Research
Papers
Religious leaders' compliance with state authority: Experimental evidence from COVID-19 in Pakistan
(previous version circulated under the title "Persuasion and public health: evidence from an experiment with religious leaders during COVID-19 in Pakistan"
World Bank Economic Review, May 2024
Supported by IPA and IGC.
Ungated version; Baseline and treatment instrument; Mystery shopper instrument
We use a Randomized Controlled Trial in Pakistan to test whether one-on-one engagement with community religious leaders can encourage them to advise congregants to comply with public health guidelines from state authorities. We test whether religious content in this engagement increases its effectiveness. We find that simple one-on-one engagement significantly improves the advice given by religious leaders to congregants on preventing COVID transmission in the mosque. Engagement was equally effective with or without explicitly religious content. Treatment effects are driven by the subsample who are already convinced of basic information about COVID at baseline, suggesting the treatment does not work by correcting basic knowledge about the disease. Rather, it may work through the effectiveness of one-on-one engagement that reinforces existing knowledge and connects it to actions that respondents can take in their role as community leaders.
Data, discretion and institutional capacity: Evidence from cash transfers in Pakistan
(with Muhammad Haseeb; supported by the British Academy and the Lahore School of Economics)
Journal of Public Economics, February 2022
Ungated version; Web appendix; Survey questionnaire
(Previous version entitled "Reforming Institutions: Evidence from Cash Transfers in Pakistan")
Administrative data is key to many government functions; but generating and maintaining it is costly and challenging in low-income countries. We study an overhaul of public assistance in Pakistan that created a national database of household assets and used the data to means-test cash transfers, eliminating discretion in their allocation. We use difference-in-differences and regression discontinuity approaches to quantify the effect of this reform. Favoritism and transfers to wealthy households dropped; we estimate welfare benefits of the reform seven times its costs. The reform improved public support for social assistance, creating a robust institution that survived political transitions.
Demand for Safe Spaces: Avoiding Harassment and Stigma
(with Florence Kondylis, Arianna Legovini, Astrid Zwager, and Luiza Andrade)
Revisions submitted to Journal of Development Economics
What are the costs to women of harassment on public transit? This study randomizes the price of a women-reserved "safe space" in Rio de Janeiro and crowdsource information on 22,000 rides. Women in the public space experience harassment once a week. A fifth of riders are willing to forgo 20 percent of the fare to ride in the "safe space". Randomly assigning riders to the "safe space" reduces physical harassment by 50 percent, implying a cost of $1.45 per incident. Implicit Association Tests show that women face a stigma for riding in the public space that may outweigh the benefits of the safe space.
Drivers of change: How intra-household preferences shape employment responses to gender reform
Revisions requested, American Economic Review
(with Chaza abou Daher, Erica Field, and Kendal Swanson)
Billions of women still face legal barriers to economic inclusion, yet it is unclear whether lifting these barriers is sufficient to enhance their economic participation. We conduct a field experiment to quantify the impact of a major legal reform - the lifting of the Saudi women’s driving ban - on women’s employment by randomizing rationed spaces in driver’s training. Two years later, women in the treatment group are 61% more likely to drive, 19% more likely to leave the house unchaperoned, and 35% more likely to be employed. However, they are also 19% more likely to require permission to make purchases. These patterns vary systematically with marital status: although physical mobility increases for all women, treatment effects on employment are only observed among never-married and widowed women, who negotiate employment with their fathers. Married and divorced women with children, over whom husbands and ex-husbands have leverage, actually \emph{exit} the labor force and experience decreased spending autonomy. We posit that these patterns reflect differences in male family members’ support for women’s employment. They provide evidence that men’s resistance to wives’ employment poses a binding constraint to female labor force participation when legal restrictions are relaxed, but also that men are more open to granting their daughters economic rights, as has been posited in the literature. The results underscore the importance of intra-household responses to gender reforms, which have the potential to counteract legal gains in women’s freedoms, and help explain why potential economic gains from lifting discriminatory laws often go unrealized.
