SEED 2026

Gift giving is a ubiquitous social practice. Using a lab-in-the-field gift exchange experiment in a birthday context, I study whether reciprocity behavior is driven by social norms or by perceived kindness, and whether reciprocity norms account for differences in sender means. Inducing experimental variation in sender endowments and the salience of sender sacrifice, I test whether reciprocity responds to the sender’s sacrifice relative to their means or primarily to its nominal value. I find that reciprocity is strongly anchored in social norms rather than perceived kindness, and that these norms largely track nominal gift values rather than sender means. As a result, prevailing reciprocity norms can impose disproportionately higher burdens on poorer individuals. Counterfactual simulations suggest that shifting reciprocity norms toward greater sensitivity to sender means, rather than eliminating reciprocity altogether, could improve equity while preserving the social function of gift exchange.

This paper studies whether collective aspiration building can enhance cooperation in groups. We first develop a theoretical model in which individuals experience a psychological loss when contributing less than a shared aspiration level, mitigating free riding incentives but also, in the process, generating multiple equilibria in group cooperation. Raising collective aspirations can therefore move groups to a higher-cooperation equilibrium. We test these predictions using a lab-in-the-field experiment with women's self-help groups, where half received a one-day “Dream Building” training designed to raise collective aspirations and group identity. Following the intervention, treatment groups contributed 30% more on average in a public good game compared to control groups, with the effect persisting, though attenuated, even three months later. These findings highlight the potential for psychological interventions to overcome free riding in groups, offering valuable insights for the design of group-based development programs.

Ethnonationalist governments adopt symbolic policies that challenge the status of ethnic and religious minorities as equal members of the nation. We propose that exposure to such policies reduces the cognitive bandwidth of marginalised groups, leading to suboptimal economic decisions. We test this hypothesis in India. About 1,800 gig workers performed data entry and information processing tasks in a field experiment. In the course of completing their tasks, workers were randomly exposed to social media content referencing exclusionary policies. The policies varied in whether they were primarily symbolic or whether they were also likely to cause direct material harm to affected minority participants. We measured decision making by allowing workers to choose between two payment contracts, one of which would objectively lead to higher earnings for a given worker. Despite no change in productivity, treated participants were significantly more likely to select economically suboptimal contracts, leading to lower earnings. We explore additional data, including cognitive tests and categorisations of mistakes, that suggest that wrong contract choice is driven by increased cognitive load. Our findings demonstrate how even purely symbolic exclusionary policies can have tangible economic consequences, potentially exacerbating marginalisation of minority groups through cognitive and psychological channels.

We examine the impact of labor market information on job search behavior and employment outcomes of job seekers in urban India. Through a cluster-randomized intervention in Delhi, over 3,000 men and women are randomized into a treatment group in which they are informed and assisted in accessing a job matching platform, supplemented by phone messages that deliver either non-personalized or personalized labor market information; or a control group in which neither assistance with the matching platform nor any phone messages is given. Month-level analysis up to a year after intervention shows that while men reduced overall job search, women's search shifted to online modes. Although any employment impacts petered out in the long run, we find evidence of a marginal shift in occupational structure towards salaried work away from casual jobs, along with higher earnings, for women in later months. These impacts were relatively stronger in the personalized messaging group. We highlight the role of expectations and (mis-)beliefs about the labor market in explaining our findings.

Can training interventions reduce poverty? In a large RCT in urban Bangladesh, we document four key facts about an NGO-run training program. First, there is minimal demand for training when participants are required to pay for the full cost of the program. Second, demand can be boosted substantially by either offering monetary discounts, or by relying on the NGO recruitment protocol, whereby NGO selectors spend significant time persuading potential trainees of the benefits of the programs. Third, while lowering training prices attracts poorer trainees on average, NGO recruitment skews selection towards less deprived individuals. Fourth, the marginal trainees attracted under the NGO recruitment protocol experience larger benefits from the program, while the marginal trainees attracted with price discounts experience lower benefits. We conclude by studying, through the help of a structural model, whether a combination of price discounts and the NGO recruitment protocol can target the program to high-deprivation, high-benefit individuals.

This paper studies how language barriers impact internal migration, the skill premium, and aggregate welfare using rich microdata from India applied to a quantitative spatial general equilibrium framework. I first document four empirical facts: (1) workers migrate less often to locations where they face high language barriers; (2) migrants with high language barriers are employed less often in speaking-intensive occupations; (3) migrants with high language barriers get a wage premium; and (4) these patterns are strongest for unskilled workers. To explain these facts, I then develop and estimate a static migration model in which heterogeneous workers sort across occupations and locations by skill and language, with wages accounting for worker selection and adjusting in general equilibrium. I show through the lens of the model how language barriers, by increasing worker sorting and selection, significantly obstruct internal migration, augment skill premium, and reduce aggregate welfare. As economies shift towards services, language barriers increasingly impede aggregate gains due to the rising prevalence of speaking-intensive occupations. In the absence of language barriers—relative to observed changes—structural change would have increased aggregate welfare by 1.9 percent. Finally, I calibrate costs of both program provision and learning languages to evaluate potential benefits of language programs for unskilled migrants. Using the calibrated model, I argue that welfare benefits of implementing language programs would outweigh costs.

