Journal of Behavioral and Experimental Finance 37 (2023) 100788

Contents lists available at ScienceDirect

Journal of Behavioral and Experimental Finance

journal homepage: www.elsevier.com/locate/jbef

Buy now, pay later (BNPL) ...on your credit card✩

∗
Benedict Guttman-Kenney a , , Chris Firth b , John Gathergood b
a
University of Chicago, Booth School of Business, Chicago, IL, USA
b
University of Nottingham, Department of Economics, Nottingham, UK

article info a b s t r a c t

Article history: We provide the first economic research on ‘buy now, pay later’ (BNPL): an unregulated FinTech credit
Received 18 July 2022 product enabling consumers to defer payments into interest-free instalments. We study BNPL using
Received in revised form 14 October 2022 UK credit card transaction data. We document consumers charging BNPL transactions to their credit
Accepted 6 January 2023
card. Charging of BNPL to credit cards is most prevalent among younger consumers and those living
Available online 16 January 2023
in the most deprived geographies. Charging a 0% interest, amortizing BNPL debt to credit cards –
JEL classification: where typical interest rates are 20% and amortization schedules decades-long – raises doubts on these
G51 consumers’ ability to pay for BNPL. This prompts a regulatory question as to whether consumers should
G28 be allowed to refinance their unsecured debt.
D04 © 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND
D18 license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
M38.

Keywords:
BNPL
Buy now pay later
Consumer credit
Consumer financial protection
Credit cards
FinTech
Household finance
Regulation

1. Introduction one or more (often four or fewer) instalments. With £2.7bn in

UK BNPL lending during 2020, the UK BNPL market is larger by
‘Buy now, pay later’ (BNPL) is an unregulated FinTech credit volume of lending than the UK payday loan market at its peak.
product enabling consumers to defer payments interest-free into Rapid recent growth in consumer use of BNPL has led UK, EU, and
US governments and consumer financial protection regulators
✩ First version: January 5, 2022. The views expressed are the authors and to consider whether and how to regulate BNPL, while Australia
do not necessarily reflect the views of the data provider. The data provider recently introduced regulatory measures.1
reviewed the paper before its release. We thank Michael Dowling & Stefan Palan Very little is known about how consumers use BNPL products,
(the Editors), an anonymous reviewer, Anthony Lee Zhang, Constantine Yannelis, despite the market’s rapid growth, and scrutiny by consumer
Karthik Srinivasan, Kilian Huber, Matthew Notowidigdo, Neale Mahoney, Pascal
financial protection regulators. At the time of writing, in Decem-
Noel, Scott Nelson, Walter Zhang, participants at Harvard Kennedy School,
Federal Reserve Bank of Kansas City, NatWest Group roundtables, RAND Behav- ber 2021, there were no relevant economics or finance research
ioral Finance Forum, Russell Sage Foundation Summer Institute in Behavioral papers studying BNPL. Searches for ‘Buy Now, Pay Later’ and
Economics, Equifax Global Buy Now Pay Later (BNPL) Working Group, NBER
Behavioral Public Economics Bootcamp, Financial Conduct Authority, & industry
participants for their comments and also to Arif Sulistiono and Fabian Gunzinger. 1 US https://www.consumerfinance.gov/about-us/newsroom/consumer-
This work is supported by the UK Economic and Social Research Council financial-protection-bureau-opens-inquiry-into-buy-now-pay-later-credit/ &
(ESRC) under grant number ES/V004867/1. ‘Real-time evaluation of the effects https://financialservices.house.gov/events/eventsingle.aspx?EventID=408594.
of COVID-19 and policy responses on consumer and small business finances’. UK: https://www.fca.org.uk/news/press-releases/fca-publishes-woolard-review-
Guttman-Kenney acknowledges support from the University of Chicago, Booth unsecured-credit-market &
School of Business and the George J. Stigler Center for the Study of the Economy https://www.gov.uk/government/consultations/regulation-of-buy-now-pay-
and the State. later-consultation.
∗ Corresponding author. EU: https://ec.europa.eu/info/sites/default/files/new_proposal_ccd_en_3.pdf.
E-mail addresses: [email protected] (B. Guttman-Kenney), Australia: https://asic.gov.au/about-asic/news-centre/find-a-media-release/
[email protected] (C. Firth), 2020-releases/20-280mr-asic-releases-latest-data-on-buy-now-pay-later-
[email protected] (J. Gathergood). industry/. UK payday loans: (Gathergood et al., 2019a).

