What Matters for Consumer Credit Choice? Evidence from the Philippine Digital Credit Market

Last registered on December 06, 2023

Pre-Trial

Trial Information

General Information

Title
What Matters for Consumer Credit Choice? Evidence from the Philippine Digital Credit Market
RCT ID
AEARCTR-0012635
Initial registration date
November 30, 2023

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
December 06, 2023, 8:26 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Primary Investigator

Affiliation
Trinity College Dublin

Other Primary Investigator(s)

PI Affiliation
University of Houston
PI Affiliation
Trinity College Dublin
PI Affiliation
Philippine Competition Commission

Additional Trial Information

Status
In development
Start date
2023-12-01
End date
2024-12-31
Secondary IDs
Prior work
This trial does not extend or rely on any prior RCTs.
Abstract
Digital credit has exploded in popularity over the last decade with the number of digital lenders growing nearly tenfold globally (Venkatesan, 2023). Although digital credit offers significant potential to advance financial inclusion by allowing previously unbanked and underbanked consumers access credit (Bharadwaj and Suri, 2020), the speed and ease of access to digital credit has raised several consumer protection concerns particularly in low- and middle-income countries (LMICs).

Digital loans are disbursed and repaid electronically, and they differ from traditional credit in several aspects: approval is nearly instantaneous, evaluation of loan application is automated, loans are processed remotely without requiring in-person interaction, and loan decisions are typically determined using “non-traditional data” such as mobile phone data (Francis et al., 2017). The most popular form of digital credit offered in the Philippines is short-term, high-interest, small amount consumer loans disbursed via mobile money platforms (e.g., FMA, 2021; Francis et al., 2017). Despite being costly, the popularity of digital credit in the Philippines suggests unmet demand for credit. However, the surge in customer complaints concerning inadequate disclosures, misrepresentations, high interest rates, and unreasonable collection practices (e.g., FMA, 2021; Tamayo, 2021), also indicates the need to build a stronger evidence base to improve consumer protection in digital lending.

The target client base, typically low-income consumers with low levels of financial literacy, and influenced by typical human behavioral biases such as weak self-control, present bias, overconfidence and limited attention, are susceptible to exploitation due to poor transparency of fees and loan terms and costly roll-over refinancing (Garz et al., 2021). Furthermore, most digital credit borrowers live precarious financial existences that allow little room for financial error and poor consumer decisions can have debilitating immediate and long-term consequences. In the short term, poor consumer decisions can worsen the borrowers’ cash flow position and induce over-indebtedness that could ultimately result in them being shut out of credit markets (Skiba and Tobacman, 2019; Melzer, 2011). In the long- term, over-indebtedness can cause asset erosion and poor psychological health (Gathergood, 2012).

In partnership with the Philippine Competition Commission, we will conduct an online survey and embed a discrete choice experiment (DCE) to provide evidence on how loan choice is affected by behaviourally-informed disclosures and alterations to the choice architecture. After passing the screening questions and answering socio-demographic questions, we will randomly assign respondents to one of the eight treatment arms or the control group. The core task involves choosing the most preferred option from six hypothetical digital credit products which vary across a number of product attributes. The control group will be presented with a set of digital loan choices, inspired by real products, and presented similarly to how they are marketed to consumers in the Philippines (with partial information and informed by specific marketing strategies), requiring potential customers to hover the mouse around to access information on nominal interest rates, processing fees, etc. The first two treatment arms involve provision of additional product information, specifically product attributes a regulator might request be presented in the fine print, These two treatment arms present the six products in random order and involve the clarification of product attributes with and without the total cost of credit (effective interest rate). The remaining six treatment arms involve a ranking one of five product attributes, followed by the final treatment which allows the participant to choose the attribute of ranking.

Our experiment allows us to explore a number of important debates in the literature. First, explored in other credit markets, is how awareness or naivete about one's own abilities and preferences affect product choice (Alcott et.al. 2022; Ausubel 1991; Campbell 2016). We measure borrowers' overconfidence and perceived time inconsistency, and examine their relationship with individual choice of digital loan products. We are specifically interested in testing how generalized overconfidence and perceived time inconsistency affect the weights that consumers place on contingent costs, such as late payment fees or the cost of repeated borrowing, which are only relevant for individuals who believe they will miss a payment or borrow repeatedly.

We anticipate that findings from this study will help improve understanding of consumer protection issues in the digital credit market globally. Across the regulators in the Philippines there is a shared sense that more needs to be done to equip consumers with tools to identify “appropriate” financial products as continued exploitation of consumers will lower willingness to access digital financial services (DFS) and erode trust in the broader financial system (e.g., Garz et al., 2021; McKee et al., 2015).
External Link(s)

Registration Citation

Citation
King, Michael et al. 2023. "What Matters for Consumer Credit Choice? Evidence from the Philippine Digital Credit Market." AEA RCT Registry. December 06. https://doi.org/10.1257/rct.12635-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
Description of interventions