With Hamna Ahmed and Dareen Latif
Revisions requested, Journal of Public Economics
The accountability relationship between a principal and agent (such as a politician and a bureaucrat) may lead to suboptimal outcomes if the principal either has limited ability to exert consequences on the agent or imperfectly observes the agent's performance. With limited observability, an increased threat of consequences may backfire, leading to multitasking or manipulation of data. We use rich administrative microdata on crime in Pakistan to investigate how police respond to (1) changes in the politician's ability to exert pressure on them for performance, using variation in political alignment of neighborhoods to the governing party over time; (2) an increase in the politician's ability to observe the agent's activity, using the rollout of a new complaints tracking database; and (3) how the two interact. We find that police are robustly less likely to record and process citizen complaints in areas that supported the governing party. We rule out the possibility that our results are explained by better crime prevention or reduced crime: logged complaints are less likely to be pursued, staffing levels and response times do not change, and criminal property crime reports appear to be ``downgraded" to non-criminal loss reports. Instead, results are consistent with a mechanism of officials under-reporting crime statistics under pressure to keep crime levels low. As a new complaints tracking database is rolled out, improving observability of police handling of citizen complaints, the effect of political alignment attenuates.
Barriers to Entry? Decomposing the gender gap in job search in urban Pakistan
with Elisabetta Gentile, Nikita Kohli, Nivedhitha Subramanian, and Zunia Tirmazee
Revisions requested, Journal of Labor Economics
Gender gaps in labor market outcomes persist in South Asia. An open question is whether supply or demand side constraints play a larger role. We investigate this using matched data from three sources in Lahore, Pakistan: representative samples of jobseekers and employers; administrative data from a job matching platform; and an incentivized resume rating experiment. Employers’ gender restrictions are a larger constraint on women’s job opportunities than supply-side decisions. At higher levels of education, demand-side barriers relax, allowing women to qualify for more jobs but at lower salaries. On the supply side, educated women become more selective in their search.
Why don't jobseekers search more? Barriers and returns to search on a job matching platform
(with Erica Field, Rob Garlick and Nivedhitha Subramanian; supported by DFID-IZA GLM-LIC, IGC, IPA, and PEDL)
Revisions requested, Journal of Labor Economics
Jobseekers face multiple barriers with potentially different implications for the level of search and returns to increasing search. An experiment on a job search platform in Pakistan shows that lowering users’ psychological cost of initiating job applications increases applications by 600%. Returns to the marginal applications induced by treatment are approximately constant rather than decreasing, in contrast with intuitive job search models. This pattern is consistent with a model in which heterogeneous psychological costs of initiating applications, potentially due to heterogeneous present bias, lead some jobseekers to miss applying to even high-return vacancies. Additional experiments and measurement reject alternative behavioral and non-behavioral explanations. Our finding of constant returns to marginal search effort, combined with limited spillovers onto other jobseekers, raises the possibility of suboptimally low search effort due to psychological costs of initiating applications.
Information gaps and de jure legal rights: Evidence from Pakistan
(with Erica Field and the Punjab Commission on the Status of Women, supported by EDI and J-PAL)
The degree of freedom that women enjoy over key life choices such as, when and whom to marry and divorce are intrinsically valuable rights with important welfare consequences. Over the last half century, there has been substantial progress in legal protections for women's rights worldwide. However, women’s de jure rights over marriage and divorce are substantially more progressive than the de facto practice of the law. Existing literature on the impact of women’s legal rights has focused primarily on de jure legislative changes. There is more limited work on what determines de facto access to these rights. In 2015, the province of Punjab, Pakistan, passed a set of legal reforms imposing penalties on families as well as marriage registrars for violations of women’s rights in marriage, including rules on the completion of the marriage contract, a binding legal contract at the time of marriage that governs divorce and financial rights in marriage. Legally the informed consent of the bride and groom to these terms is the sole requirement for the marriage contract; however, the terms of the contract are often decided by the spouses' parents, with strong influence from the religious-legal official who conducts and registers the marriage, the marriage registrar. Yet until 2017 there was no education or training requirement to become a marriage registrar; and marriage registrars were unaware of legal protections for women's rights in marriage including the new reforms. For such reforms to be effective in practice, the parties to the marriage negotiation and contracting process must be informed about the law. We collaborated with the government to conduct a randomized evaluation of the first ever mandatory training of marriage registrars. This allows us to establish whether improving the information environment can make legal reforms more effective and strengthen institutions.
