Research

Optimal transportation policies depend on demand elasticities that interact across modes and vary across the population, but understanding how and why these elasticities vary has been an empirical challenge. Using an experiment with Uber in Egypt, we randomly assign large price discounts for transport services over a 3-month period to examine: (1) the demand for ride-hailing services, (2) the demand for total mobility (km/week), and (3) its contributions to external costs (e.g. congestion). A 50% discount more than quadruples Uber usage and induces an increase of nearly 49% in total mobility. These effects are stronger for women, who are less mobile at baseline and perceive public transit as unsafe. Technology-induced reductions in the price of ride-hailing services could generate substantial benefits to users (6.1% of GDP) that would be accompanied by considerable increases in external costs (0.7% of GDP), with benefits accruing to the most affluent and costs being borne by the entire population.

We experimentally study the impact of relatively large enterprise loans in Egypt. Larger loans generate small average impacts, but machine learning using psychometric data reveals that "top-performers" (those with the highest predicted treatment effects) substantially increase profits, while profits drop for poor-performers. The large differences imply that lender credit allocation decisions matter for aggregate income, yet we find that existing practice leads to substantial misallocation. We argue that some entrepreneurs are over-optimistic and squander the opportunities presented by larger loans by taking on too much risk, and show the promise of allocations based on entrepreneurial type relative to firm characteristics.

Is the type of financial support provided to businesses more important than which businesses receive it? Loans and grants can lead to differences in optimal investments and in scope for moral hazard. We randomize 3,293 business-loan applicants into receiving a loan, cash grant, in-kind grant or nothing. All treatments equally increase income, yet there are large differences within a treatment group with impacts concentrated at the top of the distribution. Those who succeed with loans are observationally equivalent to those who succeed with grants, showcasing that owner heterogeneity is more important than the type of support received in microenterprise development.

We study how complementarities in skill may affect the returns to vocational training using a randomized controlled trial in Cairo, Egypt. Participants, who were college-educated, were either given a 4-week training in soft skills (e.g., grooming, time management), technical skills (e.g., Microsoft programs, English language), or a mix of the two (half of each). We find large differences in outcomes between the three treatments. The technical and mixed treatments do best in the short term, raising first-job income by about 15%, relative to both a control group and the soft-skill treatment. In the longer term, the mixed-skill treatment significantly outperforms the other two treatments, giving participants 20–27% higher income. The high returns for this group seem to come from climbing the job ladder to access jobs that require speaking English, which may be at higher-quality employers. Overall, the results suggest that curriculum details play an important role in the outcomes of vocational training programs and that leveraging skill complementarity can yield tangible benefits.

Hundreds of studies have examined the impacts of microfinance, finding mostly modest or disappointing results. In this chapter, instead of asking whether microfinance works on average, we study the varied impacts of microfinance. Using data from several prominent recent studies, we show that the heterogeneity in returns to microcredit, microsavings, and microinsurance is large. This means that even programs that are not effective on average could be transformational for some people. We call for researchers and policymakers to focus more on identifying those who will benefit from microfinance, and understanding why they do. Together this will improve the targeting of these interventions, increase their positive impact, and help improve the design of future products.

Aversion to "stigma"—disutility associated with a program or activity due to beliefs about how it is perceived—may affect labor market choices and utilization of social programs, but empirical evidence of its importance is scarce. Using two randomized field experiments, we show that stigma can affect consequential labor market decisions. Treatments designed to alleviate stigma concerns about taking entry-level jobs had small average effects on take-up of job assistance programs. However, using compositional analysis and machine learning methods, we document large heterogeneity in the responses to our treatments. Stigma significantly affects the composition of who takes up a program: the treatments were successful in overcoming stigma for older, wealthier, and working respondents. For other people, we show that our treatments merely increased the salience of the stigma without dispelling it. We conclude that social image concerns affect labor market decisions and that messaging surrounding programs can have important effects on program take-up and composition.

The future of travel will be characterized by changes in weather patterns and changes in transportation technology. How will these forces interact? We explore this question by utilizing a unique randomized experiment with Uber riders in Cairo, Egypt. We consider how very hot days (>35°C/95°F) affect transportation choices, how a sizeable price decrease (simulating a future with autonomous vehicles and access to cheaper transportation) changes travel, and how extreme weather interacts with these choices. We find that while travel will increase significantly in response to the price decrease, extreme weather dampens this effect by 26%. Individuals receiving subsidies also shift away from public transportation modes and towards private transportation modes, except when the public transit option is air-conditioned. These results provide important insights for policymakers when considering optimal travel policy for the future.

We study discrimination in hiring and its associated outcomes for the discriminators using a unique survey of Egyptian businesses. Discrimination against women is widespread and overt: about half (51%) of establishments directly admit that they prefer to hire men. The share varies widely across industries, from 60% in retail to only 16% in IT. Using a list randomization technique, we can rule out that discrimination against women is heavily stigmatized in Egypt, meaning that establishments are willing to admit it openly. We then provide novel suggestive evidence showing that discriminating against women is associated with hiring lower-quality workers relative to non-discriminators. We also provide guidance on the use of the list randomization technique and how to interpret it in settings with limited stigma.

