Federal Reserve Bank of Atlanta Talk About Payments Webinar

Consumer Payments Behavior in a Health Crisis: What Happened to Cash?

In the United States, COVID-19 cases and currency in circulation both surged in March 2020. Did consumer choice play a role in the increase in currency in circulation?

Data from the fall 2019 Survey and Diary of Consumer Payment Choice and follow-up surveys in spring and fall 2020 give some insights into consumer behavior in the pandemic. This Talk About Payments webinar looked at consumers' response to the health crisis:

  • Are consumers shopping and paying in person?
  • What's happening at the in-person point of sale?
  • Are consumers holding cash?

The Atlanta Fed's Kevin Foster, senior business survey specialist, and Shaun O'Brien, lead data and policy analyst in the Cash Product Office at the San Francisco Fed, joined Claire Greene of the Retail Payments Risk Forum to discuss consumer payments behavior in the pandemic. Heng Chen and Gradon Nicholls, special guests from the Bank of Canada, added insights.

Send questions about the webinar to David Lott at david.lott@atl.frb.org.

You may view previous Talk About Payments webinars.


Tim Lloyd: Hello, and welcome to today's Federal Reserve Bank of Atlanta Talk About Payments webinar. I'm Tim Lloyd, from the Federal Reserve. The views expressed in this presentation are those of the authors and do not necessarily represent the views of the Federal Reserve Bank of Atlanta, the Federal Reserve System, or the Bank of Canada.

Some data described in this presentation relies on data from surveys administered by the Understanding America Study, which is maintained by the Center for Economic and Social Research at the University of Southern California. The content of this presentation is solely the responsibility of the authors and does not necessarily represent the official views of USC or UAS.

It is now my pleasure to turn the call over to our first speaker, Claire Greene, to begin our program. Claire, the floor is yours.

Claire Greene: Thanks, Tim. Thanks everyone for joining the meeting today to talk about consumer payments behavior before and during the COVID pandemic. We'll be discussing three questions today: Are consumers shopping and paying in person? What's happening at the in-person point of sale? And are consumers holding cash?

I'll be talking with Kevin Foster from the Atlanta Fed and Shaun O'Brien of the Cash Product Office at the San Francisco Fed about data from Federal Reserve surveys—the Survey and Diary of Consumer Payment Choice and also the rapid response COVID surveys the Atlanta and San Francisco Feds conducted this spring and summer. Two special guests from the Currency Department at Bank of Canada, Heng Chen and Gradon Nicholls, will describe similar surveys of consumers in Canada: the Cash Alternative Survey and the Cash Pulse Survey. Thanks everyone  for joining.

I think you will see some similarities between the experiences of our two countries, and since Tim has handled the disclaimers, I won't be repeating them. As we speak, essential workers at the Bank of Canada and throughout the Federal Reserve System are working to ensure that currency, coin, and bank notes remain accessible to everyone. And all of us thank them for being on the job in person and following public health guidelines, making sure that we all have cash when we want it or need it.

Both Canada and the U.S. regularly collect data from consumers in our respective countries in order to understand payments, the demand for currency, consumer preferences, and aspects of household finance. The U.S. surveys, projects of the Atlanta Fed and the Cash Product Office at the San Francisco Fed, are in the field every October; you can see here October 2019 and October 2020 with the U.S. flags—so we just completed a survey at the end of last month.

Canada's Cash Alternative Survey was most recently conducted in April 2020, and you see here additional surveys of Canadian consumers in August 2019. Also, right now, Canada is conducting a consumer payments diary similar to the U.S. payments diary that we concluded in October.

Today, we are going to be focusing on results. Anyone interested in survey methodology, please get in touch and we will answer any questions, or we can respond during the Q&A. I think what this timeline shows you is that we have a very consistent history of data collection in our two countries, which has provided a very, very valuable benchmark in highly unusual economic times. As you know, lockdowns in the wake of the coronavirus pandemic began in our two countries in March, so I'm going to ask Gradon from Bank of Canada to tell us a bit about what was happening in Canada in the spring at the time of this economic crisis being precipitated by a health crisis.

Gradon Nicholls: Thanks, Claire. At the beginning of the pandemic, mobility restrictions were at their highest, leading to businesses shutting down their in-store operations. There was a drop in overall retail sales, and sales that remained shifted toward online, e-commerce modes—more than in the pre-COVID period. Based on this, we expect that there may be at least two impacts on cash usage at the point of sale. First, as consumers are spending less overall, they may be using less of all methods of payments, including cash. Second, the composition of payments—the share of payments made with cash versus other methods of payment—may also have changed as the point of sale moved to online modes.

