Full text of Economic Insights (Federal Reserve Bank of Philadelphia) : Second Quarter 2021, Volume 6, Issue 2

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Second Quarter 2021
Volume 6, Issue 2

Is Rising Product Market Concentration
a Concerning Sign of Growing Monopoly
Power?

Q&A

Why Credit Cards Played a Surprisingly Big
Role in the Great Recession

Research
Update

Regional Spotlight

Data in Focus

Contents
Second Quarter 2021

A publication of the Research
Department of the Federal
Reserve Bank of Philadelphia
Economic Insights features
nontechnical articles on monetary
policy, banking, and national,
regional, and international
economics, all written for a wide
audience.

Volume 6, Issue 2

1

Q&A…

2

Is Rising Product Market Concentration
a Concerning Sign of Growing Monopoly Power?

with Lukasz Drozd.

Leena Rudanko explains why we might want to study the data more carefully before
deciding if it’s time to use antitrust regulations to increase market competition.

The views expressed by the authors are not
necessarily those of the Federal Reserve.
The Federal Reserve Bank of Philadelphia
helps formulate and implement monetary
policy, supervises banks and bank and
savings and loan holding companies, and
provides financial services to depository
institutions and the federal government. It
is one of 12 regional Reserve Banks that,
together with the U.S. Federal Reserve
Board of Governors, make up the Federal
Reserve System. The Philadelphia Fed
serves eastern and central Pennsylvania,
southern New Jersey, and Delaware.

7

Lukasz Drozd examines the links between zero-APR credit card offers and the Great
Recession’s persistent declines in employment and output.

18

24
Patrick T. Harker
President and
Chief Executive Officer
Michael Dotsey
Executive Vice President and
Director of Research
Adam Steinberg
Managing Editor, Research Publications
Brendan Barry
Data Visualization Manager
Antonia Milas
Graphic Design/Data Visualization Intern

ISSN 0007–7011

Connect with Us
We welcome your comments at:
PHIL.EI.Comments@phil.frb.org
E-mail notifications:
www.philadelphiafed.org/notifications
Previous articles:
https://ideas.repec.org/s/fip/fedpei.html
Twitter:
@PhilFedResearch
Facebook:
www.facebook.com/philadelphiafed/
LinkedIn:
https://www.linkedin.com/company/
philadelphiafed/

Why Credit Cards Played a Surprisingly Big Role
in the Great Recession

29

Regional Spotlight: Labor Market Disparities
The data show that Black workers are overrepresented in the lowest-paying
occupations. Paul Flora examines what big business can do to help a region
address this inequality.

Research Update
Abstracts of the latest working papers produced by the Philadelphia Fed.

Data in Focus
Aruoba-Diebold-Scotti Business Conditions Index.

About the Cover
First Bank of the United States
The nation’s first congressionally chartered bank, the First Bank of the United States,
opened on South 3rd Street in 1797. Like much architecture of this era, the First
Bank was designed in the neoclassical style. The Greek and Roman republics of
antiquity were a natural inspiration for citizens of this modern democracy, and many
new buildings adopted the tall, slender columns and classical friezes being unearthed
in Pompeii and Herculaneum. However, just as those ancient republics were riven
by political rivalries, so too was the new United States, thanks in part to this very
building. Secretary of the Treasury Alexander Hamilton campaigned hard for
a national bank as a way to steady the country’s finances. His biggest opponent,
Secretary of State Thomas Jefferson, opposed any centralization of economic
power. Hamilton won the debate, but the political factions founded by the two
secretaries would clash again when the bank’s charter was up for renewal in 1811.
Illustration by Antonia Milas.

Q&A…

with Lukasz Drozd, an
economic advisor
and economist here at
the Philadelphia Fed.

How did you become interested
in economics?
After communism ended in Poland, economics was a new thing. Before, you had
socialist economics, so even professors
were not trained in what you would call
economics. It was more how to do central
planning. The Ford Foundation was bringing top U.S. economists to Poland to retrain
professors, and then they were teaching
college students too. That was my first time
studying economics. I became hooked.

Did that play a role in your decision
to attend the University of Minnesota?
Definitely. Doctorate-level education in
economics at the time was not very good
in Poland. Going abroad seemed like
the only way to get good training, and
Minnesota was renowned for studying
macroeconomics. Doctorate-level education in Poland is much better now, but
that was the reality at the time.

Lukasz Drozd
Lukasz Drozd grew up in Poland. After
graduating from the Warsaw School of
Economics in 2001, he moved to the
United States to attend graduate school at
the University of Minnesota. He’s taught
economics at the University of Minnesota, the University of Wisconsin, and
the Wharton School of the University of
Pennsylvania. Since 2015 he’s been
a member of the Philadelphia Fed Research Department, where he specializes
in many topics, including the macroeconomic implications of consumer finance.
In this issue of Economic Insights, he
writes about the role zero-APR credit
cards played in the Great Recession.

Have you ever accepted a zero-APR
offer on a credit card?
Of course! (laughs) How do you think I survived graduate school? Many graduate
students used zero-APR credit cards. What
was amazing was that we were foreigners
with no credit history and yet we were getting flooded with offers. I was puzzled back
then and I am puzzled now. I was reluctant
to use these offers, but it was me and my
wife on a single stipend. At some point it
was really difficult to make ends meet, and
zero-APR credit cards came in handy.

Do you feel like you were clear-eyed
about the risks?
I tried to be responsible, or at least that is
how I like to think about it. (laughs) The
biggest increase in my debt was when I already had a job offer and knew I would be
able to pay it back. We paid off a lot by the
time Lehman Brothers collapsed, but there
was still some debt left, and I wanted to
transfer it to another zero-APR card, but
there were no offers anymore. I squeezed
my budget as much as possible to quickly
pay that down, but I thought about other
people who were not as lucky to have a job.
That inspired my research on this topic.

Q&A

2021 Q2

Some people reading your article may
think that when researchers start off
with a concept of a free market and
then study market frictions, they miss
other issues, such as somebody wanting a product that isn’t good for them.
I did not mean it this way. At some point,
we call people adults and they should
be allowed to be as free as possible, even if
they make a “wrong” decision for themselves. If people roll over their debt, this is
not such a huge deal. Maybe some of
them get caught by fees, and they suffer,
but they brought that upon themselves.
Maybe they will learn a lesson. What is
dangerous here is that the market may
dry up—as the Great Recession episode
illustrates—and all zero-APR customers at
the same time get hammered. This can
result in a major recession, which then
affects everyone. It is also more difficult
for people to foresee such risks. The art of
regulating the markets is balancing these
risks, and what I am saying in the article
is that policymakers should take it into
consideration.

So, it might not be a problem if an
individual makes the “wrong” decision
for themselves, but if too many
people act the same way, it could
bring the whole market down.
That’s right. And in this case that is
a stronger case for a policy intervention,
because now we are less paternalistic.
(laughs) We are just saying, don’t create
problems for the rest of us, and that is fair.
Some people may lose their freedom to
get free and easy credit, but you stabilize
the market and create less vulnerability.
There is an inherent trade-off in macroprudential regulation of financial markets,
and policymakers have to carefully balance
the pros and cons.

What else are you working on?
I like to combine theory with data to
uncover something that is not easy to see.
I’m looking right now at how automation
affects the division of income between
capital and labor. It is a very exciting topic
and quite timely. I hope we will have
an opportunity to discuss it next time.

Federal Reserve Bank of Philadelphia
Research Department

1

Photo: RainervonBrandis/iStock

Is Rising Product Market Concentration a Concerning Sign
of Growing Monopoly Power?

Big firms are coming to dominate markets, but that need not imply it’s time for
government to step in.

R

By Leena Rudanko
Economic Advisor and Economist
Federal Reserve Bank of Philadelphia.
The views expressed in this article are not
necessarily those of the Federal Reserve.

2

Federal Reserve Bank of Philadelphia
Research Department

ecent evidence suggests that product market concentration has
been on the rise in the U.S. since
the early 1980s.1 This means that sales
in a broad set of markets appear to be
concentrating in a smaller share of firms.
In other words, big firms are coming to
dominate markets. This rise in concentration concerns policymakers, as it suggests
that product markets are becoming less
competitive. Healthy competition, most
economists agree, is an important feature
of a well-functioning market, allowing
consumers to get the best possible prices,
quantity, and quality of goods and
services. And to ensure that competition
prevails, government should enact and
enforce antitrust regulations.

Rising concentration has coincided with
other, related long-run changes: rising
firm profit rates and markups, weak wage
growth (and a related decline in the share
of output paid as compensation to workers), low firm investment, low productivity
growth, and a decline in firm entry.
In this article, I review recent studies
related to this rise in concentration and
consider the economic significance of
this trend. I suggest a more positive interpretation of the evidence. It may be that
firms are growing larger due to a change
in productive technologies that favors
larger firm size, as development in information technologies is making it feasible
to operate on a larger—even global—scale.
In this context, the benefits of concentrating economic activity may outweigh

Is Rising Product Market Concentration a Concerning Sign?
2021 Q2

the costs of larger firms profiting from their market power.
But to fully understand the situation, we need more detailed
analyses of specific markets.

Interpreting the Evidence

Economists often interpret market concentration as a measure
of market power. It’s a straightforward analysis: Just use sales
revenues to calculate the share of market activity accounted
for by large firms.2
The U.S. Census Bureau tracks market concentration by
industry, providing measures of industry-level concentration
with comprehensive coverage of economic activity across
the U.S. This evidence reveals increased concentration since the
early 1980s, with product markets in most industries becoming
more concentrated (Figure 1). Between 1982 and 2012, the market
share of the top four firms increased from 14 to 30 percent in
the retail trade, 22 to 29 percent in the wholesale trade, 11 to 15
percent in services, and 39 to 43 percent in manufacturing.
In utilities and transportation, furthermore, the same measure
increased from 29 to 41 percent between 1992 and 2012.3
FIGURE 1

Top Firms Have Seen Their Share of Total Sales Grow

5-year percentage point increase in share of industry sales going to 20 largest
firms in each industrial sector, 1982–2012 for retail trade, wholesale trade, services,
and manufacturing, 1992–2012 for finance and utilities and transportation.
Finance
Retail trade
Wholesale trade
Utilities and transportation
Services
Manufacturing
0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

Sources: U.S. Economic Census; Autor et al. (2020).

Rising concentration appears to be an international phenomenon. Evidence from Organisation for Economic Co-operation
and Development (OECD) sources shows measures of concentration rising between 2001 and 2012 in Europe, with a 2 to 3
percentage point increase in the share of industry sales going to
the largest 10 percent of firms.4
Before drawing conclusions from this evidence, it is good to
recall that market concentration is an imperfect measure of
market power because it represents an outcome of competition
that in turn depends on various features of the market environment. Market power refers to the ability of a firm to influence
the prices it charges, which generally leads to higher prices than
in a competitive market. Although market power is generally
associated with concentrated product markets, a very competitive product market could also raise market concentration by
preventing all but the lowest-cost providers from entering. In
other words, the relationship between competition and concentration can go either way.5
It is also important to define a product market thoughtfully
when calculating market concentration. Concentration statistics

are generally aggregated, so they ignore more-detailed product
heterogeneity as well as the geographic aspect of product markets, which can be local rather than nationwide.
Due to these caveats, I see if two alternative indicators of
market power are consistent with the suggested increase in
monopoly power.

Alternative Measure No. 1: Profit Rates

During this increase in market concentration, the average
corporate profit rate for publicly traded firms has risen substantially, from 1 percent in 1980 to 8 percent in 2016.6 The increase
has been driven by growth in the profitability of the most profitable firms, rather than by an across-the-board increase in firm
profitability. The most profitable firms have become even more
profitable, attaining profit rates of 15 percent or more.
Extending these calculations to the broader universe of firms
is challenging, because information on the balance sheets of
privately held firms is private. However, studies using moreaggregated (and hence less-detailed) data covering the broader
universe of firms show a rising share of aggregate firm profits
since the early 1980s, too.7
These calculations suggest that the share of output paid
to workers as well as the share of output paid to capital have
both declined over this period. As a result, the share of output
going to firm profits has risen. We should remain cautious in
interpreting these intriguing findings, however, as calculating
the share of output paid to capital involves making a number of
assumptions that influence the results. Firms own various kinds
of capital but do not generally report estimates of the corresponding costs of holding these assets. Moreover, a share of
firms’ productive assets—such as software and product designs—
are not even physical, making it even more difficult to assess
the corresponding costs.8
Aggregated data have the benefit of allowing us to study
the evolution of profits over a longer time horizon. (The data
on publicly held firms are less suited to this purpose because,
earlier on, fewer firms chose to become publicly traded.) Thanks
to the longer time frame, we see that even though the average
of firm profits has risen since the 1980s, today’s average is not
particularly high relative to the broader period since World War
II. From this perspective, the changes in profitability are not
so alarming.
In any case, firm profitability is also an imperfect measure of
market power. Even though there are circumstances where
a fully competitive market should drive profits to zero, there are
natural circumstances where one would expect to observe
positive profits in a competitive market—for example, when firms
invest in capital up front and recover related profits later. This
capital may be tangible, like equipment and structures (and hence
more easily measured), or it may be intangible and thus harder
to measure. The growing importance for firms of intangible capital, which is associated with the development of new technologies
for producing goods and serving customers, may contribute to
the recent changes in profit rates.

Is Rising Product Market Concentration a Concerning Sign?

