The Venmo Effect: Digital Payments and Consumer Willingness (Part 1)

The Venmo Effect: The Impact of Digital Payment Platforms on Consumer Willingness to Pay 2021 Submitted to the Department of Economics in partial fulfillment of the requirements for the degree of Bachelor of Arts with Honors Abstract Using data from an online purchasing experiment, we show that Venmo, a digital payment platform with over 40 million users, increases consumer willingness to pay (WTP). While the impact of credit cards on WTP has been documented, the role of Venmo has not previously been studied. We conduct an online RCT with undergraduate students in which participants are randomly assigned to a payment form of either Venmo, credit, or debit and then partake in ten Becker, DeGroot, and Marschak (1964) lotteries to estimate their WTP for ten low-cost consumer goods. Our main finding is that WTP with Venmo had the highest mean WTP ($1.12; s.e. ($0.037)), followed by debit ($1.07; s.e. ($0.058)), and then credit ($0.96; s.e. ($0.074)). Within Venmo, WTP is highest when using the Public setting of the Venmo app and lowest when using the Friends Only setting. Using the app’s social feeds to view other users’ transactions on the app does not increase WTP uniformly; rather, priming is found to amplify the existing effect of the user’s privacy setting on WTP. JEL classification: D12, D91, M310, Z13 Keywords: Behavioral, Mental Accounting, Preferences, Social Inclusion, Social Media, Willingness to Pay, Venmo Table of Contents 1 Introduction …...……………………………………………………….….... 1 2 Description of Venmo ………………………………….……………...….... 5 3 Literature Review …………………………………………...……………… 7 3.1 Mental Accounting Overview ……………………………..………….. 7 3.2 Willingness to Pay with Credit Cards versus Cash …………..………….. 9 3.3 The Rise of Digital Payment Platforms and Venmo ………….…………. 10 3.4 Social Influences on Spending ………………………….....………… 10 3.5 Limitations of Existing Research ……………………………………. 12 4 Theory …………………………………………………...………………... 12 4.1 Mental Accounting …………………………..………..…………… 12 4.2 Priming ………………………………………………...………… 15 4.3 Signaling ……………………………..……………...…………… 16 5 Experimental Design ……………………………………………………… 17 5.1 Participants ………………………………………………….…… 18 5.2 Treatments …………………………………………...…………… 19 5.3 Experiment …………………………………..…………………… 20 5.4 Pilot ……………………………………………...……………… 22 5.5 Empirical Strategy ………………………………………………… 24 6 Results and Discussion …………………………………………………… 26 6.1 Summary Statistics ………………………………………………… 26 6.2 Main Results ……………………………………………………… 29 6.3 Results by Item Type ……………………………………..………… 33 6.4 Discussion ……………………………………………………...… 35 7 Conclusion ………………………………………………………...……… 37 8 Appendix ………………………………………..………………………… 40 8.1 Tables and Figures ………………………………………………… 40 8.2 Survey Interface …………………………………………………… 41 8.3 References …………………...…………………………………… 48 1. Introduction Just as credit cards transformed consumer spending in the 1980s, modern digital payment platforms like Venmo are revolutionizing commerce. Seminal behavioral economics research finds that credit cards elicit a higher willingness to pay (WTP) than cash, a phenomenon called the credit card premium (Hirschman 1979; Feinberg 1986; Prelec and Simester 2001). Unlike credit cards, though, the impact of digital payment platforms on consumer spending has been unknown until now. This paper is the first to study the impact of Venmo – or any specific digital payment platform – on consumer willingness to pay. Understanding the impact of Venmo on willingness to pay is necessary given the increasing prevalence of Venmo and digital payments in general. Since their emergence in the early 2010s, digital payment platforms have skyrocketed in popularity and are now the preferred mode of transferring money to peers, especially for young people (Zhang, Tang, Zhao, Wang, Zheng, and Zhao 2017). Venmo is convenient; it can be used to pay for anything from a coffee to rent, payments are processed immediately, and funds can be transferred directly to a bank account. The convenience and popularity of Venmo have only been exacerbated by the current COVID-19 pandemic as consumers and businesses are now more likely to prefer contactless payments over methods that could transmit the virus, like cash or physical cards. Despite its popularity, the effect of Venmo on consumer spending behavior has not been studied. Anecdotally, many college students say that money spent on Venmo feels “fake.” Is Venmo causing its users to spend more money? This paper investigates the existence of the Venmo Effect, our novel theory that Venmo increases the WTP of its users. Additionally, we propose three mechanisms that may contribute to the Venmo Effect: 1) 1 mental accounting, 2) priming, and 3) signaling. First, given Venmo’s digital nature, users may not view their Venmo balance as fully fungible, making spending on Venmo less painful or guilt-inducing. Second, the social media feeds integrated in the Venmo app may prime spending generally or for specific purchases, making users more likely to