Network-based hiring is a common form of recruitment in businesses across the world. We administered a unique survey of Egyptian retail establishments to study the use of these hiring methods. We document important differences in establishments' use of ties to the owner ("connections") and to employees ("referrals") and their relationships with hiring outcomes. While all types of establishments use referrals at similar rates, use of owner connections varies widely and is most common among small informal establishments. We develop a model of hiring which predicts that connections and referrals should have heterogeneous effects on hiring outcomes depending on establishment type. Our empirical results are consistent with the model's predictions. When high-productivity establishments use connections, the practice is associated with lower-quality hires (nepotism), yet when low-productivity establishments use connections, they find more productive workers. By contrast, referrals benefit high-productivity establishments more due to network homophily.

Low utilization of household credit in developing countries may be partially due to religious considerations. In a randomized marketing experiment in Jordan, this paper estimates the effect of sharia-compliant loan features on demand for credit. To comply with Islamic law, the sharia-compliant product uses a bank fee rather than an interest payment structure, while keeping the rest of the product features very similar. Sharia-compliance increased the application rate for loans from 18 percent to 22 percent, an increase in demand that is equivalent to a 10 percent decrease in interest rates. This study also randomly varied the price of the sharia-compliant loan and finds that less religious individuals are twice as elastic with respect to price as the more religious. By comparing reasons for refusal across treatment groups, this paper estimates that survey measures that try to assess the importance of religious objections to conventional credit overestimate the importance of this type of objection.

Two for-profit Philippine banks, aiming to increase microlending to the poor, incorporated a widely used poverty measurement tool into their loan applications and tested the tool using randomized training content. Treated loan officers were provided an explanation of the tool's purpose; exhortation tying the tool to the organizations' social missions; and reassurance that these data, conditional on other characteristics, do not predict default and thus should not jeopardize incentive pay based on portfolio performance. The control group training merely labeled the tool "additional household information." The strategy backfired, leading to no additional poor applicants and potentially lower-performing loans. Descriptive evidence suggests the training exacerbated loan officer misperceptions about compensation incentives and multitasking problems.

We use tailored surveys and benchmarking in the flat-weave rug industry to better understand the shortcomings of standard productivity measures. Quantity-based productivity (TFPQ) performs poorly because of variation in product specifications across firms. Controlling for specifications aligns TFPQ with lab benchmarks. We also collect quality metrics to construct quality productivity (the ability to produce quality given inputs) and find substantial dispersion across firms. This motivates interest in multidimensional productivity, or capability. As quality productivity is negatively correlated with TFPQ, revenue-based productivity (TFPR) may perform better at capturing capabilities in settings where better firms make products with more demanding specifications.

Poverty is persistent but the evidence regarding the existence of poverty traps is mixed. We provide the first experimental test of asset-threshold poverty traps, by randomizing households in Egypt into a standard asset transfer, a half transfer, or control. Both transfer groups show large, sustained gains in assets and consumption 40 months later. By comparing how assets grow after the transfer for households at different points in the baseline asset distribution, we find no support for the presence of a poverty trap. These results suggest that scarce anti-poverty resources need not be concentrated in large asset transfers to be effective.

The canonical approach to corporate governance posits a tradeoff between maximizing shareholder profits and maximizing other aspects of shareholder welfare such as social impact. Advances in machine learning that enhance the firm's ability to target specific customers may exacerbate or ameliorate this tradeoff. We estimate these tradeoffs using data from randomized microcredit approvals in South Africa, the Philippines, and Bosnia. We examine social impact on two dimensions: credit access (i.e., reaching disadvantaged groups that typically have less access to financial markets) and impact (i.e., treatment effect on household income). Two of the three lenders could have increased average loan profit margin by 7–9 percentage points through machine learning-based targeting. Such targeting would, however, lower the social impact vis-à-vis credit access (specifically women and lower-income households would have been more excluded) but the change in impact on borrowers' income is too imprecisely estimated and thus inconclusive. To gauge the magnitude of the first tradeoff, we examine how profits would change if the bank altered who they lent to within each quintile of baseline borrower income, thus holding the distribution broadly similar but fine-tuning the targeting within income bands. Such a constraint would lower the profit gains of targeting by about half. These findings highlight the importance of quantifying tradeoffs and complementarities when deciding what to maximize.

Digital marketing enables firms to overcome information barriers and other frictions that prevent firms and customers connecting across space. However, firms may not adopt this new technology because they lack knowledge, or because they are uncertain of the returns and consider it risky. Our key research questions are whether firms can be induced to adopt this technology, and whether they will benefit from doing so. A randomized experiment with 1,575 small firms will test the relative effectiveness of a standard input-based approach (training, assistance, and subsidies) against that of a pay-for-results approach in getting firms to adopt digital marketing, comparing each treatment to a control group and to a group that receives only information. Our main outcome is adoption of this technology as measured by firms setting up a business Facebook page and carrying out paid advertising. We will also assess how this technology affects firm growth, as measured by new customers, sales, and profits.

Credit and information constraints can affect not only participation levels in different occupations, but also the types of individuals found in those occupations. I develop a model of occupational choice which shows how new information alters expectations and thus occupational choice. Credit and information constraints also interact with each other: the same information can have opposing impacts on occupational choice depending on the presence of credit constraints. Using a survey and information experiment in seven vocational high schools in Egypt, I find support for the model's predictions, including a key compositional prediction: risk-averse individuals respond more to changes in expectations of risk at both the intensive (variance of income) and extensive (probability of finding a job) margins. This differential response leads highly risk-averse individuals to shift towards "safer" occupations on average, and out of using credit to start a small enterprise.