The Poor Tax: Effects of Redistributive Pressures in Côte d’Ivoire

We offer a private, blocked savings account to regular piece-rate factory workers. These are employed full-time in formal cashew-processing factories in Côte d’Ivoire. Production is an individual manual activity (e.g., peeling cashews). Each worker receives an individual piece rate for her output—so that changes in effort translate directly into changes in worker's earnings and factory output.

In Phase 1, eligible workers randomly selected into treatment will be offered a private commitment savings account. This account pays out deposited savings in a lump sum after 9 months(at a private date). During the blocked period, deposits cannot be accessed—making it possible to save, while credibly denying transfer requests. Once the blocked period ends, workers can use the saved funds or leave them in the bank account.

This account can only be used for above-average baseline earnings, allowing workers flexibility to cover household needs and emergencies. Treatment workers will be offered the option to choose a threshold, equal to or greater than the worker’s historical average fortnightly earnings, above which any additional salary they earn fortnightly will be automatically and privately deposited into a savings account. We compare the daily attendance and productivity of control vs. treatment workers (“productivity effect”).

After the completion of Phase 1, a subset of workers will be offered a private illiquid account (similar to the one available in Phase 1). A second subset will instead be offered an account that may be used by the bank for promotional purposes and advertised to their social network. We test whether the labor supply and productivity effects observed in Phase 1 change depending on whether workers are offered a private or public account.

2017-06-01

2019-10-01

Our key outcome variables relate to the labor supply of workers. Specifically, we will look at the effect of the intervention on productivity (measured by daily output and earnings) and attendance. We will also consider quits.

To complement those key outcomes, we collect baseline data on perceptions about redistribution and its relationship to savings and the returns to effort. In addition, we will attempt to collect data on the level of transfers to members of the workers' kin network and the effects on overall savings, intra-household decision-making power and subjective well-being among workers. Finally, we will correlate revealed preference measures of self-control problems with treatment effects.

The design will be implemented in two phases. In a first phase, we individually randomize eligible factory workers. Eligibility is based on attendance and seniority at the plant. In a second phase, we experimentally vary the visibility of the accounts.

The design will be implemented in two phases.

Phase 1 – Blocked accounts.
In Phase 1, eligible factory workers (those who have been present at the factory for at least 3 months and who have recently attended the factory regularly) will be randomly divided into treatment and control. Treatment workers will be offered the option to take up a private, blocked savings account. If they take-up, they will choose a threshold level for deposits that is equal to or higher than their historical average earnings. The level of savings in the account will be known only to each individual worker. The accounts are blocked for 9 months. To increase the statistical power of the study in factories with a limited number of workers and a substantial turn-over, several waves of the intervention will be run. In each wave, treatment workers from the previous wave are offered the account again. A random subset of control workers from the previous wave, as well as of workers who were not eligible in the previous wave but are eligible in the current wave, will be offered the account. In each wave, control workers will not be offered an account (though they could open a standard checking account at the bank, as was the case before our intervention).

Phase 2 – Public vs. private variation.
In Phase 2, for a subset of workers, we will offer blocked accounts whose presence (along with the payout date) may be advertised to the worker’s kin and social networks if they choose to take-up the account, thereby lowering the value of the blocked accounts as a tool to prevent redistributive pressure.

Additional activities – We complement this design with two additional tests for self-control as a mechanism. First, using the paycycle effects test from Kaur, Kremer, and Mullainathan (2015), we will check if increases in productivity leading up to the payday at baseline predict take-up of the accounts. Second, we will attempt to randomly give some workers the chance to opt out of the direct deposit for just one upcoming paycheck, varying the number of days between the day in which the option is given and the upcoming payday. If there are self-control problems in consumption, workers will choose to opt out, especially closer to the date of the payday. In contrast, we would not expect this pattern under redistributive pressure. (Note our ability to conduct this second test is contingent on our ability to operationally enable selective opt out in the design of the blocked savings account with the factory and the bank).

Randomization is done in the office by one of the PIs or Research Assistants on the computer, prior to the start of each phase. Workers selected into treatment are invited to join sensitization sessions, led by one of the PIs or Research Assistants (including translators hired by the research team) during which they learn about the treatment and their treatment assignment.

Workers are randomized into different treatment waves at the individual level.

Yes

The productivity of at least 200 individuals will be measured across two factories. We will enroll all eligible workers across the two factories. Thus, the final sample size will be determined by total employment numbers at the factories at that time.

Over 75,600 productivity observations (200 workers * 6 workdays/week * (50 weeks of intervention + 13 weeks pre-intervention))

The shares of the sample in the different treatment arms will be approximately as follows:

(1) Phase 1: 50% Treatment, 50% Control
(2) Phase 2: Among those who receive blocked accounts, approximately 50% will receive public blocked accounts, and 50% will receive private blocked accounts.