spend. Third, Venmo may be used as a signal for wealth or social status, leading to higher WTP on Venmo for transactions that would send a positive signal to a user’s peers. We investigate which, if any, of these three mechanisms drive the Venmo Effect. To test the existence of a Venmo Effect and its potential mechanisms, we run an online experiment with college students that assesses the difference in average willingness to pay for purchases made using Venmo, debit cards, and credit cards. Participants are randomly assigned to one of five payment treatments: debit, credit, Venmo on the Public privacy setting, Venmo on the Friends Only privacy setting, and Venmo on the Private privacy setting. Half of the participants in the Venmo groups are also randomly assigned to a priming treatment that entails spending two minutes browsing the social feeds on the Venmo app, resulting in eight distinct treatment groups. All participants undergo ten incentivecompatible Becker, DeGroot, and Marschak (1964) lotteries for ten different low-cost consumer goods, which allows us to accurately measure participants’ willingness to pay for each of the ten items. A difference in means analysis of our experimental data provides two important insights about the role of Venmo: 1) a Venmo Effect exists, and 2) this Venmo Effect is driven by social influences rather than mental accounting. First, we find that the Venmo Effect does exist – the mean WTP for participants assigned to the Venmo treatment arms is 5% higher than for those assigned to debit and 17% higher than for those assigned to credit. In our fixed 2 effects regression estimation, the point estimates are consistent with a higher willingness to pay for Venmo users, though this finding lacks statistical significance. In evaluating our three proposed mechanisms, we find that signaling and priming both influence consumer willingness to pay but mental accounting does not. For the signaling mechanism, we find that participants assigned to the Friends Only setting have a 21% lower mean WTP than the Private group and a 32% lower mean than the Public group. While this impact on WTP is in the opposite direction of our expected signaling effect, it still proves that Venmo users modify their spending based on who will see their transactions. In evaluating the priming mechanism, we find that participants assigned to the priming treatment have a 15% higher mean WTP than those assigned to no priming. The direction of the priming effect is not uniform across treatment groups; rather, it depends on the Venmo privacy setting. Priming increases WTP for the Private and Public groups (by 19% and 38%, respectively) but decreases WTP by 21% for the Friends group, heightening the existing signaling effect of each privacy setting on WTP. We do not find evidence of our proposed mental accounting mechanism, which suggests that the Venmo Effect is driven purely by social factors. More specifically, to evaluate the mental accounting mechanism, we focus on participants assigned to Venmo on the Private setting without priming, as those participants should be isolated from Venmo’s social features; we find that the mean WTP for this Venmo Private No Priming group is about the same as the debit group ($1.06 versus $1.07) and slightly higher than the credit group ($1.06 versus $0.96). In addition to our novel findings about the effect of Venmo on consumer WTP, a key contribution of our research is observing a higher WTP with debit than credit. This challenges older literature on the credit card premium, which finds that WTP for credit cards 3 is higher than cash. Given that debit functions similarly to cash (money is immediately withdrawn from a bank account, purchases do not accrue rewards, etc.), we would willingness to pay with debit to fall somewhere between cash and credit. However, we find the opposite; participants assigned to the debit treatment had a 11% higher mean WTP than those in the credit treatment. This may suggest that spending preferences differ across generations or have changed over time. We segment the lottery items into four categories (Food/Drinks, Office Supplies, Toiletries, and COVID-Related Items) to determine whether different types of goods have different Venmo Effects. Across all item categories, the credit and Venmo Private No Priming treatments decrease WTP, and the Venmo Private Priming treatment increases WTP. Four of the eight treatments do not have consistent effects across all item categories. The Venmo Friends No Priming and Venmo Friends Priming treatments decrease WTP across all items, but the Venmo Friends No Priming treatment increases WTP for Office items and the Venmo Friends Priming increases WTP for Toiletry items. The Venmo Public No Priming and Venmo Public Priming treatments increase WTP across all items, but both decrease WTP for COVID-Related items and the Venmo Public No Priming treatment decreases WTP for Office items. However, these item category findings are not statistically significant. The remainder of the paper