2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

3

Alternative Measure No. 2: Markups

Recently, economists have closely observed an alternative measure of market power, the price-cost markup that firms charge
(that is, the ratio of price to the cost of producing an additional
unit of output to sell). In a fully competitive market, competition
should drive prices down to zero markup. A monopoly producer,
on the other hand, would generally set a higher price, selling
fewer units at a positive markup.
Recent studies have found that markups, like profit rates, have
indeed increased: Based on evidence on public firms, the average
markup has risen significantly, from 20 percent in early 1980 to as
high as 60 percent in 2016 (Figure 2). And as with profits, this rise
in average markups was driven by high-markup firms growing
larger and taking over a larger share of industry sales.

FIGURE 2

Average Markup Rose as the Largest Firms Took a
Greater Share of Sales
Average markup for publicly traded firms, 1980–2016
60%
50%
40%
30%

Making Sense of It All

Profit rates and markups, in addition to the increase in concentration, suggest that market competition is declining. It seems
that sales in many markets are increasingly dominated by large
firms making greater profits through higher markups (while
leaving their workers with a smaller share of the pie). This suggests that the government needs to use antitrust law to limit the
growth in market power of large firms. However, there remain
reasons to be cautious when considering this evidence.
For one, the phenomenon is affecting not just the U.S., so it is
likely not driven by U.S.-specific policies. This suggests that the
underlying causes may be technological rather than institutional.
Perhaps modern technology, most notably the development of
information technologies, favors a larger scale of operations.
There may be social costs associated with firms profiting from
their market power, but if the technology has changed to favor
operating at a larger scale, the benefits of increased firm size may
outweigh the costs.11
Although this economywide evidence helps us observe broad
patterns, to ultimately understand what is happening we must
analyze individual industries and the concrete changes affecting
them. There is substantial heterogeneity across markets, after all.
To illustrate this point, I revisit the trends in market concentration
from two alternative perspectives. One perspective defines
a market as a narrow geographic area, instead of considering total
industry sales across the U.S., while the other defines a market
in terms of a product.

20%

The Importance of Localized Product Markets

10%
0%

1980

1990

2000

Sources: Compustat North America Fundamentals
Annual via Wharton Research Data Services (WRDS);
De Loecker et al. (2020).

2010

2016

Note: The average
markup is revenue
weighted.

Again, we must be cautious in interpreting these findings due
to the assumptions behind the measurement. Firms use different
types of inputs; taking them all into account appropriately poses
a challenge, especially when seeking to calculate markups across
a broad range of industries at the same time.9
If anything, the increase in markups appears to have been
larger than the increase in profit rates. We can reconcile the
magnitudes of the two effects (that is, the size of the increases
in profits vs. markups) if we consider the increase in overhead
expenses. If a growing share of firm costs take the form of overhead, markup measures tend to grow for that reason alone. Even
in a fully competitive environment where profits remain zero
throughout, an ongoing increase in overhead requires firms to
raise markups to cover these expenses.10

4

Federal Reserve Bank of Philadelphia
Research Department

Many product markets are local. Examples include grocery stores,
and the retail sector more generally, as well as many services,
like haircuts. In these product markets, transportation costs limit
the number of providers of goods and services that individual
consumers (or firms) can choose from in practice, an issue that
economists ignore when they calculate concentration measures
using all providers nationwide. It turns out that when we redefine a market as a localized geographic area, we no longer find
rising product market concentration.12
When a recent study defined a market as all firms in a specific
industry in a specific county, it found that average market concentration fell from 1990 to 2014, even while the more broadly
defined measures of concentration rose. Local product markets
have thus seen sales spreading out among more firms over this
period, rather than the opposite.
The finding of falling concentration in more narrowly defined
product markets holds across a broad range of industries. This
means that for product markets that are truly local, such as many
markets for services and retail, the nationwide statistics are misleading. On a national level, sales may be concentrating in
a smaller number of large firms, but in local product markets we
see the opposite.

Is Rising Product Market Concentration a Concerning Sign?
2021 Q2

Notes

How can we reconcile these two opposing trends? National
sales may be concentrating in a smaller number of large firms,
but these large firms may be expanding into a growing number
of local markets served by smaller local firms. Indeed, the study
found that the expansion of the largest firms explains much of
the divergence in these trends, while local competitors persist
despite the entry of these large firms into their local markets.

1 See Council of Economic Advisors
(2016) and Autor et al. (2020).
2 The two most common measures of market concentration are
the Herfindahl-Hirschman Index—
the sum of squared market shares
across firms in the market—and
the combined market shares of
the largest firms in the market.

The Importance of Product-Level Markets

Industry-level concentration statistics also aggregate over different types of products, sometimes more appropriately viewed as
separate product markets. A recent study looked at changes
in product-level markets, focusing on the retail trade and items
generally found in grocery stores.13
The study documented a growing number of product varieties
per product category available to households. Households’
options have thus increased, whatever may have happened
to firm competition during this time. And correspondingly, aggregate household spending has also spread out across varieties,
with households taking advantage of this increase in options.
Yet the study found that individual households are concentrating their spending on a shrinking number of varieties. Even
though the product space is expanding with options, suggesting
increasing competition in these markets, individual households
are self-selecting into smaller niche markets—making it less clear
whether competition in the relevant product markets is increasing
or decreasing.
To connect these product-level observations to competition
among firms, we must connect product varieties to the
relevant firms, something the study did not attempt. However, this example highlights the need to carefully consider the
changing competitive environment in individual markets before
drawing conclusions from broader aggregate-level patterns.

3 See Autor et al. (2020).
4 See Bajgar et al. (2018) and
Criscuolo (2018).
5 See Syverson (2019). There is
corroborating evidence that the
share of output paid as compensation to workers has declined
more in industries that are more
affected by rising concentration,
which is consistent with firms in
these industries retaining greater
profits. See Autor et al. (2020)
6 See De Loecker et al. (2020).
7 See Barkai (2020).
8 See Karabarbounis and Neiman
(2018).
9 See Basu (2019), Syverson
(2019), and Traina (2018).

Conclusion

Faced with evidence of rising concentration, profits, and markups,
it is hard to avoid thinking that the economy is seeing a widespread increase in monopoly power, which calls for increased
government intervention in markets. However, this conclusion
might not be warranted. Technological change may favor a
larger scale of operations, justifying larger firm size despite
corresponding increases in market power. What’s more, aggregated evidence can mask what is actually happening. The
bird’s-eye view has its benefits, but we need to consider specific
markets in more detail before taking action.

10 See De Loecker et al. (2020).
11 See Autor et al. (2020) and De
Loecker et al. (2020).
12 See Rossi-Hansberg et al. (2020).
13 See Neiman and Vavra (2020).

Is Rising Product Market Concentration a Concerning Sign?

2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

5

References
Autor, David, David Dorn, Lawrence F. Katz, et al. “The Fall of the Labor
Share and the Rise of Superstar Firms,” Quarterly Journal of Economics,
135:2 (2020), pp. 645–709, https://doi.org/10.1093/qje/qjaa004.
Bajgar, Matej, Giuseppe Berlingieri, Sara Calligaris, et al. “Industry
Concentration in Europe and North America,” OECD Productivity Working
Papers No. 18 (2019).
Barkai, Simcha. “Declining Labor and Capital Shares,” Journal of Finance,
75:5 (2020): pp. 2421–2463, https://doi.org/10.1111/jofi.12909.
Basu, Susanto. “Are Price-Cost Markups Rising in the United States?
A Discussion of the Evidence,” Journal of Economic Perspectives, 33:3
(2019), pp. 3–22, https://doi.org/10.1257/jep.33.3.3.
Criscuolo, Chiara. “What’s Driving Changes in Concentration Across
the OECD?” Organisation for Economic Co-operation and Development
working paper (2018).
Council of Economic Advisors Issue Brief. “Benefits of Competition and
Indicators of Market Power” (April 2016).
De Loecker, Jan, Jan Eeckhout, and Gabriel Unger. “The Rise of Market
Power and the Macroeconomic Implications,” Quarterly Journal of
Economics, 135:2 (2020), pp 561–644, https://doi.org/10.1093/qje/qjz041.
Karabarbounis, Loukas, and Brent Neiman. “Accounting for Factorless
Income,” in Martin Eichenbaum and Jonathan Parker, eds., NBER Macroeconomics Annual 2018, Vol. 33, 2018, University of Chicago Press,
pp. 167–228, https://doi.org/10.1086/700894.
Neiman, Brent, and Joseph Vavra. “The Rise of Niche Consumption,”
National Bureau of Economic Research Working Paper 26134 (2020),
https://doi.org/10.3386/w26134.
Rossi-Hansberg, Esteban, Pierre-Daniel Sarte, and Nicholas Trachter.
“Diverging Trends in National and Local Concentration,” in Martin
Eichenbaum and Erik Hurst, eds., NBER Macroeconomics Annual 2020,
Vol. 35, 2020, University of Chicago Press, https://www.nber.org/
books-and-chapters/nber-macroeconomics-annual-2020-volume-35/
diverging-trends-national-and-local-concentration.
Syverson, Chad. “Macroeconomics and Market Power: Context, Implications, and Open Questions,” Journal of Economic Perspectives, 33:3
(2019), pp. 23–43, https://doi.org/10.1257/jep.33.3.23.
Traina, James. “Is Aggregate Market Power Increasing? Production
Trends Using Financial Statements,” Stigler Center for the Study of the
Economy and the State Working Paper (2018).

6

Federal Reserve Bank of Philadelphia
Research Department

Is Rising Product Market Concentration a Concerning Sign?
2021 Q2

Photo: ideabug/iStock

Why Credit Cards Played
a Surprisingly Big Role in
the Great Recession

When economists and policymakers try to understand how a credit
crunch within the financial sector affects consumers, they usually
don’t think of the credit card market. They should.

T
By Lukasz Drozd
Economic Advisor and Economist
Federal Reserve Bank of Philadelphia.
The views expressed in this article are not
necessarily those of the Federal Reserve.

welve years after the Great Recession, one of the biggest economic
disasters of the modern era, economists still debate exactly what led to its
persistent declines in employment and
output. The basic narrative is clear: The
collapse of the housing price bubble
destroyed swaths of wealth, and the
ensuing credit crunch within the financial
system tightened borrowing constraints
on firms and households, depressing
consumption and investment across the

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

economy. But this basic narrative raises
further questions. Which was more
important, the destruction of wealth or
the tightening of borrowing constraints?
How much of the decline in output was
directly caused by these initial shocks, and
how much by the subsequent, dominolike propagation mechanisms? What were
these propagation mechanisms? Finally,
what does the Great Recession teach us
about the macroprudential regulation of
credit markets?1
Federal Reserve Bank of Philadelphia
Research Department

7

Economists are still answering these
questions, but one of their key insights is
that severed access to credit played a big
role.2 This insight has spurred renewed
interest in mapping the exact mechanisms
that drove the tightening of credit to
firms and households across different
markets, and in these mechanisms’ macroprudential ramifications.
When economists and policymakers try
to understand how a credit crunch within
the financial sector affects consumers,
they usually don’t think of the credit card
market. Historically, credit card borrowing
has been small, and credit card debt
involves a soft long-term commitment of
lenders to terms—an arrangement known
to be more stable and less prone to credit
supply disruptions than other forms
of debt—so it’s not obvious how, to the
detriment of borrowers, tightening of
credit conditions within the financial
system could severely contract available
credit, force early debt repayments, or
unexpectedly hike interest payments on
outstanding credit card debt.
But, as I will explain, by 2008 the credit
card market had grown enough to have
a notable impact on aggregate consumption demand. More importantly, by 2008
a large fraction of credit card debt was
de facto short-term debt. In particular, by
2008 many credit card borrowers were
reducing their interest rate payments by
moving balances from card to card to take
advantage of the then-ubiquitous zeroAPR promotional credit card offerings.3
After Lehman Brothers collapsed in mid2008, triggering a credit crunch within
the financial sector, the zero-APR offers
that had sustained the low cost of credit
card debt vanished from the market, leading to a massive and, for many borrowers,
unexpected interest rate hike on expiring
promotional debt. As I will argue, this led
such borrowers to cut their consumption
so they could repay debt early, which
contributed to the decline in consumption
demand during the Great Recession.
Policymakers should keep an eye on
promotional lending, and perhaps even
reserve a permanent spot for credit cards
in their macroprudential policy considerations. The COVID-19 crisis reminds us that
credit card borrowing remains fragile.