https://doi.org/10.1016/j.jbef.2023.100788
2214-6350/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

‘BNPL’ on ArXiv returned no results, on SSRN returned a single BNPL lenders primarily generate revenue through merchant
law working paper on the Singapore BNPL market (Sng and Tan, fees of 3%–6% rather than charging interest or fees.5 Not all BNPL
2021), there were no relevant results on NBER Working Papers lenders charge late fees — for example PayPal does not. Non-
beyond Berg et al. (2021)’s broader review of FinTech lending, and payment of BNPL debt can still have consequences: BNPL lenders
google scholar showed no economics or finance papers in either may block new purchases by that consumer, pass unpaid debt to
published or working form (the only work appearing was from debt collectors, and missed payments may get reported in credit
other academic fields).2 files. Consumers may experience costs if their BNPL payment has
We present the first quantitative academic research to study knock-on adverse effects on their other finances such as trig-
consumer use of BNPL. Given the lack of prior literature, we gering overdraft fees, accumulating credit card interest, missing
start our paper by providing institutional details on BNPL to help payments on other bills (as found in the UK payday lending
stimulate research on this topic. market by Gathergood et al., 2019a).
One explanation for the lack of prior research is a lack of
data: for example, most BNPL is not visible in credit files. To 2.2. Economics of BNPL
help fill this gap, we offer early insights into how consumers use
BNPL, by analyzing BNPL transactions that show in UK credit card Economic theory provides mixed perspectives on the con-
transaction data.3 sumer benefits of deferred payments through credit products
BNPL is often presented as a standalone credit product in the such as BNPL. The life-cycle model implies opportunities to smooth
form of an interest-free loan repaid in one or more instalments. consumption at zero interest cost are weakly welfare improv-
At the point of purchase, the consumer provides details of the ing — especially in a high inflation environment (Ando and
account from which the payments will be taken, after which Modigliani, 1963).
loan repayments occur. However, unlike other credit products, Deferred payments – through BNPL or other credit products
consumers can choose to repay BNPL using a credit card. – also presents the possibility for welfare losses. For example,
We research whether consumers charge BNPL transactions financially unsophisticated or naïve present focused consumers
to their credit card, and the degree of heterogeneity in such mistakenly overconsuming (e.g. Allcott et al., 2022).
behaviours across consumers. We find the existence of UK con- A key BNPL product feature is decoupling the consumption
sumers charging their interest-free BNPL to their credit card. Such benefit today from the pain of paying later (e.g. Prelec and
behaviour is more common among (i) younger consumers and (ii) Loewenstein, 1998) – something especially attractive to present
biased consumers (e.g. O’Donoghue and Rabin, 1999).
consumers living in more deprived geographies. Taken together
these facts raise a concern that some consumers will enter into
2.3. BNPL on credit cards
a debt spiral. A debt spiral may occur from transforming a 0%
interest BNPL debt that amortizes over a few weeks or months
Lenders providing regulated credit are required to consider,
into credit card debt: a product that typically incurs 20% interest
at the time of credit application, the ability of a consumer to
rates and has decades-long amortization schedules. It prompts a
repay a debt – known as ability to pay in the US and affordability
regulatory question as to whether consumers should be allowed
or creditworthiness in the UK – out of their income or assets
to refinance their unsecured debt.
‘‘without the customer having to borrow to meet the repayments’’.6
This means, for example, a consumer cannot make a payment
2. Institutional details for their regulated UK mortgage with their credit card (or meet
one credit card payment with another credit card). This applies
2.1. BNPL product structures irrespective of whether a consumer repays their credit card in
full or revolves interest-bearing debt. These consumer protection
Buy Now, Pay Later (BNPL) is provided at the point of sale, regulations are designed to protect vulnerable consumers from
providing consumers the option to defer payments into one or harm such as cases where the incentives of lenders are misaligned
more interest-free instalments.4 BNPL is mainly used online with with the incentives to improve consumer outcomes.
lenders typically a third party separate to the retailer. Most BNPL As BNPL is unregulated it is not subject to ability to pay regula-
is unregulated. We analyse the UK BNPL market but the US tions: BNPL can be charged to debit or credit cards.7 Transferring
market is analogous. BNPL debt to credit cards increases the risks to the consumer
Repayment structures vary across and within BNPL lenders. and thus is a warning flag to regulators. Credit cards can be
For example, Klarna in the UK provides an option to repay in the a more costly form of credit with an average interest rate of
next thirty days as well as an option to repay in three instalments near 20% APRs unless the cardholder repays their balance in full.
thirty days apart whereas Clearpay has four payments that are Credit cards also have decades long amortization schedules if
two weeks apart. In the US, the Consumer Financial Protection the cardholder only makes the minimum payment – a costly
Bureau defines a BNPL as having four or fewer instalments but behaviour common in UK and US data (Keys and Wang, 2019;
some products have more (e.g. OpenPay in the UK offers 3–10 Guttman-Kenney et al., 2022). The burden of persistently carrying
instalments). While BNPL are primarily instalment loans, there is credit card debt may also have non-financial costs such as adverse
product innovation in structures. For example, Amazon features mental health impacts. As one BNPL lender stated ‘‘there is clearly
a credit limit (‘Instalments by Barclays’) while some debit and
credit cards enable cardholders to pay particular transactions in 5 https://www.consumerfinance.gov/about-us/newsroom/consumer-
BNPL-esque instalments (e.g. American Express’s ‘Pay It Plan It’). financial-protection-bureau-opens-inquiry-into-buy-now-pay-later-credit/.
6 https://www.fca.org.uk/publication/policy/ps18-19.pdf. See Guttman-
Kenney and Hunt (2017) & DeFusco et al. (2020) for research into UK and US
2 See Online Appendix for more details including consumer surveys, blog
ability to pay rules.
discussions, and industry reports. 7 In the UK, debit cards are the most common means of payment. 57% all
3 See Agarwal et al., 2009 for an example of early analysis of consumer use
UK payments are made on debit or credit cards. Cash has declined to from 45%
of payday loans informing subsequent causal research (e.g. Gathergood et al., to 15% of payments 2015 to 2021. The remainder of transactions are mainly
2019a; Allcott et al., 2022). bank transfers (e.g. Autopays/Direct Debits). Source: UK Finance, UK Payments
4 See Online Appendix for more institutional detail on BNPL. Markets Summary, August 2022.

