Will tapping finally kill cash?

I was at the local big-box hardware store the other day, making that very 21st-century purchase, compost bags. I tendered my $20 bill and as I received my change and receipt I briefly looked back at the two or three people waiting patiently in line behind me and had an economics epiphany. (If you’re an economist, these happen to you all the time. Not to worry, they’re not health-threatening—unless you spend too much time describing them to your spouse.)

I wondered how much time (on average) is taken up waiting for people in front of you in the checkout line to put their change in their pocket or purse, stow their receipt in their favoured place (in my case my wallet) and pick up their purchases and clear out.

A few seconds, at least.

I wondered how many hours of my now long life, maybe even days, have been spent that way, either waiting on others or causing the delay myself. It occurred to me there must be people who study the economics of checkout lines. It occurred to me mainly because these days there are people who study more or less everything.

Well lo and behold, just a couple of days later, what shows up in my email but a new study from the Bank of Canada called “How Long Does It Take You to Pay: A Duration Study of Canadian Retail Transaction Payment Times,” by Geneviève Vallée, an MA intern in the currency department at the Bank. Judging by the sophistication of the econometrics Vallée uses in exploiting her data set, technological change is occurring as rapidly in MA studies as in payments systems.

That said, her bottom line is that the oldest means of payment, cash, is still the most efficient, at least in terms of how long transactions take. Her econometric concern, as best I understand it, is whether the fact that consumers “endogenously” choose which method of payment (MOP) to use may bias our estimates of how fast each method is, a statistical problem that’s actually quite difficult. (In effect, do they choose cash only for those transactions where it’s faster so the fact that it is faster should therefore not be surprising?) In the end she finds that the estimates aren’t biased, which is reassuring, but I’m afraid I was more taken by some of the simple facts that arose along the way.

To begin with, the fact that there actually are surveys of how people make transactions. In this case, the survey being used, the “2014 Transaction Duration Study,” involved 12 observers assigned to different kinds of retail outlets to watch and time—using a stopwatch application on a smartphone—the interval between the customer being informed of the amount owing and their receiving their receipt and any change back. (So in fact the fumbling time that got me thinking about this problem didn’t actually get counted. It would have been categorized as post-payment.) The observers worked six-hour shifts—I hope they got breaks, it seems a mind-numbing job!—and generated data on 5,107 transactions.   

As for payment times, they varied between one and 180 seconds. A one-second transaction seems almost too good to be true. Change presumably wasn’t involved. Three minutes, on the other hand, seems very long, though I expect we’ve all experienced credit-card delays from time to time, some of them as a result of our own incorrect entry of our PIN. Note that if there had to be a price check or another checkout person came by for a chat or there was any interruption not associated with the making of the payment itself, the observers stopped the clock. That kind of queuing delay, though it happens all the time, doesn’t enter into this study.  

Perhaps strangely, the old technologies are the fastest technologies. As noted, cash is fastest of all, with a mean transaction completion time of just 15 seconds and a median even better than that at 11 seconds. (The mean is higher than the median, I assume, because of those fastidious outliers who insist on foraging through their change purses to provide the exact value of the transaction.) Cash does get slower the higher the value of the transaction. Scatter plots of the data that Vallée provides show a couple of cash transactions of more than $400. Counting and double-counting all that cash, unless it was tendered as $100 bills, takes time. Of course, hundreds usually have to be checked for counterfeit.

The next-fastest MOP is the old-fashioned “swipe and sign” credit card, which Canadians tend to find quaint when they run across it in the US, where it’s only just now being phased out. Signing is a bother but at least there doesn’t have to be confirmation of a PIN. The mean duration of transaction with this technology is just 19 seconds and the median is 13 seconds. (Would the longer mean be caused by people with intricate signatures or long names? Or maybe they drop the pen.) Compared to cash, those numbers are only four seconds and two seconds slower, respectively.

For their part, “chip and PIN” cards, though very modern, are a glacial 24 (mean) and 22 (median) seconds. Time of transaction isn’t everything, of course, and from the retailer’s point of view, though people take a little longer to pay with this kind of card their payment presumably is less likely to involve fraud, which certainly reduces the card company’s costs and may also benefit the retailer.

Unfortunately, the survey on which the study was based was for 2014 when contactless “tapping” cards were just coming into use. Only 197 of the 5,000-plus transactions used them and some of those may have been first use. I know my first couple of times tapping I didn’t quite know what to do so the transaction took longer than necessary. In those 197 transactions, however, paying by tapping your credit card took 16 seconds on average, with a median of 14 seconds—times that are just one second and three seconds, respectively, slower than cash.

My guess is that though swipe cards and chip cards didn’t kill cash, tapping may be the doomsday technology, not to mention paying with your cellphone. The currency department of the Bank presumably will update its surveys to gauge the adoption of these new technologies. As for me, I’ll continue to ruminate on this problem during future fumbling intervals.