FinTech Lending and Cashless Payments Pulak Ghosh Indian Institute of Management Boris Vallee Harvard Business School Yao Zeng Wharton February 2, 2021 Preliminary; click here for latest version Abstract This study provides a new perspective to understand the rise and future potential of FinTech lending by linking it to the informational role of cashless payments. We uncover both theoretically and empirically a synergy between FinTech lending and cashless payments. FinTech lenders screen borrowers more e ciently when borrowers use more cashless payments that produce transferrable and veri able information. Because borrowers expect lenders to rely on such payment information to screen them, a strategic consideration for a borrower to stand out of other borrowers then pushes more borrowers to adopt cashless payments. Using novel loan-level data from a large Indian FinTech lender who focuses on small-business lending, we nd that a larger use of veri able cashless payments (relative to cash) predicts a higher chance of loan approval, a lower interest rate, and lower default conditional on the interest rate obtained. These relationships are more pronounced for higher-quality rms. The uncovered synergy provides a plausible explanation for the joint rise of FinTech lending and cashless payments, and suggests an alternative banking model without a balance sheet or traditional banking relationships. Our ndings also provide new policy implications on data sharing and open banking. Keywords: FinTech, lending, payments, data sharing We are grateful to Dolly Yu for outstanding research assistance. We thank Tobias Berg and Shawn Cole for helpful comments. We are indebted to Indi for generously providing us access to their data. All errors are ours only. Electronic copy available at: 1 Introduction The past decade has witnessed a drastic rise of lending by FinTech companies, which was traditionally dominated by banks.1 It is well understood that banks informational advan- tage in lending stems from their relationships with borrowers from repeated lending (e.g., Diamond, 1991, Rajan, 1992) and from deposit-taking (e.g., Berlin and Mester, 1999, Puri, Rocholl, and Ste en, 2017), both helping produce borrower information inside the same bank. Without enjoying such relationships and the resulting inside information production, how can FinTech lenders compete with banks and even become dominating in some lending markets? In this paper, we provide a new perspective to understand the rise and future potential of FinTech lending by linking it to the informational role of another important nancial service: cashless payments.2 We uncover both theoretically and empirically a synergy between FinTech lending and cashless payments, the latter producing borrower information outside the lender. This synergy in producing outside information leads to a hand-in- hand rise of both FinTech lending and cashless payments, and also suggests an alternative banking model without a balance sheet and without relationships in the traditional sense. Our study builds on two simple observations and yields novel predictions. First, FinTech lenders typically use outside veri able information to assess borrowers creditworthiness be- yond the usual credit bureau inquiry. Second, cashless payment service providers collect the abundant veri able data generated through the use of their service.3 Building on these 1As per the Financial Stability Board (FSB) and Basel Committee, FinTech is de ned as technologically enabled nancial innovation that could result in new business models, applications, process, or products with an associated material e ect on nancial markets and institutions, and the provision of nancial services. In the U.S., FinTech lending has been becoming dominating in some of the most important lending markets including the mortgage markets (e.g., Buchak, Matvos, Piskorski, and Seru, 2018, Fuster, Plosser, Schnabl, and Vickery, 2019) and has also developed dramatically in small-business lending markets (e.g., Gopal and Schnabl, 2020). The rise of FinTech lending has also been particularly pronounced in developing economics (Claessens, Frost, Turner, and Zhu, 2018). 2The rise of cashless payments has speeded up since the global nancial crisis and has coincided with the rise of FinTech, with the global revenue reaching two trillion dollars (Vives, 2019). A large share of cashless payments is operated by traditional banks. Several trends have also motivated non-bank institutions to participate, including the willingness to tackle fraud, lower operating costs, and the development of novel payment technologies. 3Our study can be generalized to other easily accessible veri able data beyond cashless payments. For instance, our results are externally valid to sales data on marketplaces, which are also collected and 1 Electronic copy available at: two observations, we develop a theoretical framework showing that the interaction between FinTech lenders and cashless payments fosters the development of both technologies. In one direction, FinTech lenders become more e cient in screening high- versus lower-quality borrowers when borrowers adopt cashless payments that produce more veri able informa- tion. In the other direction, because would-be borrowers expect lenders to rely on outside veri able payment information to screen them, a strategic consideration for a borrower to stand out of worse borrowers emerges, which ultimately pushes all borrowers to adopt cashless payments. This synergy further implies that even without policies to promote data sharing or open banking such as the Second Payment Services Directive (PSD2) in Europe, borrowers may voluntarily commit to data sharing in order to improve their outcome on the lending markets. More speci cally, we build a simple model of FinTech lending and borrowers choice of payment methods to illustrate how the synergy arises in equilibrium. In the model, there is a risk-neutral rm and a risk-averse nancier, both are competitive. A rm of higher type has a better investment technology, and thus more likely to produce a product of higher quality. Only the rm is privately informed about its type. Thus, the nancier relies on its prior belief and any information available to decide whether to nance the rm. Prior to this nancing stage, each rm chooses their payment technology for the period preceding the loan application, which we call the production stage. Cash does not leave any veri able information about the production outcomes. The cashless payment service, in contrast, can record and keep veri able information about all the production outcomes and make them available for the nanciers potential use in the nancing stage. If the rm adopts cashless payments, it commits to provide whatever information being generated to the nancier in the nancing stage, consistent with the practice that FinTech lenders can easily access and use veri able information generated by outside payment service providers. Thus, the rm optimally chooses whether to use cashless payments at the beginning under the expectation potentially shared via a third party. These types of data are conceptually close as they both are directly generated by the economic activity of the rm. 2 Electronic copy available at: that the nancier will use the information generated to make nancing decisions.4 We illustrate the synergy between lending and cashless payments in two steps, each