Thread by @pauldadams

I’m super excited to share that we published new experimental research into #BuyNowPayLayer (#BNPL) understanding and behaviour today. The work is joint work with @owainservice and @CogCo_ for @CitizensAdvice. Find the report here - citizensadvice.org.uk/Global/Citizen…

🧵 (1/n)

#BNPL has seen a massive growth worldwide over the last few years, spurred on by seamless tech and online shopping habits. Some policymakers are concerned that, despite the providers claims, #BNPL could be storing up debt problems for the future

(2/n)

Despite this, little research has been conducted in this market (apart from @gk_ben recent paper) and even less on how we might improve consumer understanding of the potential risks. We fill this gap and give policymakers a sense of what interventions can achieve

(3/n)

We recreate in extreme high fidelity the online shopping and payment experience. I’ve run a lot of online hypothetical experiments in my time, and this is by far the smoothest and most realistic I’ve seen. The team at @CogCo_ did a fantastic job in creating this environment (4/n)

We then create 3 potential regulatory interventions that introduce #info/#warnings, add #friction (pop-ups) or reduce the #salience of #BNPL options. These are inspired by interventions that have worked in other financial settings like pay day lending and investments

(5/n)

We look at the impact on both stated payment choices (basically do people say they would use their credit/debit card or the BNPL option), as well as on comprehension measured across a series of 9 questions

(6/n)

Our key results are:
1. All three of our interventions increased comprehension, but unsurprisingly the Combined+ intervention had the largest point estimate relative to control

(7/n)

2. Only the combination of repeated risk warnings, additional friction and reduced salience of BNPL options was able to reduce individuals use of BNPL in this fictitious setting

(8/n)

3. We looked at a number of demographic variables to see if there was anything jumping out at us, but did not find anything too surprising. The single biggest predictor of choosing BNPL in our setting was whether they’d used it before

(9/n)

Now, we can all poke holes in a hypothetical online experiment, after all this isn’t really money on the line, but we have two things that help.

1. the realistic environment we created
2. control group behaviour seems to match what we see in the real world

(10/n)

That being said, I’d love to see this research being taken into a real natural field experiment, working with a firm like @Klarna to understand how we can help people use BNPL responsibly. Any volunteers?

(11/n)

Finally, worth saying we had some great input from experts in the field including @dilipsoman, @MoneyMindMerle, @johngathergood, @pnewall, @stefanhunt, @danielmabuse and @shlomobenartzi as well as great input from staff @CitizensAdvice (@DrNadyaAli and @Dave_MdaC)

(12/12)