Buy Now, Pay Later Cost Disclosure and Consumer Payment Choice: Evidence from a Randomized Field Experiment in Retail Stores

The intervention consists of a standardized verbal disclosure of the full cost of installment payment / buy-now-pay-later (BNPL) options at the point of sale. In treatment stores or treatment time blocks, sales staff are trained to inform customers who consider an installment or BNPL payment option of: (i) the immediate payment price, (ii) the total amount payable under the installment plan, calculated as the monthly payment multiplied by the number of months, and (iii) the implied overpayment in absolute and/or percentage terms. The disclosure is delivered before the customer finalizes the payment decision. Staff are instructed to communicate this information neutrally and not to recommend either immediate payment or installment payment.

2026-05-10

2026-06-10

Share of purchases made via BNPL/installment payment

The primary outcome is the proportion of completed transactions in which the customer chooses a BNPL/installment payment method rather than immediate payment. Immediate payment includes cash, card, or other non-installment payment methods. BNPL/installment payment includes BNPL products and any equivalent installment-based payment options offered at the point of sale.

Construction

The outcome is constructed from transaction-level sales data. For each completed transaction, the following variables are recorded:
* Transaction ID: unique identifier for each purchase
* Payment method: immediate payment vs. BNPL/installment payment
* Store ID: to distinguish stores in the between-store design
* Date and time slot: to distinguish treatment and control blocks in the within-store design
* Treatment/control status: assigned at the store or time-slot level
* Product category and transaction value, where available

At the transaction level, the outcome is defined as:
BNPL=1
if transaction is paid via BNPL/installment, and:
BNPL=0
if transaction is paid via immediate payment.

The BNPL share is then calculated as the ratio of number of BNPL/installment transactions and total number of completed transactions. This share can be calculated daily, by time slot, by store, or by treatment/control condition.

Interpretation

A statistically significant reduction in BNPL/installment share in the treatment condition relative to the control condition would be interpreted as evidence that transparent disclosure of the full installment cost affects consumers’ payment choices and reduces the likelihood of choosing BNPL/installment payment.

Secondary outcomes include:

* Average transaction value
* Total sales volume
* Number of completed transactions
* Product category composition / product shifts
* Conversion rate, if customer traffic or store-visit data are available
* Returns, if available

Overview

This study uses a randomized controlled field experiment to evaluate the effect of transparent payment cost disclosure on customer payment choice and purchasing behavior in a retail setting.

Two Complementary Designs

The study combines two randomization approaches implemented during the same study period.

1. Between-Store Design

A pair of comparable retail stores is selected and matched on key pre-intervention characteristics, including historical sales, average transaction value, traffic patterns, and historical share of BNPL/installment purchases. Within this matched pair, one store is randomly assigned to the treatment condition and the other to the control condition. The treatment store implements the intervention, while the control store continues standard operations unchanged.

2. Within-Store Design

A third store serves as its own control through time-slot randomization. Each business day is divided into two half-day blocks. For each day, one block is assigned to the treatment condition and the other to the control condition. The allocation schedule is randomized and balanced so that treatment and control blocks occur approximately equally often in the first and second halves of the day and across days of the week.


Randomization

Treatment assignment in the between-store design is determined randomly within the matched store pair. In the within-store design, treatment and control status is assigned at the half-day time-slot level, with a 50/50 treatment-control ratio maintained over the study period.

Control Condition

In control store and control time slots, standard pricing communication practices remain in place. Staff do not introduce the additional standardized disclosure of total installment/BNPL cost beyond existing practice.

Baseline Period

Data collected before the start of the intervention, prior to May 10, are used to assess baseline comparability across stores and to support difference-in-differences analysis.

Unit of Randomization

* Between-store design: store
* Within-store design: half-day time slot


Unit of Analysis

The unit of analysis is the individual transaction.

External Validity

The study is conducted across a small number of comparable retail stores in one geographic area. Because the sample of stores is limited and the intervention is implemented in a specific retail context, external validity will be interpreted cautiously. The main objective of the pilot is to obtain internally valid evidence on whether full BNPL/installment cost disclosure affects payment choice and purchasing behavior in the participating stores.

The study will be conducted in 3 retail stores. The primary unit of observation is the individual transaction. The expected number of transaction-level observations will depend on realized sales during the study period and will be reported after data collection.

Not available

Lottery

Retail stores

Yes

3 stores total. Between-store design: 1 treatment store and 1 control store. Within-store design: 1 store with approximately 60 half-day time blocks over one month, split 50/50 between treatment and control.

3 retail stores

3 retail stores; approximately 60 half-day time-block clusters in the within-store component; approximately 1,200 completed transaction-level observations over one month.