Work in progress

Effects of Limited Information about Workseekers' Skills 

Skills, Beliefs, and (Mis)Directed Job Search (with Rob GarlickLukas Hensel and Andrea Kiss)

Fieldwork completed, drafting in progress

This paper studies how inexperienced workseekers' beliefs about their skills affect their search strategies and labor market outcomes. In a large sample of young South Africans, workseekers have overconfident beliefs about their skill levels and do not perfectly observe their comparative advantage. These inaccuracies in their beliefs are relatively persistent, even for those who are most actively searching or working. An experiment that gives workseekers more information about their skills partly but persistently corrects their beliefs about their skill levels and comparative advantage. These shifts in beliefs lead workseekers to target their search toward jobs aligned with their comparative advantage advantage and substantially increase their earnings, consistent with a model of targeted search across multiple job types. The shifts in beliefs do not change search effort or reservation wages, allowing us to reject some models of endogenous search effort and of search with overconfidence.

Light Touch Psychological Interventions with Economic Effects

Aspiring to a Better Future: How a Simple Psychological Intervention Reduces Poverty (with Rob Garlick, Mahreen Mahmud, Richard Sedlmayr, Johannes Haushofer, Stefan Dercon) 


How do material and psychological constraints individually and jointly affect investment and poverty? To answer this, we design a workshop teaching techniques to raise aspirations and encourage long-term planning. We cross-randomise this with large unconditional cash transfers in a 415-village, 8,300-person, 1.5-year experiment in Kenya. The workshop increases aspirations and investment, and raises living standards, at least twice as cost-effectively as cash transfers. But combining the workshop with cash produces similar effects to cash alone, potentially because cash raises aspirations. This shows that alleviating psychological constraints can improve economic outcomes but these constraints may relax when material conditions improve.

The Future in Mind: Short and Long Run Impact of an Aspirations Intervention in Rural Ethiopia (with Tanguy Bernard, Stefan Dercon, Giulio Schinaia and Alemayehu Seyoum Taffesse)

People living in deprivation may not make long term investments if they doubt that a better life is possible. In a randomised control trial in remote, rural Ethiopia, we test if changing how people perceive their future opportunities alters investment. We show a treatment group video documentaries about the lives of role models from similar communities who escaped poverty through their own efforts in agriculture or business. Five years later, treated household heads report higher labour supply, more use of agricultural inputs and increased education spending. They have accumulated more assets, their children have attained more years of schooling, and they have increased asset wealth, housing quality and food security. We provide evidence consistent with the mechanism being an increase in aspirations. We outline a simple model of household consumption and production with reference-dependent utility, in which aspirations are defined as reference points for consumption and meeting aspirations yields utility over and above the utility of consumption. In such a model, changes in reference points increase effort and investment. We also show in survey data that individuals in treated households report higher aspirations -- hopes for the level of income and assets they hope to attain -- straight after screening and after five years. Furthermore, exposure to video or outsiders and other psychological mechanisms are unlikely to account for effects: a placebo group shown a local entertainment programme are unaffected, and there are no effects of treatment on preferences, social norms and other psychological characteristics. Exploiting randomised intensity of treatment between villages, we find little evidence of aspirations or behaviour change being transmitted between households, suggesting such interventions may need to reach individuals directly. 

Economic Effects of Mental Health Interventions

Treating Mental Health Conditions Improves Labor Market and Other Economic Outcomes in Low and Middle-Income Countries (with Crick Lund,  Marc Witte, Thandi Davies, John Walker, Johannes Haushofer, Sarah Murray, Judy Bass, Laura Murray, Vikram Patel)

Does treating mental health conditions improve labor market and other economic outcomes in low and middle-income countries? We run a systematic search for all randomized controlled trials (RCTs) which evaluate mental health treatments and measure economic outcomes in these countries. We conduct frequentist and Bayesian meta-analyses on estimates of treatment effects from 39 interventions. Treatments reduce the number of days participants cannot work by 16%, reduce the probability of being unable to work by 9 percentage points (26%), and improve qualitative measures of performance at work. They increase asset wealth and education investment. Results suggest multiple psychological and behavioral mechanisms.

Digital Delivery of Behavioural Activation to Overcome Depression and Facilitate Social and Economic Transitions of Adolescents in LMICs (DoBAt)

Intervention being piloted 

Depression is the leading cause of disability worldwide, and it is particularly problematic among adolescents given the risk for greater depression chronicity across the lifespan. Untreated depression exerts a huge economic toll as it impairs cognitive functioning, interpersonal relationships, interferes with schooling and disrupts work and productivity. These impairments have a greater impact on adolescents in low- and middle-income countries due to the additional adversities they face and the lack of available, effective treatments.

