The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Third QUARTER 2018
FEDERALRESERVE
RESERVEBANK
BANKOF
OFRICHMOND
RICHMOND
FEDERAL
Help
Wanted
Employers are having
a hard time hiring.
Not enough workers
or not the right skills?
Why Countries Avoid
Defaulting on Debt
Measuring Rural-Urban
Differences
Interview with
Antoinette Schoar
Volume 23
Number 3
Third QUARTER 2018
c o ver st o r y
8
Help Wanted
Employers are having a hard time hiring. Not enough
workers or not the right skills?
FEATURE
When Nations Don’t Pay Their Debts
Mounting U.S. debt has raised some concerns over its
sustainability. What happens when countries can’t or
won’t repay
Econ Focus is the
economics magazine of the
Federal Reserve Bank of
Richmond. It covers economic
issues affecting the Fifth Federal
Reserve District and
the nation and is published
on a quarterly basis by the
Bank’s Research Department.
The Fifth District consists of the
District of Columbia,
Maryland, North Carolina,
South Carolina, Virginia,
and most of West Virginia.
Director of RESEARCH
Kartik Athreya
Editorial Adviser
Aaron Steelman
Editor
11
Renee Haltom
S eni o r E dit o r
David A. Price
Managing Editor/Design Lead
Kathy Constant
Staff WriterS
Jessie Romero
Tim Sablik
Editorial Associate
Lisa Kenney
Contributors
DEPARTMENTs
1 President’s Message/The Supply Side of Rural Development
2 Upfront/Regional News at a Glance
3 Federal Reserve/Leaving LIBOR
6 Jargon Alert/Machine Learning
7 Research Spotlight/Did the Great Recession Increase
Skill Requirements?
15 At the Richmond Fed/What to do When Large Firms Fail
16 Interview/Antoinette Schoar
22 Economic History/Founding America’s First Research University
26 Policy Update/Tailoring Bank Regulations
27 Book Review/Information, Incentives, and Education Policy
28 District Digest/Definitions Matter: The Rural-Urban Dichotomy
36 Opinion/What Have We Learned since the Financial Crisis?
Caitlin Dutta
Joseph Mengedoth
Akbar Naqvi
Karl Rhodes
Design
Janin/Cliff Design, Inc.
Published quarterly by
the Federal Reserve Bank
of Richmond
P.O. Box 27622
Richmond, VA 23261
www.richmondfed.org
www.twitter.com/
RichFedResearch
Subscriptions and additional
copies: Available free of
charge through our website at
www.richmondfed.org/publications or by calling Research
Publications at (800) 322-0565.
Reprints: Text may be reprinted
with the disclaimer in italics
below. Permission from the editor
is required before reprinting
photos, charts, and tables. Credit
Econ Focus and send the editor a
copy of the publication in which
the reprinted material appears.
The views expressed in Econ Focus
are those of the contributors and not
necessarily those of the Federal Reserve Bank
of Richmond or the Federal Reserve System.
ISSN 2327-0241 (Print)
ISSN 2327-025x (Online)
President’s Message
The Supply Side of Rural Development
T
he more I’ve traveled in our district, the more I’ve
learned about the economic struggles in many of
our rural areas. Rural areas have been lagging badly
in recent years in both employment and output growth.
The employment-population ratio for working-age people
in our district is about 6 percentage points higher in urban
areas than in rural ones. These communities have been
hit hard by changes in the economy, including the loss of
manufacturing jobs.
The well-being of a lot of people is at stake. Moreover,
the country as a whole needs more opportunities for rural
Americans so that we can all benefit from the resulting
economic growth. Rural America isn’t as populous as it
once was, but it still makes up almost a fifth of the country’s
population, or about 60 million people. What, then, should
policymakers be doing to foster the economic development
of these communities?
Yes, geographic mobility — the movement of workers
from distressed rural areas to metro areas with more jobs
— has a role to play. But relocation may come at a steep
price in terms of family and community ties, valuable in
themselves. So we should be thinking about helping rural
workers where they already are.
I sometimes encounter arguments that distressed rural
economies are a lost cause. But I’m old enough to remember
when there was a similar pessimism about our major cities,
which appeared during the 1970s and 1980s to be doomed
to perpetual decline. They weren’t. From my perspective,
the first step in thinking about the problems of distressed
rural areas is to approach them as solvable — by good
policymaking, by markets, and by rural residents themselves.
Rural labor markets have challenges on both sides.
On the demand side, they are dominated by low-wage,
low-productivity jobs. On the supply side, workers tend,
on average, to have less education and to lack skills that are
highly valued by employers in other areas. Both are significant: Without high-wage, high-skill jobs on the market,
workers lack an incentive to invest in their skills; without
a pool of high-skill workers, an area is unlikely to attract
high-wage, high-skill jobs.
There are many forms of rural economic development
that can boost demand for local labor; depending on the
area, these may include tourism and recreation, assembly
plants, energy production, and high-value-added agriculture. But I would like to focus on the supply side here.
How do we get rural workforces the right skills? The
desire for skill acquisition is there: According to survey
data, a third of rural Americans believe they need new
skills to get or keep their jobs, with computer and technical skills being cited most often.
For many young people,
the right answer is a four-year
college degree. College grads
earn about 80 percent more
than those with only a high
school diploma, and they’re
less likely to be unemployed.
But there’s a stark urban-rural
divide in college completion:
33 percent of adults in urban
areas have a four-year degree
or higher compared to 19 percent in rural areas — and that
gap has been growing. University of Virginia research
published by the Richmond Fed has found that part of the
problem is information. Low-income rural families are less
apt to know about the college application process, college
choices, the availability of financial aid, and the return on
a college degree.
In addition, Richmond Fed research has concluded
that high school students are influenced, quite rationally,
by their beliefs about whether they’ll be able to complete
their degrees: Attending college without finishing may
mean a pile of debt without much economic reward — and
40 percent of college students don’t finish within six years.
So academic preparation is critical.
But a four-year college isn’t the right answer for everyone. There are well-paying occupations in high demand
that don’t require a degree, such as truck driving and skilled
trades. How will they get those skills? Community colleges
play a major part in delivering training (as well as preparing some students for college transfer). Apprenticeships
are a small part of the picture for now but hold promise.
And a handful of online “boot camps” for entry into coding and related fields now charge tuition in the form of
income-sharing agreements, in which students don’t pay
unless and until they get a job in their field.
Whatever the right option for a particular worker, skill
acquisition in rural areas creates a virtuous circle, benefiting
both the worker and his or her community. And it will be
critical to helping the nation’s economy grow.
EF
Tom Barkin
President
Federal Reserve Bank of Richmond
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
1
UpFront
Regional News at a Glance
By L i s a K e n n e y
MARYLAND — At a time when cybersecurity breaches seem rampant,
Maryland’s legislature has passed legislation to help small businesses avoid
them. In June, the Cybersecurity Incentive Tax Credit went into effect for
Maryland companies with fewer than 50 employees. The law allows eligible
small businesses that buy cybersecurity products or services from approved
providers to claim a state income tax credit that equals 50 percent of the
cost, up to $50,000. The program is administered through the Maryland
Department of Commerce.
NORTH CAROLINA — The Publix supermarket chain announced in October
that it will build a $400 million, 1.8-million-square-foot distribution center in
Greensboro, which will be its northernmost distribution center. The center, which
is scheduled to open in 2022, is expected to employ 1,000 people by 2025 with average annual salaries of $44,000. Greensboro was chosen over other locations thanks
to incentives, including tax breaks and training programs.
SOUTH CAROLINA — It will soon be easier to hop across the pond from the
Lowcountry. British Airways announced in October that it will start two nonstop
flights per week between Charleston and London in April 2019. South Carolina
officials estimate the economic impact of the new route could be more than $20
million per year due to job creation and increased tourism. Officials also hope it
will help draw more international companies to the state.
VIRGINIA — Less than a year after Facebook announced it would invest
$1 billion in a new data center in Henrico County, the company said in
September that it will invest an additional $750 million and build three
additional buildings, bringing the total number of buildings to five. Facebook
says the expansion will bring more than 200 permanent full-time jobs and 1,500
construction jobs. Construction is already underway on the two initial buildings
and a power substation on the 325-acre site, which are expected to open in the
first half of 2019.
WASHINGTON, D.C. — In June, D.C. voters passed a ballot initiative that
would have changed wage rules for tipped workers, gradually raising their
minimum hourly wage until it matched the standard minimum wage. But in midOctober, the D.C. Council repealed Initiative 77 by an 8-5 margin. Opponents of
Initiative 77 say the repeal will help the city’s dining scene and keep restaurant
owners from cutting hours or staff. The repeal bill voted on by the Council
does address some concerns of Initiative 77’s supporters, such as a hotline for
reporting wage theft, though these provisions must be funded in the district’s
upcoming budget.
WEST VIRGINIA — State parks across West Virginia will soon receive $60
million worth of upgrades and improvements, thanks to an early October sale of
$55.2 million worth of excess lottery revenue bonds. The projects are expected
to focus on modernizing parks and completing delayed maintenance projects,
which officials hope will boost the state’s tourism industry. The West Virginia
Economic Development Authority issued the bonds, which received AAA and
A1 ratings from S&P Global and Moody’s, respectively.
2
FEDERALRESERVE
Leaving LIBOR
The Fed has developed a new reference rate to replace
the troubled LIBOR. Will banks make the switch?
By J e s s i e Ro m e ro
I
n 2010, the British bank Barclays came under investigation for manipulating a reference interest rate called
the London Interbank Offered Rate, or LIBOR. At the
time, LIBOR underpinned more than $300 trillion worth
of financial contracts worldwide. Over the next several
years, authorities would learn that multiple global banks,
including U.S.-based institutions JPMorgan Chase and
Citigroup, were guilty of manipulating LIBOR; the banks
would end up paying more than $9 billion in fines, and more
than 20 people faced criminal charges.