As we moved into the spring and summer months, mobility restrictions due to COVID were lifted, but these trends that we saw in spending did not fully return to pre-COVID levels. Specifically, retail sales remained low and the share of retail payments completed online remained high, though not at the same levels as in April.

So as we moved into these uncertain times, starting in March and April, it became imperative to collect timely survey data so [that] we could conduct evidence-based policy at the Bank of Canada. So in collaboration with many colleagues, such as those from the Atlanta Fed, USC, [inaudible] Canada, and Ipsos, we came up with a survey program early in the year. Using our survey in August 2019 as a baseline, we then conducted a survey in April—in the beginning of the pandemic, when restrictions were highest—and then again in July, as these mobility restrictions were being lifted.

As you mentioned, we also have a survey in the field right now as the COVID numbers are again climbing, and another one planned next year in the spring. These forthcoming surveys will help us better understand the compositional aspect of cash use—that's what I was talking about before, with the share of transactions made with cash versus alternative methods. Our ongoing survey work has been imperative of policy work at the Bank of Canada, and it was only made possible with live collaboration with colleagues at other institutions.

And as we'll see a little bit from Claire's chart there, running frequent surveys concurrently with the Atlanta Fed offers us a great opportunity to gain insights into the payments landscape by looking at the similarities or differences between our two countries. So I'd just like to thank the Atlanta Fed for inviting us, myself and Heng, to speak today so we can compare and contrast our research. So I'll just say, thanks, Claire, and then I'll give it back to you.

Greene: Thanks, Gradon. We do really enjoy our collaboration with you at Bank of Canada and learn a lot from it. So, what Gradon described is, of course, I think for all of the people in the U.S. who are listening, sounds really very familiar in that we in the U.S. are also seeing similar economic effects of the health crisis. So, I would like to ask Kevin Foster from the Atlanta Fed to tell us a little bit about the rapid response surveys that we have implemented in the U.S. in response to the pandemic.

Kevin Foster: Thank you, Claire. Yes, I'll describe our survey and diary in general, first of all, and then describe how we set up these quick-response COVID surveys. So, we've been running the Survey of Consumer Payment Choice annually since October of 2008, and I myself have been working closely with the team at the Bank of Canada almost since day one. So, what Gradon said about our collaboration—it's all true.

The purpose of the survey is to measure how consumers rate payment instruments along a series of characteristics such as ease of use, cost of use, security; and the survey also measures the adoption of these payment instruments—so [by asking], do you have a credit card: yes or no? And it also measures the use of these payment instruments [by asking]—how many times in a typical month do you use a debit card, for instance.

We've been running the survey for a while, and then after a few pilot studies, we also began the Diary of Consumer Payment Choice, which we've been running annually since 2015. Both the survey and the diary are in the field every October. The Diary of Consumer Payment Choice is the only source of data for U.S. consumer cash activity, spending habits, and cash behavior.

In the past, when a hurricane struck during our diary data collection period, we were able to do a quick follow-up survey in those states that were directly impacted by the hurricane. And so when the pandemic started around March and April of 2020, we decided to use the same agility that we were able to use when we did these hurricane follow-up surveys, and we added some questions to the University of Southern California's Understanding Coronavirus in America tracking survey. Our questions appeared in the April wave and the August wave of this ongoing survey; you can visit the USC website for more information on that.

This is the same survey vendor that we use for our survey and diary, so we're able to link these responses across time because some of the same consumers that participated in our survey and diary in October of 2019 also participated in our COVID surveys in April and August of 2020; and  then again, they participated in our survey diary in October of 2020. We now have measurements for these same consumers across four different time frames.

In addition to the questions that we added to the survey, the USC coronavirus tracking survey has a wealth of other questions as well, and we can use those questions and the responses to those questions to analyze the impact of the pandemic on other aspects of the lives of our survey respondents. Some of these aspects include jobs and employment, labor market participation, financial security, and other COVID-related questions.

So that's how we stepped up our data collection during the pandemic. Back to you, Claire.

Greene: Kevin, thank you. And I'll just point out that the data is available for anyone to download on the Federal Reserve Bank of Atlanta website or at USC, and we can help you do that if you're interested. So now we're going to move away from what the data is to talking   about our three questions. And our first question is: Are consumers shopping and paying in person? When we talk about paying in person here, we include not only the purchases of goods but also purchases of services—like having a haircut or eating a meal in a restaurant—and even, perhaps, a walk-in bill pay.

