We use a cash transfer to relax households' borrowing constraints, then elicit their willingness to pay (WTP) for a remedial education program offering tutoring and life-skills training. Lottery losers were willing to pay 3,300 Tanzanian Shillings for the program, seven percent of per-capita monthly expenditures. For those identified at baseline as able to borrow, WTP increases by three percent upon winning a lottery prize of 3,200 TSh. For those unable to borrow, WTP increases by 27 percent upon winning the lottery. We conclude that borrowing constraints limit access to educational programs, and may increase inequality of educational attainment.

Researchers frequently use variants of the Becker-DeGroot-Marschak (BDM) mechanism to elicit willingness to pay (WTP). These variants involve numerous incentive-irrelevant design choices, some of which carry advantages for implementation but may deteriorate participant comprehension or trust in the mechanism, which are well-known problems with the BDM. We highlight three such features and test them in the field in rural Uganda, a relevant population for many recent applications. Comprehension is very high, and 86 percent of participants bid optimally for an induced-value voucher, with little variation across treatments. This gives confidence for similar applications, and suggests the comprehension-expediency trade-off is mild.

We propose a technique for assessing robustness to demand effects of findings from experiments and surveys. The core idea is that by deliberately inducing demand in a structured way we can bound its influence. We present a model in which participants respond to their beliefs about the researcher's objectives. Bounds are obtained by manipulating those beliefs with "demand treatments." We apply the method to eleven classic tasks, and estimate bounds averaging 0.13 standard deviations, suggesting that typical demand effects are probably modest. We also show how to compute demand-robust treatment effects and how to structurally estimate the model.

Numerous authors point to a decline in joint liability microcredit, and rise in individual liability lending. But empirical evidence is lacking, and there have been no rigorous analyses of possible causes. We first show using the well-known MIX Market dataset that there is evidence for a decline. Second, we show theoretically that commercialization - an increase in competition and a shift from non-profit to for-profit lending (both of which are present in the data) - drives lenders to reduce their use of joint liability loan contracts. Third, we test the model's key predictions, and find support for them in the data.

Motivated by recent controversies surrounding the role of commercial lenders in microfinance, and calls for regulation of the sector, we analyze borrower welfare under different market structures, considering a benevolent non-profit lender, a for-profit monopolist, and a competitive credit market. To understand the magnitude of the effects analyzed, we simulate the model with parameters estimated from the MIX Market database. Our results suggest that market power can have severe implications for borrower welfare, while despite possible enforcement externalities competition typically delivers similar borrower welfare to non-profit lending.

As predicted by loss aversion, numerous studies find that penalties elicit greater effort than bonuses, even when the underlying payoffs are identical. However, loss aversion also predicts that workers will demand higher wages to accept penalty contracts. In six experiments I recruited workers online under framed incentive contracts to test the second prediction. None find evidence for the predicted distaste for penalty contracts. In four experiments penalty framing actually increased the job offer acceptance rate relative to bonus framing. I rule out a number of explanations, most notably self-commitment motives do not seem to explain the finding. Two experiments that manipulate salience are successful at eliminating the effect, but do not significantly reverse it. Overall, loss aversion seems to play surprisingly little role in this setting. The results also highlight the importance of behavioral biases for infrequent, binding decisions such as contract take-up.

We study the relative effectiveness of contracts that are framed either in terms of bonuses or penalties. In one set of treatments subjects know at the time of effort provision whether they have achieved the bonus / avoided the penalty. In another set of treatments subjects only learn the success of their performance at the end of the task. We fail to observe a contract framing effect in either condition: effort provision is statistically indistinguishable under bonus and penalty contracts.

This paper contrasts individual liability lending with and without groups to joint liability lending. We are motivated by an apparent shift away from the use of joint liability by microfinance institutions, combined with recent evidence that a) converting joint liability groups to individual liability groups did not affect repayment rates, and b) an intervention that increased social capital in individual liability borrowing groups led to improved repayment performance. First, we show that individual lending with or without groups may constitute a welfare improvement over joint liability, so long as borrowers have sufficient social capital to sustain mutual insurance. Second, we explore how the lender's lower transaction costs in group lending can encourage insurance by reducing the amount borrowers have to pay to bail one another out. Third, we discuss how group meetings might encourage insurance, either by increasing the incentive to invest in social capital, or because the time spent in meetings can facilitate setting up insurance arrangements. Finally, we perform a simple simulation exercise, evaluating quantitatively the welfare impacts of alternative forms of lending and how they relate to social capital.