The scandal exposed serious flaws in how LIBOR was
calculated and spurred international regulators to seek
out alternative benchmarks. In the United States, this
effort has been led by the Alternative Reference Rates
Committee (ARRC), a private-sector group convened by
the Federal Reserve and other regulators. The committee
has recommended that markets adopt a new reference
rate, and although the transition is underway, there are
still about $200 trillion — 10 times the level of U.S. GDP
— worth of outstanding contracts based on the U.S. dollar
LIBOR. (The rate is also calculated for the Swiss franc,
the euro, the British pound, and the Japanese yen; before
the scandal, LIBOR was calculated for 10 different currencies.) In addition, new contracts referencing the rate
continue to be written, even though it’s likely to disappear
after 2021. Will the financial sector leave LIBOR in time?
What is LIBOR?
LIBOR is based on how much banks pay to borrow from
one another. Each day, a panel of 20 international banks
responds to the question, “At what rate could you borrow
funds, were you to do so by asking for and then accepting
interbank offers in a reasonable market size just prior to
11 a.m.?” The highest and lowest responses are excluded,
and the remaining responses are averaged. Not every bank
responds for every currency; 11 banks report for the franc,
while 16 banks report for the dollar and the pound. For each
of the five currencies, LIBOR is published for seven different maturities, ranging from overnight to 12 months. In total,
35 rates are published every applicable London business day.
About 95 percent of the outstanding contracts based
on LIBOR are for derivatives. (See chart.) It’s also used
as a reference for other securities and for variable rate
loans, such as private student loans and adjustable-rate
mortgages (ARMs). In 2012, the Cleveland Fed calculated
that about 80 percent of subprime ARMs were indexed to
LIBOR, as well as about 45 percent of prime ARMs. Prior
to the financial crisis, essentially all subprime ARMs were
linked to LIBOR.
As journalists Liam Vaughan and Gavin Finch described
in their 2017 book The Fix, LIBOR was the brainchild of
financier Minos Zombanakis. In 1969, Zombanakis helped
arrange an $80 million loan to the shah of Iran, one of the
first modern syndicated loans (loans funded by multiple
banks). The banks involved were nervous about lending at
a fixed rate when inflation was on the rise. So Zombanakis
devised a system in which the loan would be funded with
rolling deposits and the interest rate would be recalculated
every few months. Banks would report their funding costs
before every rollover, and the new interest rate would be
based on the weighted average.
Other financiers adopted Zombanakis’ formula, and in
1986 the British Bankers’ Association, in consultation with
the Bank of England, took over data collection and reporting. To discourage cheating, the association refined the formula to remove the top and bottom quartile of responses.
Around the same time, financial deregulation made
London an attractive home for the growing markets in
derivatives, bonds, and syndicated loans. These transactions
referenced LIBOR, and the rate quickly became ubiquitous
throughout the financial system. “As the swaps market
developed for banks to hedge their interest rate risk, they
needed some kind of reference rate, and LIBOR was already
in place,” says David Skeie of Texas A&M University.
In 1997, the Chicago Mercantile Exchange decided to
adopt LIBOR as the reference rate for eurodollar futures
Linked to LIBOR
LIBOR underpins
almost $200 trillion
worth of financial
contracts
$3.4 trillion
Business Loans
$1.3 trillion, Consumer Loans
$1.8 trillion, Bonds
$1.8 trillion, Securitizations
$45 trillion
Exchange-Traded
Derivatives
$145 trillion
Over-the-Counter
Derivatives
Source: Alternative Reference
Rates Committee (2018)
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
3
contracts, which were a popular way for traders to hedge
their positions against other derivatives, and LIBOR’s
position in the financial system was cemented. “Once
LIBOR had become a widely used reference point, it fed
on itself,” says Matthew Lieber, a vice president in the
Markets Group at the New York Fed. “Liquidity begets
liquidity.”
Zombanakis himself didn’t foresee how widespread
LIBOR would become. “We just needed a rate for the
syndicated-loan market that everyone would be happy
with,” he has said. “When you start these things, you never
know how they are going to end up, how they are going to
be used.”
Hindsight Is 20/20
In retrospect, the potential to manipulate LIBOR seems
obvious. But in the 1980s and 1990s, according to Vaughan
and Finch, most regulators thought it was a remote possibility. First, because the highest and lowest reported rates
were excluded, any major shift in LIBOR would require
mass collusion. Second, because each bank’s submission
was made public, it would be immediately apparent if anyone were reporting questionable numbers. As the financial system became more complex, however, smaller and
smaller movements in LIBOR were worth more and more
money. If a bank reported a rate that was thrown out, that
had the effect of pushing in rates that would otherwise
have been excluded. Even a change of a few basis points
could be worth millions of dollars.
The first hints that something was amiss were in 2007,
when the research arm of the brokerage ICAP published
some traders’ claims that the one-month LIBOR was
lower than actual borrowing costs. Around the same time,
a Barclay’s employee emailed a group including several
New York Fed officials to say that LIBOR submissions
appeared unrealistically low. The following spring, the
Wall Street Journal published two articles estimating that
banks were underreporting their borrowing costs to make
themselves appear less risky than they actually were.
Later research has supported these claims. In ongoing research, Skeie, along with Dennis Kuo, a former
researcher at the New York Fed, and James Vickery of the
New York Fed, has compared LIBOR rates between 2007
and 2009 with other measures of borrowing costs, including Term Auction Facility bids and Fedwire transfers.
While LIBOR generally tracked these other measures,
it was consistently 20 to 30 basis points below them. The
authors considered several explanations for the disparity
and concluded that it was consistent with banks trying to
avoid the appearance of financial distress.
As regulators investigated underreporting, they learned
that banks had another motivation for fudging the numbers: Beginning at least in 2003, banks had been submitting
LIBOR reports that would benefit their trading positions.
Rate submitters and traders at different banks and brokerages also conspired with each other to manipulate LIBOR,
4
E co
n
F o c u s | T h i r d Q u a rt
E r
| 20 1 8
promising each other steaks, Champagne, and Ferraris
(among other perks). Internal emails and instant messages
revealed the scheme. As one trader wrote, “Sorry to be a pain
but just to remind you the importance of a low fixing for us
today.” Another wondered “if it suits you guys on hiking up
1bp on the 6mth Libor in JPY [one basis point on the sixmonth LIBOR in Japanese yen] ... it will help our position
tremendously.” At least 11 financial institutions faced fines
and criminal charges from multiple international agencies,
including the Commodity Futures Trading Commission
(CFTC) and the Justice Department in the United States.
Separately, in 2014 the FDIC sued 16 global banks for
manipulating LIBOR, alleging that their actions had caused
“substantial losses” for nearly 40 banks that went bankrupt
during the financial crisis. The lawsuit is pending in the U.S.
District Court for the Southern District of New York.
For the past five years, LIBOR has been regulated and
administered by the United Kingdom’s Financial Conduct
Authority (FCA) and the Intercontinental Exchange
Benchmark Administration. The organizations have made
a number of changes to prevent false submissions, including developing a new, less-subjective methodology, but
post-crisis there’s another problem: Banks no longer
borrow from each other at longer maturities very often.
That means the market underlying LIBOR is very thin;
on a typical day, there are only six to seven transactions
underpinning the one- and three-month LIBOR, two to
three for the six-month LIBOR, and one — if any — for
the one-year LIBOR. As a result, banks have to make a
judgment call about what rate to report. Even if it isn’t
intentionally misleading, that judgment could be wrong.
Winds of Change
In 2013, as the investigations continued, the Financial
Stability Board (FSB), a global monitoring agency, began
reviewing whether and how to reform LIBOR. After a
year of work, the FSB issued a report calling for the development of new benchmarks. An effective reference rate,
according to the report, should meet three criteria: First,
it should minimize the opportunities for market manipulation. Second, it should be anchored in observable transactions wherever feasible. And third, it should command
confidence that it will remain resilient in times of financial stress. (The International Organization of Securities
Commissions published more detailed principles in 2013.)
The FSB asked international regulators to help engineer the transition. “Reference rates are vital to efficient
market functioning,” says Lieber. “But they affect a range
of market participants in considerably different ways,
so different types of institutions might have conflicting
incentives. This means there’s an important role for the
official sector to play in helping develop an optimal rate.”
In the United States, the Federal Reserve convened the
new ARRC in cooperation with the Treasury department,
the CFTC, and the U.S. Office of Financial Research. It’s
currently composed of around two dozen participants
from the private sector, including representatives from
banks, investment firms, trade associations, and other
financial institutions. Representatives from regulators and
other government agencies serve on an ex officio basis.
As the committee was beginning its work in 2014,
the New York Fed was also working with the Office of
Financial Research to develop several new reference rates
based on Treasury repurchases, or repos, in an effort to
create greater transparency in that market. (Repos function as short-term loans; one party sells a security with a
promise to buy it back, usually the next day.) In mid-2017,
the ARRC decided to recommend one of these rates —
the Secured Overnight Financing Rate, or SOFR — as a
replacement for the dollar LIBOR.
The committee chose SOFR for several reasons. First
and foremost, it’s based on a large volume of observable
transactions — more than $800 billion per day, much
larger than any other U.S. money market. And because it
covers multiple segments of the repo market, it can evolve
as the market evolves, according to the New York Fed. In
addition, SOFR was designed from the beginning to comply
with the new international standards for reference rates.
Some observers are concerned that changing benchmarks could create a disconnect between banks’ assets and
liabilities; because LIBOR is based on banks’ borrowing
costs, it enables them to hedge against changes in those
costs. As the scandal demonstrated, however, LIBOR is
not necessarily an accurate gauge. Moreover, banks are
no longer the only users of LIBOR. “When it comes to
floating rate loans and interest rate swaps for commercial
banks, it does make conceptual sense to have a benchmark
tied to a bank funding rate,” says Skeie. “But so much
financial intermediation is now outside of commercial
banking, and LIBOR has become the reference rate for
such a vast amount of contracts. For these other players,
SOFR is likely a much better instrument.”
Keep Calm and Trade On?