And what you can see here—which I'm sure you experienced yourself—is the dramatic drop in the percentage shares of U.S. consumers who have made at least one in-person payment in the prior month, from almost everyone in October 2019 to just one-third of U.S. consumers in April 2020. This is an important finding because research finds that the location of the payment—that is our shorthand for "in-person" or "not in-person"—matters for payment instrument choice. This finding is the foundation for a lot of what we will be reporting about payment instrument choice and also the composition of payments, as Gradon mentioned today.

So you see the drop in April; and then as the economy was reopening,... you can see that by August, six in 10 U.S. consumers had shopped or paid at least once in person in the past 30 days. Now of course, as you all know, before COVID, online shopping and paying was quite prevalent. And even though, however, there is already lots of online shopping, our U.S. surveys did identify some switching that people were making from types of payments they might typically have made in person, and moving those payments to mobile or online pay.

And I'd like to ask Shaun O'Brien to tell us a little bit about what we learned about switching in the U.S. surveys in April.

Shaun O'Brien: Sure. And Claire also, thank you for the invitation and the opportunity to come and speak here today. With respect to how people were potentially changing their shopping and their payment habits—as was discussed in some of the previous slides, really at the height of a lot of the lockdown measures, we saw a lot of people reduce their in-person payments. What we were curious about was whether or not individuals were shifting a lot of their previously in-person payments and shopping patterns to simply shifting them online, or were they just mostly reducing their overall consumption.

What we found in April when we asked if people were switching from in-person payments to online or mobile phone payments, we found that about 20 percent indicated that they were actually making that switch. The remaining number of individuals either weren't, or weren't really sure if their behavior patterns had changed. But what was interesting when we asked individuals if they had switched was we had a follow-up question asking them, if you did switch, what types of purchases are you shifting to—more of an online platform and less in-person?

Really, the two large responses we got were for restaurants and fast food and big-box stores; you really saw this shift from—especially in restaurants and fast food—where cash is, or was at least before the pandemic, used quite a bit, to where nearly 60 percent of those who said they switched had said that they had at least made some shift toward online paying, or purchasing online, in order to make those previously in-person payments.

So as the pandemic continues, it will be interesting to see how much these behaviors become sticky or permanent, or if they start to regress back to those prepandemic behaviors. So Claire, I will turn it back to you.

Greene: Thanks, Shaun. I believe in our October 2019 data, we found that fully 35 percent of payments to restaurants and fast food were made in cash. So, it is interesting that where we see this switching is among the cash-heavy merchants. And then, Shaun just used the word sticky, which frequently when we're talking about consumer behavior, we say habits are sticky. And ways of paying, and also ways of purchasing, are habits. And so once someone learns a new way—for example, to order that coffee with the mobile phone—that potentially has implications for the future, but clearly we can't say that yet.

So that's it for question number one, what happened to in-person paying and shopping? As Gradon pointed out, consumption overall dropped and in-person purchases also dropped, and any change in the choice of payments or the mix of payments that consumers are using is in the context of those two big events.

So now we're going to move on to question number two, which is: What's happening at the in-person point of sale? As you can see, this slide talks about both consumer and retailer behavior in April and at the time of the summertime surveys—so that's July for Canada and August for the U.S. And it is important to talk about both consumers and retailers, because payments are a two-sided market where each side can make a choice.

So first we'll talk about the consumers and the consumer choice to use cash. So, I'd like to ask Kevin to describe a little bit about what's happening with the use of cash for in-person payments.

Foster: Thank you, Claire. So first of all, during the pandemic we're seeing fewer overall in-person payments. We saw a 60 percentage-point drop in the share of consumers paying in person at least once when we measured from October of 2019—the Diary of Consumer Payment Choice—until the April 2020 coronavirus special survey.

But if the consumer did make an in-person payment, they basically continued to use cash at about the same rate as before the pandemic. Both in October of 2019 in the diary and April of 2020 in our first coronavirus survey, around six in 10 consumers who paid in person at least once used cash at least once. Therefore, we can conclude that the differences in the U.S. numbers have to do with the drop in in-person shopping more than any changed preference that might be revealed by using less cash while paying in person.

Back to you, Claire.

Greene: Thanks. And I think you noticed, probably, that Kevin used the phrase "at least once" a couple of times, and that's because in these rapid response surveys, we're able to ask people brief questions about what they did over a time period. And this is in contrast to the diary surveys that we have both in the U.S. and Canada, where participants write down every payment they make. And so we have a sense of, for example, cash as a percentage share of all the payments they make—which we can't do with this quick survey data but we can do with the data that was collected in October and that Bank of Canada is collecting now in November.