8

Federal Reserve Bank of Philadelphia
Research Department

The Rise of Credit Card Debt

Until the 1950s, credit cards were a form
of store credit, limited to purchases of
goods and services from a single issuing
merchant and too inconvenient to become
a major source of credit for households.
It was the success of the first generalpurpose charge card, issued by Diners
Club in the early 1950s, that inspired Bank
of America to combine a credit line with
a charge card and offer BankAmericard,
the first general-purpose credit card.
By the 1970s, more than 100 million such
cards were in circulation. Bank of America
began licensing its BankAmericard to
other banks that were issuing credit cards,
eventually spinning off BankAmericard as
a separate company called Visa.
But the revolution in payment technology did not spur a revolution in lending
right away. In the 1960s and 1970s, credit
cards were mainly used as a payment
instrument, and borrowing on credit cards
did not take off until the 1980s. What
delayed the growth of credit card lending
was the combination of high inflation and
usury laws that capped interest rates.4
With a tight cap on interest rates, and with
inflation driving up the cost of funds for
lenders, credit card lending struggled
to make a profit in the 1970s. In fact, by
the end of the decade, due to a doubledigit spike in inflation, many credit card
lenders found themselves on the brink
of collapse.5
The credit card industry was saved
in 1978, when the U.S. Supreme Court, in
Marquette National Bank of Minneapolis v.
First of Omaha Service Corporation, ruled
that if the interest rate cap in the state
where the bank is chartered is higher
than in the state where it offers its product
(in this case, a credit card), that bank may
charge a rate subject to the higher cap.
In other words, the court allowed a bank
to “export” its interest rate cap to other
states, which in the case of First of Omaha
meant that the company could issue
a credit card in Minnesota and charge an
interest rate in excess of Minnesota’s comparatively low cap of 12 percent.6
The broader implication of the Supreme Court ruling, however, was that, by
creating competition between states to
attract bank headquarters, it not only
relaxed usury laws for lucky issuers—such
as First of Omaha—but dismantled usury

laws for the credit card industry altogether.
Recognizing an opportunity for additional
tax revenue, South Dakota and Delaware
were the first states to raise their usury
laws’ ceilings on interest rates. Credit card
issuers did not wait long to relocate their
operations to these lender-friendly states,
and to this day their major offices can
be found in Wilmington, DE (for example, JPMorgan Chase), or Sioux Falls, SD
(for example, Citibank). To retain their
financial institutions, other states began
loosening their usury laws as well, and
today many states have no limit on credit
card interest rates.
Following the Marquette decision, credit
card borrowing steadily rose, notably
crowding out nonrevolving consumer
credit and gradually turning America
into a credit card debtor nation (Figure 1).
What fueled this expansion—especially in
the 1990s—was the steady spread of credit
card lending among lower-income and
riskier households. Credit card debt
per household relative to the annual median household income roughly doubled
every decade until the 2008 financial crisis,
topping 20 percent for a household with
FIGURE 1

Credit Card Borrowing Rose to
Prominence in the 1990s…

Credit card debt per family as a percentage
of median annual family income, 1984–2007
25%

20%

15%

10%

5%

0%

1984

1990

1995

2000

Sources: Board of Governors of the Federal Reserve
System (U.S.), G.19 Consumer Credit, Total Revolving
Credit Owned and Securitized, Outstanding [REVOLSL],
retrieved from FRED, Federal Reserve Bank of St. Louis;
https://fred.stlouisfed.org/series/REVOLSL, September
29, 2020. U.S. and Census Bureau, Current Population
Survey, March and Annual Social and Economic
Supplements, 2019 and earlier.

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

2007

at least one card by early 2008.7 Since much income growth over
the last several decades has occurred among the top 1 percent
of earners, and these earners do not borrow on credit cards as
much, the median rather than the mean household income
provides a better picture of how important credit card lending
had become for the majority of households.8 For low-income
households, credit cards often replaced far more expensive
options, such as “loan sharks” or payday lenders, and so the
growing availability of credit card debt has importantly contributed to the “democratization of credit” in the U.S. (Figure 2).
Although the Supreme Court ruling enabled the industry to
grow, it was, according to economic research, the convenience
of credit card debt and the rapid progress in information technology that drove the unprecedented, decades-long expansion
in credit card borrowing. Information technology affected both
the direct costs of lending and indirect costs associated with
debt collection—a less visible but equally important pillar
that sustains unsecured lending.9 By reducing lending costs that
creditors must cover to break even, technology increased the
affordability of credit card debt, fueled borrowing, and even
had a somewhat counterintuitive effect of increasing default risk
on a statistical dollar of outstanding credit card debt despite all

the progress in credit scoring technology.10 The overhaul of the
U.S. personal bankruptcy regulations in the Bankruptcy Reform
Act of 1979, which made discharging credit debt in court far easier,
and the overall increasing demand for debt by U.S. households were two other factors that contributed to the growth of
borrowing on credit cards on the demand side.
By the 2000s, credit card companies were making more money
from credit card lending than from merchant or interchange
fees. (Merchant or interchange fees are the fees paid by merchants
on each transaction settled using a credit card.) By 2003, of $95
billion in the credit card industry’s total revenues, interest revenue (that is, revenue earned from finance charges) amounted to
$65 billion, with lending-related penalty fees and cash advance fees
contributing another $12.4 billion. In comparison, merchant
fees contributed just $16 billion to revenue. Even after subtracting
$50 billion in costs and default losses, lending, though a more
costly part of the business, still came out on top in 2003. These
numbers did not change dramatically until 2008, and lending
maintained its prominent role.11 At that point, with its $1 trillion
in debt outstanding, credit card lending had grown big enough
to affect the entire economy.

FIGURE 2

...and Contributed to the Democratization of Credit in the U.S.
Growth of credit card borrowing by income quintile, 1989–2007
Card debt per cardholding family (1989=100)
500

2007
2004

First income quintile

1998

400
1995

2002

Second income quintile
300

1992

Third income quintile

2007

Fourth
income
quintile

Fifth
income
quintile

2007
2007

2007
200

100

1989

1989

1989

1989

1989

Shades indicate isolines of (fixed)
levels of credit card debt per family
0
27%
35%
45%
Share of families with at least one card

55%
65%
75%
85%
95%
Source: Board of Governors of the Federal Reserve System, Survey of Consumer Finances (SCF).

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

100%

9

The Origins of “Zero”

As the credit card market became saturated in the late 1990s
and early 2000s, competition for customers intensified.
Balance transfers and promotional-rate offers proliferated
as the leading marketing tools.12 The Marquette ruling, by
unifying regulations, set the stage for massive, nationwide
mail-marketing campaigns and permitted lenders to realize
economies of scale in marketing and processing. By the
end of the 1990s, an ever-increasing volume of mail-in offers
defined the credit card industry, and does so to this day.13
In the mid-1990s, Providian Financial Corporation
became the first issuer to drop a seemingly unprofitable
offer into people’s mail: a credit card with a zero APR on
balance transfers. This offer allowed consumers to transfer
their outstanding balance from any other credit card
account into their new Providian account ( just like any
other balance-transfer offer) and pay no interest for an
introductory period. The bank could profit later only if
consumers for some reason did not repay debt after the
promotional rate expired, or if they violated the “fine
print” of the contract, triggering a penalty rate reset.
At the time, Providian had a highly profitable credit
card business and was on the forefront of the industry’s
expansion to low-income customers.14 The new market
looked promising but risky: Lower-income customers had
lower balances and were more likely to default, making it
difficult for credit card companies to cover the fixed costs
of opening and operating their accounts. Such conditions
normally necessitate higher interest rates, but high interest
rates may also discourage borrowing, leaving lenders
exposed to default losses and bringing too little interest
income on borrowing to make a profit.
Litigation against Providian in the late 1990s, which led
to the credit card industry’s largest Office of the Comptroller of the Currency (OCC) enforcement action, offers
a unique glimpse into how the company approached the
marketing of credit cards and what led it to offer zero
APR. This evidence suggests that behavioral psychology
rather than competition was the key factor behind the
invention of “zero.”
For example, in one of 12 internal memos to Providian’s
top executives that became public in the course of litigation,
Andrew Karr—the founder of Providian, its CEO, and later
a strategic adviser to the company—described in this way
how the company planned to profit on subprime customers: “Making people pay for access to credit is a lucrative
business wherever it is practiced…. Is any bit of food too
small to grab when you’re starving and when there is nothing else in sight? The trick is charging a lot, repeatedly,
for small doses of incremental credit.”15 The memo confirmed that the company was indeed concerned that raising
interest rates to compensate for higher lending costs might
backfire, and it explained why its marketing strategy was
aimed at mitigating this issue by obscuring the true cost of
debt from borrowers—as the litigation showed.
Karr later echoed the content of this memo in a rare
interview by explaining that he suggested zero promotional

10

Federal Reserve Bank of Philadelphia
Research Department

rates to Providian executives because seeing “zero” leads
borrowers to “believe what they want to believe,” which
one can infer he saw as being conducive to increased borrowing by consumers even if competition ensues.16
Providian paid a hefty price for its aggressive practices
in the early 2000s, but the litigation was about the company’s deceptive practices, not the products themselves,
and zero APR lived on to become the hallmark of the credit
card industry in the 2000s.17 Providian’s approach may not
be representative of the industry as a whole, but recent
research shows that behavioral psychology provides a good
explanation for the widespread use of zero APR.

The Behavioral Economics of Zero APR

Zero-APR offers challenge standard economic theory
featuring rational consumers. When Boston Fed economist
Michal Kowalik and I studied a standard model of credit
card lending in which lenders can offer any introductory
promotional rate to (rational) borrowers, we found that,
under standard economic theory, rates should fully price
in the risk of default and the cost of funds, resulting in flat
interest schedules and few introductory promotions.
Although the model can generate introductory promotional offers when the default risk of a borrower is expected
to decline sharply, such occurrences are rare, and under
plausible conditions the model does not even come
close to accounting for the large volume of such offers
in the data.
The key reason is that rational consumers are best
served by prices that closely reflect the true resource cost
of lending them money—which, among other items, includes the compensation to the lender for bearing the risk
that the borrower may default under some circumstances
(default risk premium).18 In particular, when the price of
credit is too low for a period of time, as is the case with
a promotional introductory offer, credit card customers
borrow too much: The benefit that accrues to them exceeds
the cost implied for the lender by the fact that the customer
may default on this amount later on. Rational borrowers
realize that this cost must eventually be passed onto them
because lenders must break even, and for this reason they
prefer flat schedules. The key virtue of a competitive market
is that competition between lenders drives down prices to
a common break-even point, which implies that, to attract
customers, lenders must offer the product that best suits
the customer.19
So why do we keep finding zero-APR offers in our mailboxes? Research in behavioral economics may have the
answer. This research suggests that zero APR may indeed
let people “believe what they want to believe.”
The best-known piece of evidence supporting this theory
comes from an influential albeit unpublished study by
University of Maryland economists Lawrence M. Ausubel
and Haiyan Shui. In collaboration with a major credit
card issuer, Ausubel and Shui performed a unique study
of credit card marketing that involved an experiment of

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

simultaneously mailing several different offers to tease out
customer bias for promotional introductory offers. In
cooperation with the issuer, the researchers tracked the
activity on the accounts after the offers were accepted.
To assess the customer’s choice, they also calculated the
interest rate payments the customer would have faced
had they chosen a different offer.
Surprisingly, customers on average chose what the
rational model would deem a “wrong” offer. More
importantly, they were not simply accepting offers at random, possibly ignoring the offered terms; to the contrary,
customers were attracted to offers that minimized their
immediate interest payments, even if choosing such offers
cost them more later. Ausubel and Shui concluded that
consumers fail to accurately predict their future behavior,
which leads them to erroneously think that they are picking the best offer.
In particular, Ausubel and Shui have demonstrated that
the results of their experiment are consistent with naiveté
hyperbolic discounting—the leading theory of consumer
myopia put forth by Harvard economist David Laibson and
earlier shown successful in addressing several puzzling
observations in consumer credit markets. According to this
theory, borrowers have an idealistic view of their future
self, incorrectly believing that their future self will have
almost no debt and pay no interest. This idealistic view
leads them to underestimate the burden of the interestrate hike associated with the expiration of an introductory
offer. As a result, they prefer introductory offers and underestimate the significance of these offers’ high reset rates.
Ausubel and Shui also found that this theory fits the
data well for parameter values consistent with earlier work
with this model. By assuming the same parameter values,
Michal Kowalik and I showed that this theory can explain
the widespread use of zero APR in the U.S. credit card
market, where competitive lenders are free to design the
credit card offers they send to consumers.20
Of course, the fact that the leading theory of consumer
myopia may explain the U.S. credit card market doesn’t
imply that the entire population is prone to zero-APR offers.
It may be that credit card customers who did not accept
a zero-APR offer in the Ausubel and Shui study are the rational ones and only the overoptimistic found promotional
offers particularly attractive, leading to selection bias among
study respondents. Their finding only shows that there are
enough customers prone to these offers to drive promotional lending.

The Makings of a Perfect Storm

Before my work with Kowalik, surprisingly little was known
about the prevalence of promotional offerings in the U.S.
credit card market and their effect on the functioning of
the market. Data provided by the three credit bureaus lack
interest rates, and their data are the most comprehensive
commercially available source of information about credit
market activity in the U.S. Without interest rate data, we

can’t study promotional activity as carefully as we would
like, and consequently we did not know much about it.21
In our work, for the first time, we could uncover evidence
of the widespread and intricate use of promotional lending
owing to the availability of regulatory account-level
data covering the majority of the general-purpose credit
card accounts in the U.S. right before the 2008 financial
crisis—a data set large and detailed enough to characterize promotional lending in the economy as a whole.
Although we suspected some use of introductory offers to
reduce interest rate payments, what we found surpassed
our expectations.22
By 2008, the credit card market was essentially in the
grips of zero-APR offers, with a vast amount of credit card
debt being de facto short-term debt and prone to disruptions during crises. In particular, as of the first quarter
of 2008, we found that 35 percent of credit card debt held
on general-purpose credit card accounts was on promotional terms with rates close to zero, with an average
yearlong expiration of the promotional terms. Among
prime borrowers with a good credit history (that is, a credit
score above 670), the percentage was even higher: 42
percent. When we factored in a typical fee of 3 percent for
transferring funds at the time, and a rate on the promotional debt near zero, promotional accounts provided
an average discount of about 10 percentage points from the
average reset rate on those accounts—and a similar discount
vis-à-vis the average interest rate paid on nonpromotional
credit card debt. This was true for both the prime segment
and the whole market, which shows that promotional debt
importantly contributed to making credit card debt
affordable to borrowers.
Crucially, balances that fed promotional accounts before
the crisis were mainly transfers of debt from other accounts—
as opposed to debt accrued from purchases using the new
card.23 This finding implies that consumers were not only
using promotion on a massive scale but also moving
funds to reduce the interest rate paid on their credit card
debt, something we corroborated by showing that some
borrowers were chaining promotional cards to extend the
duration of promotional rates. As for the market as a whole,
this observation is key, since it implies that at the onset
of the Great Recession the affordability of credit card debt
hinged on an uninterrupted flow of promotional offers.
Three percent on zero APR may not sound like enough
for lenders to be able to break even, but lenders too could
profit on the promotional offers, since they attract borrowers who later may have to pay the reset rate on the account
when they are unable to switch to a new card or when their
rate resets early because they violated the contract’s “fine
print.” Basic economic theory implies that lenders put up
with this behavior precisely because they could break even
and borrowers preferred such offers.24 As explained earlier,
a competitive market leads to the outcome that best suits
the borrowers, and the evidence suggests that promotional
offers suited them best.