The other reason to make a switch is that LIBOR is
unlikely to exist in a few years.
Today, many banks participate in the LIBOR panel only
at the urging of the United Kingdom’s FCA. That’s because,
after the rate manipulation came to light, banks were wary
of being associated with LIBOR. And as the market grew
thinner, they became more and more reluctant to essentially
guess what rate to submit. In 2013, several banks announced
they were planning to quit the panel, and the agency (at the
time called the Financial Services Authority) sent letters
intimating that doing so would damage their relationship
with regulators. But the agency can’t legally make banks
participate indefinitely, and it’s announced that it won’t
pressure them to do so after 2021. Most industry observers
expect LIBOR to vanish at that time.
The ARRC has estimated that about 20 percent of
existing dollar LIBOR contracts mature after 2021, which
could create major headaches for the parties to those
contracts if and when LIBOR disappears. While most
contracts include “fallback language” that applies if the
underlying reference rate is unavailable, the provisions
are inconsistent, and the language is designed to address a
temporary disruption — not a permanent disappearance.
“Permanent cessation without viable fallback language in
contracts would cause considerable disruption to financial
markets,” the ARRC has warned. “It would also impair
the normal functioning of a variety of markets, including
business and consumer lending.”
The ARRC and other groups are developing guidance
to help financial institutions revise their contracts, but so
far, not much progress has been made. “It’s very complex
and costly to change,” notes Skeie. “Since you still have a
few years until the real uncertainty hits, it’s a lot easier to
not go first.”
Encouraging market participants to renegotiate existing
contracts is one challenge. Encouraging them to write new
contracts based on SOFR rather than LIBOR is another.
“Because everybody prefers to be in the high-liquidity club,
there is a coordination problem,” wrote Darrell Duffie of
Stanford University and Jeremy Stein of Harvard University.
(Stein is also a former Fed governor.) “No individual actor
may be willing to switch to an alternative benchmark, even
if a world in which many switched would be less vulnerable
to manipulation and offer investors a menu of reference
rates with a better fit for purpose.”
Many observers have voiced concern that the financial
system won’t be ready when LIBOR goes away. But in
some respects the switch is ahead of schedule. For example, the Chicago Mercantile Exchange launched SOFR
futures in May 2018, and the clearing house LCH cleared
the first SOFR swaps in July — well before the expected
timing outlined in a transition plan developed by the
ARRC. The growth of SOFR-based derivatives activity
has been encouraging, and the participation has been
diverse, says Lieber, but “we need to see more take-up
for it to become meaningful. It’s been good so far but not
sufficient.” While regulators might lead traders to SOFR,
they can’t make them use it.
EF
Readings
Duffie, Darrell, and Jeremy C. Stein. “Reforming LIBOR and Other
Financial Market Benchmarks.” Journal of Economic Perspectives,
Spring 2015, vol. 29, no. 2, pp. 191-212.
Vaughan, Liam, and Gavin Finch. The Fix: How Bankers Lied, Cheated
and Colluded to Rig the World’s Most Important Number. West Sussex:
John Wiley & Sons, Ltd., 2017.
Kuo, Dennis, David Skeie, and James Vickery. “A Comparison of
LIBOR to Other Measures of Bank Borrowing Costs.” Manuscript,
April 2018.
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
5
JargonAlert
Machine Learning
C
ustomers of online music services have long been
able to explore new music, or revisit old music,
through the services’ playlists. Whether you like
’80s pop, ’90s rap, or new country, your online music service
has had a playlist for you, handmade by music experts. But
in 2015, Spotify added something different: individually personalized playlists that each of its millions of users received
every Monday. The feature, known as Discover Weekly,
gained devotees. One wrote, “It felt like an intimate gift
from someone who knew my tastes inside and out.”
Of course, Spotify didn’t scale up its staff of human
music experts to create weekly playlists for what are now
reportedly 87 million subscribers. Discover Weekly relies
instead on a user’s past listening habits and those of others
with apparently similar tastes — and
on machine learning software that
converts this data into predictions
of what a user would like.
Music is just one of a range of
industries being affected by machine
learning technology. Machine learning is likely to improve high-tech
products in applications from
spam filtering to face recognition.
In medicine, machine learning
may improve the interpretation of
X-rays and other scans, as well as suggest diagnoses based
on detailed patient information. Within the financial sector, some applications include detecting fraud, estimating insurance risks, and analyzing investments. In some
industries, the adoption of machine learning may change
the profile of skills sought by employers and even reduce
employment numbers outright.
But what is it, exactly? Historically, it has a number of
fields in its family tree: computer science, cognitive science,
and statistics, among others. It’s sometimes said to be a
branch of artificial intelligence, or AI, but not the general,
human-like AI seen in the fictional computers of 2001: A
Space Odyssey and Star Trek. Rather, it’s a type of software
that learns from examples — that is, it autonomously constructs models based on data fed into it. The data may represent transactions, images, or anything else in digital form.
Machine learning systems fall into one of two broad categories: supervised or unsupervised. In supervised machine
learning, the system receives training data: a set of examples
and information about the correct classification of each
example. The latter is the “supervision.” For instance, the
training data could be images of furniture with information
about whether each item is, say, a chair, a desk, or a sofa.
With sufficient training data, the system would be able
6
to predict the correct category of an image of an item of
furniture it hasn’t seen before. Alternatively, the training
data could be individuals’ financial information, together
with an indicator for each individual of whether he or she
has a home mortgage default on record. The system would
use that data to build a model for predicting whether a loan
applicant is likely to default on a loan. (The person creating
the system may hold back some of the data he or she has on
hand to test the reliability of the model.)
In unsupervised machine learning, the system receives
records, such as images or financial information, but no
information on how to classify them. The task for the
system is to discover categories within the data on its own.
In both supervised and unsupervised machine learning, the potential performance of
the system improves as the system
receives more data. Commonly,
what goes into a machine learning system is an enormous dataset,
so-called “big data,” comprising
millions of observations. Indeed,
part of what has fueled the growth
of machine learning is the availability of such datasets within technology companies as a byproduct
of their operations as they capture
data on transactions and other user behavior.
One important difference between machine learning
and conventional techniques is that conventional statistical techniques produce models that can be interpreted
by humans. Someone can look at the coefficients of a
multiple regression analysis and see how it works — which
variables count positively, which count negatively, and by
how much. In contrast, complex machine learning models
are like black boxes and cannot be translated into a form
that lets humans understand the model’s workings.
Within the discipline of economics, some researchers,
such as Susan Athey of Stanford University, foresee that
machine learning may become an increasingly important
tool, transforming economic research. But for the time
being, at least, switching from conventional statistical
methods to machine learning comes at a price: Compared
to machine learning, econometrics is better suited to asking
about causation. Machine learning is about classification
and prediction. Econometrics is too, but it also lets a
researcher make inferences about whether and how one
variable among many has been influencing the phenomenon that the researcher is studying. That distinction could
erode, however, as researchers are seeking to combine
machine learning with analysis of causation.
EF
Illustration: Timothy Cook
By Dav i d A . P r i c e
Research Spotlight
Did the Great Recession Increase Skill Requirements?
W
By C a i t l i n D u t ta
hat you need to know to get a job has changed
MSAs upskilled more than firms in less hard-hit MSAs.
drastically over time in the United States.
Next, they explained investment in IT, a routineOccupations that used to employ many midbiased technology, in firms in hard-hit MSAs relative to
skill workers, such as assembly-line work or typing, now
less hard-hit MSAs. They found that firms in harder-hit
face falling employment shares.
MSAs increased their IT investment more than firms in
Much of the disappearance in routine jobs like these is
better-off MSAs. They also found that firms with more IT
attributed to routine-biased technological change — that
upskilled more than firms with less IT.
is, the introduction of technology that substitutes for
Finally, they ran the model to compare the upskillsome routine jobs and complements some more cognitive
ing in jobs denoted as routine-manual and as routineskills. Routine-biased technological change is related to
cognitive. This distinction follows a 2010 National Bureau
skill-biased technological change, the scenario in which
of Economic Research working paper by Daron Acemoglu
technology substitutes for unskilled labor. An example
and David Autor of MIT in which the authors labeled
of routine-biased technological
jobs that involve routine physchange is an ATM that can proical tasks, such as installing a
“Do Recessions Accelerate Routine-Biased
cess a check for deposit. This
car door in a car factory, as
Technological Change? Evidence from
ATM is a substitute for the
routine-manual and jobs that
Vacancy Postings.” Brad Hershbein and Lisa
worker who used to manually
involve routine mental tasks,
process checks, but it is complesuch as receptionist work, as
B. Kahn. American Economic Review,
mentary to the labor of a comroutine-cognitive. Kahn and
July 2018, vol. 108, no. 7, pp. 1737-1772.
puter programmer who would be
Hershbein found that the
hired to program the machine.
upskilling was concentrated in
While routine-biased technological change has been
routine-cognitive jobs. They also found that routine-manhappening for decades in the United States, a recent
ual jobs declined in employment share and productivity
American Economic Review article by Brad Hershbein of
while routine-cognitive jobs increased in employment
the W.E. Upjohn Institute for Employment Research
share and wages. These findings offer an explanation for
and Lisa Kahn of the Yale School of Management found
the known increase in the probability that college gradthat the process was accelerated by the Great Recession
uates will take a routine job. If routine-cognitive jobs are
of 2007-2009.
upskilling and increasing in wages, they will become more
Kahn and Hershbein analyzed a novel dataset for their
attractive to college graduates.
work: about 100 million online job postings in the United
What does this mean for the story of routine-biased
States, which included almost all of the online job postings
technological change? The authors conclude that the
from 2007 and 2010-2015. They calculated the proportion
recession encouraged upskilling by increasing demand for
of postings that had requirements in four categories: eduroutine-biased technology. This adoption of technology
cation, experience, cognitive skills, and computer skills.
meant that employers demanded fewer routine-manual
They found that a job posting was more likely to post a
workers and demanded more skills from their routinerequirement in each of the four categories after the recescognitive workers, accounting for the upskilling seen in the
sion than before the recession. From this, they inferred
original data analysis. The authors find that these effects
that after the recession, employers were more likely to
continued through 2015, after other employment indicators
require applicants to have high skills than before the
affected by the recession returned to pre-recession levels.
recession. Such an increase in skill requirements for a job
The authors don’t commit to one explanation for this
is known as “upskilling”; Kahn and Hershbein endeavored
phenomenon, but they favor the theory of Schumpeterian
to find out what caused it with a new model.
cleansing. Schumpeterian cleansing, advanced by Joseph
The model they created explains various employment
Schumpeter of Harvard University in 1939, is an effect
indicators in metropolitan statistical areas (MSAs) harder
in which bad economic times force less-productive firms
hit by the recession relative to MSAs that were less hard
to shut down, while more productive and modern firms
hit. They found that the shock of the recession raised
succeed. If this theory is the correct explanation, the
the probability of posting skill requirements more in
recession forced the closure of unproductive firms that
harder-hit MSAs than in less hard-hit ones and that this
were not using routine-biased technology, while new or
increase in skill requirements is seen within postings for
existing productive firms that were using routine-biased
a given occupation. This implies that firms in harder-hit
technology succeeded.