2
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

greater risk of consumer harm from spending on credit cards’’ not BNPL. Other BNPL lenders – most notably Clearpay – are only
(than BNPL).8 If a consumer repays their credit card in full and BNPL lenders without payment processing services.13
does not have binding liquidity constraints (i.e. not near their Although our data is a sample and subject to potential mea-
credit card limit and have liquid cash available) there is limited surement error, we feel justified using it as an informative proxy
benefit from charging BNPL to their credit card. This is because for all BNPL usage in a market that currently has very thin data
any BNPL payments that are charged within the same credit card and, of course, as a highly accurate reading of BNPL usage on
payment cycle (with its interest-free grace period that lasts 15– credit cards. The latter is especially important for regulators wish-
ing to put in place measures that help consumers avoid high costs
56 days) all come due at the same time. Any payments that
of debt, or at the extreme, entering debt spirals. We believe our
come due in subsequent payment cycles effectively extend the
data is an informative proxy given the absence of representative,
grace period before payments are due. However, this comes at
comprehensive data: a matched, transaction-level dataset of BNPL
a cost since it is also discounting the benefits of any credit
and credit cards.
card rewards points relative to if the consumer had immediately
charged the transaction to their credit card in one instant (non- 4. Results
BNPL) payment.9 Finally, if a consumer is unable to meet their
BNPL payments on time, transferring BNPL debt to their credit 4.1. BNPL use on credit cards
card may be simply postponing an inevitable default and risk
shifting the credit losses from the BNPL lender to the credit card Our first main finding is documenting the existence of charg-
lender. There is evidence of this: Capital One deemed the risks so ing of BNPL transactions to credit cards. BNPL transactions are
great it has banned consumers from charging BNPL to their global commonly present on credit cards: 19.5% of UK credit cards active
credit cards.10 (i.e. with any transaction present) in December 2021 have a
transaction by a BNPL firm charged to their credit card during
2021.
3. Data Some cardholders have multiple BNPL transactions on their
credit cards. This may be due to multiple BNPL purchases or BNPL
We use anonymized UK credit card transactions data sourced purchases split into multiple instalments. While each individual
from multiple banks and credit card issuers.11 These data are BNPL transaction is typically small (a median value of £19.65 and
created to be a real-time leading indicator used by industry to 96% are £100 or less), the total amount of BNPL transactions per
track spending (in aggregate and also of particular firms, sec- card during 2021 by credit cardholders using BNPL is non-trivial:
tors, and regions) and are provided to us under an academic li- the median value is £157 and 17.6% have spent £500 or more.14
cence.12 Our data include approximately one million credit cards What fraction of all BNPL is charged to credit cards? In the
held by UK consumers between December 2018 & September US this is estimated to be 22%.15 For the UK, a consumer survey
estimates 26% of BNPL consumers paid for BNPL using their
2022. For context, the median UK credit cardholder holds one
credit card and 42% using any type of borrowing (e.g. credit card,
credit card (Gathergood and Guttman-Kenney, 2016; Gathergood
overdraft, payday loan).16 The incidence of such practices may
et al., 2019b). The Online Appendix contains additional descrip-
