Fintech, Bigtech, and Financial Inclusion

Loretta J. Mester
President and Chief Executive Officer
Federal Reserve Bank of Cleveland

Fourth Annual Fintech Conference

November 9, 2020
 1

Introduction

I thank the Federal Reserve Bank of Philadelphia and the other organizers for inviting me to speak at the

fourth annual Fintech Conference. I spent most of my career at the Philadelphia Fed, so it is always a

pleasure to participate in one of its events. Since the beginning of this conference series, the discussions

have consistently been very topical, and the agenda for the next two days does not disappoint on that

score. The conference will cover many of the hot issues confronting practitioners, academics, and

policymakers as financial system innovation proceeds at a rapid pace. Today I will discuss the

implications of digitalization for financial inclusion and some steps that need to be taken to ensure that

digitalization helps to foster inclusion rather than promote exclusion. The views I will present today are

my own and not necessarily those of the Federal Reserve System or my colleagues on the Federal Open

Market Committee.

Financial Inclusion Has Benefits

It is probably best to start off with a definition of financial inclusion, and the World Bank provides us

with a workable one: financial inclusion refers to individuals and businesses having access to useful and

affordable financial products and services that meet their needs and that are delivered in a responsible and

sustainable way. 1 Much of the literature measures financial inclusion as households’ and businesses’ use

of formal financial services from banks or other traditional providers, so-called mainstream financial

services. With the entry of fintech providers, that definition is expanding because fintech financial

services are moving into the mainstream.

The World Bank’s Global Findex database provides data on how people in 140 economies access

accounts, make payments, save, borrow, and manage risks. It includes data on formal and informal

financial services and the use of financial technology to conduct financial transactions. 2 The data indicate

1
The World Bank, Financial Inclusion, https://www.worldbank.org/en/topic/financialinclusion/overview.
2
World Bank (2017).
 2

that across the globe financial inclusion has risen. The share of adults with an account at a financial

institution or through a mobile money service rose from 62 percent in 2014 to 69 percent in 2017.

Nonetheless, there is still a large portion of the population that does not have access to or use financial

services. As of 2017, an estimated 1.7 billion adults around the world are unbanked or underserved.

Women lag behind men; the poor lag behind the wealthy; and developing economies lag behind

developed economies in the shares having formal accounts.

In the U.S., the level of financial inclusion is high: 93 percent of adults have an account at a bank or other

type of financial institution or use a mobile money service. But disparities exist: there is a 13 percentage

point gap between those in the richest 60 percent of households and those in the poorest 40 percent.

Federal Reserve data indicate that, compared to whites, Blacks and Hispanics in the U.S. are less likely to

have a bank account and more likely to rely on alternative financial services such as money orders and

check cashing services. 3 And compared to small firms with white ownership, those with Black ownership

were half as likely to have obtained financing from a bank in the past five years, relying more on online

lenders, which, according to the survey respondents, provide less satisfactory service. 4

To address these types of gaps, many countries have set financial inclusion as a formal target. 5 This is

reasonable because numerous studies have documented the contribution of a well-functioning financial

system and higher levels of financial inclusion to longer-run macroeconomic goals of output and

3
According to a 2020 supplement to the Board of Governors’ Survey of Economic Well-Being, 14 percent of Black
adults and 10 percent of Hispanic adults do not have a bank account, compared to 3 percent of white households.
(See Board of Governors of the Federal Reserve System (May 2020), Table 11, p. 28.) Unbanked is defined as not
having a checking, savings, or money market account. Underbanked is defined as having a checking, savings, or
money market account but having used one of these alternative financial services: money order, check cashing
service, pawn shop loan, auto title loan, payday loan, paycheck advance, or tax refund advance.
4
According to the Fed’s most recent Small Business Credit Survey of firms with 1 -499 employees, 46 percent of
firms with white ownership, 32 percent of firms with Hispanic ownership, and 23 percent of firms with Black
ownership had obtained financing from a bank in the past five years; 19 percent of firms with white ownership, 22
percent of firms with Hispanic ownership, and 27 percent of firms with Black ownership had obtained financing
from an online lender in the past five years. (See Federal Reserve System (2020), p. 9.)
5
Sahay, et al. (2015).
 3

productivity growth. 6 On the macro level, healthy financial markets and institutions allow for more

efficient allocation of capital and better monitoring and broader diversification of risk, which can enable

higher levels of overall growth. At the micro level, access to savings and credit via financial

intermediaries connects households and businesses to economic opportunities that would not be available

otherwise. A sound financial system can spur entrepreneurship and support the competitive forces that

drive productivity in the economy.