EF
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
7
HELP
WANTED
Employers are having
a hard time hiring.
Not enough workers or
not the right skills?
By Jessie Romero
B
efore every meeting of the Federal Open Market
Committee, the Fed publishes a new Beige Book,
a compilation of qualitative economic information
from each Federal Reserve district. In the most recent
one, the Richmond Fed’s business contacts reported that
“labor demand strengthened and job openings increased
as employers struggled to find qualified workers.” The
language would have been familiar to regular readers:
Six years earlier, the Beige Book had noted that “[Fifth]
District employment improved somewhat, but both manufacturers and professional services firms continued to
report problems finding qualified workers.”
It’s not surprising that employers are having a hard time
finding workers today, when the unemployment rate is the
lowest it’s been in nearly five decades. But why were they
having trouble finding workers in 2012, when the unemployment rate had been stuck above 8 percent for several years?
Many people attributed persistently high unemployment after the Great Recession to “skill mismatch” — the
idea that the people looking for work didn’t have the
qualifications employers were seeking — and there was
considerable concern that such mismatch would be a
permanent feature of the labor market. Today, however,
things look quite different: Many lower-skill occupations,
once the hardest hit, are now in high demand, and employers are increasingly willing to train. Is skill mismatch a
thing of the past?
8 E CO N F O C U S | T H I R D Q U A RT E R | 2 0 1 8
It’s Getting Hot, Hot, Hot
In September 2018, the unemployment rate dropped to
3.7 percent — its lowest reading since December 1969.
At the same time, the Congressional Budget’s Office
estimate of the “natural” rate of unemployment, which is
widely viewed as the benchmark for full employment, was
4.6 percent. (Even in a healthy economy, there will always
be some level of unemployment as workers transition
between jobs. The natural rate is the lowest rate that can
be maintained without accelerating inflation.)
That’s not the only indication the labor market is tight.
In 2000, the Bureau of Labor Statistics (BLS) began tracking data on labor market turnover, including job openings.
In April of this year, for the first time ever, there were
more vacancies than there were people looking for work,
and the gap has continued to grow. (See chart.)
Qualitative data also suggest it’s hard to find workers. In recent surveys of business activity in Maryland
and the Carolinas conducted by the Richmond Fed,
the monthly indexes that measure employers’ ability to
find workers reached their lowest readings ever. (The
surveys began in 2008.) Nationally, nearly 40 percent of
small-business owners reported having unfilled job openings in September, according to a survey conducted by the
National Federation of Independent Business; the previous peak was 34 percent in 1999.
“A few years ago, our contacts talked about not being
able to find people with specific skills,” says Sonya Waddell,
the Richmond Fed’s director of regional research. “Now,
they talk about not being able to find anyone at all.”
Labor market tightness isn’t evenly distributed across
industries, however. The job openings rate for accommodation and food service workers was 6 percent in August 2018,
for example, while the rate for educational services was just
3.2 percent. Economists at ZipRecruiter, an online recruitment firm, analyzed responses to job postings and found 118
applicants for every administrative position advertised but
just 12 responses per truck driving job and nine per nursing
job. Even within industries there is variation; in the Census
Bureau’s Quarterly Survey of Plant Capacity Utilization,
just 3.5 percent of textile manufacturers reported an “insufficient supply of labor” as a constraint in the second quarter
of 2018. But 32 percent of wood manufacturers were constrained by their inability to find workers.
There are geographic differences as well. Across
Virginia as a whole, the unemployment rate has averaged
3.1 percent in 2018, well below the national average. But
Making the Match
How large a role did skill mismatch actually play in the labor
market during and after the Great Recession? Although
it no longer appears to have been the primary factor driving unemployment, some research suggests its role was
There are more job openings than people looking for work
16,000
14,000
12,000
Thousands
10,000
Job Openings
8,000
6,000
Number of
Unemployed
People
4,000
2,000
2017
2018
2015
2016
2013
2014
2012
2011
2010
2009
2007
2008
2005
2006
2003
2004
2001
0
2002
Baffled by Beveridge
Still, 7.7 percent unemployment is a significant improvement from the end of the Great Recession, when
unemployment in Scotland County topped 17 percent.
Nationally, the unemployment rate reached 10 percent
in October of 2009 and remained above 7 percent until
the end of 2013. Historically, high unemployment has
been associated with few job openings (because employers aren’t interested in hiring) and low unemployment
with plentiful job openings, a relationship known as the
Beveridge curve. But as the economy began to recover in
2009 and firms started posting jobs, the unemployment
rate remained several percentage points higher than the
Beveridge curve would have predicted.
The position of the Beveridge curve is determined by
how efficiently the labor market pairs available workers
with available jobs, what economists call “matching efficiency.” Multiple factors influence matching efficiency,
including employers’ recruiting processes, how people
search for jobs, and policies such as unemployment insurance or at-will employment. The rightward shift of the
Beveridge curve after 2009 suggested that overall matching efficiency had declined significantly. (See chart.)
Skill mismatch made intuitive sense as an explanation
for this decline. Roughly half of the job losses resulting
from the 2007-2009 recession were in construction and
manufacturing, and it seemed reasonable to assume that
unemployed roofers and forklift drivers were not finding
(or even looking for) jobs in the industries that fared relatively better, such as education and health care. And even
as manufacturers, for example, did begin to look for new
employees, they frequently said they were unable to find
applicants with the necessary skills and training.
In the short term, skill mismatch was a product of the
recession. But many observers also viewed it as a symptom
of longer-term trends in technology and education that
were operating to the detriment of lower-skilled workers
— and were unlikely to reverse. “In simple terms, the skills
people have don’t match the jobs available,” said Dennis
Lockhart, former president of the Atlanta Fed, in a 2010
speech. “Coming out of this recession there may be a more
or less permanent change in the composition of jobs.”
Tightening Up
NOTE: Shaded areas denote recessions.
Source: Bureau of Labor Statistics via Federal Reserve Economic Data (FRED)
A Bend in the “Beveridge Curve”
After the Great Recession, the unemployment rate was high relative to the
number of job openings
Job Vacancy Rate
in some western and southern counties, the rate has been
around 6 percent; in many northern counties, it’s averaged
about 2.5 percent. In North Carolina, average county
unemployment rates for 2018 range from 7.7 percent in
Scotland County, which has lost several thousand manufacturing jobs over the past two decades, to 3.1 percent in
Buncombe County, home to tourist destination Asheville.
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
Unemployment
Job Openings
0
1
2
3
4
5
6
7
8
9
10
11
Unemployment Rate
Pre-Great Recession (Jan. 2007-Nov. 2007)
Great Recession (Dec. 2007-June 2009)
Post-Great Recession (July 2009-Present)
Source: Bureau of Labor Statistics via Federal Reserve Economic Data (FRED)
nontrivial. In a 2014 article, AyŞegül Şahin of the University
of Texas at Austin, Joseph Song of Bank of America Merrill
Lynch, Giorgio Topa of the New York Fed, and Giovanni
Violante of Princeton University found that mismatch
across occupations and industries could account for up to
one-third of the rise in unemployment between 2006 and
2009. The authors speculated that the remainder could be
explained by weak demand for labor and extended unemployment benefits, among other culprits.
Regis Barnichon of the San Francisco Fed and Andrew
Figura of the Federal Reserve Board also have found a role
for mismatch. In a 2015 article, they measured mismatch as
dispersion in the labor market, or how much variation there
is in the tightness of different submarkets, such as the market for nurses versus the market for construction workers.
More dispersion indicates more mismatch. They calculated
that rising dispersion contributed to about one-third of the
decline in matching efficiency between 2008 and 2012.
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
9
The other factor driving the decrease in matching
efficiency was a change in the composition of job seekers.
In general, during recessions, the pool of unemployed
workers becomes more concentrated with people who
have a lower likelihood of finding a job, such as workers
on a permanent layoff or who have been unemployed for a
long time. This was especially true in the Great Recession,
when employers were much less likely to use temporary
layoffs than in previous downturns and long-term unemployment reached unprecedented levels.
Barnichon and Figura’s study covered 1976 through 2012,
and they found that dispersion and composition effects
increased during all the recessions during that time period.
What was unique about the Great Recession was how
large those effects were and how long they lasted. Even
after the severe recession in 1981-1982, matching efficiency
rebounded fairly quickly. But after the Great Recession
ended, it remained historically low three years later.
Other research, however, suggests that the decline in
matching efficiency wasn’t especially large compared to
previous recessions. In a 2017 article, Andreas Hornstein
of the Richmond Fed and Marianna Kudlyak of the San
Francisco Fed studied not only unemployed workers,
but also people out of the labor force — that is, people
unable to work or no longer looking for work. (A person
who has not looked for work during the past four weeks is
technically considered out of the labor force rather than
unemployed.) Although those out of the labor force are
less likely to transition into employment than those who
are unemployed, they are a much larger group in absolute
terms. According to previous research by Hornstein,
Kudlyak, and Fabian Lange of McGill University, people
out of the labor force account for about two-thirds of new
transitions to employment.
