The main point for me to start this blog was to be able to disseminate academic work to an audience that was much wider than academia. I’ve recently finished my PhD and once again realised how long it would take for my findings to be published in what is essentially a non-accessible format. So I thought “fuck it. I’ll do it myself."
Now obviously the blog format is not the best format for this type of publication. My thesis was 183 pages long. It’s a beast of its own. Now if you’re super tempted to read this thesis cover to cover, I won’t stop you. Just pop me an email or reach out via social media, I’d be happy to send the whole thing over, or read it here in a shared doc. But let’s see how you like the summary first!
The Context As I was finishing my MSc in Behavioural and Economic Science, and was preparing to enter the PhD, a major shift in the payment landscape had occurred: contactless payment methods, despite being introduced in the early 2000s, gained increasing popularity. Within the UK, 2016 was the year that they started gaining traction, and from thereon, they gained traction across the globe. Now, they represent the majority of in-store transactions in the UK, the Eurozone and Australia (Campbell, 2015; Statista, 2020b; WestPac, 2017). A trend which was further greatly accelerated by the physical distancing measured deployed during the Covid-19 pandemic (Financial Conduct Authority, 2021; Statista, 2020c). As with the global proliferation of payment cards, the increased uptake of contactless payment methods and the resulting replacement of cash may not be universally beneficial (Rosenberg, 2005). Cash is often used as a budgeting tool or for constraining expenditure (Doyle et al., 2017), activities that may become harder with contactless payment. We have seen issues with payment methods before. There is a long list of research focusing on “issues” as seen with credit cards. As compared to cash, credit cards were found to be associated with increased willingness-to-pay and increased spending (Feinberg, 1986; Hirschman, 1979; Prelec and Simester, 2001; Soman, 2003; Tokunaga, 1993), less accurate expenditure recall (Gross and Souleles, 2002; Raghubir and Srivastava, 2008; Srivastava and Raghubir, 2002), reduced impulse control leading to more frequent spending (See-To and Ngai, 2019; Thomas, Desai, and Seenivasan, 2011) and debt accumulation (Gross and Souleles, 2002). It has also been found that those who pay by credit card feel less attached to the products they buy (Shah et al., 2016), and that the expected use of credit cards for the purchase of a product is associated with increased focus on the product’s benefits, rather than its cost (Chatterjee and Rose, 2012).
Don’t think you’re safe using debit cards either; similar effects have been found comparing cash and debit cards, with significantly higher willingness-to-pay with debit cards, even after controlling for cash-on-hand constraints, spending type, price familiarity and consumption habits (Runnemark, Hedman, and Xiao, 2015).
So there is research to indicate that any payment method that deviates from cash makes you worse on a number of dimensions. Good to know, but we’ve moved on from credit and debit cards. They are clearly so late 1900s, and we live in the now. And now we are dealing with contactless payment methods. So I thought I’d update the research.
Chapter 1: The Effect of Contactless on Expenditure Recall
One key issue with credit cards: people seemed to forget how much they spent on them often leading to a nasty surprise. This has even been found to hold true for a single transaction (See-To and Ngai, 2019; Thomas, Desai, and Seenivasan, 2011). Is that true for contactless too?
We tested this in two studies. In Study 1 we observed behaviour. We let people do their shopping in the on campus grocery store and asked them to fill in a survey about the experience afterwards, checking which payment method they used, asking them to estimate how much they paid, and then some other questions. We also asked them to give us their receipt (we did need to check the amount paid etc.). About 3000 people did the survey. And what did we find? People are best at recalling their expenditure using cash (as expected!) and worst at recalling their expenditure when using PIN-verified credit cards (huh?!). Contactless fell right in the middle. Well there goes that theory… There was a small redemption: compared to cash, expenditure recall accuracy with contactless was 6.6% lower.
In Study 2 we decided to kick things up a notch. For transparency: this study was conducted 3 years later as a result of a revise & resubmit, so things had changed a bit (I’m talking about the pandemic, obviously). We recruited participants online, randomly allocated them to one of three payment methods (PIN-verified debit, contactless debit and cash – the most popular methods in Study 1) and asked them to go grocery shopping at their own leisure, come back to the survey later with a receipt and then they were done. We over-recruited, had an okay drop-out rate (to be expected for a study with multiple parts) and then had to massively cut down on non-usable answers (oh the life of a researcher). In the end we found that recall with contactless was again lower than for cash (5.9%), but there was significant difference between contactless and PIN-verified debit, again. I argued that this may be due to the fact that contactless and PIN-verification can be construed as features rather than methods themselves (the method remains to be a card) and that the feature isn’t strong enough in and of itself to make a difference.