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

11

The Perfect Storm

The September 2008 collapse of Lehman Brothers, by triggering
a panic within the financial sector, set the stage for a perfect storm
in the credit card market. Starved for liquidity, and expecting a
recession that would harm consumers, the financial sector tightened the supply of credit to firms and households, whereupon
many credit card borrowers suffered because of their heavy
reliance on the constant flow of promotional offers to reduce
interest payments.
The data show that preapproved and prescreened promotional
balance-transfer offers had fallen more than 70 percent by mid2008 (Figure 3), suggesting that many credit card borrowers who
had previously hoped to transfer balances onto a promotional
account might have had trouble getting a new card during the
crisis.25 Consistent with the decline in mail-in offers, promotional
balance transfers dived, falling 70 percent by early 2009 (Figure
4). Not surprisingly, the fraction of promotional debt began to
decline, bottoming out in 2011 at about half of its precrisis value
of 35 percent. This was true for all accounts in our sample as well
as just those with a good credit history (Figure 5).
Kowalik and I further investigated to what extent the deteriorating financial health of the lenders might have driven the decline,
which is a proxy for the impact of the crisis on each individual
lender’s financing conditions. We analyzed how the county-level
credit card lender health index, which we constructed, correlates
with the decline in the share of promotional debt and balance
transfers in each county. If a credit card issuer has a large
presence in a U.S. county, and if its financial health worsens
more than that of creditors in other counties, we should see
a larger decline in balance transfers and promotional debt in that

county relative to other counties. This we did see, indicating
that the financial sector’s credit crunch was in part responsible
for the declining share of promotional balances.26
Of course, other factors may have also contributed to the
decline in the availability of promotional credit card offers, and
our research design does not allow us to quantitatively assess
the relative importance of those factors. The most straightforward reason is that lenders might have discontinued promotional
offers because they themselves feared a recession-related spike
in defaults on credit card debt due to falling incomes and employment. Credit card debt is unsecured, which is one reason why
default rates spike during recessions. By reducing credit during
a recession, banks can avoid losses from rising defaults.

Connecting the Dots

The second half of 2008 was a turning point for credit card
borrowing overall.27 Credit card debt, despite rising steadily for
decades, fell markedly relative to median household income and
other types of consumer debt (Figure 6). In our work, Kowalik
and I have hypothesized that the decline in credit card borrowing relative to the previous trend was driven by the collapse of
promotional offerings, which then led credit card customers to
either default on debt more frequently or make early debt repayments, contributing to the decline of aggregate demand during
the Great Recession.
It’s difficult to assess exactly
See How Chaining of Zerohow much the collapse in proAPR Offers May Amplify
motional offerings contributed
a Recession.
to the decline in credit card

FIGURE 3

FIGURE 4

FIGURE 5

Recession Brought an End to the
Abundance of Zero-APR Offerings…

…Promotional Balance Transfers
Collapsed…

…and the Share of Promotional
Card Debt Began to Shrink…

12,000

50%

50%

40%

40%

30%

30%

Number of mail-in preapproved credit card solicitations
with a promo balance transfer offer, in millions,
2007–2013

10,000

Promotional balance transfers as a percentage of
credit card debt outstanding, annualized, 2008–2013

Promotional credit card debt as a percentage of credit
card debt outstanding, all accounts and accounts
with at least a 670 credit score, 2008–2013

8,000
6,000
20%

20%

10%

10%

Credit score
670+
All accounts

4,000
2,000
0

Jan
2008

Jun
2009

Source: Mintel Compremedia
Inc., Direct Mail Monitor Data.

12

Oct
2013
Note: Gray bar
indicates recession.

Federal Reserve Bank of Philadelphia
Research Department

0%

Jan
2008

Jun
2009

Source: Federal
Reserve, Y14M.

Oct
2013
Notes: The OCC/Y14M sample
includes six largest banks,
eight banks in total; gray bar
indicates recession.

0%

Jun
Jan
2008 2009

Source: Federal
Reserve, Y14M.

Oct
2013
Notes: The OCC/Y14M sample
includes six largest banks,
eight banks in total; gray bar
indicates recession.

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

borrowing or consumption demand. In
the data, both the collapse in offerings
and the decline in borrowing or consumption involve changes that triggered
the recession and changes that were the
product of the recession. For example,
such a decline may have been partly due
to a hike in defaults on credit card debt
triggered by job losses during the Great
Recession, which was part of a feedback
mechanism rather than the trigger.
To isolate the contribution of the withdrawal of promotional offers, Kowalik and
I used an economic model of the credit
market that replicates what happened
during the Great Recession. Using the
model, we asked, what would have happened had fairly priced promotions held
steady during the recession?
The results we found were troubling.
According to the model, there would have
been no decline from the precrisis trend
in the ratio of median personal income
to credit card debt per adult. Indeed, the
ratio would have gone up (Figure 6).
But was the collapse in promotional
offerings enough to affect consumption
demand across the economy? To find out,
we also compared the model’s ratio of aggregate consumption to disposable income

to the same ratio in the data. This ratio
is an imperfect proxy for consumptiondepressing factors other than declining
income, which may be a product of
the recession itself and not a trigger. We
estimated that, according to our model,
peak-to-trough, the decline in the availability of promotional offerings contributed
to about a quarter of the decline in this
ratio from 2009 through 2011.28

The COVID-19 Crisis:
A Silent Alarm?

Fast-forward to 2020 and both balancetransfer activity and zero-APR offers have
not rebounded to their respective 2008
levels (Figure 7), which has made the
credit card market more stable. We do not
know why the decline has persisted for
so long after the recession, but the most
prosaic explanation may be the right one:
Having had a bad experience with zero
APR, borrowers avoided such offers after
the Great Recession. Nonetheless, promotional activity and balance transfers did
not disappear and may rise again in the
future, which raises the question: How
has promotional credit card lending fared
during the more recent COVID-19 crisis?

... which Turned the Decades-Long
Borrowing Boom into a Bust

25%
24%

Promotional balance transfers as a percentage of
credit card debt, annualized, 2018–2020
50%

40%

30%

20%

10%

0%

Feb
2018

Feb
2020

Source: Federal Reserve System, Y14M.
Notes: The data in Figure 4 pertain to a smaller sample of eight banks and are not directly comparable to
data in the figure; gray bar indicates recession.
FIGURE 8

... and the Share of Promotional
Debt Also Began to Shrink

Note: Model predictions
are approximate due to
minor differences in data
formatting and sources.
For detailed analysis, see
my work with Kowalik
(2019).

50%

40%

30%
Credit
score
670+

23%
22%

20%

All
accounts

21%
Model Prediction:
Just recession

20%
19%

10%

0%

18%
17%
16%
15%

Oct
2020

Promotional credit card debt as a percentage of credit
card debt outstanding, all accounts and accounts
with at least a 670 credit score, 2018–2020

FIGURE 6

Actual and model-predicted credit card debt per
adult as percentage of median personal income,
2001–2014

FIGURE 7

The COVID-19 Recession Had
a Similar Effect on Balance
Transfers…

2001

2008

Feb
2018

Feb
Oct
2020 2020

Source: Federal Reserve System, Y14M.

Data
Model Prediction:
Recession and
2014 Zero-APR crisis

Note: Gray bar indicates recession.

Sources: Board of Governors of the Federal Reserve System, G.19 Consumer Credit, Total
Revolving Credit Owned and Securitized, Outstanding [REVOLSL], FRED, Federal Reserve Bank
of St. Louis; https://fred.stlouisfed.org/series/REVOLSL. U.S. and Census Bureau, Current
Population Survey, March and Annual Social and Economic Supplements, 2019 and earlier.

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

13

How Chaining of Zero-APR Offers May Amplify a Recession
Here is how credit card borrowers chain
promotional zero-APR offers: First, they
charge purchases on their zero-APR credit
card. Then, before the card’s new, higher
base rate kicks in, they apply for another
zero-APR card and transfer the debt to the
new card. In effect, they are extending
the duration of the promotional interest rate.
For economists, there is nothing unusual
about “chaining” of promotional credit card
offers. It’s just another instance of borrowing
via rolling over short-term debt obligations—
a widespread practice across the economy.
However, this type of borrowing is known
to be vulnerable to disruptions of the credit
supply and may trigger or contribute to
a recession, which is why it is monitored

and regulated as part of macroprudential
policies. (See Endnote 1 for an explanation
of macroprudential regulation.)
Here is how it happens. Consider a situation
where a borrower takes out a long-term
loan and borrows for two periods from Bank
A using two different strategies. In the first
situation (Case I), debt does not become due
until Period 3, and Bank A cannot request
funds early. In the second situation (Case II),
the borrower “chains” lenders by repaying
Bank A with funds borrowed from Bank B
in Period 2. Both cases lead to the same
outcome when credit flow is uninterrupted:
The borrowers borrow in the first period
and repay in the third, effectively borrowing
funds for a duration of two periods. But the

second case (Case II) is vulnerable to a credit
supply disruption and the first is not. Say, for
example, that in Period 2, banks decide not
to lend as much, so that the borrower in Case
II has a hard time finding another lender
(Figure 9). This borrower will be forced to
repay debt early and cut down on their
spending on purchases of goods and services. Alternatively, the borrower, unable
to make the payment, will default on their
debt, in which case Bank A will be hurt and
will possibly reduce the credit supply to
other customers, which will hurt their consumption (or investment). In both situations,
if banks, amid a recession, withdraw funds
from the market to reduce their losses, they
may amplify that recession due to reduced
consumption or investment demand.

FIGURE 9

Chain Lending and How It Might Amplify a Recession
PERIOD 1
Case I
Long-term
Lending

$

PERIOD 2

PERIOD 3

Bank A loans to Borrower
for two periods to spend
Bank A on goods and services
$

$

Case II
Chain
Lending

Borrower
spends on
goods and
services

$

Bank A

Bank A loans to Borrower
for one period to spend
on goods and services

Bank B loans to
Borrower for one
period, which
Borrower uses to
repay Bank A

$

Borrower repays
Bank A in Period 3

$
$

Borrower repays
Bank B in Period 3

and Borrower
spends on goods
and services

But What Might Happen If the Credit Supply Is Disrupted in Period 2?
Because the Borrower does not need a loan
in Period 2, the outcome is the same.
Case I
Long-term
Lending

Case II
Chain
Lending

14

$

$

$

Bank A

Bank A loans to Borrower
for one period to spend
on goods and services

Federal Reserve Bank of Philadelphia
Research Department

$

If Bank B rejects the loan application, Borrower has
two options that will both be recessionary:
Option 1
Reduce spending on goods and services to
repay Bank A’s loan early

Option 2
Default on loan repayment, forcing Bank A to
reduce loans to other borrowers and hurting
creditworthiness in the economy

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Notes

The answer to this question is important because it
helps us address another question: How vulnerable is
promotional lending during a recession not triggered
by a financial crisis?
Credit markets fared well during the crisis, but as
for promotional credit cards, the data from the
first half of 2020 are troubling because it suggests that
promotional offerings might have been similarly
depressed, and the overall impact of this development
was lower because the starting volume was lower.
In particular, the data for the first half of 2020 show
a modest 4 percentage point decline in the share
of promotional debt, which fell from about 22 percent
prior to the Great Recession to about 18 by October
2020 (Figure 8). Worryingly, the decline in promotional balance transfers is almost as striking as during
the Great Recession, falling by over 50 percent peak
to trough, albeit from a volume that is less than a third
of that at the onset of the Great Recession (Figure 7).
As more data become available, we will be able
to examine this crisis more closely, but the early
indication is that promotional credit card borrowing
is vulnerable during recessions that do not involve
a financial crisis.

1 Macroprudential regulation of credit markets is
an approach to regulation guided by the principle
of mitigating risks to the financial system
and the economy as a whole. Stress testing
of banks to ensure their resilience in times of
distress is an example of macroprudential
regulation implemented in the aftermath of the
Great Recession by the Dodd–Frank Wall Street
Reform and Consumer Protection Act of 2010.
2 For an accessible discussion, see the Economic Insights article by my colleague Ronel Elul.
See also the work by Gilchrist, Siemer, and
Zakrajsek; Mondragon; and Aladangady. The
study by Mian and Sufi initially suggested
a modest role for credit markets.
3 The annual percentage rate (APR) refers to
the annual rate of interest charged to borrowers for carried-over balances after the credit
card statement closes. In a zero-APR offer, the
credit card holder pays no interest on charges
to their credit card for an introductory period.
Thereafter, a new APR kicks in for the outstanding balance and all future charges.

Conclusion

The 2008 financial crisis taught us that the proliferation of zero APR on balance transfers can threaten
economic stability. The COVID-19 crisis reminds
us that a significant fraction of debt still originates as
promotional transfers, and nothing prevents that
fraction from rising again. At the very least, then, the
volume of zero-APR debt and balance transfers should
be carefully monitored. The credit card market is
now large enough to affect the whole economy, and
policymakers should keep it in mind when they craft
their regulatory agendas.
Laissez faire theory holds that, if both sides of
a market transaction decide to use a particular credit
instrument, this credit instrument is likely socially
beneficial, and the government shouldn’t regulate it.
But the research points to the role of flawed human
psychology in the rise of zero-APR offers, and this
should raise concerns about the application of the
laissez faire principle. What’s also worrisome is that
the way lenders break even falls outside of the contract. For example, consumers may get hit with the
reset rate when they cannot find another offer, or
when they violate the contract’s “fine print,” thus
exposing themselves to an imminent and unexpected
rate hike on debt. The contract doesn’t specify how
much they will pay for borrowing—a departure from
how most loan contracts are written. Such an arrangement is conducive to abuse and predatory practices.

