We investigate gender discrimination in an online marketplace in Pakistan. Employing buyer profiles that signal gender, we experimentally engage in transactions with sellers on the platform. We find no evidence of discrimination in pricing or product quality, suggesting that digital marketplaces may neutralize traditional economic biases. However, significant gender differences persist in non-price interactions. Female buyers are significantly more likely to receive unsolicited messages and friend requests following transactions, primarily from male accounts. Linguistic analysis further reveals that male sellers exhibit greater verbosity, enthusiasm, and flirtatiousness towards female buyers. While these interactions may not constitute overt harassment, in conservative and patriarchal settings, such unsolicited contact, regardless of intent, can carry reputational and social costs for women. Our findings highlight that online marketplaces, even as they remove discrimination on economic outcomes, may pose subtle barriers to equitable participation.

We study possible labor market discrimination by workers towards their out-group employers as manifested via social preferences (altruism and reciprocity). We run a well-powered, model-based, lab-in-the-field experiment, recruiting 6,000 white American worker subjects from Amazon’s M-Turk platform for a real effort task. We also hire trainer subjects, our stand-in for employers, from a university. To worker subjects, we randomly (and unobtrusively) reveal the racial identity of the trainers, who may be white or black. We find evidence that white workers may discriminate against white employers based on altruism (working harder due to concern for the employer’s well-being). However, they may discriminate based on reciprocity in favor of their white employers (working harder because of a small gift). From the perspective of black (white) employers, altruism evokes a stronger (weaker) effort response, but gift-giving has no (positive) effect. The combined effect of altruism and reciprocity on worker effort is the same for black and white employers.

This chapter examines Pakistan’s tax system’s structural challenges and reform opportunities, highlighting its inefficiencies, inequities, and reliance on distortionary taxes. It advocates for a scientific approach to tax policy, emphasizing evidence-based decision-making, data-driven interventions, and institutional reforms. The chapter provides actionable pathways to improve tax compliance, enhance equity, and promote sustainable economic growth.

Point-of-sale (POS) devices promise to strengthen tax compliance by reducing information asymmetries through real-time transaction reporting. To spur voluntary adoption, a tax authority in Pakistan offered firms in the hospitality sector a reduced sales tax rate if they integrated with the tax authority. From December 2021 to June 2024, 234 businesses (189 integrating existing devices and 45 receiving tax authority-supplied units) linked to the tax authority. Exploiting this staggered adoption, we apply a doubly robust staggered difference-in-differences event-study on monthly VAT return data. We find a large immediate surge in reported sales and tax payments, driven almost entirely by extensive margin, but these gains vanish within a few months. By contrast, the intensive margin sees smaller yet persistent increases in reported sales and value added, while conditional tax liabilities remain flat due to the preferential rate. Integrated firms also exhibit a sustained 20 percent drop in reported input purchases, suggesting potential shifts toward informal sourcing due to the incentive structure for device adoption. Finally, matching POS records to tax returns uncovers widespread underreporting and incomplete device use, indicating significant unrealized revenue potential.

This paper analyzes the impact of the 2022 floods on agricultural supply networks in Punjab, Pakistan. We unlock high-frequency agriculture supply chain data on quantities and merge it with spatial remote-sensing data from February 2022 to December 2022 to assess disruptions to the movement of agricultural commodities between “source” supply regions and “destination” consumption districts, where a lot of the demand is driven in large urban centers. Using an event study design, we find statistically significant pre-event anticipatory effects that increase supply by 35%, followed by a reduction of up to 34% in the overall quantities compared to the baseline in the aftermath of the floods. We document heterogeneity over crop types and district sizes. Vegetables show the highest susceptibility to such shocks, whereas the trajectory of grains is suggestive of state interventions. We also estimate optimal road networks and show that flooding of roads does not significantly disrupt the supply lines. We test the robustness of our findings by generating a spatial flood risk profile for Pakistan and demonstrate that anticipatory effects exist only in high-risk regions, which also show relatively quicker recovery. This study contributes to the economics literature by offering new policy-relevant insights on the response of agricultural networks to natural disasters in the context of developing countries.

This paper estimates the causal impact of messages displayed on dynamic message signs adjacent to roads on reported near-to-sign crashes and crash severity. We match accident reports to displayed messages using minute-level time and location metrics, along with hourly data on traffic and weather conditions in Vermont from June 2016 through the end of 2018. We evaluate the impact of safe driving messages (behavior messages) differently from those that provide information about current or future road or weather conditions (information messages) and evaluate their impact on the number and severity of crashes. Using a mixture of difference-in-difference, regression discontinuity, and ordered logit, we show that behavior messages causally decrease the number of accidents by about 45% and the number of vehicles involved in accidents by 30%, while information messages do not, and neither causally impact whether or not an accident induced an injury or fatality. Dynamic Message Signs are popular due to easing the mental burden of traveling with drivers and appear to provide modest near-to-sign improvements in driving as measured by the number of crashes and crash severity.

In this paper, we examine the effect of COVID-induced lockdowns and compliance on the potato supply chain in Punjab, Pakistan. By combining granular price level information from various sources and primary survey data of over a thousand farmers, we show that the middlemen were the clear winners in the supply chains during COVID-induced lockdowns. During COVID, the middlemen retained a higher share of prices, both upstream and downstream of the vertical supply chain. Besides bearing the burden of reduced price shares, we also show that the farmers who lacked market linkages and depended on unlicensed agents were affected the most. This effect is accentuated when farmers receive additional services from middlemen, such as credit and transport.