During the Great Recession, the entire pool of nonemployed people shifted more toward people out of the labor
force. Once Hornstein and Kudlyak accounted for this
change, the decline in efficiency looked comparable to
declines in previous recessions. “If the composition of the
search pool shifts toward groups who always have a lower
job finding rate, average search effectiveness declines,”
says Hornstein. “This shows up as reduced ‘matching efficiency’ even though the ‘effectiveness’ of the labor market
in matching vacancies and unemployed has not changed.”
Love the One You’re With
A few years ago, employers might not have been willing to hire an applicant who didn’t check every box
— but they’re changing their tune as the labor market has
tightened. In the September Beige Book, most districts
reported that employers in their regions were devoting
more resources to training. In a survey conducted in early
2017 by the Wall Street Journal and the consulting group
Vistage International, two-thirds of the businesses surveyed said they were spending more or significantly more
time training new employees than they had a year ago.
Employers also have been expanding their applicant
pool — for example, by relaxing skill requirements. The
labor-market research firm Burning Glass Technologies
recently analyzed 15 million online job postings and found
that the number of jobs requiring a college degree fell from
34 percent in 2012 to 30 percent in 2018, and the number
requiring three or more years of experience fell from
29 percent to 23 percent. Amazon, the country’s
second-largest employer after Walmart, advertises that its
hiring process requires “No resume. No interview.”
In addition, anecdotal evidence is growing that employers are more amenable to former offenders. The New York
Times recently profiled a company that is hiring inmates
as apprentices even before they are released; similar
stories have been reported in Los Angeles, Boston, and
Allentown, Pa., to name just a few. In a recent speech,
Richmond Fed President Tom Barkin noted that he had
spoken with an employer in the Fifth District who had
relaxed its views on employees with criminal backgrounds.
Will this continue? In the short term, the economic outlook is rosy. But productivity growth — the ultimate determinant of long-run economic growth — has lagged during
the past decade, which suggests the gas currently fueling the
economy could be stimulus whose effects might dissipate
over the next few years. In addition, although the Beveridge
curve has largely looped back to its pre-recession position, it
still remains further to the right than it was for much of the
postwar era. According to research by Thomas Lubik of the
Richmond Fed and Luca Benati of the University of Bern
(Switzerland), with each successive recession since the 1950s,
matching efficiency has gone down — the unemployment
rate implied by a given job vacancy rate has increased. A
likely explanation for these successive rightward movements
is technological change whose effects on the labor market
are hastened by recessions. A large body of research has
documented how such change has tended to benefit workers
with more skills and more education. These forces might be
masked by a hot economy for a time, but if things cool off,
some workers, especially the more recent entrants to employment, might once again find themselves without a match. EF
Readings
10
Barnichon, Regis, and Andrew Figura. “Labor Market Heterogeneity
and the Aggregate Matching Function.” American Economic Journal:
Macroeconomics, October 2015, vol. 7, no. 4, pp. 222-249.
Lubik, Thomas A., and Karl Rhodes. “Putting the Beveridge
Curve Back to Work.” Federal Reserve Bank of Richmond
Economic Brief No. 14-09, September 2014.
Hornstein, Andreas, and Marianna Kudlyak. “How Much Has
Job Matching Efficiency Declined?” Federal Reserve Bank of San
Francisco Economic Letter No. 2017-25, Aug. 28, 2017.
AyŞegül Şahin, Joseph Song, Giorgio Topa, and Giovanni L.
Violante. “Mismatch Unemployment.” American Economic Review,
November 2014, vol. 104, no. 11, pp. 3529-3564.
When Nations Don’t
Pay Their Debts
What happens when countries can’t or won’t repay
By Tim Sablik
T
reasury bonds in the United States are widely
considered among the safest financial assets in
the world. But in 2011, a political standoff over
the debt ceiling prompted some to call that safety into
question. Rating agency Standard & Poor’s downgraded
U.S. debt for the first time from the flawless AAA to the
merely excellent AA+, a rating it maintains today.
To be sure, the downgrade does not mean the United
States will face a debt crisis anytime soon. Indeed, the
other two major rating agencies, Moody’s Investors
Service and Fitch Ratings, still rate U.S. debt as triple-A.
But in the wake of the political standoff over the debt,
policymakers and researchers have discussed what might
happen if the United States ever did default. Recent examples from other countries could provide some clues.
In 2010, a crisis over Greece’s debt created hardship for
the nation and the rest of the European Union. Closer to
home, Puerto Rico announced in 2015 that it would not
be able to pay its debts, resulting in economic pain for the
island territory and some uncertainty in the United States
as Congress rushed to implement a solution.
Such episodes are actually fairly common throughout
history. In their 2009 book This Time is Different, which
surveys 800 years of financial crises, Harvard University
economists Carmen Reinhart and Kenneth Rogoff found
that most countries that have borrowed have at some point
struggled to repay what they owe. Even the United States,
which has a strong reputation for always paying its debts,
defaulted early in its history following the War of 1812.
And President Franklin Roosevelt’s suspension of the gold
standard in 1933 and subsequent revaluation of the dollar
also represented a default of sorts because those actions
substantially changed the value of the dollars used to repay
previous debt contracts.
The ever-present possibility of sovereign default raises a
question: How are countries able to borrow huge amounts
in the first place? It’s a puzzle many economists have
attempted to solve. Their research sheds light on what
happens to governments that default and helps explain
why many of them do honor their debts — eventually.
The Burden of Debt
The weight of public debt can become harder to bear the
more it piles up. Several studies have documented a negative correlation between rising public debt and economic
growth. While correlation does not necessarily imply
causation, it is easy to see how public debt could harm the
economy. As debt increases, the required interest payments on that debt become a larger share of the budget,
crowding out other spending. This has become a concern
in the United States as public borrowing has grown to
unprecedented levels.
“Right now, our debt-to-GDP ratio is the highest it
has ever been except for a few years around World War
II,” says William Gale, a senior fellow at the Brookings
Institution and co-director of the Urban-Brookings Tax
Policy Center.
In June 2018, the Congressional Budget Office reported
that the amount of federal debt held by the public was
78 percent of GDP, and it is projected to reach nearly
100 percent within the next decade. (See chart on next
page.) As a result of growing debt and rising interest rates,
federal spending on servicing the debt is slated to soon
surpass several other major categories of government
spending, such as the military and Medicaid. As the government devotes more resources to interest payments, it
leaves less money for everything else.
Mounting public borrowing can crowd out private borrowing as well. As the government issues more debt, it may
eventually be forced to offer higher interest rates in order
to attract new investors. Rising interest rates make it more
expensive for private firms to borrow. They must either
offer higher interest payments on their own debt, find
other ways to finance their investments, or shelve projects
until rates fall. To the extent government borrowing crowds
out private investment, it may reduce overall productivity,
which is the ultimate driver of long-run economic growth.
“My late colleague Charles Shultz used to say that deficits are not the wolf at the door, they’re more like termites
in the woodwork,” says Gale. “They eat away at the foundation of the economy.”
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
11
U.S. Debt Moving Toward Historical Peak — and Beyond
Federal Debt Held by the Public as a Share of GDP
160
Actual Projected
140
PERCENT OF GDP
120
100
80
60
40
20
0
1790 1810 1830 1850 1870 1890 1910 1930 1950 1970 1990 2010 2030
NOTE: Vertical line indicates 2018.
Source: Congressional Budget Office
There is no consensus among economists about when
public debt becomes a problem for economic growth.
But it is clear that as a country accumulates debt, sooner
or later it becomes more expensive to continue borrowing. High debt levels can prompt creditors to wonder if
the borrowing nation will ever be able to repay its debts.
That concern translates into higher interest rates on the
nation’s debt to reflect the higher risk of default. In addition to making existing debt more costly, this can limit the
government’s ability to borrow during future emergencies.
Historically, federal debt has risen during economic contractions to fund government stimulus programs. During
the last recession, federal debt held by the public rose from
35 percent as a share of GDP to 52 percent. In the past, debt
levels have tended to fall during economic expansions. But
nearly 10 years after the end of the Great Recession, federal
debt continues to rise and shows little sign of changing
course. This may leave less room to fund a fiscal expansion
to stimulate the economy during a future recession.
Given the costs associated with large levels of public
debt, countries might be tempted to simply renege on
what they owe. But history suggests the costs of doing so
are often much higher.
Enforcement
King Philip II of Spain defaulted on his country’s debt
payments four times during his reign from 1556 to 1598.
Embroiled in war for much of his rule, it is little wonder
the monarch accumulated sizable debts. Less clear is how
he was able to continue borrowing from private banks
after repeatedly demonstrating his unwillingness to repay
what he owed. Can creditors actually punish a sovereign
nation for defaulting?
Private debt is typically secured by some type of collateral, which exposes the borrower to a cost should they fail
to repay. If a borrower defaults on a mortgage or car loan,
for example, creditors can claim the underlying house or
car to recoup the lost value of the loan. But when a nation
defaults, it is less simple for creditors to lay claim to that
nation’s assets.
12
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
“It’s not hard to get a legal judgment against a country
that is in default validating that they owe you money,” says
Mark Wright, research director at the Minneapolis Fed.
“The problem is actually collecting.”
In the case where the creditors are sovereign nations
themselves, they may be able to use diplomatic or military
pressure on defaulters to collect what they’re owed. This
sort of “gunboat diplomacy” was more common at the
turn of the 20th century than it is today. In a 2010 article,
Kris James Mitchener of Santa Clara University and Marc
Weidenmier of Chapman University documented a number of episodes from 1870 to 1913 where creditor nations
took military action against delinquent borrowers. For
example, a group of European nations imposed a naval
blockade on Venezuela in late 1902 to early 1903 over
delinquent debts.