Now, Bank of Canada also asked questions about cash use, and they approach this question a little differently. Our question is in the context of whether or not you paid in person; Bank of Canada asked directly: Have you paid in cash in the past week? And you can see that for Canada, the probability of having used cash at least once in the past week did increase from April to the summertime, from 36 percent to 54 percent. If I had to guess, I would say it was related to the economy opening up and more people being out and about.

Now we're going to move on to the other side of the transaction: the retailers' choice to accept or refuse cash—or maybe more mildly, to discourage its use. As I'm sure you remember, in and around March 2020, various retailers announced that in fear of transmission of COVID-19, they were discouraging the use of cash at the in-person point of sale. So, I'd like to ask Gradon to tell us a little bit about what consumers in Canada were experiencing in terms of merchant acceptance at this time.

Nicholls: Thanks, Claire.  Just to provide a little bit more context on the consumer numbers, I would confirm what you're saying: that we do see an increase in the usage of cash in the past week but also of other methods of payment as well. So, I think we would draw the same conclusion as the U.S., that we really see an increase in purchases at the point of sale overall which is driving increasing usage of all methods of payment.

We did also ask a question, from the consumer perspective, on merchant acceptance. What we find is that in the summer of 2020, about 9 percent of consumers stated they were not able to use cash at a store or business. And that sounds a bit infrequent—less than 10 percent of consumers saying this. But actually, if we compare it to stats that we have on merchant acceptance in the past—for example, we had a survey in 2018 where we found that merchant acceptance of cash was about 96 percent, so only 4 percent of businesses saying that they did not accept cash.

So in this context, the fact that 9 percent are saying they weren't able to use cash at the point of sale is quite high. We also asked a weaker version of this question: Did you see a sign at a store or business saying that they did not accept cash? And so, what we do find is that about 16 percent of consumers said they had seen a sign.

And I would just end by saying that these numbers are a slight decrease compared to April. So for example, 12 percent of people said they weren't able to use cash in April, which declined to around 9 percent in summer.

Thanks, Claire.

Greene: Thanks, Gradon. And I'd like to ask Shaun to respond to these comments by Gradon, because the U.S. numbers appear to be small but potentially may be important. What do you think about that?

O'Brien: One thing to point out in this particular slide [is] you'll notice that in April, the wording is how many consumers experienced a merchant who had refused to accept cash? And then if you look in the summer for our August survey, you'll see that the wording gets switched slightly to "card preference." So really, to answer Claire's question about 7 percent and how might that relatively small number be impactful: well, the contrast and really the context for this is thinking back to 2019, when there was a lot of discussion about different retailers choosing to not accept cash and local governments potentially deciding to enact—or actually enacting—laws requiring retailers to accept cash so as to not exclude individuals who didn't have bank accounts.

So when we think about retailers thinking about opening up without cash, and the response to be, no, you're not allowed to do that—and then, in the earliest part of the pandemic, 7 percent of individuals report seeing at least one merchant say that they're not going to accept cash, is really quite a drastic change from what we saw before the pandemic. The reason that we had switched the wording and why we didn't ask individuals specifically if merchants had refused cash in the summer was obviously we had the pandemic, and we were still interested in how consumers perceived merchants' acceptance of different payment instruments. But the other thing that was going on was in the U.S., there was a coin circulation problem where, as individuals decided to stay home longer and make any purchases that they might have made in person online or just reduced their shopping, they stopped bringing coins in to depository institutions, to banks, to grocery stores that had kiosks and deposit their coins.

So we found that the available supply of coins was not enough to overcome that circulation issue, and so we were curious also if merchants would start to ask individuals to not use cash because maybe the lack of coins might influence them to say, "Maybe we can steer individuals away from cash and toward cards." So, we just asked consumers: in the last 30 days, how often have you seen a merchant request that you use a card? And what we found was that about a little more than 40 percent of individuals reported seeing merchants ask for a card to be used, at least some of the time—which, again, is in the context of there's a coin circulation issue. And again, before the pandemic, merchants asking for cards was not nearly as common as 42 percent of individuals seeing such signs at least some of the time.

So Claire, I'll turn it back to you.

Greene: Thanks. That's a very interesting point about the coins. So that's cash use and acceptance at the point of sale, and we're still at the in-person point of sale, where we consumers have a choice to tap and pay (or not), and retailers have a choice to enable terminals to accept NFC transactions (or not). Both the U.S. and Canada surveys make it possible to distinguish between a contactless payment made by tapping a card and a contactless payment made with a mobile phone, from the consumer's point of view.

