NEW UPDATE: Completed trials may now upload and register supplementary documents (e.g. null results reports, populated pre-analysis plans, or post-trial results reports) in the Post Trial section under Reports, Papers, & Other Materials.
Working Together or Never: Labor Supply Externalities and the Unemployment Trap
Initial registration date
July 30, 2020
August 03, 2020 12:33 PM EDT
Other Primary Investigator(s)
Additional Trial Information
When social networks generate redistributive pressures, labor supply decisions are undertaken in a social context. Individuals who optimize labor supply do so with the understanding that some of their earnings may be subject to taxation within the network; if they are capable of avoiding this tax by not working, then labor supply will be distorted downwards. If the unemployed exert greater redistributive pressure, then this insight leads to the theoretical possibility of multiple equilibria in networks: networks may be characterized by equilibria where few work and where the incentives to work are poor, or alternatively by equilibria where many work and the incentives to work are high (Hoff and Sen 2011). While experimental evidence indicates that important redistributive pressure exists for windfall gains (e.g. Jakiela and Ozier 2015) we know much less about whether these pressures generate unemployment in practice.
In this project, we test for the presence of externalities in labor supply and multiple employment equilibria, using a hiring experiment implemented in Côte d’Ivoire with a large agri-processing company. Registration Citation
To test for the presence of labor supply externalities and multiple employment equilibria, we will randomly vary the intensity of job offers across redistributive networks – offering actual positions at a local cashew-nut processing factory.
If redistributive pressure is leading to multiple equilibria, we anticipate that saturating a network with positions will lead to greater take-up of our job opportunities (and greater labor supply on the intensive margin). Other mechanisms may also contribute to greater take-up of joint offers (with similar multiple equilibria in labor supply). As an illustration, network complementarities in the return to leisure, or in commuting, may be relevant. To alleviate redistributive pressures faced by a potential worker, providing job opportunities to network members is an option. Another is to provide direct cash payments to those network members, corresponding in size and frequency to the transfers they would have received from that potential worker – making those payments public to the latter so that they substitute for the pressures she faces.
While a random subset of networks will be treated with the first option, another will be treated with the second option. If redistributive pressures are indeed driving the observed labor complementarities, the effects should be similar in both cases.
Intervention Start Date
Intervention End Date
Primary Outcomes (end points)
Our key outcome variables relate to the employment status of the individuals who have been offered the jobs. Specifically, we will look at the effect of the intervention on job take-up and retention, as well as on productivity (measured by daily output) if possible.
Primary Outcomes (explanation)
Secondary Outcomes (end points)
We will attempt to collect data on transfers among individuals in the network. However, since surveys currently need to be conducted via phone, it is possible that it will prove infeasible to collect this information.
Secondary Outcomes (explanation)
We will start with a set of potential applicants for the jobs offered at Olam’s new factory, opening in the summer of 2020. They will be randomized into three groups:
- C – Control: job offered to the potential applicant
- T1 – Jobs in Network: job offered to the potential applicant + job offered to at least 2 members of the potential applicant’s network
- T2 – Cash in Network: job offered to the potential applicant + cash payments to at least 2 members of the potential applicant’s network
We will first survey potential applicants, primarily to elicit their redistributive network members, and inform them of their treatment assignment. We will then survey their network members, and inform them of their treatment assignment (if any).
Experimental Design Details
Randomization is done in the office by one of the PIs or Research Assistants on the computer.
The randomization is done at the level of the index individuals – or equivalently, at the level of redistributive sub-networks.
Was the treatment clustered?
Sample size: planned number of clusters
120 (Determined by the number of job slots in the factory in the initial hiring wave. We will attempt to expand the sample by enlisting partnerships with additional factories or waves of hiring).
Sample size: planned number of observations
Sample size (or number of clusters) by treatment arms
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
INSTITUTIONAL REVIEW BOARDS (IRBs)
Innovations for Poverty Action IRB-USA
IRB Approval Date
IRB Approval Number