Evidence on the effectiveness of such direct intervention is mixed. Moreover, it isn’t an option available to
private creditors. But in a 2011 article entitled “Lending
to the Borrower from Hell,” Mauricio Drelichman of the
University of British Columbia and Hans-Joachim Voth
of the University of Zurich described how a coalition of
private bankers did exert power over King Philip II of
Spain: They cut him off from future borrowing.
Most of King Philip’s loans came from the same group
of Genoese bankers, giving them considerable power over
the monarch’s future credit. According to Drelichman and
Voth, the bankers would refuse to lend until the monarch
resumed payments on his past debts. “The king’s borrowing needs were so high that he would eventually have to
settle with the Genoese coalition,” the authors wrote.
Even in modern times, the pain of credit market
exclusion remains a very real cost for governments facing
default. In a 2018 paper, Anusha Chari and Ryan Leary
of the University of North Carolina at Chapel Hill and
Toan Phan of the Richmond Fed found that as Puerto
Rico’s debt crisis worsened, borrowing became increasingly expensive. This in turn hurt employment growth and
increased the cost of capital.
Private lenders may also be able to use legal proceedings to enforce sovereign debt contracts. While it was
long believed that creditors had little legal power over
sovereigns, a recent paper by Julian Schumacher of the
European Central Bank, Christoph Trebesch of the Kiel
Institute for the World Economy, and Henrik Enderlein
of the Hertie School of Governance argued that lawsuits
against defaulting nations have become much more common over the last several decades.
After Argentina defaulted in 2001, a hedge fund that
held some of the country’s debt refused to accept a
restructuring deal and instead filed a lawsuit to demand
full repayment. U.S. courts ordered Argentina’s bond
trustee not to process payments to its other creditors
who had agreed to the debt restructuring until it paid
the holdouts who had not. The injunction resulted in
Argentina defaulting on its restructured debt in 2014 and
loss of this reputation negatively affects a government’s
ability to borrow in the future.
Even setting aside the reputational costs, it’s unclear
that attempting to inflate away debt is always effective.
Some scholars have pointed to the elevated inflation
of the years immediately following World War II as
instrumental in easing America’s wartime debt burden.
Building a Reputation
Indeed, Joshua Aizenman of the University of Southern
Another long-term cost defaulting sovereign nations may
California and Nancy Marion of Dartmouth College estiface is damage to their reputations, which can affect the
mated in a 2011 paper that inflation was responsible for
terms they receive from credit markets in the future.
reducing the postwar debtThe incentive to rebuild
to-GDP ratio by more than
that reputation can explain
a third over the course of a
why, even in the absence
One of the things that
decade.
of direct enforcement,
puzzles researchers is that some
But Aizenman and
governments that have
Marion argued that it is
defaulted will restructure
countries are able to borrow a lot
unlikely such an intervendebt agreements with credwithout defaulting while others can
tion would work as well
itors and seek to prove
today. Average maturity for
themselves as trustworthy
only borrow very little.
U.S. debt was more than
borrowers once again.
twice as long in the late
In a pair of 2017 arti— Mark Wright, research director at the
1940s than it is today, makcles, Phan of the Richmond
Minneapolis Fed
ing it more susceptible to
Fed showed how sovereign
surprise inflation. Today,
debt acts as a reputational
rising inflation would be
signal to investors. Foreign
met
with
creditor
demands
for
higher interest rates or
creditors in particular do not have full information about
inflation-indexing
on
future
debt
securities, limiting the
the government they are lending to. Default signals that
power
of
inflation
to
diminish
the
debt burden. Thus,
the government is unreliable, which will dissuade foreign
inflation
doesn’t
necessarily
help
the
debtor government
investment. When governments restructure and repay
get ahead.
their debts after a default, they are signaling improved
“There is also some evidence that countries that run
political and economic conditions in order to attract new
high inflation to escape debt end up destroying their finanforeign investment. Phan showed that, in theory, some
cial markets, and it can take a long time to recover from
countries may even borrow not because they need the
that,” says Wright.
money but because they want to send these positive signals to investors.
The Breaking Point
“Historically, we’ve seen that countries in default typAs history shows, attempting to escape sovereign debt
ically don’t borrow a lot, or if they do borrow, it is at very
through default or strategic inflation rarely pays off. But
high rates,” says Wright. “That suggests they are facing
what happens when default becomes inevitable rather
worse terms as a result of the default. But is it because
than a choice?
everyone sees that they are unlikely to repay because they
Predicting when a country will be unable to susjust defaulted and their economy is not doing very well? Or
tain its debts is fraught with difficulty. Although the
is it because they are being punished?”
debt-to-GDP ratio is an oft-reported metric of public
Economists disagree about which of the two explanaindebtedness, it is not necessarily the best indicator of
tions drives the market response to default. What is clear
debt sustainability. For example, Greece’s debt-to-GDP
is that defaulting countries lose access to markets until
ratio was 126 percent when its debt troubles began in
they are able to restructure their debts and rebuild their
late 2009. Meanwhile, Japan’s debt-to-GDP ratio surreputations, and Wright’s research suggests this can take
passed 200 percent in the same year and has remained
a long time — roughly seven years on average.
above that threshold for nearly a decade with no signs of
Reputation may also explain why attempting to
impending default.
lighten the load of debt issued in a country’s own cur“One of the things that puzzles researchers is that some
rency by engineering inflation or currency devaluation is
countries are able to borrow a lot without defaulting while
rarely successful in the long run. Phan’s research shows
others can only borrow very little,” says Wright.
that the reputational costs of strategically inflating away
The spread between the interest on a sovereign’s debt
debt are similar to those of defaulting. Countries that
and a risk-free rate can be a sign of impending crisis. For
devalue their currencies to escape debt lose credibility
example, as the Greek crisis intensified, the yield on Greek
with regard to monetary stability and independence. The
ultimately prompted a new settlement with the holdout
creditors. The legal rulings that led to that injunction
were somewhat controversial, however, so it’s not clear
that future creditors would necessarily have the same
success.
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
13
bonds increased from 3 to 9 percentage points higher than
the relatively riskless German bonds. But this spread typically only spikes when a default crisis is imminent, leaving
little time to prepare.
The strength of a country’s economic growth relative
to the growth of its deficits can be another signal of future
difficulties. While current economic growth in the United
States is strong and is projected to remain so, government
revenues remain too small to prevent public debt from
increasing, says Gale. Still, that in itself may not necessarily be a concern.
“I don’t see anyone pricing in a default premium into
the U.S. debt for economic reasons anytime soon,” says
Gale. “We’re a strong country, a safe place to invest, we
print our own currency, and our inflation rate is low.”
But political standoffs over the debt ceiling could be
a different story. After the 2011 political battle led to the
S&P downgrade, Congress again fought over the debt
limit in 2013. In a 2015 report studying the aftermath of
the event, the Government Accountability Office found
that interest rates on some Treasuries did increase, resulting in slightly higher federal borrowing costs.
Predicting the likelihood of sovereign default may be
next to impossible, but history shows the costs of such
episodes. Once lenders re-evaluate a borrowing nation’s
creditworthiness on the basis of new information, the
adjustment can lead to swift and significant economic
consequences.
EF
Readings
Aizenman, Joshua, and Nancy Marion. “Using Inflation to Erode
the U.S. Public Debt.” Journal of Macroeconomics, December 2011,
vol. 33, no. 4, pp. 524-541.
Schumacher, Julian, Christoph Trebesch, and Henrik Enderlein.
“Sovereign Defaults in Court.” European Central Bank Working
Paper No. 2135, February 2018.
Phan, Toan. “A Model of Sovereign Debt with Private
Information.” Journal of Economic Dynamics and Control, October
2017, vol. 83, pp. 1-17.
Tomz, Michael, and Mark L.J. Wright. “Empirical Research on
Sovereign Debt and Default.” Annual Review of Economics, August
2013, vol. 5, pp. 247-272.
Economic Brief
Economic Brief publishes an
online essay each month about
a current economic issue
December 2018, EB18-1
2
Have Yield Curve
Inversions Becom
e More Likely?
Wissuchek, and Alexa
nder L. Wolm
By Renee Haltom, Elaine
an
The recent flattenin
g of the yield curv
e has raised conc
is around the corn
erns that a recessio
er. Such concerns
n
stem partly from
inversions have prec
the fact that yield
eded each of the
curve
past seven recessio
factors affect the
ns. However, othe
yield curve’s shap
r
e besides the expe
the economy. In part
cted future health
icular, a low term
of
premium — as has
recent years — mak
been observed in
es yield curve inve
rsions more likel
recession has not
y
even if the risk of
increased at all.
Recent changes in
the yield curve have
raised
questions about whet
her a recession is
likely in
the near term.
December 2018
Have Yield Curve Inversions Become More Likely?
Yield curve inversions have preceded each of the past
seven recessions, so the recent flattening of the yield
curve has fueled speculation that another recession
might be imminent. But the latest Economic Brief shows
how the low term premium (compensation for holding
long-term rather than short-term bonds) could mean
that yield curve inversions are more likely even if the risk
of recession has not increased.