Control (Shrouded attributes and random order of products, random order) The control group mirrors how consumers are currently presented with digital loan options in the Philippines. Products are described with short lines taken from typical advertisements, without explicitly presenting detailed information about product attributes up-front. The respondent needs to hover the mouse over the advertisement to access information on attributes such as nominal interest rates, processing fees, etc. As is typical in the market the total cost of credit is not presented with the product attributes. Using Qualtrics’s randomizer function, the six products are presented in a random order and they are randomized evenly.
Treatment 1 (Information treatment, without cost of credit, random order) This experimental arm lists attributes of the six products in tabular format. However, instead of presenting effective interest rates, respondents are shown nominal interest rates and processing fees. As such, a total of six attributes are shown per product, namely: monthly nominal interest rate, processing fee, late fee per day, application requirements, time to disburse, and probability of approval. Using Qualtrics’s randomizer function, the six products are presented in a random order and they are randomized evenly.
Treatment 2 (Information treatment, with cost of credit, random order) This experimental arm lists the attributes in a tabular format. Unlike treatment 1, effective interest rates are included instead of nominal interest rates and processing fees. As such, a total of five attributes are presented, namely: effective interest rate, late fee per day, application requirements, time to disburse, and probability of approval. Using Qualtrics’s randomizer function, the six products are presented in a random order and they are randomized evenly.
Treatment 3 (Ranking treatment, by effective interest rate) Using the same set of attributes from T2 (effective interest rate, late fee per day, application requirements, time to disburse, and probability of approval), this experimental arm lists the attributes in a tabular format and the products are ranked by effective interest rate (%).
Treatment 4 (Ranking treatment, by late fee per day) This experimental arm lists the attributes in a tabular format and the products are ranked by late fee per day (₱).
Treatment 5 (Ranking treatment, by time to disburse) This experimental arm lists the attributes in a tabular format and the products are ranked by time to disbursement (hours).
Treatment 6 (Ranking treatment, by application requirements) This experimental arm lists the attributes in a tabular format and the products are ranked by the number of documentary requirements during the application process.
Treatment 7 (Ranking treatment, by probability of approval) This experimental arm lists the attributes in a tabular format and the products are ranked by probability of approval.
Treatment 8 (Ranking treatment, personal ranking) This experimental arm lists the attributes in a tabular format, allowing respondents to decide how the products are ranked based on a specific attribute (e.g., by effective interest rate, or by late fee per day, etc.). First, on the landing page, respondents are asked how they want to rank the products. Then, products are presented based on the attribute they chose. For this treatment arm, a back button is enabled as an option to allow the respondent to change how products are ranked. This treatment arm mimics the loan comparison tools online.

Respondents are randomly assigned to either control group or one of the treatment groups, as per the design in figure 1.
Intervention Start Date
2023-12-01
Intervention End Date
2023-12-31

Primary Outcomes

Primary Outcomes (end points)
Primary Outcomes
Effective interest rate (%) - The dependent variable is the value of the effective interest rate of the product chosen.
Late fee per day (₱) - The dependent variable is the value of the late fee per day of the product chosen.
Time to disbursement (hours) - The dependent variable is the value of the time to disbursement of the product chosen.
Application requirements (number of documents) - The dependent variable is the value of the application requirements of the product chosen.
Probability of approval (%) - The dependent variable is the value of the probability of approval of the product chosen.
Willingness to pay estimates
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
Recruitment and sampling
A survey firm in the Philippines is contracted to conduct the online survey and recruit 4,000 respondents. The following criteria must be satisfied to be eligible to take part in the survey: (1) must be a Filipino citizen with age between 18 and 65 years and (2) intends to apply for digital credit in the next 12 months.

The first part of the survey involves questions on demographics and DFS usage. Participants are then randomised into one of nine experimental arms as described above. Varying the type of disclosure/information architecture, respondents are asked to choose one their most preferred product from six hypothetical, but largely representative, digital credit products. Each product has 5 attributes (with 3 levels each) to compare, specifically: effective interest rate (5%, 15%, and 25%), late fee per day (₱10, ₱30, and ₱50), time to disbursement (1 hour, 24 hours, and 48 hours), application requirements (1 document, 2 documents, and 3 documents), and probability of approval (30%, 60%, and 90%).

A range of other questions in areas of overconfidence, financial literacy, patience, risk taking, liquidity constraints, awareness about time inconsistency and present bias are then presented to respondents.

A willingness to pay exercise is conducted for each product attribute.

Finally, we intend to capture use of digital credit products and their attributes 12 months post the on-line experiment.

See pre-analysis plan for further details.

Experimental Design Details
Not available
Randomization Method
Randomisation will be conducted at the point of entry into the online survey.
Randomization Unit
Individual
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
4,000 Philippine consumers interested in taking out digital credit in the next 12 months
Sample size: planned number of observations
4,000 Philippine consumers interested in taking out digital credit in the next 12 months
Sample size (or number of clusters) by treatment arms
444 per experimental arm.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB

Institutional Review Boards (IRBs)

IRB Name
Trinity College Dublin
IRB Approval Date
2023-10-25
IRB Approval Number
N/A
IRB Name
Ateneo de Manila University
IRB Approval Date
2023-11-30
IRB Approval Number
AdMUREC_23_065
Analysis Plan

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information