Adoption and Impacts of Digital Payment Technologies: Evidence from Informal Transit

Owners and drivers will be invited to use the digital payments technology. QR codes will be installed in taxis to allow passengers to pay securely. We will recruit owners and drivers in two ways: marketing meetings in collaboration with taxi associations and listing of interested drivers and owners throughout the city by hired enumerators. Owners and drivers will be invited to participate in the study to test the product before its official launch. In collaboration with a partner company, I designed a base product with three options for randomly varying the amount of information available to the owners on their drivers. The randomly selected control group will be offered nothing and on a wait list.

2022-03-29

2022-12-31

The analysis will include three main sets of outcomes:
(a) Technology adoption: I will measure adoption barriers on two dimensions: take-up and usage across the three options, as key variables of interest, using both survey and administrative data.
(b) Impacts on business productivity, profits, growth, and financial inclusion, including leakages, costs, quantities (passengers), prices, and business growth, savings, credit, money transfers to other sources, using both survey and administrative data.
(c) Contracts and relationships between owners and drivers, including trust, repayment, beliefs, value of the relationship, contract changes. Owners driving their own taxi alone, with no other drivers will be included and randomly offered one of the 3 options to study hiring of a new driver as an outcome.

Endline surveys will be conducted with studied drivers after about 3-6 months. In addition, I will conduct high-frequency bi-weekly phone surveys with drivers in all groups. Mystery passengers will also monitor technology adoption and collect data like prices. In addition, passengers will be able to reach Wave’s call center with any complaints about drivers’ compliance with digital payments.

The partner plans to offer discounts to passengers to increase take-up. The latter discounts will be randomized and studied as secondary outcomes.

Owners and drivers will be invited to use the digital payments technology that aims at reducing transaction costs. In collaboration with the partner company, I designed a base product with three options for randomly varying the amount of information available to the owners on their drivers:
A) No Visibility: The owners can only receive payments from drivers at no fee.
B) Limited Visibility / Limited Liability: The owners get notified each day via text every night of their drivers’ total collection up to the estimated daily usage of digital payments for taxis. If the total balance is above average, the owners do not get notified of the surplus.
C) Full Visibility: The owners get notified and can observe all the drivers’ transactions. They also receive an SMS about the total transactions of the day. This option is currently the only one offered to other types of businesses accepting digital payments.
D) Control group: Waitlisted for the time of the experiment and followed using both survey and administrative data.

The partner company plans to reach further taxi owners/drivers outside of this sample by the end of the pilot study.

The goal of these three treatment arms is to quantify one potential barrier for drivers to use the technology, which I call the “moral hazard” barrier to technology adoption. A short baseline survey will be conducted with participants to assess their willingness-to-pay (WTP) for their randomized option, incentivized using the BDM method, where a price is drawn, and they are invited to complete their purchase only if the price is below their WTP. Most of the sample (95%) will be “surprised” and randomized into a group irrespective of their WTP. Only 5% of respondents will actually be treated using the BDM mechanism. Thus, the proposed design will collect incentive-compatible measures of WTP (hence the 5% who are actually implemented) while preserving the randomization and thus impact estimates for the vast majority of the sample (the 95%). All participants will be asked to rank the three options. Under some assumptions about baseline beliefs (which will be measured), Option B might be a middle ground pleasing both parties, conditional on passenger take-up: it allows drivers to credibly signal “bad days” to their owners while not revealing the overall surplus on “good days”.
Surveys will be conducted with both owners and drivers to obtain their characteristics and baseline measures of the outcome variables described in the following section (e.g., productivity). We will also measure risk aversion in the field in an incentivized manner (5% of the participants will get their choices realized).

Randomization done by computer.

The unit of randomization is the taxi business, i.e., the interested taxi owners. The owners and their drivers will be randomized into one of the six groups, i.e., 3 options and a control group, also each assigned to one of the three options to elicit their willingness-to-pay for it (waitlisted) (3 + 3).

I stratify the randomization by anticipated dimensions of heterogeneity, i.e.,
-- Proxy for baseline knowledge/beliefs (business type, number of taxis, and length of the relationship).
-- Proxy for baseline risk aversion, i.e., whether the taxi drivers circulate throughout the city or mostly wait for passengers in a fixed place.
-- Proxy for baseline taxi-like digital usage

The randomization will be done in several batches and listing surveys will continue to happen in addition to the baseline survey.

No

N/A

The target sample is 1,200 owners along with their drivers.

240 owners and their drivers in each of the three treatment arms. 480 owners and their drivers in the control group (sub-divided into three options to elicit willingness-to-pay).

Pre-Analysis Plan - Adoption and Impacts of Digital Payment Technologies