November 2018
The Differing Effects of the Business Cycle on
Small and Large Banks
economy. The Feder
al Open Market Comm
ittee
(FOMC) acknowledg
ed this in the minu
tes from
its September meet
ing:
The yield curve is
a graph depicting
yields on
U.S. Treasury bond
“A
few participants offere
s at multiple matu
d perspectives on
rities. One
can visualize yield
the term structure
curve behavior over
of e st rates and
time by
differencintere
what
plotting shorter-ter
a poten
the
dual,tial
invers
resi
m Treasuries and longe
ion of the
sured; it is a signal about
r-term
rage of ex- yield curve might
Treasuries, as show
directly mea
and the aveeconomic prospects in light
n in
Figure be
ratethe
m
in
1
on
-ter
the
lock
follow
to
of
long
page. When
orderthe two series
ween the ing
s.historical regularity that an inverted
betcloser
rate
ing to pay in
if move
rm
will
that
rt-te
yield
are
es
s
curve
sho
toget
re
has often preceded
the belocurve
w argu
what investor
becomntly
pected futu her,
the
analysisyield
sions in thecom
ent — onset of recesn rece es flatter. Figure 1 shows
return. The thatas
pon
has bee
d States
it yield
first Unite
a long-term
curve’s slope
. On the one hand,
low the
the
as
s
that
has
e
n
stay
been
an
inverted—
ium
declining
es the
nt to not
since
mak
t it has bee
curve
early
ges2014.
s yield
could indicate an
the term prem
s sug
It is importa
rm rate
increase
of
ular measure
d risk of n time,
ected short-te
will become
ion;
on the other hand,
— indeed, pop
any giverecess
average exp
e inversions
the. At
low
level ofard
downon average
then yield curv
or prem
n has not
As theofyield
iums in recent
d curve flat
e upw term
curve
negative —
risk recessio
has flattened inyiel
yearsslop
e can
— reflecting, innd
recen the yiel
relacentra
t even if thequestions
arou part,
hs,d curv purch mov
es
l
have intensified abou course,t mont
bank
more frequen
asset
ases — could temp
,
t its predictive
ent short rate
all.
er the reliability of
power. An inverted
But on average
d as the curr
increased at
the
rt rates.the
yield curve, or war
shoslope
yield curve
ater
a situatexp
pe?
ion in
ected futureof future of
as an indicator
which
Sharates
longve’sterm
be neither gre
tive to
d Cur
s willecono
are lower than
mic activi way
1
ines the Yiel
short d future short rate
sed on the
rates, may
n focust
. An intuitivety.”
What Determ
expecte-term
thatThe
nt attentiosugge
marke
ent short rate
in- ts expec
or
curr
the recesion
maj
t
the
pe.
a
and
reces
of
sha
e
erst
than
its
and thus
n lower intere
less
This Econothe absenc
To und
break dow
rates innor
featu rateRichm
s
the future. ut this is that in mic Brief
bondsthas
it helps to Indee
ond Fed
Treasury
d, anminver
y, interestres
reseathe
yield curve,
to think abo
nom
ted yield rage
asses
a long-ter
sing
how one ers to rcheco
red on each
prece
ded
is the ave curve has stru
al changethe
entseven
of the
ctur
around a longof these factors,
pon
past
terest rate offe
e ium,
term prem
recessions (also show
first com
may affect the frequ
d to fluctuat
ents. TheFigure
d to prevail
n inexpecteyield
ency of
two compon
would be
are expecte
curve inversions. The
s that 1).
cy,
rate
poli
y
rm
term premium refers
.
of the short-te
ected monetar
term average to the extra compensat
Exp
d.
e
e
bon
ion invest
d curv
of the At the same
influenc
ors demand (in
y, will
that the yiel
terms oft,highe
over the life
time,
other thing
r intere
of the econom
poin the fact
st rates) term
healthyield
the
if a recessions influence
curve
n the previous
suggests theto hold longermple,
term
e
besid
and thus the
exa
Give
asset
slop
es
For
the
s
rathe
ard
future
r than short
strength of the has had an upw
s not
ent heavily.
er-ter
short-term
m assets.
rage. This doe
If the term prem
this compon
usually
expect lower
on ave
ium is negat
would
er itthe
investors may
been positive
ive,
nev
represents
all else equal
is expected,
premium has
negative whe
future, which
current
premium is
s in2the
EB18-1
- Federal Reserve
interest rate
just noted, the
mean the term
d curve.
e of the yiel Bank of Richmond
n — since, as
ent than
slop
dow
es
mom
the
n
ce
slop
d
give
redu
yiel
in any
ld be higher
mium. As
short rate cou
s.
is the term pre
Page 1
re short rate
component
s demand to
The second
expected futu
sation investor
cently
not
the compen
premium can
Fla�ened Re
noted, this is
lds
bonds. The term Yie ld Curve has
yie
erm
ry
er-t
The
hold long
th Treasu
on
an4d 3-m
y 201
e Earlry
sincasu
ar dTre
-yetene
10Flat
Has
Yield Curve
Figure 1: The
1818
1616
1414
1212
October 2018
Inequality in and across Cities
Percent
1010
88
66
44
22
0
To access the Economic Brief and other research publications,
visit: www.richmondfed.org/publications/research/
14
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
1965
1970
1975
Year Treasury
1980
Yieldear
Note10-y
2010
2005
2000
1995
1990
Yield
Treasury Bill
sury Bill Yield
e-Month Trea3-Month
ThreYield
Note
Treasury
2015
1985
tics
Tenm, Haver Analy cs
ral Reserve Syste er Analy�
the Fede
rnors of erv
e Board / Hav
Board of Gove
eral Resreces
Source:
sions.
rce : Fed
Sou
areas indicate
Note: Shaded
Page 2
ATTHERichmondFED
What to Do When Large Firms Fail
B y Re n ee H a l t o m
Highlighted Research
“On the Measurement of Large Financial Firm
Resolvability.” Arantxa Jarque, John R. Walter, and
Jackson Evert. Working Paper No. 18-06R, February
2018 (revised July 2018).
T
he financial crisis of 2007-2008 confronted policymakers with the question of how to handle large firms
that get into financial trouble. During the crisis, some
failing firms went through bankruptcy, but others were
rescued by emergency loans or other forms of support
from the government.
There are costs to either choice: Bankruptcy may leave
a substantial mess in terms of costs on other financial
market participants or the overall economy. For example,
there could be “fire sales,” when large quantities of assets
are sold quickly to raise funds, causing asset prices to fall.
Costs also could arise through “contagion,” when firms
have a financial or operational relationship such that the
failure of one disrupts others. Bailouts, on the other hand,
minimize those spillovers, but they create potentially
more costs in the future by providing an incentive to take
risks in the first place.
It’s not an easy choice, and how policymakers make
the decision has historically not been transparent. Two
Richmond Fed economists, Arantxa Jarque and John
Walter, aided by former research associate Jackson Evert,
have proposed a tool that could help. Jarque and Walter
created a framework for weighing the trade-offs using
objective metrics.
“Many aspects of the potential costs of a firm’s failure
are hard to measure, for example, the likely magnitude of
fire sales,” explains Walter. “But it is reasonable to think
those hard-to-measure costs are correlated with characteristics that we can objectively measure, such as a firm’s use
of financing tools that may be most subject to fire sales.”
The researchers combined various firm characteristics
— such as their connections to other firms and reliance on
certain types of debt contracts — into an overall “impact
score” that represents the costs of a firm’s failure. In principle, this allows a comparison between the impact score
from bankruptcy and the impact score from bailouts. If
the score under bankruptcy is lower, that firm is “resolvable” in the sense that a hypothetical policymaker would
not choose bailouts. But if the bankruptcy score is higher,
one implication could be that regulators and firms may
want to consider changes to avoid bailouts.
Their score design accounts for the fact that policymakers may have different views on how the financial
system works. That may influence whether they prefer
bankruptcy to bailout. Jarque and Walter illustrated how
these differences of opinion may affect a policymaker’s
decision by computing the score for different hypothetical
policymakers — for example, one who believes firm size is
the most important variable and one who doesn’t believe
fire sales are important.
Overall, the framework provides a tool that could help
policymakers choose between bankruptcy and bailout.
Such a tool also could make the decision more transparent
to the public and hold policymakers accountable, which
were concerns many observers raised during the 20072008 crisis.
As they dove into the research, Walter says he was
fascinated to learn in detail how large, globally systemic
institutions differ from one another in their financial
structure and activities. “It was challenging to very carefully think through which financial characteristic of a
firm might produce which impacts on the financial system — for example, which items are related to fire sales
and which to contagion. The academic literature is still
working through these issues.”
There remains more they would like to do with the
score. “Many of the measurable characteristics that we put
in the score were not measured for these firms back when
they got in trouble,” Jarque says. “This prevents us from
using past failures to learn about the views of past policymakers and from validating our score by comparing firms
that failed and those that didn’t. We would like to explore
a simplified version of the score that would allow us to use
historical data in this way.”
The work adds to a body of work at the Richmond Fed
on the effects of large firm failures and the “too big to fail”
problem. Walter helped create the “Bailout Barometer,” a
measure of the share of the financial system that has benefited from bailouts — one gauge of future risk. And Jarque
has studied living wills, the plans large financial firms
have been required to make describing how they could be
wound down without government support in the event of
failure. Living wills are another tool for minimizing bankruptcy costs and avoiding bailouts.
All this work supports a better understanding of financial stability. “Our bank examiners, our analysts who
work with banking data, and many other people around
the Fed System and at the Federal Deposit Insurance
Corporation do tremendous work in monitoring large
systemic financial institutions,” Jarque says. “We learned
a lot from talking to them and reading about the evolution
of their approach to evaluating living wills, for example. It
is inspiring for future research.”
EF
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
15
Interview
Antoinette Schoar
Editor’s Note: This is an abbreviated version of EF’s conversation with Antoinette Schoar. For additional content, go to our
website: www.richmondfed.org/publications
16
EF: There has been a lot of talk recently about declining business dynamism in this country — that is,
fewer businesses are being opened or closed. There’s
concern that this may be lowering productivity and
economic growth. Do you see a problem here?
Schoar: There is lots of research showing that the number
of people who are employed in small and young firms has
gone down dramatically over the last two decades in the
United States. Also, the number of small businesses that
are being started is going down. It’s very concerning for
the United States, which has always prided itself on entrepreneurship. This trend is very strong in the data, but I
also believe it’s not the full story.
It’s not that there are no startups in the United States,
especially on the high-tech side. If anything, the United
States is where startup financing like venture capital or
angel finance is really the most vibrant.
A second trend that is very concerning is that the
way small businesses exit has changed dramatically over
the last 20 years. It used to be that the large fraction
of them went IPO. Now, the vast majority are sold to
companies and end up being small divisions of a much
larger company. In the long run, we might be worried if it
means the whole economy becomes more concentrated.