vary across BNPL lenders and over time.17
tive analysis on these data. The patterns of transactions we observe are consistent with
Each transaction record provides details including the spend- a variety of potential economic models of consumer behaviour.
ing amount and tagged information on the type of spending. Each One plausible model is where overconfident consumers mis-
transaction has an anonymized card-account identifier to enable predict their future ability to repay BNPL on-time and, when
tracking over time. For each card-account we observe the card- payments come due, defer payments by charging them to their
holder’s age range and geography (‘postcode sector’). Postcode credit card — potentially transferring them interest-bearing credit
sectors are very granular geographies: there are over 11,000 post- card balances or defaulting on this debt.
code sectors in the UK with each sector containing approximately
3,000 addresses. As these are a dataset of credit card spending 4.2. BNPL usage over time
they do not include when or whether the consumer made credit
card repayments, their cost of borrowing, whether they made Fig. 1 shows rapid growth in the value of BNPL spending on
payments via debit cards, or have unpaid BNPL payments due. credit cards between January 2019 and December 2021: increas-
BNPL transactions are tagged by us via a field that records a ing 21.4 times. To help interpret our time series of BNPL spending
on credit cards, Fig. 1 also includes an index of BNPL payments
transaction’s payment processor (e.g. Afterpay, Klarna). We do not
on debit cards — using a separate sample from the same data
categorize PayPal as a BNPL lender in our analysis as we cannot
provider. Both series follow similar trends.
distinguish BNPL from its large, established non-BNPL business
of processing payments. Klarna are the largest BNPL lender in the
13 Square is a payment processing business that recently acquired Afterpay
UK but they also provide some payment processing services so
(known as Clearpay in the UK). As they have different names we can differentiate
we may be tagging some transactions processed by them that are Afterpay/Clearpay’s BNPL from Square’s payment processing in our transactions
data.
14 Understanding repeated use of BNPL is an important avenue for future
8 https://www.cityam.com/buy-now-pay-later-firm-klarna-claims-credit-
research. For example, repeating analyses similar to that done on payday
cards-pose-the-real-risk/. loans to understand consumer patterns of BNPL borrowing and how commonly
9 There is also an effort cost of the consumer having another financial they ultimately end up in default or a cycle of persistent debt. https://files.
intermediary to interact with. consumerfinance.gov/f/201403_cfpb_report_payday-lending.pdf.
10 https://www.reuters.com/article/us-capital-one-fin-payments/capital-one- 15 https://www.creditkarma.com/about/commentary/consumers-rely-on-buy-
stops-risky-buy-now-pay-later-credit-card-transactions-idUSKBN28H0OR. now-pay-later-amid-record-inflation-use-credit-to-pay-it-off.
11 See Baker and Kueng (2022) for a review of household financial transaction 16 https://www.citizensadvice.org.uk/about-us/about-us1/media/press-
data. releases/two-fifths-borrowed-to-pay-off-buy-now-pay-later/.
12 These are similar to data used in the US by Chetty et al. (2020) to track the 17 One anonymous UK BNPL lender reported ‘‘less than 10% of its cus-
COVID-19 pandemic in real-time. See Vavra (2021) for a review of using private tomers repay using a credit card’’. https://www.ukfinance.org.uk/system/files/
sector administrative micro data for tracking the COVID-19 pandemic. UK-Finance-response-to-the-HMT-BNPL-consultation-FINAL-060122.pdf.