The use of observational methods remains common in program evaluation. How much should we trust these studies, which lack clear identifying variation? We propose adjusting confidence intervals to incorporate the uncertainty due to observational bias. Using data from 44 development RCTs with imperfect compliance (ICRCTs), we estimate the parameters required to construct our confidence intervals. The results show that, after accounting for potential bias, observational studies have low effective power. Using our adjusted confidence intervals, a hypothetical infinite sample size observational study has a minimum detectable effect size of over 0.3 standard deviations. We conclude that — given current evidence — observational studies are uninformative about many programs that in truth have important effects. There is a silver lining: collecting data from more ICRCTs may help to reduce uncertainty about bias, and increase the effective power of observational program evaluation in the future.

Simple decisions can reveal two layers of preference. Suppose a hiring manager always chooses a woman over an identically-qualified man, but always chooses a man over a woman with different qualifications. Intuitively, these choices reveal an explicit preference for women, but an implicit preference for men. We define an implicit preference for an attribute as one whose influence increases whenever that attribute is mixed with others ("dilution"). We prove two representation theorems, and provide three two-layer decision-making models that exhibit implicit preferences. We give extensive guidance for applications, and present evidence of implicit risk preferences, implicit selfishness, and implicit discrimination.

We revisit two clinical trials that randomized depressed adults in India (n=775) to a brief course of psychotherapy or a control condition. Four to five years later, the treatment group was 11 percentage points less likely to be depressed than the control group. The more effective intervention averted 9 months of depression on average over five years and cost only $66 per recipient. Therapy changed people's beliefs about themselves in three ways. First, it reduced their likelihood of seeing themselves as a failure or feeling bad about themselves. Second, when faced with a novel work opportunity, therapy reduced over-optimistic belief updating in response to feedback and thus reduced overconfidence. Third, it increased self-assessed levels of patience and altruism. Therapy did not increase levels of employment or consumption, possibly because of other constraints on employment in the largely female study sample.

Agriculture in low-income countries is characterized by misallocation of land across farmers and fragmentation--the separation of farms into smaller plots--which increases costs and limits the use of increasing returns technologies. We argue that a carefully-designed set of trading rules can address these problems, and test this using two lab-in-the-field experiments with smallholder farmers in Uganda and Kenya. First, with survey data, we document that agricultural land markets are thin, prone to exposure risk, and suffer from coordination frictions. These characteristics typically hamper decentralized trade. Market design may improve outcomes by thickening markets, finding chains, and enforcing conditional contracts, but right-in-theory designs may be unfamiliar and hard for farmers to understand. Our first experiment, conducted in Uganda, simulates status quo land markets and confirms severe inefficiency. In a second phase of the experiment, we find that a centralization intervention can eliminate direct exposure losses that arise through trade and promote consolidation but does not reduce misallocation. We then test whether designs that are more tailored to the land trade problem, but potentially harder to understand, can further improve outcomes. Our second experiment, in Kenya, finds that a computerized package exchange, which allows traders to specify a sequence of conditional trades as a single transaction, performs particularly well. Our results suggest that improved market design can reduce frictions and lead to important productivity gains among smallholder farmers, and may help unlock gains from complementary programs like titling.

Major depressive disorder (MDD) is one of the most prevalent mental illnesses worldwide. Existing evidence suggests that it has both economic causes and consequences, such as unemployment. However, depression has not received significant attention in the economics literature. In this paper, we present a simple model which predicts the core symptoms of depression from economic primitives, i.e. beliefs. Specifically, we show that when exogenous shocks cause an agent to have pessimistic beliefs about the returns to her effort, this agent will exhibit depressive symptoms such undereating or overeating, insomnia or hypersomnia, and a decrease in labor supply. When these effects are strong enough, they can generate a poverty trap. We present descriptive evidence that illustrates the predicted relationships.