This study investigates whether informal markets remain stuck in low formalization due to self-reinforcing dynamics, which we call the “informality trap.” We conduct a randomized field experiment among restaurants, targeting two channels: bringing informal businesses into the tax net and improving compliance among already registered firms. By experimentally varying the enforcement intensity across areas, we aim to understand whether improving the formalization of some businesses can trigger broader shifts across other businesses in the same area. The study will generate evidence on the dynamics of market formalization, the role of spillovers in tax compliance, and strategies to expand the tax base in developing country settings.

Economic growth in developing countries is often limited by the state’s inability to raise tax revenue. In many countries, tax administration systems rely on infrequently updated and out-of-date property tax valuations, and tax officials often employ significant discretion when assessing properties. These factors can lead to errors that could increase tax leakages or lower citizen trust in the state. This study addresses this challenge in two steps: first, by developing a computer vision algorithm that can use property images to predict property assessments and second, by testing how well the algorithm performs in identifying properties for reassessment.

Farmers in low-income countries typically suffer from low productivity. Most farmers remain subsistence farmers and can hardly cover their living expenses from farming. These farmers generally receive low returns on their investments, which prevents them from investing in technologies or crops that can offer the potential for higher returns. At the same time, consumers in these countries typically pay relatively high prices for agricultural produce. This creates a puzzling scenario of high consumer prices but low producer prices. In this research, we study the reasons for this price divergence and specifically explore the role of information asymmetries as well as output market linkages as a barrier that prevents farmers from getting better returns on their productive investments. We also explore the general equilibrium effects and how these interventions can change domestic supply chains’ structure and the intermediaries’ role. This work is funded by the IGC and J-PAL.  

The economic literature on labor market discrimination has been mostly focused on the demand side. This paper shifts the focus from the demand side (employers) to the supply side (workers) for a more nuanced understanding of labor market discrimination. I outline a simple principal-agent model of the labor market and study the contract design when workers have differential social preferences depending on the employer’s group identity (e.g., race). I show that workers’ identity-based social preferences imply that the workers are more productive (work harder) for the same identity employers. Workers’ productivity differential provides the incentive for nondiscriminatory employers to recruit workers from their own group and pay higher wages to own-group workers. The assortative matching of employers and workers leads to segregation of the labor force when there are enough same-group employers. Under-representation of employers of one group leads to adverse labor market outcomes for the workers of that group in terms of wages. I show that this has implications for interpreting the existence and source of discrimination in the labor markets. Specifically, I demonstrate that what is traditionally understood as discrimination by employers may, in fact, be a rational response to the worker’s differential social preferences towards the employer’s group identity. I also show that ignoring workers’ social preferences (and employers’ beliefs about them) may lead to misleading conclusions about the sources of discrimination.

Tax expenditures—preferential treatments such as exemptions, deductions, and reduced rates—are widely used by governments to pursue social and economic objectives without direct outlays. In many agrarian economies, agriculture emerges as a major beneficiary of these concessions, often resulting in substantial foregone revenues and uneven benefit distribution. Yet in many countries, the true incidence of these concessions is poorly understood: Do they translate into lower market prices for staple commodities, or are the benefits captured by input suppliers, distributors, or retailers? This ongoing study develops a unified, data‑driven framework to answer this question.

Digital marketplaces promise to lower barriers for entrepreneurs, yet women often encounter hidden obstacles that hinder their participation and growth. This study first analyzes administrative Facebook Marketplace data—tracking listing rates, buyer engagement, pricing, and completion outcomes by seller gender—to document real‑world disparities. Building on these insights, we then launch a large‑scale randomized audit experiment in Pakistan, posting identical products under male and female profiles with varied information levels across markets of different thickness. By measuring both pecuniary outcomes (response rates, offers, prices, sales) and non‑pecuniary interactions (unsolicited messages, harassment), we disentangle statistical from taste‑based discrimination. Funded by PEDL, CEPR, and LUMS, our ongoing research will generate rigorous evidence to guide targeted interventions and platform policies that promote gender equity in online entrepreneurship.

Consumer receipt lotteries—programs that reward shoppers with entry into prize draws for submitting formal sales invoices—have been adopted in many countries to promote the issuance of proper receipts and boost tax compliance, yet rigorous evidence on their broader impacts remains limited. In this study, we conduct the first randomized field experiment of a consumer lottery in the hospitality sector of Khyber Pakhtunkhwa, Pakistan, assigning business eligibility to participate in the lottery. Leveraging administrative VAT data, firm and consumer surveys, and mystery-shopping audits, we comprehensively examine both the direct and indirect effects of the intervention. Specifically, we assess the impact of the lottery on formal invoice issuance, reported revenues, and potential spillovers to non-participating firms, as well as changes in taxpayer–official interactions, rent-seeking behavior, tax morale, and perceptions of state capacity. We also investigate the motivations underlying consumer participation and the welfare trade-offs involved. By unpacking these channels, our experiment provides new evidence on the mechanisms, externalities, and cost-effectiveness of consumer-driven monitoring as a complement to traditional tax enforcement in low-capacity settings.