3
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

Fig. 1. Value of BNPL transactions on active credit cards (black line) and active debit cards (yellow line), 2019–2022.
Notes: UK credit and debit card transactions. BNPL is buy now, pay later. Each series is 28 day moving averages of value of BNPL spending on those cards indexed
to 1 in January 2019. 19.5% of UK credit cards active (i.e. with any transaction present) in December 2021 have a transaction by a BNPL firm charged to their credit
card during 2021, while 15.3% and 12.1% have charged in the last six and three months leading up to December 2021.

BNPL spending on credit cards as a percent of all credit card Regulators may be interested in evaluating BNPL’s role as gate-
spending averaged 1.2% and 1.6% during 2021 and December way debt products for young consumers who are inexperienced
2021. This rapid growth lines up with official estimates which users of financial products. BNPL may potentially be harmfully
size the value of all UK BNPL lending during 2020 at £2.7bn - leading young consumers into a debt spiral of taking on in-
‘‘more than tripling in 2020’’: in our data the 2020 value of BNPL creasingly expensive forms of debt. Or conversely, BNPL may be
transactions is 3.4 times its 2019 levels.18 beneficially enabling young consumers to learn to prudently use
December 2021 is the peak in BNPL spending on credit cards low cost credit, improve their credit access, and avoid relying on
by transactions value and the index is 5.9 times December 2019 higher cost products.
and 2.0 times its December 2020 levels. There are local seasonal
peaks aligning to the timing of payments coming due for ‘Black 4.4. BNPL usage by area deprivation
Friday’ and Christmas spending. BNPL on credit cards grew during
the first wave of COVID-19 (March–April 2020). These were a We seek to understand the vulnerability of consumers using
period when UK consumption sharply fell. BNPL on credit cards BNPL. Regulators’ risk assessments consider the vulnerability of
growing during this period may be a function of consumers consumers using financial products (e.g. giving higher welfare
shifting towards spending online during this period. Our real- weights to more deprived consumers).21 From a consumer wel-
time data also reveals a 2022 slump in BNPL on credit cards
fare standpoint, a regulator may be more worried about potential
following worsening macroeconomic conditions: a period when
consumer detriment from a (ultimately) high-interest product
BNPL lenders’ valuations also fell.19
sold to lower income, less financially-capable consumers than
a zero-interest product sold to consumers wealthy consumers.
4.3. BNPL usage by age
While we cannot observe income and other consumer charac-
What heterogeneity by consumers’ age is there in BNPL usage teristics at the individual level, we can draw upon postcode
on credit cards? Fig. 2 finds the use of BNPL on credit cards identifiers of the cardholder’s address to match in measures at
is more prevalent among younger consumers. 84% of overall the local level.
transactions by BNPL firms on credit cards is by consumers aged To evaluate this aspect we aggregate data to the Local Author-
18 to 49. While we do not observe credit card repayments or ity District – areas of local government – and merge in the 2019
interest in our data, we know more generally that younger credit English Indices of Multiple Deprivation (IMD). IMD is the official
cardholders in the UK are least likely to repay their credit card government measure providing a relative ranking across districts
balances in full and so are more likely to incur interest costs.20 of deprivation. IMD is constructed from 39 indicators to ensure it
captures a broad range of resources.22
18 https://www.fca.org.uk/news/press-releases/fca-publishes-woolard-review- Panels in Fig. 3 rank each authority by its IMD on the x-axes