4 Usury laws govern the maximum amount of
interest that can be charged on a loan.
5 High levels of fraud and defaults also contributed to low profits during this early period.
See Evans and Schmalensee (page 72) for
more details.
6 According to the court's unanimous opinion,
the National Bank Act of 1864 created a path
toward a national consumer lending economy.
7 See Livshits, MacGee, and Tertilt; Drozd and
Serrano-Padial; and Athreya, Tam, and Young
for detailed analyses of the growth of credit
card borrowing in the U.S. Jaromir Nosal and
I provide an analysis of how a decline in the
fixed cost of lending leads to an expansion in
access to lending.
8 According to data from the Survey of Consumer Finances (SCF), the mean credit card
debt per household whose income is close to
the median (that is, between the 40th and
59th percentiles of income) has been almost
identical to the overall mean credit card debt
per household between 1989 and 2007. This is
not true for income. In the same data source,
income per household close to the mean was
lower by 50 percent in 1989 and by 70 percent

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

15

in the 2000s. This shows that income is more concentrated at the top of
the income distribution than debt, and hence the burden of debt for the
majority of households is best captured by using median income instead
of mean income. For more details on the income growth among top earners, see the Economic Insights article by my colleague Makoto Nakajima.
9 See my work with Ricardo Serrano-Padial for more details on the connection between debt collection and credit card lending.
10 See my work with Ricardo Serrano-Padial. “Default risk” measures
the fraction of debt that lenders expect will not be paid back because
some credit card borrowers may default, and debt may be deemed
nonrecoverable. Because credit card debt is unsecured, and debt can
be discharged in court, default risk is substantial on credit cards. One
measure of default risk is the so-called charge-off rate on a credit card
debt portfolio: the fraction of debt charged off the creditor’s books after
180 days of being delinquent during a period, net of any recovered and
previously delinquent debt over the same period.
11 See the article by James J. Daly. In their monograph, Evans and
Schmalensee report very similar numbers in the credit card market for
the preceding year.
12 In 1991, Capital One became the first issuer to introduce a balancetransfer offer.
13 Evans and Schmalensee report that, by the 2000s, 75 percent of
credit accounts were initiated via prescreened offers.
14 The company was known to use advanced (for that time) modeling to
thoroughly understand the behavior of its customers. See online post by
Andrew Becker.
15 The memos were published by the San Francisco Chronicle after a yearlong legal battle with Providian to make them public. Excerpts of the 12
released memos can be found in the Chronicle article by Sam Zuckerman.
16 The interview appears in the 2004 PBS Frontline documentary “Secret
History of the Credit Card.” The documentary can be found at https://
www.pbs.org/wgbh/pages/frontline/shows/credit/.
17 Providian settled in 2000 for $105 million after already reimbursing
customers at least $300 million. The company was sold to Washington
Mutual in 2005 for approximately $6.5 billion. Its credit card portfolio
at the time amounted to 10 million card holders.
18 In the case of credit cards, the risk of default is significant given
the unsecured nature of credit card debt. Borrowers may default on unsecured debt by filing for bankruptcy. Since the borrower does not have
to offer collateral as potential compensation to the lender, the lender is
at risk of never receiving payment on the principal amount owed. And,
even if the borrower does not file for bankruptcy, their (usually) small
amount of debt may make debt collection prohibitively costly for the
lender, leading to a widespread phenomenon of “informal bankruptcies.”
For more details, refer to my work with Ricardo Serrano-Padial.

16

Federal Reserve Bank of Philadelphia
Research Department

19 Consider a situation in which a borrower is encouraged to draw an
additional dollar of debt because of a low promotional interest rate. Suppose this borrower will default on this additional dollar of debt when they
lose their job. In a competitive market, the borrower must compensate
the lender by paying more interest in the future for the additional risk of
default because the lender must break even on average. In the model, the
additional benefit from the dollar when the borrower becomes unemployed outweighs the cost of paying more interest when the borrower
keeps their job—an effect that makes introductory offers suboptimal for
rational borrowers.
20 The evidence that Ausubel and Shui found has been confirmed in
other studies, which point to similar biases in investing and saving
behavior. For example, in a closely related study, Agarwal et al. show
that credit card customers prefer low-annual-fee cards, even though
they end up later overpaying in interest in excess of the fee.
21 Promotional lending can be studied using proprietary account-level
data, but such data are typically not available at a scale that allows
researchers to see how borrowers transfer balances across accounts
and lenders. Prior to the Dodd–Frank Act, the OCC was the only institution
we knew of that possessed an account-level data set covering a large
fraction of U.S. credit card accounts. The Federal Reserve System later
acquired this data set for its stress testing. The numbers reported in this
article come from this merged data set.
22 These data are collected by the Federal Reserve System under
Dodd–Frank to help the Fed conduct stress testing of banks. The data are
available for economic research conducted within the Federal Reserve
System, providing new insights into the inner workings of credit markets.
23 See figures in my work with Kowalik.
24 Our data does not allow us to calculate lender costs on the account
level, and it is not possible to precisely assess profitability of zero-APR
accounts. Initially, lenders do lose money on zero-APR accounts in the
data, but over time we did not find any indication that these accounts
are less profitable than comparable accounts.
25 Prescreened offers mailed out by credit card issuers are the main tool
of customer acquisition in the credit card market, so the number of mailedout solicitations is a reliable measure of the credit card industry’s hunger
for new customers. Evans and Schmalensee report that in the early 2000s
about 75 percent of credit accounts were initiated via prescreened offers.
26 Using a different approach, Keys, Tobacman, and Wang reach a similar
conclusion.
27 Credit card borrowing takes place when a credit card holder does
not pay back the balance in full after the credit card statement closes
and “rolls over” the outstanding balance to the next billing cycle
(partly or fully).
28 Consumption demand was an important factor in the Great Recession.
Mian and Sufi have shown that the decline in consumption was key to
explaining the fall in aggregate demand.

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

References
Agarwal, Sumit, Souphala Chomsisengphet, Chunlin Liu, and Nicholas
S. Souleles. “Do Consumers Choose the Right Credit Contracts?”
Review of Corporate Finance Studies, 4 (2015), pp. 239–257, https://doi.
org/10.1093/rcfs/cfv003.
Aladangady, Aditya. “Housing Wealth and Consumption: Evidence From
Geographically-Linked Microdata,” American Economic Review, 107:11
(2017), pp. 3415–3446, https://doi.org/10.1257/aer.20150491.
Aruoba, S. Borağan, Ronel Elul, Şebnem Kalemli-Özcan. “How Big Is the
Wealth Effect? Decomposing the Response of Consumption to House
Prices,” Federal Reserve Bank of Philadelphia Working Paper No. 19-6
(2019), https://www.philadelphiafed.org/consumer-finance/mortgagemarkets/how-big-is-the-wealth-effect-decomposing-the-response-ofconsumption-to-house-prices.

Haltom, Renee, and John A. Weinberg. “Does the Fed Have a Financial
Stability Mandate?” Economic Brief EB17-06, Federal Reserve Bank of
Richmond, June 2017.
Keys, Benjamin, Jeremy Tobacman, and Jialan Wang. “Rainy Day Credit?
Unsecured Credit and Local Employment Shocks,” unpublished manuscript (2019).
Laibson, David, Andrea Repetto, and Jeremy Tobacman. “Estimating
Discount Functions with Consumption Choices Over the Lifecycle,” NBER
Working Paper 13314 (2007), https://doi.org/10.3386/w13314.
Laibson, David. “Golden Eggs and Hyperbolic Discounting,” Quarterly
Journal of Economics, 112:2 (1997), pp. 443–478, https://doi.org/10.1162/
003355397555253.

Athreya, Kartik, Xuan S. Tam, and Eric R. Young. “A Quantitative Theory of
Information and Unsecured Credit,” American Economic Journal: Macroeconomics, 4:3 (2012), pp. 153–183, https://doi.org/10.1257/mac.4.3.153.

Livshits, Igor, James C. Mac Gee, and Michèle Tertilt. “The Democratization of Credit and the Rise of Consumer Bankruptcies,” Review of
Economic Studies, 83:4 (2016), pp. 1673–1710, https://doi.org/10.1093/
restud/rdw011.

Ausubel, Lawrence M., and Haiyan Shui. “Time Inconsistency in the
Credit Card Market,” University of Maryland Department of Economics
working paper (2005).

Mian, Atif, and Amir Sufi. “What Explains the 2007–2009 Drop in Employment?” Econometrica, 82:6, (2014), pp. 2197–2223, https://doi.org/
10.3982/ECTA10451.

Becker, Andrew. “The Battle over ‘Share of Wallet’,” Public Broadcasting
Service (PBS), Frontline, November 23, 2004, https://www.pbs.org/wgbh/
pages/frontline/shows/credit/more/battle.html.

Mondragon, John. “Household Credit and Employment in the Great
Recession,” Kellogg School of Management at Northwestern University,
mimeo (2018).

Daly, James J. “Smooth Sailing,” Credit Card Management, 17:2 (2004),
pp. 30–33.

Nakajima, Makoto. “Taxing the 1 Percent,” Federal Reserve Bank of
Philadelphia Economic Insights (Second Quarter 2017), pp. 1–10, https://
www.philadelphiafed.org/the-economy/macroeconomics/taxing-the-1percent.

Drozd, Lukasz, and Michal Kowalik. “Credit Cards and the Great Recession:
The Collapse of Teasers,” unpublished manuscript (2019), https://www.
lukasz-drozd.com/uploads/4/3/1/8/43183209/drozd-kowalik-v31.pdf.
Drozd, Lukasz, and Jaromir Nosal. “Competing for Customers: A Search
Model of the Market for Unsecured Credit,” unpublished manuscript
(2008).

Zuckerman, Sam. “How Providian Misled Card Holders,” San Francisco
Chronicle, May 5, 2002, https://www.sfgate.com/news/article/HowProvidian-misled-card-holders-2840684.php.

Drozd, Lukasz, and Ricardo Serrano-Padial. “Modeling the Revolving
Revolution: The Debt Collection Channel,” American Economic Review,
107 (March 2017), pp. 897–930, https://doi.org/10.1257/aer.20131029.
Elul, Ronel. “Collateral Damage: House Prices and Consumption During
the Great Recession,” Federal Reserve Bank of Philadelphia Economic
Insights (Third Quarter 2019), pp. 7–11, https://www.philadelphiafed.org/
the-economy/macroeconomics/collateral-damage-house-prices-andconsumption-during-the-great-recession.
Evans, David S., and Richard Schmalensee. Paying with Plastic: The Digital
Revolution in Buying and Borrowing. Cambridge, MA: The MIT Press, 2005.
Gilchrist, Simon, Michael Siemer, and Egon Zakrajsek. “The Real Effects
of Credit Booms and Busts: A County-Level Analysis,” discussion paper,
mimeo (2017).

Why Credit Cards Played a Surprisingly Big Role in the Great Recession
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

17

Photo: Faina Gurevich/iStock

Regional Spotlight

Labor Market Disparities

A region’s big businesses can help monitor racial progress in the labor market.

By
Paul R. Flora
Manager of Regional
Economic Analysis
Federal Reserve Bank
of Philadelphia.
Thank you to Sydney
Lodge, who performed
the initial data collection and analysis for
this research.
The views expressed
in this article are not
necessarily those of
the Federal Reserve.

18

I

n the wake of last summer’s Black Lives
Matter protests, Black CEOs forcefully called
on large corporations to act. As Darren
Walker, president of the Ford Foundation and
a member of the board of Pepsi, told the New
York Times, “Boards should hold themselves and
management accountable for specific objectives
around recruitment, retention and promotion
of African-Americans and other minorities…. Only
when companies and management are accountable in ways that are quantifiable will we see real
systemic transformation of corporate America.”1
Data from large Philadelphia companies do
show that when compared with the region’s
population distribution, non-Hispanic Black
workers are underrepresented in high-wage
occupations and overrepresented in lowwage occupations.2 Philadelphia is not unlike
other regions in this respect.3

Federal Reserve Bank of Philadelphia
Research Department

Of course, responsibility for these disparities
does not lie solely at the door of the business
community. Labor market disparities have many
causes. Some of these causes may reach back
to a person’s early life experiences or to those of
prior generations. Past lack of access to neonatal
health care, insufficient pre-K and K-12 education,
or lack of career-training opportunities may
limit an individual’s life prospects. Moreover,
historic patterns of discrimination in employment, housing, lending, and criminal justice
have lowered the incomes and wealth of prior
generations. These patterns can lower the
human capital (and incomes) of subsequent
generations and reduce intergenerational wealth
transfers. These channels
can account for some
See Why Racial
current labor market
Inequality
disparities.
Matters

Regional Spotlight: Labor Market Disparities
2021 Q2

A Straightforward Measure of Inequality

Fortunately, we already have a straightforward measure of workplace diversity for
See Enforcement
most of a region’s larger businesses: occuEfforts Motivate
pational data by race and ethnicity (known
Basic Data
as EEO-1 data) from the Equal Opportunity
Collection
Employment Commission (EEOC).5
EEO-1 data show that Black workers are underrepresented in
Philadelphia’s higher-paying occupations and overrepresented
in lower-paying occupations. According to Census data, Philadelphia’s Black population accounted for a little over 20 percent
of the region’s total population (ages 15 to 74) in 2018, but just 10 to
12 percent of managers, professionals, and craft workers in the
EEO-1 data were Black.6 In Philadelphia, these occupations command annual average salaries of $140,000, $86,000, and $59,000,
respectively.7 And Black workers held just 4 percent of executive
positions, which pay an average of $251,000. In contrast, Black
workers held 35 percent of laborer positions, which pay $35,000
on average, and nearly half of service worker positions, which
pay an average of only $31,000.