So, what you can see here on the right-hand side of this slide is that in the U.S., a very, very small share of in-person card payments were contactless in one way or another in October 2019. The data for Canada, on the left hand of this slide, shows differences in consumer behavior between April and this summer, just as we saw in the prior slide on cash use and acceptance. And in Canada... in the past week, more consumers were likely to have tapped and paid in the summer, and also more consumers were likely to have made a mobile phone payment in the summer.

And you can see that these numbers are quite different, April and the summer, compared to what we saw in 2019 in the U.S. So, I'd like to ask Heng to tell us a little bit about why tap and pay is so much more prevalent in Canada.

Heng Chen: Thank you, Claire. The reason why tap and go is prevalent is Canada can be understood from the timing of the introduction of the technology and also the truth in the market story. So, we'll first look at the financial institution side. Canada's card was first introduced in Canada in 2006 as MasterCard PayPass, and in 2007 as Visa payWave.

From the consumer side, the diffusion of credit cards to the card holders actually happened gradually in conjunction with replacement of current magnetic stripe cards with chip credit cards, over the 2007-2015 period. And also in the end, for the merchant side, concurrent business development was the gradual development of contactless payment terminals at merchants, driving the success of contactless payments. Actually in fact, the contactless terminals are now the default option—many requiring a new device, and then merchants have to disable this function if they don't want to accept contactless payments.

So in the end, back to the contactless debit card: it was first introduced by a few players in 2011, but some big important financial institutions just adopted it in 2018—so that actually, to take away, is the timing and also the truth in the market from consumers, financial institutions, and the merchant.

So I will turn it back to Claire.

Greene: Thanks, Heng. So that's very interesting. In Canada, people have just had way more time to get used to tapping and paying. Perhaps you know that, as reported in the Federal Reserve Payments Study, just 2 percent of in-person, general purpose card payments were made with a chip card back in 2015. So Heng just told us that consumer cards had graduated from between 2007 and 2015 to being all chip cards—so just a much longer adoption period.

And also following up on this idea, different ways of doing things are habits, and we all need a little bit of time to learn to adopt a new habit. And there's been more time for that on all sides of the market: the financial institutions, the merchants, and the consumers.

So, I'll just sum up our question number two, which is: "What is happening at the in-person point of sale? So in the short term, in-person shopping is certainly not back to normal. In the short term, the pandemic appears to have accelerated the shift to online shopping, and we still do not know if this is a long-term development (we do not have data that shows that).  And in the short term, the pandemic appears to have given some retailers new license, let's call it, to steer consumers toward cards.

So what we're seeing, in other words, is that both consumers and retailers are learning new ways to do things that potentially could be long lasting. So that's it for question number two.

Question number three is: Are consumers holding more cash? As you probably have heard, for both of our countries currency in circulation has skyrocketed during the pandemic, and here are the elements of currency in circulation. First, currency held outside the country; for the U.S., that's more than half by value. For Canada, there has been an increase in demand for Canadian bank notes outside Canada since 2015, but currency held elsewhere for each of our respective countries is outside the scope of this presentation. There are two papers in your references list that address this topic.

For currency held inside our respective countries, we're interested today in cash held by consumers—and that includes, conceptually, two kinds of consumer cash holdings: what we consumers have for spending, let's say, in our pocket, purse or wallet—that's how we ask it in our survey question, and what we might have put in a safe place for emergencies or to save for some other purpose in particular. When we ask about that in the U.S surveys, we call it cash stored elsewhere, and we mention as examples in your home, your office, or your car. And then, in addition to holdings by consumers, we'll also talk about depository institution vault cash, which has a consumer aspect because that's what depository institutions keep on hand so that when you and I get up to the teller window or the ATM, the cash we want is in inventory.

You might expect that with people staying home and the images we saw in the news of salesclerks spraying cash with cleaning products, that the demand for currency would go down. And in fact, that is exactly the opposite. And here are two years of data to show you how unusual 2020 is. So this chart shows for each year—2020 and 2019, with a starting point of January 1—the percentage change over the year from January 1 in the amount of currency in circulation.

And what you see for the U.S. over the first two and a half months or so of 2020 is that the change in currency in circulation from January 1 was moving pretty much in lockstep with the change in currency in circulation from January 1, 2019—that's the blue dotted line. And looking at Canada—the red line–you see a similar behavior pattern, where the change in currency in circulation starting January 1, 2019, through mid-February 2019 is pretty much the same as the change that we see from January 1, 2020, through mid-February 2020.

And then, of course, at the start of the pandemic in March, currency in circulation in the U.S. increased quite dramatically and also in Canada. Just comparing that, in the lower left-hand side of this chart, to the 2019 data for both of our countries, you can see that by mid-year 2019, U.S. currency in circulation had increased about 2 percent from January 1. And then if you look at U.S. currency in circulation in 2020 mid-year, you see the percentage change of about 10 percent.  So, that's the big picture overall—quite a large change for both of our countries.