unsecured-credit-market & https://www.gov.uk/government/consultations/
where a ranking of 1 is most deprived and 315 the least. Panels
regulation-of-buy-now-pay-later-consultation. A and B use the sample of cards active in December 2021 and
19 https://www.ft.com/content/483451db-9221-4ca4-83a6-b4ddc6bfcfbb. panels C and D those active in both January and December 2021
20 Financial Conduct Authority’s (FCA) nationally-representative UK Financial (results are consistent). We measure BNPL usage on credit cards
Lives survey reports the proportion of UK consumers who revolve credit card
debt (conditional on holding any credit card) by age: 30% of 18–24, 46% of
21 https://www.fca.org.uk/publication/consultation/cp21-13.pdf & https:
25–34, 46% of 35–44 to 37% of 45–55, 22% of 55–64, 10% of 65+. This survey
does not contain any information on BNPL usage or the number of credit cards //www.fca.org.uk/publication/finalised-guidance/fg21-1.pdf.
held preventing a like-for-like comparison. See Fulford and Schuh (2015) for 22 https://assets.publishing.service.gov.uk/government/uploads/system/
estimates of revolving behaviour across the life cycle for US consumers. uploads/attachment_data/file/833951/IoD2019_Technical_Report.pdf.

4
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

Fig. 2. BNPL usage on active credit cards by age, 2021.

Notes: UK credit card transactions. BNPL is buy now, pay later. The sample of active credit cards are defined as those with any BNPL or non-BNPL transactions in
both January and December 2021. Panel A shows whether any BNPL spending is recorded on an active credit card as a percent of all active credit cards. Panel B
shows the value of BNPL spending recorded on active credit cards as a percent of the total value of spending on active credit cards.

the y-axes: Panels A and C use the percent of credit cards with charging of BNPL to credit cards. The β coefficients can be inter-
any transaction by a BNPL firm in the last 12 months, Panels B preted as follows: moving from the least to most deprived local
and D use the share of the total value of spending on credit cards authority is associated, on average, with higher BNPL use: 28 to
that by a BNPL firm in the last 12 months. The size of each dot is 30% (4.4 to 4.6 pp) as measured by fraction of cards, and 58 to
each district’s share of population. 66% (0.5 pp) measured by the fraction of spending.23 Given this
We plot a linear regression (Eq. (1)) to describe the uncon- relationship it is unclear how much BNPL lenders restrict credit
ditional (non-causal) relationship between IMD and BNPL usage. supply — something visible in regulated credit (e.g. Agarwal et al.,
There is one observation (d) per district in England (315 in total) 2018). While we do not observe whether these credit cards are
where Yd are our measures of BNPL usage (0–100) and IMDd is repaid in full or accumulate fees or interest (or at what rate), we
know more generally that credit cards in the most deprived areas
the ranking of IMD (1–315 where a higher value is less deprived).
of England are those least likely to repay in full, be eligible for 0%
We weight each observation by its district’s ONS population and
credit card deals, and instead have higher interest rates.24
cluster standard errors by regions of the UK to allow for spatial
correlation across districts.
23 Results robust to using income instead of IMD: BNPL use is 30% higher as
Yd = α + β IMDd + εd (1) measured by fraction of cards, and 68% higher as measured by the fraction of
spending moving from the highest to lowest income local authority.
By both measures of BNPL usage we find the most deprived 24 Financial Conduct Authority’s nationally-representative Financial Lives sur-
areas have statistically significantly (p-values < 0.001) higher vey reports the proportion of UK consumers who revolve credit card debt

