Technological Change and Financial Innovation in Banking: Some Implications for Fintech (Federal Reserve Bank Atlanta Working Paper 2018-11) by
W. Scott Frame, Larry Wall, and Lawrence J. White
comments
Financial intermediation has changed dramatically over the past 30 years, due in large part
to technological change. The paper first describes the role of the financial system in a modern economy
and how technological change and financial innovation can affect social welfare. We then survey the
empirical literatures relating to several specific financial innovations, broadly categorized as new
production processes, new products or services, or new organizational forms. In each case, we also
include examples of significant fintech innovations that are transforming various aspects of banking.
Drawing on the literature on innovations from the 1990s and 2000s informs what we might expect from recent developments.
The authors state
Financial intermediation has changed dramatically over the past 30 years, due in large part to
technological change arising from advances in telecommunications, information technology, and
financial practice. This technological progress has spurred financial innovations that have altered
many financial products, services, production processes, and organizational structures. To the extent
that such financial innovations reduce costs or risks, social welfare may be improved. Of course, many
financial innovations fail owing to fundamental design flaws or simply being replaced by better
alternatives.
A good example of technological change that has been dramatically reshaping the financial
services industry is the ongoing shift from relying on human judgment to automated analysis of
consumer data. This has taken what had been largely local markets for banking services and opened
them up to nationwide competition from other banks and nonbank financial institutions. For
example, retail loan applications are now routinely evaluated using credit scoring tools built using
comprehensive historical credit registry databases. This automated approach eliminates the need to
have a local presence to make a loan and substantially reduces underwriting and compliance costs for
lenders, and the resulting data can be leveraged to improve further their risk measurement and
management. Such a reliance on hard information also makes underwriting transparent to third parties
and hence facilitates secondary markets for retail loans through securitization, which allows nonbank
firms that lack deposit funding to compete via capital market financing.
Given the growing importance of technology to financial services, it is perhaps not too
surprising that the latest trend has been for technology-based firms to offer financial services, a
development that is often called “fintech”. Many fintech firms combine automated analysis of retail customers with more user-friendly interfaces to provide services that are more convenient, and
sometimes lower cost, to consumers. For example, “marketplace lending” platforms have emerged
as a new organizational form that attracts borrowers with a simplified loan application process,
leverages credit scoring tools to analyze these applications, and then matches creditworthy borrowers
directly to investors. Furthermore, in some jurisdictions, machine learning (artificial intelligence) is
now being leveraged to further improve retail loan risk measurement.
Another set of recent technological developments are being touted as having the potential to
have an even more fundamental impact on the financial system, potentially eliminating the need for
trusted third parties such as banks. Whether and to what extent blockchains and cryptocurrencies will
disrupt the existing financial system remains to be seen, as the technology is too new and immature
to draw firm conclusions. However, the potential benefits of cryptocurrencies and blockchain
technology are sufficient to attract considerable interest from tech-knowledgeable individuals, large
financial organizations, and even major governments.
This chapter surveys the research literatures pertaining to several specific financial innovations
that have appeared in recent decades that were specifically driven by technological change. Particular
attention is paid to innovations that may provide insights into the prospects for certain widely
discussed fintech applications. To set the stage, we begin by providing some additional clarity about
what is meant by financial innovation.
One of the more interesting discussions, for instance in relation to Australia's emerging Consumer Data Right regime (eg noted
here), is
Marketplace lenders, which match consumers and small firms with
lenders/investors using online platforms, have been popping-up all over the world. In the United
States, these lending arrangements generally work in the following way: First, borrowers apply on the platform and are subject to automated underwriting based on standard criteria (such as a credit score)
plus additional information and assigned a proprietary risk rating. Second, institutional investors
purchase loans in bulk from the marketplace lenders, principally based on the risk ratings. The online
marketplaces themselves generally have no direct exposure to the credit risk of the loans through their
platforms, as they do not typically hold the loans or otherwise retain an interest in them or guarantee
their performance. Instead, marketplace lenders principally generate revenue from loan origination
and servicing fees. Marketplace lending is growing rapidly, but it remains a very small part of the $3.3
trillion U.S. consumer lending market.
Much of what constitutes marketplace lending is actually not new. As discussed above, for
many years, larger banks and finance companies have used credit registry data, credit scores, and
borrower income information as inputs for statistical models to estimate risk and price consumer
loans. However, marketplace lenders appear to be increasingly supplementing their models with
additional information. Jagtiani and Lemieux (2018) find that LendingClub’s credit scores had an 80
percent correlation with FICO scores in 2007, but that the correlation drops about 35 percent for
loans originated in 2014-15. The authors suggest that the change is likely due to a combination of
LendingClub using alternative data and machine learning as the platform gains more experience with
consumer lending. In complementary research that uses information from Prosper (which is a prominent marketplace lender), Balyuk and Davydenko (2018) discuss that lender’s use of secondary
screening to identify suspicious applications and to verify automatically some borrower-provided
information. The authors report that this additional screening has led to cancellation of 27 percent of
the previously accepted loan applications since 2013.
Vallee and Zeng (2018) observe that, while the fintech platforms are using their own models
to grade loans and determine credit spreads, informationally sophisticated investors may be able to
differentiate credit quality within these ratings grades. The authors derive a model allowing for such
a split in investor sophistication, which results in a trade-off for the platform in terms of the
contribution of sophisticated investors in improving loan quality but also creating adverse selection
for less sophisticated investors. The volume-maximizing solution for the platform is to provide
intermediate levels of screening and information to investors. Consistent with their model, the authors
find that loans purchased by more informationally sophisticated investors were less likely to default
for the universe of investments made through Lending Robot from 2014-2017. They also observe
that one marketplace lender, LendingClub, reduced the amount of information it provided to investors
and this caused a reduction in the ability of sophisticated investors to “cherry-pick” loans with lower
default rates.
Beyond marketplace lenders specifically, there has been a general increase in online lending.
According to Fuster, Plosser, Schnabl, and Vickery (2018), fintech mortgage lenders have increased
their market share from two to eight percent between 2010 and 2016. The authors find the biggest
benefit provided by fintech lenders is an average reduction in the time from application to closing of
10 days (20 percent) after controlling for borrower and loan characteristics. They also find that fintech
lenders can scale up the volume of mortgages they process more readily than can other lenders.
The information technology underlying such an automated approach to underwriting is subject to significant scale economies (large fixed costs and very low marginal costs), which provides
strong incentives to grow large quickly. This suggests that the consolidation of the marketplace
lending industry is very likely. Moreover, as marketplace lenders become more successful, they are
likely to find themselves facing increased competition from incumbent consumer lenders.
The paper complements the ACFS
International competition policy and regulation of financial services report noted
here.