As of 2018, Black workers
were underrepresented
in Philadelphia’s higherpaying occupations and
overrepresented in lowerpaying occupations.
According to the EEO-1 data, Philadelphia’s lack of diversity is
comparable to that of six other regions of a similar size.8 The
patterns of over- and underrepresentation across the 10 broad
occupational categories look very similar in all seven regions
(Figure 1). Just like in Philadelphia, Black workers in these other
regions are overrepresented in low-wage occupations and underrepresented in high-wage occupations.

FIGURE 1

Philadelphia's Black Workers Are Overrepresented in
Occupations with Low Average Salaries
This pattern is comparable to that of six regions of similar size.
Percentage of Black workers (ages 15–74) in 10 broad occupational categories
with mean annual salary, Philadelphia MSA, 2018

Percent of Black Workers by Occupation Relative to Percent of Black Population

Higher pay →

Philadelphia

← Lower pay

Moreover, although a lack of diversity within a region’s largebusiness community, defined as firms with at least 100 employees,
may indicate ongoing underlying problems, such disparities
might still exist even if we removed all of these causes.4
Still, a call to focus on workplace diversity within a region’s
business community may serve two valuable functions. First,
providing firms with a benchmark against which to compare
themselves may encourage them to participate in regional efforts
to address the underlying causes of these disparities. Second,
a well-designed benchmark could serve as a useful metric of
overall progress if the region were to adopt a comprehensive plan
to address the many complex underlying problems that engender racial and ethnic inequality.

Percent Black Population

Executives $251k
Managers $140k
Professionals $86k
Craft Workers $59k
Technicians $53k
Sales Workers $47k
Office Workers $43k
Operatives $39k
Laborers $35k
Service Workers $31k

0%
10%
30%
40%
← Underrepresented Overrepresented →
Representation

Philadelphia

50%

Executives

50%
40%

Service Workers

Managers

30%
20%

Laborers

Professionals

10%

Operatives

Craft Workers

Office Workers

Technicians
Sales Workers

Other MSAs Relative to Philadelphia MSA
Standardized for differing ratios of percent Black population.
Atlanta

Boston

Houston

Miami

Phoenix

Washington

Sources: U.S. Bureau of Labor Statistics (BLS), U.S. Census Bureau, and U.S. Equal Opportunity
Employment Commission (EEOC).

Regional Spotlight: Labor Market Disparities
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

19

Broad Groupings May Mask
Greater Workforce Inequality

Researchers have long used EEO-1 data to
study labor market inequality. For example,
using this data, sociologists Philip Cohen
of the University of Maryland and Matt
Huffman of the University of California,
Irvine, found that significant underrepresentation of Blacks in management jobs
was more likely for firms operating in
regions with a high proportion of Black
workers in the labor market.9
However, EEO-1 data do not reveal how
race and ethnicity are distributed within
each category’s wide range of salaries.
Salaries for sales workers, for example, are
highly skewed (Figure 2). Retail salespersons and cashiers are the two largest
subcategories of sales workers, representing 54 percent of all sales workers. They
are also the two lowest-paid subgroups.
With an average salary of $30,810, retail
salespersons earn just $7,600 more
than the $23,240 drawn by the lowestpaid cashiers.

The typical worker in these two subgroups, however, earns far less than
the average $47,056 salary for the sales
worker category. Meanwhile, the typical
sales representative—the next largest
subcategory, with 18 percent of all sales
workers—earns more than twice that of
retail salespersons and more than three
times that of cashiers. If sales representatives are more likely to be white, and
if retail salespersons and cashiers are more
likely to be Black, then there may be
further racial disparities even within this
broad category. More-refined occupational
categories would help researchers quantify
racial inequality within each category.
Still, the EEO-1 data have the significant
advantage of being a full count of large
employers in a region. For the Philadelphia
region, EEO-1 data counted nearly 1.3
million employees in 2018, capturing over
40 percent of the 2.9 million regional
employees estimated by the Bureau of
Labor Statistics. The next best alternative
is occupational data from the American

Community Survey, but its 2018 sample
size from the Philadelphia region contained only 36,200 households. The EEO-1
data, despite their limitations, provide the
best starting point for tracking economic
inequality by race and gender—but greater
occupational detail would make them
even better.

Large Businesses Can Be More
Transparent

Few firms release their EEO-1 reports. Just
Capital, a nonprofit that supports corporate responsibility to the public at large,
reports that “as of January 2021, only 6.3%
of America’s largest corporations disclosed the type of intersectional data that
could be derived from an EEO-1 Report.”10
(The report does note that out of 931
companies, the number of firms disclosing
their employment diversity had nearly
doubled from 32 in December 2019 to 59
in January 2021.)
As one example of transparency, the

FIGURE 2

Why Racial Inequality Matters
The current pandemic-induced recession has provided
a grim reminder that recessions increase hardship for
low-wage workers. Because minorities are disproportionately represented in low-wage occupations, they
are more vulnerable to the negative impacts of recessions. As Mellody Hobson, the co-chief executive of
Ariel Investments and a board member at JPMorgan
and Starbucks, told the New York Times last summer,
“We’ve been disproportionally affected in layoffs and
unemployment.”22
Subsequent research from the Federal Reserve Bank
of Philadelphia supports Hobson’s claim. In September 2020, the fourth in a series of COVID-19 surveys
of U.S. consumers found that “Black respondents,
those who earn less, younger respondents, and women all continue to report experiencing more adverse
[economic] effects.”23
These survey results were further corroborated by
research using monthly data for the three states of
the Third District. This research found that “three
groups of workers with no more than a high school
diploma—Black men, Black women, and Hispanic
women—have experienced far worse outcomes
during the current downturn.”24

20

Federal Reserve Bank of Philadelphia
Research Department

Average Salaries Vary Within the Sales Workers Category

Without more-refined occupational categories, it's impossible to know if Black
workers are overrepresented in each category's lower-paying occupations.
Average annual salary and number of workers by subcategory within the sales workers occupational category, out of 265,320 sales workers, Philadelphia MSA, May 2019
Average Annual Salary
$100,000
Misc. group 1
(incl. financial sales agents)
$80,000

Sales
reps

Misc. group 2
(incl. insurance agents)
$60,000
Retail sales supervisors
$40,000

Retail
salespersons

Misc. group 3
(incl. telemarketers)

Cashiers

$20,000

$0

0
10k
20k
30k
40k
Number of Workers in Sales Occupation

Source: U.S. Bureau of Labor Statistics.

Regional Spotlight: Labor Market Disparities
2021 Q2

50k

60k

70k

80k

Federal Reserve Bank of Philadelphia releases its own EEO-1 data
as part of an annual Office of Minority and Women Inclusion
report. The 2020 report revealed that Black employees make up
about 15 percent of the total workforce at the Philadelphia Fed.11
While the racial and ethnic representation is more balanced than
among all the region’s businesses, a similar pattern of over- and
underrepresentation is present.

Alternatively, Large Businesses Can Collectively
Track Progress

Firms that are reluctant to release their EEO-1 reports can nonetheless collectively track progress—ideally on an annual basis
and with greater occupational detail than the EEO-1 data provide.12 For example, signatories to Boston’s 100% Talent Compact,
designed in partnership with the Boston Women’s Workforce
Council (BWWC) to close gender and racial wage gaps, pledge to
share their anonymized EEO-1 data for aggregate analysis.13
While the Talent Compact ensures anonymity and also requests pay data, firms are only asked to submit data every other
year and only for the 10 EEO-1 occupational categories.14 This
prompted the Talent Compact to warn against comparing one
year’s results against prior-year results15 and to further warn that
wage gaps in any category may be overstated.16
For most large businesses, these data are already available

in-house. And the Talent Compact’s success indicates a willingness on the part of some large businesses to participate. Adding
greater occupational detail would provide a far more accurate
picture of equal opportunity in the region, and submitting data
annually would increase the metric’s value for tracking.
However, while tracking the data is needed to gauge progress,
ultimately program efforts will be needed to make progress.
And these programs would likely benefit by paying close attention
to research into the causes of inequality.

Making Progress

A long literature across multiple disciplines examines the “stubborn persistence of racial differences in socioeconomic outcomes.”17 This literature attempts to identify the main causes of
and most effective remedies for these unequal outcomes.
For example, Harvard economist Roland G. Fryer Jr. argues
that most of the racial differences in socioeconomic outcomes
would be greatly reduced if educational opportunities and
school quality were equalized from early childhood through
high school.18
Economists Patrick Bayer of Duke University and Kerwin Kofi
Charles of Yale note that as earnings inequality has risen in
the U.S., the gap has widened further among Black men.19 Welleducated Black men at the top of the earnings distribution have

Enforcement Efforts Motivate Basic Data Collection
The EEOC is charged with enforcing laws that
prohibit discrimination against employees
and job applicants on the basis of race, color,
religion, national origin, sex (including
pregnancy, transgender status, and sexual
orientation), age (40 or older), disability, or
genetic information. To assist its investigations
into specific allegations of discrimination,
the EEOC has collected mandated, basic data
from most private and public employers
since the mid-1960s. Because EEO-1 data are
protected by confidentiality requirements,
however, the data have been heavily aggregated to obscure individual firm reports.
As a result, publicly available measures of
occupational employment diversity are not
nearly as precise and robust as they might be.
In September 2016, the EEOC decided to begin
collecting more detail on a new EEO-1 form—
revised to include employees’ earnings and
hours worked by pay band. Known as Component 2 pay data, the additional confidential
detail provides greater insight into a firm’s
pay patterns. The EEOC is collecting this data
so that it can ensure pay equity on the basis
of sex, race, and ethnicity.

In August 2017, the Office of Management
and Budget (OMB) under the new administration of President Trump blocked this effort.
However, the National Women’s Law Center
and other plaintiffs took the OMB to court and
compelled it to allow the EEOC to resume this
data collection.
Component 2 data were collected only in 2018
and 2019; these data are now being evaluated
by an expert panel convened by the National
Academies of Sciences, Engineering, and
Medicine. The panel will assess the quality
and value of the compensation data for
various uses and will recommend improvements to the methodology.
Although nondisclosure rules prevent the
public disclosure of details from the additional
data, the panel may identify new summary
measures that provide greater insight for
a region while preserving employer and employee confidentiality. Meanwhile, further
collection of Component 2 data is suspended
pending the panel’s recommendations.

Regional Spotlight: Labor Market Disparities
2021 Q2

At the state level, California recently passed
legislation to effectively require what the EEOC
has suspended. States normally maintain
their own body of equal employment opportunity laws, which complement or augment
federal laws. In September 2020, California
Governor Gavin Newsom signed a law that
requires employers to annually submit
a report identical to an EEO-1 with Component
2 pay data, beginning in 2021.
However, pay data attached to the current
10 broad occupational categories offer little
insight into potential pay inequities for the
same reasons that the broad groupings make
it difficult to quantify occupational employment diversity.
Moreover, one can’t easily test for the
presence of pay inequities, much less explicit
discrimination. Statistically significant pay
inequities by race or gender may be suggestive
of discrimination, but they are not definitive
without data on additional individual employee
characteristics such as education, tenure,
and performance assessments.

Federal Reserve Bank of Philadelphia
Research Department

21

Notes

benefited, as have well-educated white men, while
lower-skilled Black men face ever fewer job options
and contend with higher incarceration rates than
their white counterparts. Because Black men are
overrepresented in the lower and middle portions of
the earnings distributions, they write, “race-neutral
economic changes and related public policy decisions that improve the prospects of all workers in the
lower and middle portions of the earnings distributions will have the side effect of reducing racial
economic inequality.”
Harvard economist Raj Chetty and his coauthors
have documented that Black-white income gaps
persist in the U.S. primarily because of significant
differences in the outcomes of Black and white
men from families with similar incomes.20 They
recommend creating greater opportunities for Black
children—especially for Black boys—and fostering
substantial improvements in neighborhood environments to increase upward mobility and narrow the
outcome gap.
Harvard sociologists Mario L. Small and Devah
Pager have gone beyond the traditional economic
models of racial discrimination, which focus on
preferences for discrimination or on statistical
discrimination. They argue that discrimination also
arises from unconscious bias and from racism that
has become integrated into organizational practices
or been written into laws.21 For example, they have
shown that the negative wealth effects from the
redlining practices of the past persist across generations in minority neighborhoods. They also note
that the cumulative effect of everyday discrimination
has negative consequences for physical and mental
health, as does the perception (or presumption) of
discrimination when the reality itself may be uncertain. Economic research into these and other
examples of institutional discrimination, they argue,
could uncover still more opportunities to address
the underlying causes of inequality.
This and related studies suggest that remedies for
these and other factors are needed. If these efforts
succeed, improvement should be evident in overarching measures of inequality, such as occupational
diversity in a region’s large businesses. And the
perception of discrimination should diminish.