Transport, urban labor markets, and women's mobility: Experimental evidence from urban Pakistan
(with Erica Field; supported by DfID-IZA; IGC; Gates-J-PAL; PEDL; NSF, 3IE and ADB)
In many contexts with conservative norms or high crime, female workers may face greater restrictions on their physical mobility within the city. This limits women's opportunities in the labor market and the pool of workers that firms can attract and retain. In this study, we test the overall impact of transport to work on men, women, and the differential impact for women of women's-only transport. We experimentally vary access to a transport service, and study how this affects job search, employment outcomes, and employers' access to a broader pool of candidates. For women, the comparison between women's-only and mixed-gender transport services allow us to quantify the benefit of a reduction in cost, an improvement in safety and social acceptability, and the two combined, on women’s mobility and labor supply.
Infrastructure investments, public transport use and sustainability: Evidence from Lahore, Pakistan
(with Hadia Majid and Ammar Malik; supported by the IGC, IFPRI-PSSP, 3IE and ADB)
In the developing world, private vehicle ownership is growing rapidly, increasing mobility but increasing congestion and pollution. To address these challenges, hundreds of cities in developing countries have built mass transit systems in recent years. Yet most research on urban transport has focused on developed countries. We collect microdata to study the impacts of a transit line in Lahore, Pakistan, using as a comparison group areas that are similar at baseline and slated for future transit routes. We make four contributions. First, we quantify the impact of transit on commuting: access to the new transit line reduced both the time and cost of commuting, and increased public transport use in treated areas by 30%. Second, we examine the patterns of use of transit. At baseline, public modes were primarily used by low-income, low-educated commuters; we find that mass transit benefited low- and middle-income commuters, and they report higher willingness to pay. This implies that transit can be made more financially sustainable through better targeting subsidies. Third, we test for responses in other markets. We find that the real estate prices did not change overall, but decrease in the center relative to the periphery. We also see a quantity response in the land market: transit stops increased nearby building cover. This elastic supply response allows transit to increase public transport accessibility on two margins: first, by directly connecting areas of existing activity, and second, by concentrating economic activity near stops. Fourth, we examine externalities averted due to riders switching to transit. We find a reduction in travel time on private modes, suggesting a reduction in congestion. We also find evidence of a reduction in particulate matter air pollution observable in satellite data.
Political favoritism and production distortions
(with Sadia Hussain and Ignacio Rodriguez-Hurtado)
Draft available on request.
We document political influences on the procurement of wheat in Pakistan using panel administrative and farm survey data. Using a panel fixed effects approach, we document that wheat procurement increases in governing party constituencies when the government procurement price is high relative to world prices. We document that this political distortion leads to a production response: when political changes expand their access to the government program, farm households respond by increasing the proportion of their land dedicated to planting wheat, and selling more wheat on the market.
Work in Progress
Intended and Unintended Consequences of a De Facto Inheritance Reform
(with Sabrin Beg, Erica Field and Jeremy Lebow; supported by EDI, NBER and the World Bank)
De jure reforms to improve women’s legal rights are often not enforced in practice. In this study, we examine the effects of a land records reform in Punjab, Pakistan on de facto land rights of females, who are legally entitled to a share of parental land. This reform digitized and centralized land records and created a biometric verification requirement intended to limit females' exclusion from the inheritance process. Exploiting the staggered rollout of the reform across Punjab as well as the quasi-random timing of father/husband deaths, we find that it significantly increased the probability of female inheritance from 13% to 22%. However, we also find evidence of unintended consequences for younger women: they marry earlier and without consenting to their choice of spouse, marry lower quality spouses, are more likely to have children by their early twenties, and are more likely to drop out of school if they have large, expected increases in inheritance. These responses may be due to intrahousehold resource reallocation away from daughters in response to a forced inheritance transfer. The unintended marriage effects could also be driven by female land inheritance being used as a substitute for more liquid forms of dowry, as well as attempts to keep land in the family, either by marrying daughters off younger to exclude them from family negotiations or through consanguineous marriages.