Using information on universe of firms for both formal and informal firms from Ethiopian Manufacturing Enterprise Survey (1996–2010) matched with tariff and geo-referenced road network data from the Road Sector Development Program (RSDP), we estimate how trade liberalization and road infrastructure can jointly shape women's labor market outcomes in Ethiopia's manufacturing sector. We document two key findings: (i) a one percentage point decline in input tariffs increases the female employment share by 0.96 percentage points, driven partly by male-to-female labour substitution; and (ii) road access and trade liberalization are complementary: firms with above-median road quality experience a 1.32 percentage point larger increase in female employment share per unit tariff reduction. In terms of mechanism, drop in input tariffs led to expansion of activity towards female-intensive industries and this led to increase in female employment share. This increase is higher for firms with more than 50% share in female ownership. Our findings are robust to controls for local infrastructure, multinational presence, and output tariff changes, and provide the first evidence that road connectivity amplifies the gender-equalizing effects of trade reforms in Sub-Saharan Africa.

This paper examines the socioeconomic and welfare impacts of Kenya's 2019 demonetisation policy. Using a difference-in-differences design with Afrobarometer survey data, we document a stark urban–rural divide: demonetisation significantly accelerated mobile money adoption and improved crime perceptions in urban areas, but imposed severe, asymmetric income shocks on cash-dependent households. To rationalise these findings, we construct a Two-Agent New Keynesian (TANK) model featuring a cash-in-advance constraint and an informal sector subject to a working capital constraint. The model reveals that demonetisation triggers a “liquidity vacuum”, as informal firms hoard cash to survive, causing a severe recession. Furthermore, material welfare calculations demonstrate that digital infrastructure dictates economic resilience; rural cash-only households suffer the most catastrophic lifetime welfare losses because cash and digital finance remain technological complements rather than substitutes in agricultural economies. These findings underscore that the success of monetary modernisation hinges on pre-existing digital institutional capacity.

We examine how political shifts affect household economic sentiment and spending in identity-polarized settings. Using panel data on over 178,000 Indian households, we find that sentiment about personal finances—and, to a lesser extent, the national economy—predicts expenditure, even after accounting for income changes. Using close state elections, we show that Muslims become markedly more pessimistic than Hindus about national economic conditions following victories by the Hindu-nationalist party, but exhibit relatively smaller differences in personal financial sentiment, and no detectable divergence in expenditure. A Bayesian learning framework explains the insulation of consumption from politically induced sentiment shocks through the limited transmission of macro-beliefs to individual behaviour in high-volatility environments.

Global supply chain disruptions have become ubiquitous, driven by uncertainties arising from political conflicts, trade policies, and natural disasters. Yet little is known about firms' adaptation strategies in response to such shocks. We exploit the sudden disruption in the coal market caused by the Russia–Ukraine war and, using administrative firm-transaction data, examine how coal-procuring firms in India reorganized their supply chains in response to the subsequent surge in global coal prices. We find that firms more exposed to the shock reshore their coal input sourcing within the country, deepening existing domestic connections in the short run, while expanding their domestic supplier networks over time.

I show novel evidence using French administrative data that automation reduced wage markdowns by 3% — reflecting increased labour market power — and increased concentration by 20%, with effects intensifying over time. To quantify the aggregate implications, I develop a general equilibrium model with oligopsonistic labour markets, endogenous automation adoption, and firm entry. Using the reduced-form evidence to discipline the model, I find that median welfare would be approximately 13% higher, and aggregate output 2% higher, had automation occurred without raising labour market power.

This study examines theoretically and empirically the relationship between R&D outsourcing and the intensive and extensive margins of R&D. We develop a theoretical model in which heterogeneous firms decide whether to undertake R&D activities, whether to outsource R&D tasks and the volume of both activities. We test the main predictions of the model using Spanish firm-level data, by using a staggered adoption difference-in-difference (DiD) approach combined with matching methods. Our results show that R&D outsourcing positively affects both a firm's internal and total R&D investment, consistent with our theoretical prediction when the elasticity of substitution between different sources of knowledge is small enough. We also find that the results differ by a firm's export status and the nature of the outsourcing (i.e., domestic or international). At the industry level, we find that in industries where R&D outsourcing profitability is higher, fewer firms invest in R&D. This is consistent with R&D outsourcing being a relatively expensive activity that benefits the most productive innovative firms at the expense of the least productive innovative ones.

This paper examines the effectiveness of forest protected areas (PAs) in a dynamic setting. Using georeferenced data, we investigate whether the establishment of PAs in the Brazilian Amazon immediately accelerates deforestation in surrounding areas, and whether logging activities gradually shift back into the PAs over time. We develop a spatial dynamic optimization model to show that the dynamic displacement of deforestation is fundamentally driven by the nonrenewable nature of trees in the Amazon.

A core challenge in developing countries is the lack of productive and well-paying jobs, with informality being a persistent feature of labour markets. In the lecture, I will first describe the stylised facts on informality, and discuss the causes of high informality. I will then provide a framework of analysis that takes into account the heterogeneity of informal work in developing countries, and apply it to India subnationally. I will also discuss whether policies such as minimum wage increases in societies with high informality and low enforcement capabilities are effective in reducing inequality using India as an example. Finally, I will discuss what the possible implications of AI may be for labour markets in developing countries.