5
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

Fig. 3. BNPL usage on active credit cards by local area deprivation, 2021.
Notes: Credit card transactions in England, Ministry of Housing, Communities & Local Government (MHCLG), Office for National Statistics (ONS) data. BNPL is buy
now, pay later. Data aggregated to Local Authority District (LAD) level based on cardholder postcode sector. Cardholders in England across 315 LADs since there is
no official standardized UK-wide index of multiple deprivation. Due to small populations, City of London is merged with Westminster and Isles of Scilly merged
with Cornwall. Size of dot is share of ONS England population estimates and the linear regression is weighted by these shares with standard errors clustered by UK
region. Active credit cards are those with any BNPL or non-BNPL spending. Panels A and C are percent of the number of active credit cards in a LAD which have
any BNPL spending. Panels B and D are percent of the value of credit card spending in a LAD on BNPL. Deprivation ranks by English Indices of Multiple Deprivation
(2019) - more details: www.gov.uk/government/statistics/english-indices-of-deprivation-2019.

While this relationship indicates BNPL’s potential risks for reg- constrained consumers in poorer areas with high MPCs to in-
ulators to consider, its welfare effects are unclear. BNPL may be crease their consumption or use BNPL to save money instead of
welfare improving enabling consumption smoothing for liquidity- borrowing on higher cost credit (e.g. overdrafts, payday loans).
However, BNPL may be welfare decreasing if BNPL leads to in-
(conditional on holding any credit card) is 52% for the most deprived decile creased borrowing on higher cost credit or consumers in poorer
of IMD compared to 20% in the least deprived decile of IMD. This survey does
not contain any information on BNPL usage or the number of credit cards held
areas are less financially sophisticated or making naïve mistakes
preventing a like-for-like comparison. (e.g. overoptimism and/or present focused in their ability to repay
6
B. Guttman-Kenney, C. Firth and J. Gathergood Journal of Behavioral and Experimental Finance 37 (2023) 100788