Can biometric verification get more cash to poor women? Evidence from Pakistan
(with Muhammad Haseeb, Amen Jalal, and Bilal Siddiqi; supported by the Gates Foundation-World Bank ID4D initiative)
Many countries target low-income women for cash transfer programs but struggle to ensure that female beneficiaries personally receive and retain control over these funds. Biometric verification may address this issue by requiring beneficiaries to personally withdraw funds after authenticating their fingerprints. While this may reduce unauthorized withdrawals and capture of funds by other household members, it may also increase time and money costs of withdrawal and cause unintended exclusion of eligible beneficiaries. We exploit the staggered rollout of biometric identity verification in Pakistan’s Benazir Income Support Programme to assess its impact on the delivery of cash to low-income women. We find that biometric verification increases women’s control over how to spend the cash. In areas where the new system does not require additional payment agent involvement, beneficiaries also report receiving slightly more cash under the new system. However, we also find an increase in reports of side payments paid involuntarily to access the cash transfer, and a decline in beneficiary satisfaction with the payment system. We explore mechanisms for the effects including mobility constraints faced by female beneficiaries and market power of payment agents.
What happens when cash transfers suddenly stop? Quasi-experimental evidence from new BISP eligibility criteria in Pakistan
(with Nasir Iqbal, Amen Jalal, and Mahreen Mahmud; funded by IFPRI and IPA)
A growing body of evidence shows mostly positive impacts of cash transfers on a range of outcomes. However, there is limited work, empirical or theoretical, on what happens when long running unconditional cash transfers stop. Understanding how households cope when cash transfers stop, either because they no longer meet the eligibility criterion because their economic position has improved or cuts to the funding pot result in a more stringent eligibility criterion, is crucial. This is because cash transfer programmes are costly and may not be expected to provide support permanently. In this study, we use a regression discontinuity approach to examine the impact of the discontinuation of cash transfers on households in Pakistan who have been receiving transfers over a ten year period.
Incentivizing NGOs: A Field Experiment in Pakistan
(with Hamna Ahmed and Simon Quinn; supported by PPAF, IGC and NSF)
Donors and developing country governments have experimented with alternative ways to deliver services and assistance to communities and to individuals. One such approach is to deliver funding for basic services through non-governmental organisations. Incentive issues have been studied extensively in government bureaucracies. However, similar issues also arise when public spending is channelled through the non-government sector, which may reduce state effectiveness in policy delivery; yet there is very little research on these issues. In addition, we know very little about how public spending through non-government organisations affects service delivery through the traditional bureaucracy. Under what circumstances does it complement or bolster mainstream public service delivery, providing information or exerting citizen pressure on the government? Under what circumstances does it “let government off the hook,” substituting for services that would have been provided by the state? In this project, we are collaborating with the Pakistan Poverty Alleviation Fund to study these issues. PPAF receives support from both donor and Pakistani government sources, and funds local volunteer-run NGOs across Pakistan to provide public services in a wide range of sectors. We are collaborating with PPAF on a Randomized Controlled Trial with 836 volunteer local NGOs across Pakistan. As a part of the trial, PPAF is randomly varying the reporting obligations and incentives for those NGOs. The results of the RCT and related analysis on a rich panel dataset on these NGOs, their communities and local government offices will help shed light on how monitoring and incentives can be improved in the NGO sector; how the staff of a complex organization (PPAF and its partners, who disburse the funds and support the local NGOs) respond to and manage new information; and on the relationship between publicly-funded NGO sector and traditional state institutions.