Now we're just going to take a step back and look at the start of the lockdowns due to the pandemic, and ask Shaun to tell us a little bit about what happened to U.S. currency in circulation in mid-to-late March.

O'Brien: Sure; thanks, Claire. So as the pandemic came to the U.S.—and what really started was after a national emergency was declared by the president on March 13th, what we saw was the payments out of the Federal Reserve System out to depository institutions really started to increase quite drastically. And at first you can see that line really start to increase fairly substantially. And what was happening at that point is really as a result of that declaration, and really the uncertainty that went on with the pandemic in general, we found that depository institutions were stacking up and loading up their vaults with inventory in preparation for the response that they imagined consumers would have to this type of situation.

So you really do see that runup, and most of that runup has again to do with the depository institutions increasing their vault cash. But if you look just beyond March, it's not as if the institutions held that cash and then the rate of change of currency in circulation was relatively consistent. We still saw a drastic increase of currency in circulation as things continued throughout the summer. And so that early March period of inventory buildup is really just a precursor to the demand that consumers had with respect to currency in circulation—consumers really demanded currency in that domestic market.

We've seen from a lot of the public reporting data that those depository institutions continue to hold a really high level of cash as they're really planning for any type of contingency where individuals may once again turn back to additional cash to supplement what they've already taken out. And so that's why, really, you're seeing this drastic increase in the U.S. market for currency.


Greene: Thanks, Shaun. And as Shaun did just mention, this is public data, and there are links to information about bank vault cash in your references. So, I'd like to ask Heng if you were seeing this similar anticipatory behavior in Canada.

Chen: Thank you, Claire. The Bank of Canada supplies financial institutions with the banknotes they need to meet the public demand through the Bank Note Distribution System, which is the BNDS. The financial institutions can withdraw notes from the Bank Note Distribution System—BNDS—to meet the demand for cash, or they can deposit surplus notes back to the Bank of Canada. So in this state, the note circulation increased significantly in the early months of the pandemic, and it led to the growth of note circulation—though, materially, only in July.

To understand the situation, we will decompose "note distribution" into withdrawals and deposits,...  BNDS system.  Since the change in the note circulation equals the value of notes withdrawn from the Bank of Canada, which is the withdrawal of bank notes less deposits made to the bank, we will actually look at both withdrawals and deposits respectively.

From early in the period, the cash withdrawals from the Bank of Canada and their peers did increase. Withdrawals are actually now normalized by the middle of summer, but deposits of cash returned to the Bank of Canada have tended to remain abnormally low, likely due to the reduction of note-protecting capabilities due to the pandemic,   and also potentially consumers and merchants do not return the notes and are actually holding the notes for the recovery reason. So this will actually have us continue to monitor the BNDS system in the future.

And that actually concludes my part. Thank you, Claire.

Greene: Thank you. That's very interesting to hear that large denominations were a big part of the withdrawals, and also that, while withdrawals spiked and then returned to a more normal level, that the deposits are still quite low. So this idea that Shaun also mentioned of precautionary holdings, just in case—let's face it, we're all a little nervous and on edge these days, so that extends through our interests in hanging on to currency.

Now I'm going to talk a little bit about, to the right of this yellow box here, and we're going to talk specifically about consumer holdings in the long run—well, not the long run, in the six months run, we'll say. So here's what happens with consumers; this is for the U.S., and these are distributions of total cash holdings in October and in the spring. And that is the percentages of U.S. consumers holding these different amounts of cash.

So first, let's look at people who have $25 or less, and that does include people who report having zero cash. You can see that fell from 43 percent in October to 34 percent of consumers in the spring—so just about one-third reporting that they had $25 or less. And then on the other side, people holding $100 or more; in October, about one-third of U.S. consumers reported having $100 or more, and in the spring, it was almost half of consumers who said they had $100 or more. So generally, in the early days of the pandemic, more U.S. consumers are holding cash. And of the consumers who are holding cash, they are holding more.

And this slide that I'm reporting here is the sum of the two types of cash I mentioned, on person and held elsewhere. I'd like to ask Gradon how these cash holdings in the U.S. compare to what you're seeing in Canada.

Nicholls: Thanks, Claire. Yes, I just wanted to talk a bit first about how we measure and report cash holdings, because cash holdings is a variable with a lot of dispersion, and by that I mean there are a lot of people at either end of a distribution. There's a mass of people who have no cash on hand, and there are also a few individuals who have a very large amount of cash on hand, or at least they report saying they have a very large amount of cash on hand (like $300, or into the thousands or more).