proceeds as follows. Section 2 provides an overview of Venmo, its use, and how it compares to other similar digital payment platforms. Section 3 outlines existing literature on the effect of payment type on willingness to pay, particularly focusing on mental accounting and social influences on spending. Section 4 explains the behavioral economic theory which underlies the Venmo Effect and our predictions. Section 5 describes the features of our experiment designed to test the effect of Venmo on willingness 4 to pay; we describe the participants, specific experimental conditions, pilot, and our empirical strategy for the data collected. Section 6 details the results of our experiment and empirical analysis; we share our main results across the treatment groups and how those results vary for different types of goods. Section 7 concludes. 2. Description of Venmo Venmo is a digital payment platform that allows users to send and receive payments to and from other Venmo users and select retailers. A Venmo transaction requires the initiator to indicate the user they want to send to or request from, enter the dollar amount of the transaction, and write a caption for the payment. A sent payment is immediately transferred; a requested payment must be first approved by the user who receives the request in order for the money to be transferred. All money received is added to a user’s Venmo balance, which can then be transferred to a bank account. To send money, one can use their Venmo balance or link a bank account, debit card, or credit card. Venmo is generally free to use; fees are only incurred when using a credit card (3% of the transaction amount) or when initiating an instant bank transfer (1% fee capped at $10, no fee for a 1-3 business day transfer). Venmo is only available for use in the United States. While it’s becoming more common for retailers to accept Venmo, most Venmo transactions are peer-to-peer transfers. Zhang et al. (2017) find that the majority of Venmo transactions fall into the categories of food/drinks or transportation, with less popular categories being utilities and entertainment. Venmo differs from other digital payment platforms like PayPal and Zelle as it includes a social media feed or a “social awareness stream” (SAS). In the Venmo app, users see three different social feeds: a Public feed, a Friends feed, and a Private feed. The default setting is Public, but users can change their privacy setting for all future transactions or for 5 one individual transaction. According to the Venmo website, Public transactions “will be shared on the Venmo public feed and anyone on the internet will be able to see it” and Friends Only transactions “will only be shared with your Venmo friends and with the other participant’s Venmo friends” (Venmo 2016). For Private transactions, however, Venmo “will not share the transaction anywhere other than your own personal feed and, if it’s a payment to another user, the feed of the other person in the payment” (Venmo 2016). As of 2016, these Private transactions made up approximately 51% of all Venmo transactions since Venmo became public to users in 2012 (Zhang et al. 2017). Venmo users are active and enthusiastic participants in the Venmo social media feeds. Venmo users are generally inclined to add friends on the app; half of Venmo users have at least 40 friends and 30% have over 100 friends (Zhang et al. 2017). While Venmo users describe themselves as indifferent or neutral towards the social feed in surveys, they report putting a lot of thought into the captions of their transactions on Venmo (Caraway, Epstein, and Munson 2017). Venmo is incredibly popular, particularly with young people. In 2019, Venmo had over 40 million active accounts (Statista 2020). Over time, Venmo has been growing in total payment volume; from July 2019 to July 2020, volume increased by 54% from $24 billion to $37 billion (Statista 2020). Further, Venmo is also growing in user engagement, as average transaction frequency tripled from 2012 to 2016 (Zhang et al. 2017). Young people are much more likely to use digital payments and Venmo specifically. A 2017 survey finds that 65% of millennial respondents had used a mobile payment and 44% had used Venmo (Acker and Murthy 2018); similarly, Pew Research (2018) found that people in Generation X or younger made up 74% of survey respondents who use mobile payments. 