Price Transparency in the Branchless Banking Market: Evidence from Branchless Banking Agents in Indonesia

The intervention will involve provision of information on the predicted level of future competition for branchless banking agents in a given location. This information will be presented during on-line survey to agents (right after the administration of baseline survey).

2023-07-10

2024-12-31

Our main outcome variable aims at measuring agents’ willingness to disclose information about the bank’s official prices to their customers.

To measure our main outcome variable (i.e., agents’ willingness to disclose the bank’s official price list to customers), after the administration of treatment, we will present agents with an incentivized elicitation exercise in which all participants of the experiment will be presented with a choice between two different marketing plans that the bank is planning to implement. To incentivize the choice, each plan will have a different cost and every agent will receive an endowment of IDR25,000 which can be used to buy one of these plans (the endowment will be provided on top of the compensation given for responding to the survey). The plans differ in whether information about the bank’s official prices will be disclosed directly to the agent’s customers or sent to the agent, which preserves the agent’s ability to shroud prices and engage in price discrimination. The two plans are described below:
Plan A. The agent will receive a WhatsApp message from the bank containing the bank’s official price list for the most widely used types of branchless banking transactions. Agents will be told that they are free to forward this message to clients at their discretion. The cost of this plan will be set at IDR25,000.
Plan B. The bank will directly send a WhatsApp message to all of the agent’s clients containing information about the bank’s official price list for the most widely used types of branchless banking transactions. The cost of this plan will be set at IDR20,000 (therefore, the agent gets to keep IDR5,000 from the endowment).
After the agent makes a choice, we will follow through and implement the plan chosen. Note that plan A allows the agent to keep the option of shrouding information and to price discriminate between clients. Agents who have a preference for keeping this option will have to take the costly action of selecting the most expensive plan. Plan B instead is equivalent to full transparency, and will not require agents to pay a premium. We have piloted this survey experiment to ensure that agents understand the differences between the two options and to verify that there is sufficient take-up of both plans under the relative prices in the current survey. The results from the choice experiment will allow us to determine whether higher expected levels of competition lead to a higher willingness to disclose the bank’s official price list (price transparency), or alternatively, whether shrouding can be sustained at different levels of competition.
Before the treatment, we will also elicit agent’s beliefs about their expected levels of competition in the near future. Right after the treatment, we will elicit their posteriors. This information will allow us to test the effectiveness of the information treatment.

Using surveys with the agent’s clients, we will also be able to further confirm how the agent’s actions change after they have made their choices. In particular, the survey will include questions on whether the agent posts prices in her shop, and whether she forwarded the WhatsApp message to clients. We will additionally ask about the client’s trust in the agent and observe in the administrative data the frequency with which clients make transactions. Note, however, that this analysis will not be causal, given that perceptions will be influenced by the agent’s choice in the survey experiment, which in turn is determined by treatment assignment.

In our experiment, we will explore how an increase in expected competition affects the willingness to become more transparent. We will introduce experimental variation at the individual level in the expected level of competition by informing randomly selected agents in a treatment group about the predicted level of future competition in their local area. The remaining agents in the sample will be assigned to the control group and receive a placebo message that is orthogonal to the intervention.
The risk of spillovers is minimized by the fact that agents typically do not communicate with each other and the bank does not organize meetings or trainings where agents meet each other.

Not available

Randomization will be done through STATA.

Individual agents.

No

9,500 branchless banking agents (expected response rate: 25-30%)

9,500 branchless banking agents (expected response rate: 25-30%)

Control group: 4,000 agents
Treatment group (high competition): 4,000 agents
Treatment group (low competition): 1,500 agents

To ensure the soundness of our experimental design, we have conducted a series of power calculations to identify the Minimum Detectable Effect sizes for our experiment. The calculations are made under the assumptions that: 1) there is one treatment group, 2) there is one control group, 3) there is the same number of agents in each group, 4) size is set at 5% and power is set at 80%. The sample size of the experiment will depend on the number of agents responding to the survey (out of a total of 9,500 agents in the experimental site). Although the response rate is expected to be around 25-30%, even in a scenario with a 17.5% response rate, the MDE is of 0.09 standard deviations.