Randomization for Causality, Ethnography for Mechanisms: Illiquid Savings for Liquor in an Autarkic Society
How to cite
Zinman, Jonathan; Karlan, Dean; Godoy, Ricardo. 2021. "Randomization for Causality, Ethnography for Mechanisms: Illiquid Savings for Liquor in an Autarkic Society." AEA RCT Registry. AEARCTR-0008669. https://doi.org/10.1257/rct.8669-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
What should researchers do when confronted with surprising results? Financial access innovations usually leave “temptation” spending unaffected or reduced. However, we found that promotion of savings lockboxes in a largely autarkic society increased alcohol consumption and blood pressure, despite no one reporting intentions to save for alcohol. To probe mechanisms that could explain this pattern, we then used ethnographic methods, including direct observations of drinking (“scans”) and debriefing interviews to discuss the earlier trial results. We learn that sponsoring drinks confers prestige, but the stigma attached to drinking by outsiders likely discouraged reporting intentions to save for it.
Experimental design
The research team, including CBIDSI, worked with the governing council of Tsimane’ to select a sample frame of villages. Our best estimate suggested there were about 120 Tsimane' villages with at least eight households at the time of the study. We ruled out sampling from villages that were too costly to reach, too unsafe, or that contained people from other ethnic groups. This left 61 villages for our sample. Even these villages are far from the nearest market— a median of about 4 hours travel time. They tend to be even farther from the NGO office in the market town of San Borja where savings could be deposited.
All households in a village were eligible for participation in the study. To enhance participation, brief paid advertisements on the missionary radio station promoted the study shortly before the research team would arrive in each study village. The missionary radio station transmits news and announcements three times a day in Tsimane', and can be heard across the study area.
Within each of the 61 villages, CBIDSI and IPA publicly randomly assigned all the households residing in the village at the time of the visit to treatments or to a control group. These random assignments and baseline surveys took place from March through June of 2011. (For more details on the randomization mechanics and baseline survey see Appendix A and Section 1-D.)
All treatment households received a savings lockbox from CBIDSI that is small, portable, wooden, pest-resistant, and has a slit at the top of the box to insert money (Appendix B). N.b. that the lockbox was not linked to any sort of bank account or to any other service.
The main study arms, randomly assigned in roughly equal proportions within each village, are:
Treatment 1 (T1: Lockbox with Key). Savings box with a key so recipients could open the savings box at any time.
Treatment 2 (T2: Lockbox without Key). This treatment was the same as Treatment 1, except T2 households did not receive a key. Instead the key was kept in the CBIDSI office in San Borja, which, as noted, was several hours of travel time away from most villages in our sample. Our intent was to provide a strong commitment device to facilitate saving for a lumpy expenditure.
Control Group. The remaining households in a village served as a control and received a placebo intervention of six metal plates. The six plates had about the same monetary value as a savings box (~ PPP US$11).
We also randomized which household head received the savings box or plates. Heads received the box or plates after completing the baseline survey, which followed immediately after the random assignment. Due to the high prevalence of two-headed households in the population, and our randomization of which head got the box or plates, we did not attempt to survey most single heads and instead gave them a gift after the public randomization. Our analysis sample thus contains only households headed by two adults.
Design details
Cited by
No citation data available for this registration yet.