1 Gelles (2020).
2 For the remainder of this paper, “Black” will refer to
non-Hispanic Black.
3 Unless otherwise noted, “region” and “metro area” refer
to an official metropolitan statistical area (MSA). Analysis in
this article is based on data for each MSA as delineated in
the Office of Management and Budget Bulletin No. 13-01,
issued February 28, 2013. This article truncates these official
names to the names of their largest principal cities.
4 For example, a preference within one demographic group
for working in a small business or owning one’s own business
might lower the proportional representation of that demographic in large businesses.
5 EEO-1 reports must be filed with the EEOC each year by
employers with at least 100 workers. The data alone cannot
prove hiring discrimination or pay inequities.
6 General population data were drawn from 2018 Census
Bureau estimates.
7 Salaries reported in this article were drawn from 2019
Bureau of Labor Statistics wage data by occupation for the
Philadelphia metro area.
8 In addition to Philadelphia, we examined the next three
larger regions (Houston, Miami, and Washington, D.C.), and
the next three smaller regions (Boston, Miami, and Phoenix)
based on population.
9 Cohen and Huffman (2007).
10 Vaghul (2021).
11 Office of Minority and Women Inclusion (2020).
12 A guarantee of anonymity might encourage reluctant
firms to share their data as part of such an effort.
13 There were 38 signatories at the launch of the Compact
in 2013; as of February 2021, there were 250.
14 Data are shared with Boston University’s Hariri Institute
for Computing, which employs a secure multiparty computation process to ensure anonymity and protect private
information while it analyzes data for wage gaps.
15 Since firms are only asked to submit data every other year,
but the report is issued annually, the composition of the
sample changes year to year, thus weakening its value as
a tracking tool. Specifically, “comparing reporting cycles
should not be done because of the variation in the number

22

Federal Reserve Bank of Philadelphia
Research Department

Regional Spotlight: Labor Market Disparities
2021 Q2

of employees represented and the types of jobs they fill” (Boston Women’s
Workforce Council Report, 2019).

Perspective,” Quarterly Journal of Economics, 135:2 (2020), pp. 711–783,
https://doi.org/10.1093/qje/qjz042.

16 As an example, “Our sample likely includes the overrepresentation
of women professionals in lower paying professions. This means that
even if there were full wage equity in lower paying professions (such as
nursing), and full wage equity in male-dominated higher paying positions
(such as physicians), our sample might still reflect a larger wage gap
than exists in the entire Boston workforce” (Boston Women’s Workforce
Council Report, 2019).

Cohen, Philip N., and Matt L. Huffman. “Black Underrepresentation in
Management Across U.S. Labor Markets,” Annals of the American Academy
of Political and Social Science, 609:1 (2007), pp. 181–199, https://doi.org/
10.1177%2F0002716206296734.
Equal Employment Opportunity Commission. Job Patterns for Minorities
and Women in Private Industry (EEO-1), https://www.eeoc.gov/statistics/
employment/jobpatterns/eeo1.

17 Bayer and Charles (2018).
Fryer Jr., Roland G. “Racial Inequality in the 21st Century: The Declining
Significance of Discrimination,” in Orley Ashenfelter and David Card, eds.,
Handbook of Labor Economics, Volume 4b. Amsterdam, the Netherlands:
Elsevier B.V., 2011, pp. 855–971, https://doi.org/10.1016/S01697218(11)02408-7.

18 Fryer (2011).
19 Bayer and Charles (2018).
20 Chetty et al. (2020).

Gelles, David. “Corporate America Has Failed Black America,” New York
Times, June 6, 2020.

21 Small and Pager (2020).
22 Gelles (2020).

National Academies of Sciences, Engineering, and Medicine. Panel to
Evaluate the Quality of Compensation Data Collected from U.S. Employers
by the Equal Employment Opportunity Commission through the EEO-1
Form. https://www.nationalacademies.org/our-work/panel-to-evaluatethe-quality-of-compensation-data-collected-from-us-employers-by-theequal-employment-opportunity-commission-through-the-eeo-1-form.

23 Akana (2020).
24 Wardrip (2021).

References
Akana, Tom. “CFI COVID-19 Survey of Consumers—Wave 4 Tracks How
the Vulnerable Are Affected More by Job Interruptions and Income
Disruptions,” Federal Reserve Bank of Philadelphia Report (2020), https://
www.philadelphiafed.org/consumer-finance/consumer-credit/cfi-covid19-survey-of-consumers-wave-4-updates.
Bayer, Patrick, and Kerwin Kofi Charles. “Divergent Paths: A New Perspective in Earnings Differences Between Black and White Men Since
1940,” Quarterly Journal of Economics, 133:3 (2018), pp. 1459–1501,
https://doi.org/10.1093/qje/qjy003.
Boston Women’s Workforce Council. 2019 Boston Wage Gap Report (2019).
Bureau of Labor Statistics. May 2019 Metropolitan and Nonmetropolitan
Area Occupational Employment and Wage Estimates, Philadelphia-Camden-Wilmington, PA-NJ-DE-MD, https://www.bls.gov/bls/blswage.htm.
Bureau of the Census. Annual County and Resident Population Estimates
by Selected Age Groups and Sex: April 1, 2010 to July 1, 2018, https://
www.census.gov/data/tables/time-series/demo/popest/2010s-counties-detail.html.
California. Senate Bill 973, https://leginfo.legislature.ca.gov/faces/billTextClient.xhtml?bill_id=201920200SB973.
Chetty, Raj, Nathaniel Hendren, Maggie R. Jones, and Sonya R. Porter.
“Race and Economic Opportunity in the United States: An Intergenerational

National Women’s Law Center. Resource Collection: EEO-1 Pay Data
Collection, https://nwlc.org/resources/resource-collection-eeo-1-pay-datacollection/.
Office of Management and Budget. Revised Delineations of Metropolitan
Statistical Areas, Micropolitan Statistical Areas, and Combined Statistical
Areas, and Guidance on Uses of the Delineations of These Areas, OMB
Bulletin No. 13-01 (February 28, 2013), https://www.whitehouse.gov/
sites/whitehouse.gov/files/omb/bulletins/2013/b13-01.pdf.
Office of Minority and Women Inclusion. 2020 Report to Congress on the
Office of Minority and Women Inclusion, Federal Reserve Bank of Philadelphia, 2020, https://www.philadelphiafed.org/-/media/frbp/assets/
institutional/about-us/philadelphia-fed-omwi-report-2020.pdf.
Small, Mario L., and Devah Pager. “Sociological Perspectives on Racial
Discrimination,” Journal of Economic Perspectives, 34:2 (2020), pp.
49–67, https://doi.org/10.1257/jep.34.2.49.
Vaghul, Kavya. “A Small Fraction of Corporations Share Diversity Data,
but Disclosure Is Rapidly on the Rise,” Just Capital, January 19, 2021.
Wardrip, Keith. “Who’s Employed in the Early Months of the COVID-19
Recession? An Analysis by Education, Race, Ethnicity, and Gender,” Federal
Reserve Bank of Philadelphia Report (2021), https://www.philadelphiafed.
org/community-development/workforce-and-economic-development/
whos-employed-in-the-early-months-of-the-covid19-recession.

Regional Spotlight: Labor Market Disparities
2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

23

Research Update

These papers by Philadelphia Fed economists,
analysts, and visiting scholars represent
preliminary research that is being circulated
for discussion purposes.

A Survey of Fintech Research and Policy Discussion

The views expressed in these papers are
solely those of the authors and should not
be interpreted as reflecting the views of
the Federal Reserve Bank of Philadelphia
or Federal Reserve System.

The Behavioral Relationship Between Mortgage
Prepayment and Default

The intersection of finance and technology, known as fintech, has
resulted in the dramatic growth of innovations and has changed the
entire financial landscape. While fintech has a critical role to play in
democratizing credit access to the unbanked and thin-file consumers
around the globe, those consumers who are currently well served
also turn to fintech for faster services and greater transparency.
Fintech, particularly the blockchain, has the potential to be disruptive
to financial systems and intermediation. Our aim in this paper is to
provide a comprehensive fintech literature survey with relevant
research studies and policy discussion around the various aspects of
fintech. The topics include marketplace and peer-to-peer lending;
credit scoring; alternative data; distributed ledger technologies;
blockchain; smart contracts; cryptocurrencies and initial coin offerings;
central bank digital currency; robo-advising; quantitative investment
and trading strategies; cybersecurity; identity theft; cloud computing;
use of big data, artificial intelligence, and machine learning; identity
and fraud detection; anti-money laundering; Know Your Customers;
natural language processing; regtech; insuretech; sandboxes; and
fintech regulations.

An implication of the dual trigger theory of default is that mortgage
borrowers who experience an unexpected financial reverse will
prepay their mortgage rather than default if their equity in the house
is positive. We test this idea with a new data set created by matching
mortgage servicing records and credit bureau records to classify
prepayments by what happens subsequently. In particular, we can
identify a subset of prepayments that seems consistent with the dual
trigger theory. If the theory is correct, these prepayments should
exhibit similarities to defaults in the data set rather than other prepayments. We test this idea and find that these prepayments are in
fact more closely related to defaults than to other prepayments.
However, our data also support a role for strategic default. Understanding these relationships may be critical in predicting mortgage
default when house prices decline after a long period of increases.
While our work is only a first step in this direction, we believe that
a better understanding of how prepayments may be driven by financial reverses would be valuable for participants in and regulators of
mortgage markets.

WP 20-21 Revised. Franklin Allen, Imperial College London; Xian Gu,
Durham University; Julapa Jagtiani, Federal Reserve Bank of Philadelphia Supervision, Regulation, and Credit Department.

WP 21-12. Arden Hall, Federal Reserve Bank of Philadelphia Supervision, Regulation, and Credit Department; Raman Quinn Maingi,
New York University.

24

Federal Reserve Bank of Philadelphia
Research Department

Research Update
2021 Q2

Is a Friend in Need a Friend Indeed? How Relationship Borrowers Fare During the COVID-19 Crisis
We analyze loan contract terms, investigating whether relationship
borrowers fare better or worse than others in times of need, using
the COVID-19 crisis as a quasi-natural experiment. COVID-19 is superior
to prior crises for such analysis because its public health and government restrictions shocks directly harm borrowers, rather than banks.
Our data set includes Y-14Q, covering syndicated and nonsyndicated
loans and small and large firms, unlike some other data sets. We
find the dark side of relationships dominates across four relationship
measures, 14 COVID-19 shocks, and PPP participation. There are
limited pockets of bright-side findings associated with smaller firms
and smaller banks.
WP 21-13. Allen N. Berger, University of South Carolina and Wharton
Financial Institutions Center European Banking Center; Lars Norden,
Getulio Vargas Foundation; Gregory F. Udell, Indiana University;
Christa H.S. Bouwman, Texas A&M University, ECGI Wharton Financial
Institutions Center; Raluca A. Roman, Federal Reserve Bank of Philadelphia Supervision, Regulation, and Credit Department; Teng Wang,
Federal Reserve Board of Governors.

“Sort Selling”: Political Polarization and Residential
Choice
Partisanship and political polarization are salient features of today’s
society. We merge deeds records with voter rolls and show that political
polarization is more than just “political cheerleading.” Descriptively,
homeowners are more likely to sell their homes and move when their
next-door neighbors are affiliated with the opposite political party.
We use a novel, new-next-door-neighbor identification strategy along
with rich demographic control variables and time-by-geography fixed
effects to confirm causality. Consistent with a partisanship mechanism, our results are strongest when new next-door neighbors (i) are
more likely to be partisan and (ii) live especially close by. Our findings
help explain increases in political segregation, improve our understanding of residential choice, and illustrate the importance of political
polarization for economic decision-making.
WP 21-14. W. Ben McCartney, Purdue University and Federal Reserve
Bank of Philadelphia Consumer Finance Institute Visiting Scholar;
John Orellana, Federal Reserve Bank of Philadelphia Supervision,
Regulation, and Credit Department; Calvin Zhang, Federal Reserve
Bank of Philadelphia Supervision, Regulation, and Credit Department.

Factor Models with Local Factors—Determining the
Number of Relevant Factors
We extend the theory on factor models by incorporating “local” factors
into the model. Local factors affect only an unknown subset of
the observed variables. This implies a continuum of eigenvalues
of the covariance matrix, as is commonly observed in applications. We
derive which factors are pervasive enough to be economically important and which factors are pervasive enough to be estimable using the
common principal component estimator. We then introduce a new
class of estimators to determine the number of those relevant factors.
Unlike existing estimators, our estimators use not only the eigenvalues
of the covariance matrix but also its eigenvectors. We find that
incorporating partial sums of the eigenvectors into our estimators leads
to significant gains in performance in simulations.
WP 21-15. Simon Freyaldenhoven, Federal Reserve Bank of Philadelphia Research Department.

Owner-Occupancy Fraud and Mortgage
Performance
We identify occupancy fraud—borrowers who misrepresented their
occupancy status as owner-occupants rather than investors—in
residential mortgage originations during the housing bubble. Unlike
previous work, we show fraud was broadly based and appeared in the
GSE market and bank portfolio loans, not just private securitization;
accounting for that fraud increases the effective investor share by
more than one-third. Occupancy fraud allowed riskier borrowers to
obtain lower interest rates, and we show that fraudulent borrowers
performed substantially worse than similar owner occupants and
declared investors, constituting nearly one-sixth of the share of loans
in default by the end of 2008. Their defaults were also much likelier
to be “strategic.”
WP 19-53 Revised. Ronel Elul, Federal Reserve Bank of Philadelphia;
Aaron Payne, University of Pennsylvania; Sebastian Tilson.

Research Update

2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

25

The Trade-Comovement Puzzle
Standard international transmission mechanism of productivity shocks
predicts a weak endogenous linkage between trade and business
cycle synchronization: a problem known as the trade-comovement
puzzle. We provide the foundational analysis of the puzzle, pointing
to three natural candidate resolutions: i) financial market frictions;
ii) Greenwood–Hercowitz–Huffman preferences; and iii) dynamic trade
elasticity that is low in the short run but high in the long run. We
show the effects of each of these candidate resolutions analytically and
evaluate them quantitatively. We find that, while i) and ii) fall short of
the data, iii) goes a long way toward resolving the puzzle.
WP 20-01 Revised. Lukasz A. Drozd, Federal Reserve Bank of Philadelphia Research Department; Sergey Kolbin, Amazon; Jaromir B.
Nosal, Boston College.