In addition to these benefits, the distributional impacts of finance are worth considering. A body of work

has shown that finance can expand economic opportunities for those at the bottom of the income and

wealth distributions and does not merely benefit those at the top. 7 For example, access to credit allows

access to education, which can have profound effects on an individual’s economic well-being, and it also

allows households to build wealth through homeownership, which remains the most significant asset on

many U.S. families’ balance sheets.

Digitalization Is Transforming Financial Services

Digitalization of financial services is transforming the industry and expanding the range of financial

service providers to include not only fintech lending and payments firms but also so-called bigtech firms,

including technology, social media, search platform, and e-commerce companies. Consumer demand is

one factor driving the rise in these services. EMarketer estimates that in 2019, 1.9 billion people

worldwide purchased goods online. 8 In 2020, despite the global pandemic, retail e-commerce sales are

estimated to rise to over $3.9 trillion and yearly sales growth has been in the 15 to 25 percent range over

the past three years. 9 As the costs of computing have declined and the demand for new services that offer

6
Levine (2009).
7
Levine (2009) and Demirgüç-Kunt and Levine (2009).
8
See United Nations (2020) and Lipsman (2019).
9
See Lipsman (2019) and Cramer-Flood (2020).
 4

speed and convenience at a lower price has risen, new firms have entered the financial services space.

According to the U.S. Treasury, from 2010 to the third quarter of 2017, more than 3,330 new technology-

based firms serving the financial services industry have been founded. The global market capitalization

of fintech firms grew to $22 billion in 2017, 13 times what it was in 2010. Lending by these firms

accounts for over 36 percent of personal loans in the U.S., up from under 1 percent in 2010. New tools

and techniques, including machine learning and artificial intelligence, are now routinely being applied in

finance.

Digitalization Has the Potential to Increase Financial Inclusion

Like past financial innovations, digitalization holds the promise of increasing the efficiency, productivity,

and inclusiveness of the financial sector, thereby increasing the economic welfare of households and

businesses. For example, a wide body of research indicates that tools that increase a household’s savings

behavior can have a substantial impact on household welfare. 10 Digital tools that give households and

small businesses across the income spectrum the ability to track and understand their spending and

savings patterns can improve their ability to manage their finances, allow them to avoid more costly

sources of credit, and increase their savings rate. There is evidence, reflected in several countries’

experiences, that access to savings accounts not only increases savings but also results in households

shifting more of their spending to education and to healthier food. 11 Digitalization also provides

customers with tools to search for and compare financial services across various vendors to determine

which is more likely to meet their specific needs.

On the credit side, digital lending platforms hold the potential for more objective credit decisions, helping

to guard against personal prejudices influencing those decisions. Bigtech companies have access to large

amounts of data, for example, from e-commerce platforms or search engines. Models to assess credit risk

10
Karlan, et al. (2016).
11
Karlan, et al. (2016).
 5

based on these alternative data hold the potential to increase credit access for households and small firms

that do not necessarily have a long credit history, audited financial statements, or collateral, which are

traditionally used to underwrite and monitor credit risk. And there is some research to back this up. A

study co-authored by Julapa Jagtiani, one of the organizers of this conference, found that consumer loans

made by LendingClub, a large fintech lending platform, reached areas with fewer bank branches and

lower-income borrowers. 12 Research published by the Bank for International Settlements found that the

internal credit scoring model built by and applied to small merchant customers of Mercado Libre, a large

e-commerce platform in Latin America now offering financial services, has outperformed (at least in the

short run) models based on credit bureau ratings and traditional borrower characteristics. 13 And Mercado

Libre was able to provide credit to merchants who would have been assessed as high risk by the credit

bureau.