of which highlights one direction of the synergy. We start by showing how cashless payments improves lending outcomes by highlighting two complementary informational e ects of ver- i able payment records. First, when a rm adopts the veri able cashless payment service, the payment records help reveal the quality of the rms technology. This information- revealing e ect allows the nancier to better screen the rm and achieve more e cient nancing outcomes. Particularly, the information-revealing e ect bene ts rms of better types, and is stronger when the rm can establish more payment records or the payment records are more veri able. Second, veri able payment records also directly reduce the nancing risk that the nancier bear by reducing the variance when the nancier makes inference about the rm type, and this risk-reducing e ect bene ts both the nancier and all rm types. The overall impact of adopting cashless payments on lending outcomes thus depend on the combination of the two e ects. We then show how lending based on outside veri able information fosters the adoption of cashless payments. The rm, expecting to be screened by the nancier who can access the outside payment information, optimally adopts the the veri able cashless payment service. This is true even if the information-revealing e ect dominates and low-type rms may be hurt by revealing the quality of their technology. Although surprising at the rst glance, this force stems from an intuitive strategic consideration among di erent rm types. When high rm types adopt cashless payments, the nancier will rationally update its belief and expect any rm using cash to be of a low type. Thus, a relatively low-type rm would nd it optimal to stand out from even lower-type rms by adopting cashless payments. Ultimately, all rm types adopt cashless payments.5 Therefore, the rise of FinTech lending and its reliance on outside veri able payment information in turn fosters the rise of cashless 4To focus on the informational role of cashless payments, we abstract away from their other bene ts and costs (i.e., safety or convenience yields). 5This logic resembles the unraveling argument rst analyzed in an information disclosure context by Milgrom (1981). Indeed, when the nancier can e ciently use outside veri able information to screen rms, committing to generate such information by adopting cashless payments can be interpreted as an information disclosure decision for the rms. 3 Electronic copy available at: payments. To test our model predictions, we use novel loan application-level data from one of the largest Indian FinTech lenders, Indi , which focuses on small business loans. The lending context in India and broader emerging economics is a relevant laboratory to test our model predictions, as the joint rise of FinTech lending and cashless payments has been particu- larly pronounced in these economies. The data we exploit in the study is also particularly suited to our analysis as it includes borrowers detailed payment information disaggregated at the payment level. We rst develop a methodology to classify each payment appear- ing on bank statements into cash and cashless payments. Within cashless payments, we can further break down between information-intensive and information-light methods of payments, depending on whether payments are partly aggregated, and whether the pay- ment counter-party is known in the statements. Being able to access such payment-level information not only allows Indi to potentially screen the borrowers more e ciently, but also uniquely allows us as econometricians to test how payment information with varying level of veri ability a ects lending outcomes. The data also includes a wealth of granu- lar information including applicants business characteristics as well as their credit bureau data, which allows us to precisely pinpoint the role of outside payment information vs. traditionally openly accessible data. Equipped with this data and measures of use of cashless payments, we study whether such use correlates with loan screening outcomes on both the extensive and intensive mar- gins, controlling non-linearly for a wealth of applicant characteristics such as business size, age, 3-digit zip code, and owner credit score. First, we nd that a higher use of cashless payments (and relatively, a lower use of cash) is associated with improved borrowing out- comes: applicants relying heavily on cashless payments are more likely to obtain a loan, and when doing so obtain a lower interest rate. At the same time, we nd that such bor- rowers also get signi cantly lower rate from the FinTech lender than from previous loans with traditional banks. Second, this bene t is particularly pronounced for cashless pay- ment users that present a low level of risk, as proxied by the volatility of their revenues. 4 Electronic copy available at: These relationships appear to be more pronounced when focusing on more veri able cash- less payments, such as individual internet transfers, as opposed to less veri able ones, such as mobile payments that aggregate payments and do not provide information about the transacting counter-party. Finally, turning to loan default, we nd that within loans charg- ing the same interest rate, borrowers that use more cash transactions are more likely to default. This suggests that the use of cashless payments indeed helps the FinTech lender to price loans more e ciently, leading to more e cient capital allocation among di erent borrowers. To gain causal identi cation on the impact of cashless payments on loan application outcomes, we utilize a unique institutional setting in India. We instrument the reliance on cash payments with an indicator variable for the borrower banking at a reserve chest bank, that is, a bank branch that distributes new banknotes and collect damaged old ones. Banking at such an establishment after the demonetization is indeed predictive of a higher use of cash, as chest bank clients had better access to the new banknotes during the initial shortage of cash, and therefore switched less to other means of payment. When instrumenting the use of cash with this plausibly exogenous variation, our previous result is strengthened: we nd that cash use negatively impacts the likelihood of obtaining a loan, with a magnitude that is economically signi cant. Taken together, our theoretical framework and empirical results provide a new perspec- tive to understand the interaction between the rise of FinTech lenders and the development of cashless payments. In one direction, our ndings of cashless payments improving lend- ing e ciency provide direct evidence in support of policies that promote data sharing and open banking. In the other direction, the prediction of a universal adoption of veri able payment service is consistent with the trends of many economies increasingly switching to cashless payments hand-in-hand with the rise of FinTech lending, particularly in develop- ing economies such as China, India, or Kenya. The synergy we document also provides an economic rationale for the recent trends of FinTech lenders directly o ering payment ser- 5 Electronic copy available at: vice, and payment service rms and marketplaces o ering credit.6 Despite expanding their scope, these institutions remain fundamentally di erent from traditional banks as they do not accept deposits and are not regulated as banks. These developments suggests the emer- gence of an