and that these effects emerge even before the treatment is delivered. The effects are not uniform: the control group responds more strongly to the assignment method than the treated group. These findings suggest that the choice of assignment procedure is consequential and that failing to account for it can threaten the interpretation and generalizability of standard RCT treatment effect estimates.

This paper argues that retail firms' dual role as providers of goods and credit constrains firm size in poor countries. We advance this argument in four steps. First, store credit is prevalent: in an original 1,050-firm survey, 27% of sales are on credit (at zero interest), and 61% of firms sell on credit. Second, credit sales face information frictions: in transaction data from an Indian bookkeeping app, default rates exceed 20% and are higher for small firms; stores withhold future sales to incentivize repayment (consistent with moral hazard); and credit limits rise over time (consistent with screening on types). Third, public 'store credit scores' relax both frictions: combining in-app event studies with a customer profiles experiment, we find that scoring induces repayment from existing customers and expands credit sales to customers outside firms' social networks. Finally, we use these facts to discipline a model in which heterogeneous firms grow through customer acquisition, and evaluate the effects on the firm size distribution of unbundling sales from credit through consumer credit scores or credit cards.

Many group identities that influence economic behavior are imperfectly observed. Individuals and institutions often conceal identity markers to limit discrimination. Yet concealment also creates uncertainty about group membership, hampering coordination in social interaction. To study this tradeoff, we paired high-and low-caste men for collaborative data entry work in North India. We randomly assigned each mixed-caste pair to either be: (i) introduced by full names, making caste common knowledge; (ii) introduced by first names only, making caste disclosure a choice; or (iii) instructed not to disclose caste. The two concealment conditions substantially reduce the accuracy of beliefs about a partner’s caste and confidence in those beliefs. They also weaken workplace relations, lowering trust, willingness to interact, and perceived productivity—consistent with identity helping structure social coordination. Evidence on mechanisms shows that identity concealment inhibits authentic interaction, making workers less able to express their “true selves,” while certainty about a partner’s identity is associated with stronger workplace ties. Concealment leaves a sizable caste disparity in higher-status role assignment intact, suggesting minimal impacts on discrimination. We conclude that where group identities are socially entrenched, reducing their legibility may undermine intergroup relations.

We examine the importance of norms of revenge and punishment in perpetuating armed conflicts. Our analysis leverages the ‘culture of honor’ hypothesis from social psychology, which posits that traditional herding practices generate moral systems conducive to revenge-taking. We find that the descendants of herders (i) experience more frequent civil and non-civil conflicts; (ii) are more likely to be involved in conflicts motivated by retaliation; and (iii) exhibit a greater emphasis on revenge-taking in contemporary surveys and historical data. Our evidence suggests that a traditional form of subsistence generated a functional morality that continues to shape conflict across the globe today.

Experiments deliver credible treatment-effect estimates but, because they are costly, are often restricted to specific sites, small populations, or particular mechanisms. A common practice across several fields is therefore to combine experimental estimates with reduced-form or structural external (observational) evidence to answer broader policy questions such as those involving general equilibrium effects or external validity. We develop a unified framework for the design of experiments when combined with external evidence, i.e., choosing which experiment(s) to run and how to allocate sample size under arbitrary budget constraints. Because observational evidence may suffer bias unknown ex-ante, we evaluate designs using a minimax proportional-regret criterion that compares any candidate design to an oracle that knows the observational study bias and jointly chooses the design and estimator. This yields a transparent bias-variance trade-off that does not require the researcher to specify a bias bound and relies only on information already needed for conventional power calculations. We illustrate the framework by (i) designing cash-transfer experiments aimed at estimating general equilibrium effects and (ii) optimizing site selection for microfinance interventions.

Many of the world's poor have experienced trauma. We argue that memories of this trauma interfere with the process of future simulation, diminishing the ability to see how actions today can improve outcomes tomorrow. We introduce guided mental experiencing (GME) -- an intervention in which participants mentally simulate pathways between their actions and desired economic outcomes -- as a response, and study GME's impact in two RCTs. In a population of Eritrean refugees in Ethiopia, GME increases the ability of refugees to see a positive future, increases their intent to stay in Ethiopia, increases labor force participation and improves self-reported welfare. In a population that has experienced violence and poverty in Colombia, a traditional entrepreneurial training program reduces the ability to imagine a future in business and worsens economic outcomes. Integrating GME into entrepreneurial training restores future thinking and removes these negative economic effects. The largest gains accrue to the most traumatized participants in our samples.