Innovations in payments services also hold the potential to increase inclusion. Across the globe there is a

move toward faster, cashless payment systems, which can lower costs and provide more secure

transactions than cash. Early evidence from several countries shows that the move to digital payments

has had a positive impact on the well-being of individuals by strengthening their ties to other financial

services. 14 There is also evidence that the use of digital systems by governments to make transfer

payments to individuals can significantly reduce the cost of distribution and the amount of fraud. 15

The Federal Reserve’s FedNowSM service, which is currently being built, will be an around-the-clock

service whereby payments can be originated, cleared, and settled within seconds. 16 The service is

12
Jagtiani and Lemieux (2018).
13
See Bank for International Settlements (2019) and Frost, et al. (2019).
14
Karlan, et al. (2016).
15
Karlan, et al. (2016).
16
More information on FedNow is available on the Fed’s website at https://www.frbservices.org/financial-
services/fednow/index.html.
 6

expected to provide clear public benefits in the form of safety, efficiency, and accessibility of instant

payments. By lowering the cost of making payments in a secure way, FedNow can help promote

financial inclusion by drawing more people into the financial system. This service, coupled with a

directory service with accurate information on where to route payments for final distribution to

households and businesses, could also make distribution of government benefits more efficient and solve

some of the challenges the government faced when distributing pandemic relief payments earlier this

year.

Steps Need to Be Taken So That Digitalization Lives Up to Its Potential

While digitalization holds a lot of promise to bring more households and businesses into the financial

system, there is no guarantee this will happen. Indeed, digitalization could create more exclusion and

increase disparities rather than close the gaps. But steps can be taken to help ensure that digitalization

lives up to its potential. Let me discuss five of these.

Close the digital divide

First, the digital divide in the U.S. needs to close so that more people can take advantage of digital

financial services. Among communities with a population of 100,000 or more, the city of Cleveland has

among the lowest home broadband access in the nation. 17 Only about 69 percent of households have

broadband subscriptions compared to over 84 percent in the city of Philadelphia and over 86 percent for

the nation as a whole. In the meantime, mobile phone technology, which has wide distribution, could

significantly improve financial system access for unbanked adults. 18

17
The data are available in Table GCT2801, U.S. Census American Community Survey (2019). In the cit y of
Cleveland, 69.3 percent of households have a broadband internet subscription. Only Lakewood Township, Ocean
County, N.J. has lower internet connectivity, at 58.9 percent. The number for the U.S. is 86.4 percent and the
number for the city of Philadelphia is 84.2 percent.
18
World Bank data indicate that two-thirds of unbanked adults globally have a mobile phone. See Demirgüç-Kunt,
et al. (2018), p. 91.
 7

Improve financial literacy

Second, steps must also be taken to improve the reach and the effectiveness of financial literacy

programs. Digitalization increases the number of financial services providers and the types of services

being offered. But to take advantage of these, people need to be able to assess their value. One study of

188 financial literacy programs indicated that they had not, on average, increased financial knowledge or

resulted in better financial choices. But the research also suggested that programs that included simple

rules-of-thumb were more effective.19 Businesses also have to understand the costs and benefits of credit

coming from a fintech firm versus a traditional banking relationship, which may prove to be more stable

in an economic downturn. Increasing the ability of the consumers of financial services to evaluate new

services will help ensure that these products add value.

Build trust

Third, in order to increase inclusion, steps must be taken to build trust between potential customers and

the financial services industry. One Federal Reserve analysis indicated that one reason people do not

have a transaction account is the lack of trust in financial institutions. 20 Effective consumer protection

regulations encompassing all providers of financial services would give potential customers more

confidence to use the services. Trust also depends on providers of financial services adequately

protecting their customers’ accounts against cybersecurity breaches, fraud, and data leakage. With the

changes in technology and rapid expansion of available data, the methods for protecting the data will need

to adapt. The traditional way of keeping data private, by taking away names or otherwise anonymizing it,

no longer works in a world rich with multiple data sources that can be cross-referenced to de-anonymize

the data and reveal identities. 21 New methods have to be evaluated. An example is the technique used to

achieve differential privacy by carefully adding some statistical noise to the data to maintain individual