6 3. Literature Review 3.1 Mental Accounting Overview Under the assumption of the fungibility of money, a rational consumer would not change their behavior when using different forms of payment. However, there are systematic ways in which people act as if money is not fungible. Under mental accounting, people mentally segment their wealth into different accounts with different purposes (Thaler 1985; Shefrin and Thaler 1988; Thaler 1999). Mental accounting theory is applicable to our area of interest, form of payment; different payment methods may be assigned to different mental accounts, leading to a higher willingness to pay (WTP) for some payment types. Thaler (1999) identifies three critical components of mental accounting: 1) how outcomes are perceived and experienced, 2) the assignment of activities to specific accounts, and 3) the frequency with which accounts are evaluated. The second component – the assignment of activities to specific accounts – is most relevant to our research and warrants a deeper dive into the different ways that one can assign activities to various mental accounts: budgeting, the location of funds, or source of income (Thaler 1999). First, one can budget. While budgeting is necessary, it can lead to irrational purchasing decisions if a person does not make a purchase because they’ve exhausted the purchase category’s allocated budget, even if a different category still has remaining funds. For example, Heath and Soll (1996) find that participants who were told they received a $50 parking ticket were more likely to spend money on admission to a play than participants who were told they spent $50 on a basketball game; this is because the basketball game and play ticket fall into the same budget category in the participants’ minds. A second manifestation of assigning certain activities to different mental accounts is based on the location of funds. 7 This is most commonly seen in savings; people put parts of their wealth in accounts that are off-limits, like a home or a retirement account with tax penalties. Third, and most applicable to Venmo, consumers can assign spending to different mental accounts based on the way income is received. Unexpected income gains are more likely to be spent on hedonic purchases than expected income gains (Arkes, Joyner, Pezzo, Nash, Seigel-Jacobs, and Stone 1994). Further, consumers tend to spend at a similar level of frivolity as the source of the income; for example, O’Curry (1997) finds that people are likely to spend an income tax refund seriously but winnings from an office football pool frivolously. Building on this, Levav and McGraw (2009) find that the emotional label attached to a gain in income dictates how the money will be spent; more specifically, if consumers receive money in way that is associated with negative emotions, they will attempt to cope by spending it on “virtuous” or utilitarian products. In a slightly more complex understanding of mental accounting, Prelec and Loewenstein (1998) introduce pain of payment in their double entry mental accounting model; for every purchase made, consumers experience some level of pain from spending money and some amount of positive utility from the purchase. If the pain of payment is separated temporally or mentally from the consumption of the purchase (called weak coupling), the product can be consumed as if it were free, leading to consumers being more willing to spend (Prelec and Loewenstein 1998). Thus, payment forms with weaker coupling – like credit cards – have a lower pain of payment and therefore elicit higher WTP. In addition to pain of payment, Morris (2018) introduces the idea that trackability may play a role in feelings of pain associated with spending; consumers may prefer to use less trackable 8 payment methods like cash for purchases they associate with financial irresponsibility in order to lessen the guilt they feel. 3.2 Willingness to Pay with Credit Cards versus Cash Existing literature, both in the lab and the field, shows that consumers have a higher willingness to pay when using credit cards than when using cash, called a “credit card premium.” Hirschman (1979) finds significantly higher spending with credit cards than cash. Even just the presence of credit card stimuli, like logos, leads to higher willingness to pay (Feinberg 1986). In response to criticism that Feinberg (1986) primarily tests hypothetical transactions, Prelec and Simester (2001) incorporate real-money transactions into their analysis of credit cards, finding that credit cards do causally increase consumer willingness to pay, sometimes up to 100%. In addition to the literature on point-of-purchase, research also shows that people tend to remember cash transactions better than credit card transactions (Srivastava and Raghubir 2002). This ability to recall past spending can impact willingness to pay; Soman (2001) finds that that people are better at remembering past spending with checks than past spending with credit cards, leading to lower salience and vividness with credit cards and thus a higher willingness to pay. In addition to eliciting a higher WTP, credit cards also impact consumer perceptions and evaluations of products; Chatterjee and Rose (2012) find that consumers primed with credit cards focus on the benefits of a potential purchase while those primed with cash focus on the costs. These proposed mechanisms behind the credit card premium vary from classical conditioning (Feinberg 1986) to transparency of payment (Raghubir and Srivastava 2008) and memory processes (Soman 2001), but all fall under the umbrella of psychological factors and do not account for external social factors on spending. 9