And while these outliers are a small group, they can have a large impact on the average. It's difficult to distinguish whether these are true outliers or not, so to be robust we use a robust statistic, which is the median. We usually focus on the median when we're looking at trends and cash holdings over time. And so what we find is that median cash holdings rose from $70 in August 2019 to $85 early on in the pandemic in April, so about a $15 increase at the median.  Then we saw it fall back again to $70 in summer of 2020.

So this trend is very consistent with the U.S. findings. It also corresponds with Heng's analysis in the previous slide, where we see within the BNDS, we see withdrawals from Bank of Canada's cash that are stabilized as we move into the summer. I think one key difference we see from the U.S. findings is that while we see on the intents of margin more people holding cash, on the extents of margin, we see fewer cash holders early on in the pandemic. Specifically, we see that pre-COVID, about 80 percent held at least some cash on hand; this dropped to about 72 percent in April 2020, and then returned to 80 percent in the summer.

So if we take this all together, we see that early on in the pandemic, there were few cash holders compared to prepandemic, but those who did hold cash tended to hold more at the median. But, as I said, we see this trend return to pre-COVID levels as we moved into this summer. Thanks.

Greene: Thanks; that's interesting. So that's different from what we saw in the U.S., because I  believe that in October 2019, we found that 78 percent of U.S. consumers held some—$1 or more—and then at the start of the pandemic, that went up to 85 percent. It's also important, I think, to just reiterate that point you made about the median. We certainly do find in our survey data that some people really do want to hold quite a lot of cash. And when we report median data—and also when computing—any kind of averages usually get a haircut around 99 percent or 98 and a half, something like that.

So now, we're just going to talk a little bit about motivation, and why consumers might have more cash. Well, the first reason is lost opportunity to spend, because perhaps like me, you were stuck at home. Cash that came in early March in the expectation of dining out later in the month, for example, could still have been sitting in consumers' wallets or at home during the April/May survey period. That buildup of cash would be due really to consumers' changing mix of activities—so just going back to the point Gradon made at the beginning about the drop in consumption overall. And then the drop in in-person consumption—this changing mix, more than it would necessarily have to do with their desire to have more cash.

And then the second reason people might have more cash is simply that they have more money, whether it happens to be in cash or in a bank account or stored on a prepaid card, for example.  Prior research using this data does show that people with a higher income or more wealth do tend to have more cash. Kevin has done some research on cash holdings related to economic conditions of individuals, so I'd like to ask him to comment on that a bit.

Foster: Thank you, Claire. So we built some models that estimated different aspects of holding cash and how much cash you might hold, based on some of the extra questions that I mentioned that we were able to use in the UAS coronavirus tracking survey by USC—they have their own questions, we just piggybacked on their survey. So we're able to use their questions in our analysis of cash behavior.

And so, the first three models all go together. To me, at least, they lead to one conclusion that I'll discuss. But the first model estimated that for those consumers who received unemployment benefits in the past 14 days, they were more likely to have cash stored at home. And if they had cash in their wallets, they had more cash in their wallet conditional on having cash in the wallet. The second model said that if the consumer had no job at all, then they also had more cash in their wallet, conditional on having cash in the wallet.

And finally, if you've gotten a job in the past 14 days—meaning at some point in the past 14 days you did not have a job—you're also more likely to have cash stored in the home. So in all three of these cases, a commonality of these results is that if you're receiving unemployment benefits, you don't have a job; if you tell us you don't have a job, you don't have a job; and if you got a job in the past 14 days, you didn't have a job at some point. And in all those cases, these people either had more cash, or they were more likely to have any cash at all.

So these results might be explained by the $600 CARES Act benefit that we saw Congress had passed earlier in the year—I think it expired at the end of July. It could be seen as evidence that this benefit actually did what it was intended to do, which is to put extra cash into the pockets of the people who needed it most during the early part of the pandemic.

And another interesting result, which might seem a little obvious, but it's also a little sad and disappointing, is that if you missed a mortgage payment or rent payment in the past 30 days, you're less likely to have cash in your wallet or cash stored at home. And if you do have any cash, you're likely to have less of that cash. So if you don't have any cash in the house, you're going to skip your mortgage or rent payment.

So those are some of the results that we were able to find using the other parts of the coronavirus tracking survey. Back to you, Claire.