19
Karlan, et al. (2016).
20
Hogarth, et al. (2004), as cited in Raskin (2011).
21
Kearns and Roth (2020) discuss how then-Massachusetts governor William Weld’s medical record was identified
by combining the anonymized medical records released by the state with voter registration data.
 8

privacy while maintaining statistical accuracy. 22 The U.S. Census Bureau plans to apply differential

privacy techniques to the statistics released as part of the 2020 census. 23

Test the algorithms before deployment

Fourth, it must be recognized that just because it is an algorithm does not mean it is immune from

producing discriminatory underwriting and pricing decisions. Algorithms should be tested for bias before

they are deployed. Credit and pricing models produced by algorithms through machine learning are

necessarily complex because they are designed to reveal relationships in the data that are not revealed by

standard modeling. But their opacity and complexity – their black-box quality – make it more difficult to

identify these relationships and more difficult to uncover statistical discrimination or disparate impact and

to enforce fair lending laws. 24

Well-meaning algorithms can have unintended effects. Algorithms are trained on past data that may have

reflected biased decisions or may not be representative of the entire population. With more pieces of data

available, there is more risk that data correlated with race or gender could be used in discriminatory ways.

This means machine learning algorithms could exacerbate disparities by reinforcing past decisions.

Kearns and Roth point out the need for what they call ethical algorithms, which are built to balance the

accuracy of the model, for, say, assessing credit risk, with other desirable goals. 25 For example, the

developer could build the algorithm so that it has low error rates in identifying creditworthy borrowers

and the error rate for one racial group is not disproportionately higher than that for another.

At present, much of the auditing of algorithms occurs in an ad hoc way after the algorithm has been put

22
Differential privacy is achieved when no one, regardless of additional data, is able to determine better than
random guessing whether any individual’s data were used to construct a data release. See Kearns and Roth (2020).
23
See U.S. Census (2020).
24
Morse and Pence (2020) examine the ways in which technology may augment or deter discrimination in
household finance. See also Abuhamad (2019).
25
Kearns and Roth (2019, 2020).
 9

into place and has had the opportunity to have a disparate impact. 26 A well-known example is the Apple

Card: gender discrimination was alleged by some customers about the algorithms used by Goldman

Sachs, the issuer of the card, after the card was deployed. 27 Who should do the testing is an open

question. A report from the Brookings Institution’s Artificial Intelligence and Emerging Technologies

Initiative recommends that the developer test the algorithm for disparate impact. Companies that use such

algorithms could engage independent third parties to evaluate them, with appropriate restrictions to guard

the firms’ intellectual property. Regulators may also have to play a larger role in testing algorithms.

Since adequate data sets are needed for testing, the rules pertaining to data collections should be

reviewed. The Equal Credit Opportunity Act restricts creditors from discriminating in credit transactions

against credit applicants on any prohibited basis, including race and gender. The collection of data on

applicants’ personal characteristics for home purchase and refinance loan transactions is required and the

data must be publicly disclosed. The Dodd-Frank Act also requires small business lenders to collect and

report race and gender data to federal regulators. But for other nonmortgage loans, creditors’ collection

of these data is prohibited except for the creditors’ own use to conduct a limited self-test of their

compliance with the law. 28 The original prohibition was based on the notion that the data might be used

in discriminatory ways. But experience with the collection of mortgage data suggests that this has not

been the case and that the data have made it easier to monitor compliance and enforce the law. Extending

mandatory data collection of personal characteristics for nonmortgage credit applicants may allow for

better testing for compliance. But it would also entail costs for creditors that could partially be passed on

to borrowers. With more entities offering credit, it is time to undertake a new review of the costs and

benefits of data collection for different types of nonmortgage credit.29

26
Kleinberg (2018).
27
MacCarthy (2019).
28
Data collected for such a self-test are not available to government agencies or credit applicants for use in
examinations or litigation. See Taylor (2011).
29
The Equal Credit Opportunity Act does not address data collection. In 1995 and again in 1998, the Board of
Governors of the Federal Reserve System put out for public comment proposed amendments to Regulation B, which
 10