Household Mortgage Refinancing Decisions Are
Neighbor Influenced
Can social influence effects help explain regional heterogeneity in
refinancing activity? Neighborhood social influence effects have been
shown to affect publicly observable decisions, but their role in private
decisions, like refinancing, remains unclear. Using precisely geolocated
data and a nearest-neighbor research design, we find that households are 7 percent more likely to refinance if a neighbor within 50
meters has recently refinanced. Consistent with a word-of-mouth
mechanism, social influence effects are weaker when neighbors are
farther away and nonexistent for nonoccupants. Our results illustrate
the importance of the proximate community for household wealth
accumulation and the transmission of monetary policy.

Which Lenders Are More Likely to Reach Out to
Underserved Consumers: Banks versus Fintechs
versus Other Nonbanks?
There has been a great deal of interest recently in understanding the
potential role of fintech firms in expanding credit access to the underbanked and credit-constrained consumers. We explore the supply side
of fintech credit, focusing on unsecured personal loans and mortgage
loans. We investigate whether fintech firms are more likely than
other lenders to reach out to “underserved consumers,” such as minorities; those with low income, low credit scores, or thin credit histories;
or those who have a history of being denied for credit. Using a rich
data set of credit offers from Mintel, in conjunction with credit information from TransUnion and other consumer credit data from the
FRBNY/Equifax Consumer Credit Panel, we compare similar credit
offers that were originated by banks, fintech firms, and other nonbank
lenders. Fintech firms are more likely than banks to offer mortgage
credit to consumers with lower income, lower credit scores, and those
who have been denied credit in the recent past. Fintechs are also
more likely than banks to offer personal loans to consumers who had
filed for bankruptcy (thus also more likely to receive credit card offers
overall) and those who had recently been denied credit. For both
personal loans and mortgage loans, fintech firms are more likely than
other lenders to reach out and offer credit to nonprime consumers.
WP 21-17. Erik Dolson, Federal Reserve Bank of Philadelphia Supervision, Regulation, and Credit Department; Julapa Jagtiani, Federal
Reserve Bank of Philadelphia Supervision, Regulation, and Credit
Department.

WP 21-16. W. Ben McCartney, Purdue University and Federal Reserve
Bank of Philadelphia Consumer Finance Institute Visiting Scholar;
Avni M. Shah, University of Toronto.

26

Federal Reserve Bank of Philadelphia
Research Department

Research Update
2021 Q2

Piercing Through Opacity: Relationships and Credit
Card Lending to Consumers and Small Businesses
During Normal Times and the COVID-19 Crisis
We investigate bank relationships in a rarely considered context—
consumer and small-business credit cards. Using over 1 million
accounts, we find during normal times, consumer relationship customers enjoy relatively favorable credit terms, consistent with the
bright side of relationships, while the dark side dominates for small
businesses. During the COVID-19 crisis, both groups benefit, reflecting
intertemporal smoothing, with more benefits flowing to safer relationship customers. Conventional banking relationships benefit consumers
more than credit card relationships, with mixed findings for small
businesses. Important identification issues are addressed. The
Coronavirus Aid, Relief, and Economic Security (CARES) Act consumerdelinquency reporting impediments reduce the informational value of
consumer credit scores, penalizing safer borrowers.

Bayesian Estimation of Epidemiological Models:
Methods, Causality, and Policy Trade-Offs
We present a general framework for Bayesian estimation and causality assessment in epidemiological models. The key to our approach is
the use of sequential Monte Carlo methods to evaluate the likelihood
of a generic epidemiological model. Once we have the likelihood,
we specify priors and rely on a Markov chain Monte Carlo to sample
from the posterior distribution. We show how to use the posterior
simulation outputs as inputs for exercises in causality assessment. We
apply our approach to Belgian data for the COVID-19 epidemic during
2020. Our estimated time-varying-parameters SIRD model captures
the data dynamics very well, including the three waves of infections.
We use the estimated (true) number of new cases and the time-varying
effective reproduction number from the epidemiological model as
information for structural vector autoregressions and local projections.
We document how additional government-mandated mobility curtailments would have reduced deaths at zero cost or a very small cost in
terms of output.
WP 21-18. Jonas E. Arias, Federal Reserve Bank of Philadelphia
Research Department; Jesús Fernández-Villaverde, University of
Pennsylvania and Federal Reserve Bank of Philadelphia Research
Department Visiting Scholar; Juan F. Rubio-Ramírez, Emory University
and Federal Reserve Bank of Philadelphia Research Department
Visiting Scholar; Minchul Shin, Federal Reserve Bank of Philadelphia
Research Department.

WP 21-19. Allen N. Berger, University of South Carolina, Wharton
Financial Institutions Center, European Banking Center; Christa H.S.
Bouwman, Texas A&M University, ECGI, Wharton Financial Institutions
Center; Lars Norden, Getulio Vargas Foundation; Raluca A. Roman,
Federal Reserve Bank of Philadelphia; Gregory F. Udell, Indiana
University; Teng Wang, Federal Reserve Board of Governors.

How Resilient Is Mortgage Credit Supply?
Evidence from the COVID-19 Pandemic
We study the evolution of U.S. mortgage credit supply during the
COVID-19 pandemic. Although the mortgage market experienced
a historic boom in 2020, we show there was also a large and sustained
increase in intermediation markups that limited the pass-through of
low rates to borrowers. Markups typically rise during periods of peak
demand, but this historical relationship explains only part of the large
increase during the pandemic. We present evidence that pandemicrelated labor market frictions and operational bottlenecks contributed
to unusually inelastic credit supply and that technology-based lenders,
likely less constrained by these frictions, gained market share. Rising
forbearance and default risk did not significantly affect rates on
“plain-vanilla” conforming mortgages, but it did lead to higher spreads
on mortgages without government guarantees and loans to the riskiest
borrowers. Mortgage-backed securities purchases by the Federal
Reserve also supported the flow of credit in the conforming segment.
WP 21-20. Andreas Fuster, Swiss National Bank and CEPR; Aurel Hizmo,
Board of Governors of the Federal Reserve System; Lauren LambieHanson, Federal Reserve Bank of Philadelphia Consumer Finance
Institute; James Vickery, Federal Reserve Bank of Philadelphia Research
Department; Paul Willen, Federal Reserve Bank of Boston and NBER.

Research Update

2021 Q2

Federal Reserve Bank of Philadelphia
Research Department

27

Macroeconomic Forecasting and Variable Ordering
in Multivariate Stochastic Volatility Models

Aging and the Real Interest Rate in Japan:
A Labor Market Channel

We document five novel empirical findings on the well-known potential
ordering drawback associated with the time-varying parameter
vector autoregression with stochastic volatility developed by Cogley
and Sargent (2005) and Primiceri (2005), CSP-SV. First, the ordering
does not affect point prediction. Second, the standard deviation
of the predictive densities implied by different orderings can differ
substantially. Third, the average length of the prediction intervals is
also sensitive to the ordering. Fourth, the best ordering for one variable in terms of log-predictive scores does not necessarily imply the
best ordering for another variable under the same metric. Fifth,
the best ordering for variable x in terms of log-predictive scores
tends to put the variable x first while the worst ordering for variable
x tends to put the variable x last. Then, we consider two alternative
ordering invariant time-varying parameter VAR-SV models: the
discounted Wishart SV model (DW-SV) and the dynamic stochastic
correlation SV model (DSC-SV). The DW-SV underperforms relative to
each ordering of the CSP-SV. The DSC-SV has an out-of-sample forecasting performance comparable to the median outcomes across
orderings of the CSP-SV.

This paper explores a causal link between aging of the labor force
and declining trends in the real interest rate in Japan. We develop
a search/matching model that features heterogeneous workers with
respect to their ages and firm-specific skills. Using the model, we
examine the long-run implications of the sharp drop in labor force
entry in the 1970s. We show that the changes in the demographic
structure induce significant low-frequency movements in per capita
consumption growth and the real interest rate. The model suggests
that aging of the labor force accounts for 40 percent or more of the
declines in the real interest rate observed between the 1980s and
2000s in Japan. We also examine the impacts of other long-term
developments such as a slowdown of TFP growth and higher shares
of female and nonregular workers.

WP 21-21. Jonas E. Arias, Federal Reserve Bank of Philadelphia
Research Department; Juan F. Rubio-Ramírez, Emory University and
Federal Reserve Bank of Philadelphia Research Department Visiting
Scholar; Minchul Shin, Federal Reserve Bank of Philadelphia
Research Department.

Capital Buffers in a Quantitative Model of Banking
Industry Dynamics

WP 21-23. Shigeru Fujita, Federal Reserve Bank of Philadelphia
Research Department; Ippei Fujiwara, Keio University, Australian
National University.

We revisit the measurement of Employer-to-Employer (EE) transitions
in the monthly Current Population Survey. We detect sharp increases in
the incidence of missing answers to the relevant question starting
in 2007, when the U.S. Census Bureau introduced the Respondent
Identification Policy. We show evidence of nonresponse selection by
both observable and unobservable worker characteristics that
correlate with EE mobility. We propose a selection model and a procedure to impute missing answers, thus EE transitions. Our imputed
EE aggregate series restores a close congruence with the business
cycle after 2007, including the COVID-19 recession, and exhibits no
downward trend since 2000.

We develop a model of banking industry dynamics to study the quantitative impact of regulatory policies on bank risk-taking and market
structure as well as the feedback effect of market structure on the
efficacy of policy. Since our model is matched to U.S. data, we propose
a market structure where big banks with market power interact with
small, competitive fringe banks. Banks face idiosyncratic funding
shocks in addition to aggregate shocks, which affect the fraction of
performing loans in their portfolio. A nontrivial bank size distribution
arises out of endogenous entry and exit, as well as banks’ buffer
stock of net worth. We show the model predictions are consistent
with untargeted business cycle properties, the bank-lending channel,
and empirical studies of the role of concentration on financial stability.
We then conduct a series of policy counterfactuals motivated by
those proposed in the Dodd–Frank Act (size- and state-dependent
capital requirements and liquidity requirements). We find that
regulatory policies can have an important impact on banking market
structure, which, along with selection effects, can generate changes
in allocative efficiency and stability.

WP 21-22. Shigeru Fujita, Federal Reserve Bank of Philadelphia
Research Department; Giuseppe Moscarini, Yale University and
Federal Reserve Bank of Philadelphia Research Department Visiting
Scholar; Fabien Postel-Vinay, University College London.

WP 21-24. Dean Corbae, University of Wisconsin–Madison, NBER,
and Federal Reserve Bank of Philadelphia Research Department
Visiting Scholar ; Pablo D’Erasmo, Federal Reserve Bank of Philadelphia
Research Department.

Measuring Employer-to-Employer Reallocation

28

Federal Reserve Bank of Philadelphia
Research Department

Research Update
2021 Q2

Data in Focus

Aruoba-Diebold-Scotti
Business Conditions Index
The Philadelphia Fed collects, analyzes, and shares useful data
about the Third District and beyond. Here’s one example.

I

n the mid-2000s, three economists—
S. Borağan Aruoba of the University
of Maryland, Francis X. Diebold of the
University of Pennsylvania, and Chiara
Scotti of the Federal Reserve Board of
Governors—joined forces to build a framework for constructing a timely measure
of daily economic activity based on somewhat less timely information from the
U.S. government. Before then, business
owners, heads of household, and policymakers had to wait weeks or months for
a clear picture of the economy to develop,
leaving them in the dark when making
important, time-sensitive decisions. By
providing a timely snapshot of the entire
economy, the Aruoba-Diebold-Scotti
(ADS) index quickly illuminates current
conditions for these decision makers.
The average value of the ADS index is
zero. Positive values indicate betterthan-average conditions; negative values
indicate worse-than-average conditions.
These values are comparable across time:
Two quarters that are several years apart
and that both had a value of −1 were
equally below average. The index reproduced above shows the stunning drop
in business conditions in the early months
of the COVID-19 pandemic and their sharp

ADS Index

Jun 17

Jun 15

Jun 10

Jun 4

Jun 3

May 28

May 27

May 20

May 14

May 13

10
5
0

−5
−10
−15
−20
−25
−30

2020:01:01

Source: Federal Reserve Bank of Philadelphia AruobaDiebold-Scotti Business Conditions Index.

rebound a few months later. Because the
ADS index is recomputed frequently
(and always using the latest information
from the U.S. government), the graph
reproduces, at its tail end, the small revisions among the most recent index values
that arise when the U.S. government
releases more information.

2021:01:01

2021:06:12

Learn More
Online: https://www.philadelphiafed.org/
surveys-and-data/real-time-dataresearch/ads
E-mail: patrick.doelp@phil.frb.org and
tom.stark@phil.frb.org

Connect with Us
Explore a New Philadelphia Fed Website User Experience
The Philadelphia Fed launched its new website. Visit www.philadelphiafed.org to see how our
improved user-centric navigation and content gives you an easier way to find the data and information
you want—current bookmarks will need to be reset under the new navigation system—and explore
everything the Philadelphia Fed has to offer to help communities thrive.

PRESORTED STANDARD
U.S. POSTAGE
PAID
PHILADELPHIA, PA
PERMIT #583

Ten Independence Mall
Philadelphia, Pennsylvania 19106-1574

ADDRESS SERVICE REQUESTED

From Philadelphia, Pennsylvania, the birthplace of American finance, comes Economic Insights.

2016 Q1

2017 Q2

2018 Q1

2019 Q1

Featuring articles about how Americans work, invest, spend…and so much more.

2020 Q1

2020 Q2

2020 Q3

2020 Q4