Greene: Thanks, Kevin. And then, of course, people also get cash for emergency preparedness.  Kevin and I heard a presentation in September where a representative of one of the credit bureaus referred to the striking similarity between the impact of Hurricane Harvey in Texas in 2017 and the national economic impact of COVID. Now, this person was talking about credit capacity and delinquencies, not necessarily the context of holding cash. But we have some interesting findings connecting emergency preparedness for hurricanes and perhaps some of the emergency preparedness behaviors at the start of the pandemic.

So I'll ask Kevin to tell us a little about that.

Foster: Thanks, Claire. One thing that the hurricanes have in common with COVID as natural disasters or emergencies compared to something like an earthquake [is] we don't know when an earthquake is coming, so you can't really prepare for that in the same way. But as you see the news media reports of hurricanes approaching or that COVID is getting worse and worse, the consumer can go out and prepare themselves for that. So we can ask these consumers about their cash holdings both before and after the natural disaster, and in both cases—the hurricane surveys and the COVID surveys—we saw that some consumers increased their cash holdings by a significant amount.

So in our case of the coronavirus here, only about 12 percent said they increased, but they more than doubled the amount of cash that they had. So when people had a need to get prepared, they really go out there and get extra cash and prepare themselves. Back to you, Claire.

Greene: And it's interesting that when we looked at the hurricane preparedness, it was pretty similar shares. It was somewhere in the neighborhood of 12 to 18 percent of people who were the sort of "preparedness" types getting ready for the hurricane.

Foster: Exactly, the same percentages, yes.

Greene: Now, I wanted to show this slide because it's a little bit different from most of the questions we ask, in that we aren't asking people about things that they specifically do; and we also are asking—for example, in the case of the question from Bank of Canada, what they might think about doing in the future. And these are different from our regular questions because we try as much as possible to learn about people's behavior from what they actually do, what economists call revealed preferences.  Just for example, if you were to use a survey question on me, you might ask me at 11:30, "Are you going to eat two cookies for lunch?" And I would say, "Oh, no, I'm going to have a salad."

And then along about three o'clock in the afternoon, you could come back and ask me, "Did you eat two cookies for lunch?" And perhaps the answer would be different, so that is the value of asking these questions about what people actually do. In our surveys, it is possible to relate these more sentiment-style questions to what consumers have done, and I want to be able to move on to a little time for questions so I'm not going to describe these in any more detail—except to say, as I think I have already, that you can look at the data yourself.

So a final warning, from all of our presenters, is to be very careful about extrapolating changes in the recent periods and assuming that they are going to be having a permanent effect; these very possibly could be temporary adjustments. Earlier I was talking to Kevin about what he would want to know next year when we conduct our survey again, and his response was, "Wait one year, and then I will give you a better answer."

So, as Yogi famously said, "it ain't over until it's over."

The goal of this research is to bring the voice of the consumer into policy decisions and product innovation as the payment system is changing, and you can access reports and data. There's a list in your handouts and also a list of other Federal Reserve surveys that give insights into consumers. I would particularly recommend the Survey of Household Economics and Decisionmaking for its COVID updates. I work in the Retail Payments Risk Forum, where we blog every Monday at Take On Payments.

Thank you for your attention, and I will turn it back to Tim. We'd be happy to answer any questions.

Lloyd: Thank you, Claire. I think we've got time for maybe one, maybe two questions, and I'll just start. This first question is for Shaun: What happened to the U.S. currency in circulation since July?

O'Brien: So what we've seen since July is that the value of currency in circulation has increased past $2 trillion. And so the trend that we saw in that chart earlier really has continued, where the overall demand for cash continues to increase. I know earlier I had mentioned vault cash holdings and how that correlates to currency in circulation and really, what we've seen is depository institutions are not responsible for that. They've held a fairly consistent level of inventory after that initial buildup. So really, that demand, that continued increase of currency in circulation past the $2 trillion mark is really being driven by consumers and their want for holding cash as a store of value in these really uncertain times.

Lloyd: Okay, I think we've got time for one last question. This one is for either Gradon or Heng: Has Canada experienced a similar coin distribution interruption problem?

Nicholls: As far as we know, we haven't heard any reports from the Royal Canadian Mint regarding any such disruption—the Royal Canadian Mint being the ones who mint Canadian coins. So as far as we know, the answer is no.

Lloyd: OK. Well, I think we're right at time, and I just want to make sure to thank Claire and all of our presenters today for sharing their expertise with all of us. A survey should now be available to you (it should show up in your player page), and we ask that you just take a little time to fill that out. You only need to fill it out once, but please take a few minutes to do so.

And thank you for joining us today. This concludes today's Federal Reserve Bank of Atlanta Talk About Payments webinar. Thank you for joining us, and have a great rest of your day.