Rethink the regulatory framework

Fifth, a rethinking of the regulatory framework is needed to ensure that the financial innovations led by

digitalization are a net positive. Such innovations do not reduce the need for risk management, financial

regulation and supervision, and good governance, although the form each takes is likely to be different

than what has been effective in the past. Existing regulatory and supervisory structures will need to adapt

to keep up with the new ways that financial services are being delivered and the new players delivering

them. Under the principle, “If it walks like a duck and quacks like a duck, it’s a duck,” a regulatory

approach that shifts the focus from the type of institution offering the service to the type of activity would

likely be more effective in fostering the stability of the financial system and limiting regulatory arbitrage.

The extension of some regulations to new service providers is already occurring; for example, bigtech

firms’ payments services are subject to know your customer rules.

But the entry of fintech and bigtech providers of financial services also raises new issues. Data produced

by bigtech firms are what make it attractive for these firms to enter into financial services in the first

place. But their ability to control the data also makes it harder for other firms to enter the market, thus

limiting competition. There are different approaches to regulating data usage in order to limit the market

power of bigtech firm entrants and facilitate market contestability without stifling entry.30 Two examples

are the European Union’s General Data Protection Regulation and the open banking regulations, which

are in place in many countries. Both approaches allow certain data to be shared directly with third parties

but place restrictions on the type of data and the kinds of authorizations that need to be garnered in order

to allow for data sharing.

implements the ECOA, to eliminate the data ban and allow for voluntary collection of personal characteristics of
applicants for nonmortgage credit. The final rules did not drop the ban. See Braunstein (2008), Government
Accountability Office (2008), Taylor (2011), and Abuhamad (2019).
30
See Frost, et al. (2009) for further discussion.
 11

The ability of bigtech firms to produce and to capture the benefits of big data depends on network effects.

Benefits to users grow when there are more users of the services, and as the number of users increases,

more data are produced that can be garnered for use by the bigtech firm for the development of new

services, pricing, and marketing. For example, a firm that may have begun as an e-commerce firm may

find it beneficial to offer payment services as an adjunct to generate more e-commerce business, but the

transactions data can then be used to offer more services. 31 Once a bigtech firm is well established, it

becomes more difficult for other firms to enter. This adds concerns about market power to data privacy

concerns. Issues of pricing, cross-subsidization, product tying, and other anti-competitive practices that

lead to less innovation, not more, deserve increased attention from financial services regulators.

With the entry of bigtech and fintech firms into financial services, the public policy approach will need to

change to include a more holistic blending of financial regulation, antitrust policy, and data privacy

regulation. There will need to be cooperation across these types of regulators within each country, and

this may entail creating structures that allow for more formal or systematic coordination across different

types of regulators. The growth of digitalization also calls for a modernization of the antitrust laws and

policy, which is occurring in several countries. 32 The global nature of the bigtech firms entering financial

services means that effective international coordination among regulators and supervisors through the

Financial Stability Board and other international entities will be critical to ensuring that the benefits can

be captured and the risks managed. 33

Finally, it is important that regulators and those responsible for enforcing fair lending and consumer

protections increase their expertise regarding the technologies being used in the marketplace. This is

31
Bank for International Settlements (2019) provides a nice description of network externalities in bigtech’s entry
into financial services.
32
Bank for International Settlements (2019) discusses some of the implications of bigtech for traditional competition
policy.
33
The Financial Stability Board designated fintech as a key area of focus in its 2020 work program. See Financial
Stability Board (2019).
 12

beginning to happen as financial system regulators and supervisors are starting to use big data and data

analytics to improve their own assessments of banking and financial stability risks and to understand

trends in the industry more generally. Several countries have set up innovation hubs and regulatory

sandboxes to encourage innovation in financial services. The Fed has been researching new technologies

and innovations in financial services for some time. The System Payments Researchers, co-chaired by

Fumiko Hayashi of the Kansas City Fed and Bob Hunt of the Philadelphia Fed, are producing research on

fintechs and digital currencies. The Board of Governors has set up a technology lab where researchers

across the System are building and testing a range of distributed ledger platforms to understand their

potential benefits and tradeoffs. 34 The Boston Fed is working with the Massachusetts Institute of

