Evaluating Transfer Size in a Conditional Cash Transfer Program for Seasonal Migrants in Indonesia

Fieldwork by:
Location:
Timor Tengah Selatan, NTT
Sample:
775 households in five villages
Chronologie:
2015 - 2018
Target group:
  • Farmers
  • Rural population
  • Families and households
  • Migrants
Outcome of interest:
  • Earnings and income
  • Employment
Intervention type:
  • Information
  • Nudges and reminders
  • Conditional cash transfers
  • Unconditional cash transfers
AEA RCT registration number:
AEARCTR-0003045
Partners:

While conditional cash transfer (CCT) programs may encourage desirable behaviors, like sending children to school, CCT programs may also lead to unintended and sometimes undesirable outcomes if the conditions change recipients’ actions in ways that lower their welfare. Researchers conducted a randomized evaluation to test the impact of a CCT program conditional on seasonal migration and an unconditional cash transfer (UCT) program on rural households’ decisions to migrate and migration season earnings in West-Timor, Indonesia. Relative to the UCT, the CCT was most effective at raising migration season earnings when the transfer size was just enough to cover migration expenses. However, when the transfer size exceeded migration travel costs, relatively lower-income households migrated for the purpose of collecting the transfer and overall program impact lessened.

Policy issue

Conditional cash transfer (CCT) programs are a popular intervention aimed at reducing poverty or achieving other social goals by tying behavioral conditions to cash transfers. Conditions attached to most CCT programs are designed to encourage human capital investments, and examples of common conditions include school enrollment and attendance or health checkup visits for children.

While CCT programs may encourage desirable behavior, they can lead to unintended and sometimes undesirable outcomes if the conditions change recipients’ actions in ways that lower their welfare. For example, conditions tied to parent public works program participation could result in an increase of children’s domestic work and decrease in school attendance to compensate. Transfer size may be an important lever to reduce unintended outcomes, but there is little evidence on the impact of different transfer sizes on such outcomes. How does varying the transfer size in a CCT program aimed at incentivizing seasonal migration in rural Indonesia affect the decision to migrate and migration season earnings, and what are the tradeoffs compared to a UCT program?  

Context of the evaluation

Rural areas in West-Timor, Indonesia experience pre-harvest seasonal poverty and hunger, which sometimes turns into famine. During the pre-harvest season, some households send individuals to obtain employment in urban centers to cope with the seasonal income shortfall. One reason that households do not participate in seasonal migration is the risk of not finding a job after incurring travel expenses. UCT or CCT programs could facilitate seasonal migration for low-income households that are not comfortable with this risk. 

This study took place among households in the Timor Tenggah Utara (TTU) Regency in West-Timor. Eligible households had to have at least one household member aged 21 or older and could not own land exceeding 200 are (2 hectares). In West-Timor, maize and rice are staple crops, with the spring harvest taking place from March through May. Rural work opportunities are scarce from May until November, and September is the most common month for people to begin migration. 

Seasonal Migrant workers in Indonesia working in construction in urban environment.
Photo: Anggie Rizka Safitri | Shutterstock

Details of the intervention

Researchers evaluated the impact of a CCT and UCT program on rural households’ decisions to migrate and their migration season earnings. Eligible households were offered either a UCT or CCT with the condition of the CCT being migration to a destination of their choice within West-Timor.

Households were randomized into one of four intervention groups:  

  1. UCT Group: Households received IDR 150,000 (US$10) with no conditions attached.
  2. CCT Group 1: Households received IDR 300,000 (US$20) when they took up the offer to migrate.
  3. CCT Group 2: Households received IDR 150,000 (US$10) when they took up the offer to migrate.
  4. CCT Group 3:  Households were told they would receive IDR 150,000 (US$10) when they took up the offer to migrate, but were “surprised” at their destination city with a bonus of IDR 150,000 (US$10), receiving IDR 300,000 (US$20) in total.

The CCT payment was divided into two installments: half paid at the village of origin, and the other half paid at the destination city after “checking in” with a program officer. This was to ensure that the condition of migration was met. The third CCT group was designed to differentiate between the effect of the transfer size on the decision to migrate and the effect on post-migration decisions such as searching longer for a job at the destination. The lower CCT amount (IDR 150,000) was enough to cover the transportation costs to common migration destinations.  

Researchers collected information on whether or not the household accepted the offer to migrate, total household income earned during the migration season, and household food security. Researchers surveyed households when they were offered the CCT in July 2017, and followed up with households during the planting and weeding season from December 2017 to February 2018.

Results and policy lessons

Although the CCT program induced migration and increased migration season earnings relative to the UCT program, when the transfer size exceeded migration travel expenses, relatively lower-income households migrated to collect the transfer and experienced lower migration season income.

Migration decision: The CCT was more likely to be spent on seasonal migration, while the UCT was more likely to be spent on non-farm capital and food consumption. Households offered the larger CCT (initial offer of IDR 300,000) were more likely to participate in migration than households offered the lower CCT (initial offer of IDR 150,000). The larger CCT offer raised the rate at which migrants checked in at a destination by about 11 percentage points, a 26 percent increase from a base of 42 percent.

Migration season income: The lower CCT offer raised migration season household income compared to the UCT. However, among households with lower incomes at the start of the intervention, migration season income was less for those offered the larger CCT than for those offered the lower CCT. This suggests that the larger CCT encouraged relatively lower-income households to send a migrant merely to collect the larger transfer at the destination, even if it was not the best financial decision for the household. These migrants may have lacked the skills to join the seasonal migrant job market.

Taken together, these results demonstrate how an increase in transfer size beyond what was needed for recipients to comply with the migration condition resulted in unintended consequences that reduced the impact of the program overall.