Quick bonus round: the main driver for expenditure recall in both studies was the number of items purchased, which constantly had a ~30% impact. And we also tested for the pain of paying (main theory in explaining the differences in payment methods) and it had no variation across conditions, no mediation and unsurprisingly, no impact on expenditure recall. So there’s that too.
Chapter 2: The Effect of Contactless on Personal Finance Management
Expenditure recall is one of the many things than could be affected by a payment method. Other things, as you should’ve gathered from the list in the context section, are the amount and frequency spent, overdraft fees, debt accrued, savings etc. etc. So that’s exactly what I measured in this chapter.
We had a dataset from a third party, a financial aggregator app based in the UK, in which we could identify who started using contactless, and when, because we had all of their transactions associated with the accounts they had signed into this app. From this, we looked at exclusively at those who used contactless payment methods, took one year of data before contactless was used, and one year after (per user), and exclusively looked at users who had a contactless and non-contactless account. That’s the sample (more edits obviously described in the complete thesis).
So what did we find? Well, when we take an account view, we see that the onset of contactless usage significantly increased monthly spending and the number of transactions, reduces overdraft usage (but not significantly so), has no effect on various forms of debt and much more counterintuitively: significantly increases cash usage and savings. This all happens on the account that now has contactless payments on it. When looking at the same user, but checking out their non-contactless account, we find absolutely nothing. Zip. Zilch. Nada. So where is all this additional money coming from?! Btw, like good semi-economists, we did control for income, so it’s not that, I promise.
After some further investigation (which describes most of my PhD) – we find that more money is also moving into the contactless enabled account. Just not from the non-contactless account. And here we hit the main limitation on the head: consumers sign accounts into this app, but if they don’t sign up all of their accounts, we cannot see them. So arguing that we have a full picture of a consumer is a bit of a stretch. Research is not without its limits.
So in conclusion: contactless does seem to lead to increased transactions and spending, but also to increased cash usage and savings, as well as increased transfers into the account. This seems to strongly indicate that the novel payment drives more activity towards the account, at the expense of other accounts (which we were unable to identify in the data).
Chapter 3: The Effect of Mobile Payments on Personal Finance Management
Did I mention I was “updating” the research on payment methods? Well, halfway into my PhD people became less obsessed with their payment cards and even more obsessed with their phones. ApplePay usage skyrocketed and various other “Pays” followed. Mobile was the new contactless, and I had to make sure to keep up!
So what did I do? I did exactly the same thing as I did before with contactless payments (my creative juices don’t flow constantly, sometimes you have to build on a previous idea). We again looked at the third party financial tracker data, identified those who used mobile payments, half a year of data before and after that point in time, one mobile active account, and one non active account and Bob is your auntie.
What did we find this time around? The exact same results as in the previous chapters… BUT! Not driven by internal transfers. This time round, we see the results being driven by the non-mobile account: this account now has less transactions and less spending on it. Partially explaining the increase we see on the mobile account. Interestingly enough, whereas all these effects persisted on the contactless user level, they don’t on the mobile user level. On the mobile user level we only see a significant increase in the number of transactions and savings, the latter now only being significant at the 1% level and having decreased in value (~28 pounds on a monthly basis).
It has to be mentioned: your output (results) is only ever so good as your input (data). I do have some serious questions with the excellence level of this data, but this is something that I’m sure many a researcher has grappled with.
Conclusions
First, I should probably mention that there’s another chapter in my thesis. There’s an entire fourth chapter dedicated to analyzing the underlying distribution of numerical representations and their effect on personal finance management and expenditure estimation. This chapter will not be mentioned here as it really doesn’t align that well with the rest of the work, and doesn’t necessarily involve payment methods. Again, if you desperately want to know more about it, just reach out.
Second, there’s a lot of payment theories out there, most notably the pain of paying (Zellermayer, 1996), transparency (Soman, 2001) and decoupling (Raghubir & Srivastava, 2002). If I had to explain those too in this blog this article would be even longer, so I didn’t. If you’d want to know more about those, read my literature review in the thesis. It’s a damn good review, if I do say so myself.
Third and foremost, the main conclusion to draw from this work is that we’re a far cry from figuring out what effect certain payment methods have on how we use them, and how our financial situation changes as a result of them. The newer the method, the less (academic) work there is, and most payment providers aren’t exactly eager to share their findings. We’re running into similar issues with researching online and BNPL payments. I also have a much wider overarching interest in the perception of money domain, and I do think these payments will have an effect on how we perceive money, and all the adverse consequences that can come with that.
But that is maybe best left for another day.
* I’m not posting a long reference list here, but if you’d want to have the reference list for this summary, again, I’m more than happy to send it to you!