Technology to experiment with digital currency technologies, and the New York Fed has established an

innovation center, in partnership with the Bank for International Settlements, to identify and develop in-

depth insights into critical trends and financial technology of relevance to central banks. 35

Conclusion

In summary, fintech, bigtech, algorithms, machine learning, and digitalization are rapidly transforming

the financial services industry. These innovations hold the promise of increasing financial inclusion for

the benefit of individuals, businesses, and the overall macroeconomy. But as with past innovations, they

also present challenges that will need to be addressed to ensure that the promised benefits are attained and

risks are well managed.

34
Brainard (2020).
35
Bank for International Settlements (2020).
 13

References

Abuhamad, Grace M., “The Fallacy of Equating ‘Blindness’ with Fairness: Ensuring Trust in Machine
Learning Applications to Consumer Credit,” Master of Science in Technology and Policy Thesis,
Massachusetts Institute of Technology, June 2019.
(https://dspace.mit.edu/handle/1721.1/122094)

Bank for International Settlements, Annual Economic Report, June 2019.

Bank for International Settlements, “BIS Innovation Hub to Expand to New Locations in Europe and
North America,” June 30, 2020.
(https://www.bis.org/press/p200630a.htm)

Board of Governors of the Federal Reserve System, “Report on the Economic Well-Being of U.S.
Households in 2019, Featuring Supplemental Data from April 2020,” May 2020.
(https://www.federalreserve.gov/publications/files/2019-report-economic-well-being-us-households-
202005.pdf)

Brainard, Lael, “An Update on Digital Currencies,” at the Federal Reserve Board and Federal Reserve
Bank of San Francisco's Innovation Office Hours, San Francisco, CA (via webcast) August 13, 2020.
(https://www.federalreserve.gov/newsevents/speech/files/brainard20200813a.pdf)

Braunstein, Sandra F., “Equal Credit Opportunity Act,” testimony before the Subcommittee on Oversight
and Investigations, Committee on Financial Services, U.S. House of Representatives, July 17, 2008.
(https://www.federalreserve.gov/newsevents/testimony/braunstein20080717a.htm)

Cramer-Flood, Ethan, “Global ECommerce 2020,” eMarketer, June 22, 2020.

Demirgüç-Kunt, Asli, and Ross Levine, “Finance and Inequality: Theory and Evidence,” Annual Review
of Financial Economics 1 (2009), pp. 287-318.
(https://www.annualreviews.org/doi/abs/10.1146/annurev.financial.050808.114334)

Demirgüç-Kunt, Asli, Leora Klapper, Dorothe Singer, Saniya Ansar, and Jake Hess, “The Global Findex
Database 2017: Measuring Financial Inclusion and the Fintech Revolution,” World Bank Group, 2018.
(https://openknowledge.worldbank.org/bitstream/handle/10986/29510/211259ov.pdf)

Federal Reserve System, Federal Reserve Small Business Credit Survey: 2020 Report on Employer
Firms, 2020.
(https://www.fedsmallbusiness.org/medialibrary/FedSmallBusiness/files/2020/2020-sbcs-employer-firms-
report)

Financial Stability Board, “FSB Work Programme for 2020,” December 17, 2019.
(https://www.fsb.org/wp-content/uploads/P171219.pdf)

Frost, Jon, Leonardo Gambacorta, Yi Huang, Hyun Song Shin, and Pablo Zbinden, “BigTech and the
Changing Structure of Financial Intermediation,” BIS Working Paper No. 779, April 2019.
(https://www.bis.org/publ/work779.htm)

Government Accountability Office, “Fair Lending: Race and Gender Data Are Limited for Nonmortgage
Lending,” Report to Congressional Requesters, GAO-08-698, June 2008.
 14

(https://www.gao.gov/products/GAO-08-698)

Hogarth, Jeanne M., Chris E. Anguelov, and Jinkook Lee, “Why Don’t Households Have a
Checking Account?” Journal of Consumer Affairs 38(1) (2004), pp. 1-34.
(https://doi.org/10.1111/j.1745-6606.2004.tb00463.x)