Institutional capacity as an organizational challenge: a field experiment in Pakistan
(with Hamna Ahmed and Simon Quinn; supported by PPAF, IGC and NSF)
Large organizations, such as firms or bureaucracies, are often structured as complex hierarchies. Theory suggests two features of an organizational hierarchy may matter for its performance: information flow within the hierarchy; and divergent preferences of the members of the hierarchy. However, we have limited empirical evidence on how either affects organizational capacity and performance. In this study, we shed light on these issues through a novel field experiment involving a large donor organization and over 800 recipient community organizations across Pakistan. The design allows us to test how each part of a large, complex organization (the donor) responds to new information on performance (of recipient organizations) on key performance indicators, and how the responses of both donor and recipients to new information and incentives relate to organizational characteristics of theoretical importance, including divergence of preferences between members of the organization; communication costs between parts of the organization; and decentralization of decisionmaking authority.
Misreported Applicant Qualifications: Implications for Job Search and Hiring
(with Erica Field, Rob Garlick and Nivedhitha Subramanian; supported by DFID-IZA GLM-LIC, IGC, and PEDL)
Misinformation in the job market, such as candidates exaggerating or fabricating information on their CVs, can reduce firms' willingness to hire job candidates, particularly outside their existing networks. It can in turn reduce jobseekers' willingness to invest in their skills if they anticipate they cannot distinguish themselves from the competition who they anticipate may be exaggerating on their CVs. These problems may reduce total employment and earnings and inhibit skills investments, ultimately reducing economic growth. In this project, we run two experiments as part of a job search assistance service in Pakistan, Job Talash, to test whether improving the flow of information available to firms through a centralized reference checking service can change job search, hiring behavior and increase skills investment in jobseekers. The findings from this project will be useful for public policies in Pakistan and in other developing countries to help increase employment and improve skills, ultimately contributing to inclusive economic growth.
Total recall: duration versus frequency of surveys for measuring labor market outcomes
with Erica Field, Rob Garlick, and Nivedhitha Subramian; supported by IPA
There is substantial variation across developing country surveys in labor market measurement. Different surveys appear to use different recall periods, definitions of employment and search, and levels of detail in prompts. This variation may reflect open questions about how best to collect labor market data in these settings. For example, there is recent evidence that labor market measures are sensitive to changes in recall periods and detail of employment questions, with mixed evidence on changes in survey medium and use of proxy respondents. Labor modules based on Living Standards and Measurement Surveys have been carefully designed to measure labor in agricultural settings. But these are less tailored to for urban labor markets. National Labor Force Surveys often use short surveys that imperfectly capture nuances like occasional work, irregular payment periods, unpaid family work, and multiple jobs. We carry out a series of survey methods experiments in the context of a large panel study in Lahore, Pakistan. We will use these experiments to understand how variations in survey frequency, recall periods, and questionnaire design influence measures of job search, employment, and earnings.
Interactive Mobile Support for Community Health Workers Promoting Breastfeeding: Investigating the Impact on Maternal and Infant Health in a Developing Country Context
(with Erica Field, Zohra Kurji, Laila Ladak, Zahra Shaheen, Nicola Singletary, Laura Stilwell and Zoha Waqar; supported by USAID-DIV, Weiss and FID)
It is estimated that hundreds of thousands of child deaths worldwide could be averted annually if optimal breastfeeding practices were achieved, particularly in resource-constrained settings where children are at highest risk of dying from diarrheal disease in infancy; however, in many countries, misinformed beliefs about breastfeeding are common and exclusive breastfeeding rates are low. In this project, we will collaborate with UNICEF and the Government of Punjab, Pakistan, to design an intervention designed to strengthen the knowledge and practice of community health workers and improve their ability to offer rural patients effective and timely clinical advice on breastfeeding, and use a Randomized Control Trial to test its impact in the field on maternal and child health outcomes.