This three year project will use smartphones to deliver a digital intervention for adolescents, supported by local lay counsellors, to reduce depression and facilitate successful transitions to adulthood. We will adapt a tailored psychological therapy, Behavioural Activation (BA), among adolescents (15 to 19 years old) living in rural South Africa and Uganda. We will test the effectiveness of delivery of BA for reducing depression (primary outcome), and possible mechanisms, principally executive function and social cognition. As secondary outcomes, we will assess risk-taking behaviours and a range of human capital outcomes. Further, the proposed work will produce relevant measures of executive functions and social cognition through language translation and cultural adaptation and evaluate their reliability and validity in a rural context.

The team comprises a multidisciplinary group of psychologists, psychiatrists, neuroscientists and economists from South Africa (University of the Witwatersrand), Uganda (BRAC),  as well as from the UK (University of Oxford, University of Exeter, University College London) and the USA (University of California, Los Angeles, UCLA).

Economic and Political Effects of Unconditional Cash Transfers

This paper is from one large village-level randomized controlled trial in rural Kenya. The trial was conducted in collaboration with GiveDirectly, an NGO which uses mobile payments technology to send donations to extremely poor families in the developing world. Core funding for the trial is from an anonymous donor, GiveWell, the Gates Foundation and the Accelerating Adolescent Achievement Hub (UKRI GCRF). The survey fieldwork was run by Innovations for Poverty Action.

Cash Transfers and Community Participation in Public Affairs:  A Village-Level Randomized Controlled Trial in Kenya (with Michael Walker) 

We provide causal evidence on how a large NGO-run cash transfer programme affects household civic and political participation and household requests for resources from local politicians. We randomly allocated the roll-out of GiveDirectly’s unconditional cash transfer programme, which provides transfers to poor households meeting a basic means test in treatment villages, across over 1,000 villages in Western Kenya. We collect survey data from over 10,800 households (both those poor enough to be eligible to receive the transfer and ineligible households) and 1,200 local leaders during the period after the 2017 election. Receiving a cash transfer does not affect household's voter turnout, vote choice, or favourability ratings of candidates. Voters (correctly) do not attribute the programme to local leaders. We find cash recipient households shift out of lower-level, "subsistence'" political engagement to higher-value group engagement. Namely, recipient households decrease private exchanges of patronage with politicians by attending fewer rallies (for which they receive small payments), making fewer requests for private support, and receiving fewer offers to sell votes. Recipient households join more community groups, and increase their contributions to group fundraisers; groups in cash villages make more requests to local leaders and receive more funding from outside sources. These group fundraisers raise significantly more than private requests. However, ineligible households in cash villages make more low-level exchanges. We find no changes in processes to allocate public goods funding. Results suggest aid programmes have few effects on electoral outcomes in this context, neither altering vote choices nor reducing the overall presence of clientelist relationships. They also suggest poor people’s involvement in local processes may be limited by the time and monetary cost of participation, as there is a strong participation gradient by wealth for households in control villages. As well as reducing poverty, transfer programs may enable increased participation in such processes.

Voters' Beliefs and Voter Turnout

Everybody Loves a Winner: A Field Experiment Providing Information on Polls in South Africa (with Brynde Kreft)

I show voters process information from pre-election polls using crude heuristics: they are overly swayed by whether a party is just winning a poll, compared to just losing. I conducted a field experiment with 2,023 low-income registered black voters from Johannesburg in South Africa’s 2016 elections. I provided two treatment groups with information from two different IPSOS polls. The polls both predict a close election and provide statistically indistinguishable predictions about vote shares. In one, the challenger party is ahead by a tiny margin; in another, the incumbent party is barely ahead. The difference in effect of the two polls is the effect on voters of a party being the narrow leader in a poll, compared to being the narrow loser. Supporters of the party just ahead in the polls are 10 percentage points more likely than a placebo group to turn out to vote (using verified measures) and 12 percentage points more likely to vote for their party (in self-reported measures); there are no effects on voters who learn their party is just behind. Voters who learn a party is just winning rate the party more highly, are more likely to support the party, and believe the party will win a higher share of the vote, compared to if they learn that party is barely losing, although the information about party quality and likely vote in both polls is similar. These effects imply that how the media frames poll results may have large effects on electoral outcomes.