Jagtiani, Julapa, and Catharine Lemieux, “Do Fintech Lenders Penetrate Areas That Are Underserved by
Traditional Banks?” Journal of Economics and Business 100 (2018), pp. 43-54.
(https://doi.org/10.1016/j.jeconbus.2018.03.001)

Karlan, Dean, Jake Kendall, Rebecca Mann, Rohini Pande, Tavneet Suri, and Jonathan Zinman,
“Research and Impacts of Digital Financial Services,” NBER Working Paper 22633, September 2016.
(http://www.nber.org/papers/w22633)

Kearns, Michael, and Aaron Roth, The Ethical Algorithm: The Science of Socially Aware Algorithm
Design, Oxford University Press, 2019.
(https://global.oup.com/academic/product/the-ethical-algorithm-9780190948207?cc=us&lang=en&)

Kearns, Michael, and Aaron Roth, “Ethical Algorithm Design Should Guide Technology Regulation,”
report from Brookings Institution’s Artificial Intelligence and Emerging Technology Initiative, January
13, 2020.
(https://www.brookings.edu/research/ethical-algorithm-design-should-guide-technology-regulation/)

Kleinberg, Jon, Jens Ludwig, Sendhil Mullainathan, and Cass R. Sunstein, “Discrimination in the Age of
Algorithms,” Journal of Legal Analysis 10 (2018), pp. 113-174.
(https://doi.org/10.1093/jla/laz001)

Levine, Ross, “Finance, Growth, and Opportunity: Policy Challenges,” paper prepared for the Banco de
Mexico’s International Conference, “Challenges and Strategies for Promoting Economic Growth,”
October 31, 2009.
(https://faculty.haas.berkeley.edu/ross_levine/Papers/2010_Finance_Lessons_mexico.pdf)

Lipsman, Andrew, “Global Ecommerce 2019,” EMarketer, June 27, 2019.

MacCarthy, Mark, “Fairness in Algorithmic Decision-Making,” report from Brookings Institution’s

Morse, Adair, and Karen Pence, “Technological Innovation and Discrimination in Household Finance,”
Finance and Economics Discussion Series 2020-018, Board of Governors of the Federal Reserve System,
February 2020.
(https://doi.org/10.17016/FEDS.2020.018)

Raskin, Sarah Bloom, “Economic and Financial Inclusion in 2011: What It Means for Americans and Our
Economic Recovery,” remarks at a New America Foundation Forum, Washington, D.C., June 29, 2011.
(https://www.federalreserve.gov/newsevents/speech/raskin20110629a.htm)

Sahay, Ratna, Martin Čihák, Papa N’Diaye, Adolfo Barajas, Srobona Mitra, Annette Kyobe, Yen Nian
Mooi, and Seyed Reza Yousefi, “Financial Inclusion: Can It Meet Multiple Macroeconomic Goals?”
International Monetary Fund Staff Discussion Note No. 15/17, September 2015.
 15

(https://www.imf.org/en/Publications/Staff-Discussion-Notes/Issues/2016/12/31/Financial-Inclusion-Can-
it-Meet-Multiple-Macroeconomic-Goals-43163)

Taylor, Winnie, “Proving Racial Discrimination and Monitoring Fair Lending Compliance: The Missing
Data Problem in Nonmortgage Credit,” Review of Banking & Financial Law, 31 (2011), p. 199-264.
(https://www.bu.edu/rbfl/files/2013/09/ProvingRacialDiscrimination.pdf)

United Nations, U.N. Secretary-General’s Task Force on Digital Financing of the Sustainable
Development Goals, “People’s Money: Harnessing Digitalization to Finance a Sustainable Future,”
August 2020.
(https://unsdg.un.org/resources/peoples-money-harnessing-digitalization-finance-sustainable-future)

U.S. Census Bureau, “Disclosure Avoidance and the 2020 Census,” 2020.
(https://www.census.gov/about/policies/privacy/statistical_safeguards/disclosure-avoidance-2020-
census.html)

World Bank, The Global Findex Database, 2017.