BNPL debt). The effects of such naïvete has previously been eval- References
uated in credit cards (e.g. Meier and Sprenger, 2010) and payday
loans (e.g. Allcott et al., 2022). By increasing indebtedness, BNPL Agarwal, S., Chomsisengphet, S., Mahoney, N., Stroebel, J., 2018. Do banks pass
through credit expansions to consumers who want to borrow? Q. J. Econ.
may cause broader distress (e.g. adverse mental health, missing
133 (1), 129–190.
other bills) that may outweigh consumption benefits. Agarwal, S., Skiba, P.M., Tobacman, J., 2009. Payday loans and credit cards:
New liquidity and credit scoring puzzles? Am. Econ. Rev. Pap. Proc. 99 (2),
5. Conclusions 412–417.
Allcott, H., Kim, J.J., Taubinsky, D., Zinman, J., 2022. Are high-interest loans
predatory? Theory and evidence from payday lending. Rev. Econom. Stud.
We document that a large minority of UK consumers charge 89 (3), 1041–1084.
BNPL to credit cards, especially younger consumers and those Ando, A., Modigliani, F., 1963. The ‘‘life cycle’’ hypothesis of saving: Aggregate
living in more deprived areas. This raises doubts on these con- implications and tests. Amer. Econ. Rev. 53 (1), 55–84.
sumers’ ability to pay for BNPL and prompts a regulatory ques- Baker, S.R., Kueng, L., 2022. Household financial transaction data. Annu. Rev.
Econ. 14, 47–67.
tion as to whether consumers should be allowed to refinance
Berg, T., Fuster, A., Puri, M., 2021. FinTech lending. NBER Working Paper No.
their unsecured debt. Further research is required to measure 29421.
how binding liquidity constraints are for consumers using BNPL, Chetty, R., Friedman, J.N., Hendren, N., Stepner, M., The Opportunity Insights
the effects of BNPL on consumers, whether BNPL is substitut- Team, 2020. The economic impacts of COVID-19: Evidence from a new public
ing for other payment mechanisms or forms of lending, and database built using private sector data. NBER Working Paper No. 27431.
DeFusco, A.A., Johnson, S., Mondragon, J., 2020. Regulating household leverage.
BNPL’s competitive effects on these established products. Given Rev. Econom. Stud. 87 (2), 914–958.
the current interest by policymakers in the BNPL market, further Fulford, S., Schuh, S.D., 2015. Consumer revolving credit and debt over the life
research could fruitfully evaluate the effects of potential BNPL cycle and business cycle. Federal Reserve Bank of Boston Working Paper No.
regulations. 15-17.
Gathergood, J., Guttman-Kenney, B., 2016. Can we predict which consumer credit
users will suffer financial distress? FCA Occasional Paper No. 20.
CRediT authorship contribution statement Gathergood, J., Guttman-Kenney, B., Hunt, S., 2019a. How do payday loans affect
borrowers? Evidence from the UK market. Rev. Financ. Stud. 32 (2), 496–523.
Gathergood, J., Mahoney, N., Stewart, N., Weber, J., 2019b. How do individuals
Benedict Guttman-Kenney: Conceptualization, Data curation,
repay their debt? The balance-matching heuristic. Amer. Econ. Rev. 109 (3),
Formal analysis, Funding acquisition, Investigation, Methodology, 844–875.
Project administration, Resources, Software, Supervision, Valida- Guttman-Kenney, B., Adams, P., Hunt, S., Laibson, D., Stewart, N., 2022. The
tion, Visualization, Writing – original draft, Writing – review semblance of success in nudging consumers to pay down credit card debt.
& editing. Chris Firth: Conceptualization, Data curation, Formal University of Chicago Working Paper.
Guttman-Kenney, B., Hunt, S., 2017. Preventing financial distress by predict-
analysis, Funding acquisition, Investigation, Methodology, Project ing unaffordable consumer credit agreements: An applied framework. FCA
administration, Resources, Software, Supervision, Validation, Vi- Occasional Paper No. 28.
sualization, Writing – original draft, Writing – review & edit- Keys, B.J., Wang, J., 2019. Minimum payments and debt paydown in consumer
ing. John Gathergood: Conceptualization, Data curation, Formal credit cards. J. Financ. Econ. 131 (3), 528–548.
Meier, S., Sprenger, C., 2010. Present-biased preferences and credit card
analysis, Funding acquisition, Investigation, Methodology, Project
borrowing. Am. Econ. J. Appl. Econ. 2 (1), 193–210.
administration, Resources, Software, Supervision, Validation, Vi- O’Donoghue, T., Rabin, M., 1999. Doing it now or later. Amer. Econ. Rev. 89 (1),
sualization, Writing – original draft, Writing – review & editing. 103–124.
Prelec, D., Loewenstein, G., 1998. The red and the black: Mental accounting of
Appendix A. Supplementary data savings and debt. Mark. Sci. 17 (1), 4–28.
Sng, A., Tan, C., 2021. Buy now pay later in Singapore: Regulatory gaps and
reform. SSRN Working Paper No. 3819058.
Supplementary material related to this article can be found Vavra, J., 2021. Tracking the pandemic in real time: Administrative micro data
online at https://doi.org/10.1016/j.jbef.2023.100788. in business cycles enters the spotlight. J. Econ. Perspect. 35 (3), 47–66.