That’s a big debate. It’s not so clear yet whether these
firms have almost natural monopolies, in which case we
should be worried about rent extraction, or whether it’s
P h o t o g r a p h y : Evg e n i a E l is e e v a
Antoinette Schoar, an economist at the Massachusetts
Institute of Technology, is known for uncovering
surprising trends in corporate finance, but her original economic interests lay elsewhere. “I grew up in
Germany, but my father is originally from Iran,” she
says. “Seeing the differences in income inequality and
poverty between those countries, I felt this is something I want to understand.”
But upon arriving at the University of Chicago for
her Ph.D., Schoar realized a wide range of economic
decisions — affecting issues ranging from labor markets to development to economic growth — ultimately
run through finance. Her academic adviser, Sherwin
Rosen, suggested she talk to colleagues at the business
school, and the rest is history.
Schoar’s body of work is as wide-ranging as the field
of corporate finance itself. A particular focus has been
entrepreneurship: New firms have become an increasingly important source of growth and productivity, but
data on them have historically been scarce. Schoar’s
work has shed light on the many ways new firms get
funded and the managerial capital that investors bring
to startups, as well as the role of management styles
generally in a firm’s success. She has documented that
the so-called “subprime” housing crisis centered largely
on middle- and upper-middle-class households. And
Schoar’s recent work has branched out to consumer
credit, finding that credit card firms target moreshrouded offers to less-sophisticated consumers. She
discusses all these topics and more in this interview.
Schoar co-chairs the National Bureau of Economic
Research’s program on corporate finance. She
was previously a co-organizer of the NBER’s
Entrepreneurship Working Group. In 2009 she won
the Kauffman Prize Medal for research in entrepreneurship. She also co-founded ideas42, a nonprofit
that uses social sciences research to solve social
problems.
Successful academics need to be excited by the
research itself, Schoar says. “And with the freedom
you have of designing what you do, the exciting people
you can work with, the great students … I really feel
very privileged.”
Renee Haltom interviewed Schoar in her office at
MIT in September 2018.
technological innovation giving
some big firms an advantage.
EF: Is access to credit for new
firms part of the problem?
pricing is still competitive. But
for Amazon and other large
firms, it really is not about how to
price one widget versus another
but rather having more and more
data about how consumers shop,
how their preferences manifest.
For the consumer, it’s great — it
keeps prices down and gives free
access to all the search functions. Disrupters would make that industry less efficient.
It’s not a good trend that we’re seeing
fewer IPOs and many more acquisitions.
It’s not at the startup level where the
pipeline is broken; where something
seems to be changing is that these small
firms don’t become the next Google, the
next disruptive big firm.
Schoar: I would say credit is
not the culprit here. One has
to be very careful in differentiating between the startups that
are new, disruptive technologies — think about Boston,
Silicon Valley — and the kind of small businesses that are
not necessarily disrupting existing firms.
Post-2008, credit to small businesses did initially plummet. Lots of small businesses went bankrupt, and the flow
of new ones into the economy dropped. But the rate of
startup creation recovered relatively quickly. Venture
capital at the coasts, where there is a lot of entrepreneurship, recovered, and these areas have been very vibrant. If
anything, the startup economy in the sense of disruptive
financing is very deep in the United States. Some people
have said there was even an oversupply of entrepreneurs.
Where I worry is growing firms beyond the startup
level. That’s why it’s not a good trend that we’re seeing
fewer IPOs and many more acquisitions. It’s not at the
startup level where the pipeline is broken; where something seems to be changing is that these small firms don’t
become the next Google, the next disruptive big firm.
Some of it might be financing, but I feel a lot of it is that
the structure of industries is changing. Venture capitalists
call it “escape velocity”; many firms don’t have the escape
velocity to become standalone. It’s much better for them
to just be acquired and benefit from the fact that big existing firms have a big network of customers.
But what I worry about is that for entrepreneurs, it’s
not great if there are few exit options — say, being bought
by Google, Facebook, and Amazon. If these three firms
determine price, for entrepreneurs it may mean that their
valuations will be depressed. Growing up in Germany, I have
seen a market where there are few IPO opportunities, where
entrepreneurs know that the only exit options are a few large
firms like Siemens or Bosch that dictate the price at which
you can sell out. In the long run, this reduces incentives for
entrepreneurs. American venture capital firms came into
Germany and really shook up the dynamism.
To me, the real sticking point in the United States is
that access to data is becoming more difficult for small
startups that want to disrupt a market. In the modern digital world, the quality of the machine learning algorithms
that you can set up depends on how much data you have
and how good they are. This network effect story means
that it becomes tougher in those industries for newcomers
to disrupt incumbents.
I feel policy should be more mindful about this in
the United States. Right now, if you talk to people who
think about cartel enforcement, they look at whether
EF: Do angel investors have a special role in facilitating these coveted high-growth startups? Your work
with Josh Lerner of Harvard Business School on angel
investors has been some of the first on the topic.
Schoar: In the United States over the last decade, we’ve
seen many new online models of angel financing. In
research with Josh, we show that the impact of angels is
very positive on the firms where they invest. We test this
by looking at firms that were just on the cusp of being
accepted versus rejected by angels, the idea being that
these firms are probably quite similar, and we compare
their ultimate outcomes.
Getting financing from angels has a very positive impact
on your survival, growth rate, and revenues three to five
years out. What we were very surprised about is that it
doesn’t seem that it is the funding that the angels facilitate;
the two sets of firms were equal in the amount of funding
they received over the next five years, so it’s not the case
that without an angel you don’t get more capital. We think
it’s the advice, connections, and help the angels are giving
that really makes the difference — giving you a sense of
when to grow the business, who to hire at each stage, course
correction, all that. It’s actually much more the human or
“managerial” capital that comes with the angel.
We did another paper using similar data with angels
around the world, in Europe and South America. Again, we
found angels have a very big and positive impact, but there
a lot of it was because of financing. There we found that if
you didn’t get angel financing, it was much less likely for you
to get follow-on funding. On top of that, we found in most
countries outside the United States, the firms that get angel
funding are much older and already mature; they are already
cash-flow-positive compared to the United States.
EF: As informational frictions decrease, do you think
angels will become even more important as a source of
financing for new businesses? In terms of freer access
to information raising the marginal value of angels’
unique expertise identifying the highest-potential
startups?
Schoar: Some informational frictions have reduced
because of technology, but a lot of the judgment about the
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
17
Antoinette Schoar
quality of an entrepreneur, the subtle
data on private equity returns. The
➤ Present Position
differences in the quality of the busipaper got a bit of notoriety because
Michael M. Koerner (1949) Professor
ness, are still very difficult to decide
we found three things that were very
of Entrepreneurship, Massachusetts
for investors.
stunning and counterintuitive in
Institute of Technology; Co-Director,
I think the rise of angels comes
finance.
Corporate Finance Program, National
from the fact that we have more and
First, as you said, we found that
Bureau of Economic Research
more people in the United States
there was persistence in returns
➤ Education
who were successful entrepreneurs
even over quite long time periods,
Ph.D. (2000), University of Chicago;
and made some money and now have
on both the good end and the bad
Diploma in Economics (1995),
a combination of skills required to
end. Partnerships that had good
University of Cologne
understand what an entrepreneur
performance tended to have good
needs. They are often still very young,
performance from one fund to the
➤ Selected Publications
so they have the energy to want to do
next, and funds that were in the
“Managing with Style: The Effect of
Managers on Firm Policies,” Quarterly
more than just sit at home. After
top 25 percent persistently stayed
Journal of Economics, 2003 (with
firms like Google and Facebook went
there. But partnerships that were
Marianne Bertrand); “Private Equity
public, you had a wave of people leavin the bottom had several funds in
Performance: Returns, Persistence
ing these firms who were maybe 35,
the bottom. Persistence basically
and Capital Flows,” Journal of Finance,
who made a lot of money, and who
means predictability, and that’s obvi2005 (with Steven N. Kaplan); “The
turned themselves into angels.
ously very different from public asset
Market for Financial Advice: An
It spurs entrepreneurship — in
classes where you have no predictAudit Study,” NBER Working Paper,
certain pockets of the country, this
ability. That’s bizarre.
2012 (with Sendhil Mullainathan and
activity feeds on itself. These sucOn top of that, we found that in
Markus Noeth); “Loan Originations
cessful early entrepreneurs become
venture
capital and private equity,
and Defaults in the Mortgage Crisis:
angels, and they support the ecosysthe
relationship
between perforThe Role of the Middle Class,”
tem of entrepreneurship. In emergmance
and
fund
flow is concave
Review of Financial Studies, 2016 (with
ing markets like India and China, it
when
everywhere
else
it’s convex; in
Manuel Adelino and Felipe Severino);
“House Prices, Collateral, and Selflooks very similar — in places where
other words, the best funds in venEmployment,” Journal of Financial
you have a lot of entrepreneurship,
ture capital and private equity don’t
Economics, 2015 (with Manuel Adelino
the process has a positive loop.
grow as quickly as the medium-good
and
Felipe
Severino);
“Do
Credit
Card
Our research looked at some of the
funds. If you look at the mutual fund
Companies Screen for Behavioral
most successful angel groups in the
industry, it’s exactly the opposite.
Biases?” NBER Working Paper, 2016
country, and it would be interesting
There’s a ton of research over more
(with Hong Ru)
to have an even wider lens on all the
than two decades finding that mutual
different angels who are active in the
funds that perform slightly better get
United States and in other countries and see how much
massive increases in fund flows, and the ones that are in
heterogeneity there is.
the middle might see outflows.
In particular, in the United States, if the benefit of
Why this was particularly puzzling at the time is
angels really comes from the managerial capital they’re
that the top venture capital firms are what people call
bringing, there’s probably a lot of differences between
“oversubscribed,” meaning lots of investors would love
people, and so it would be good to see the distribution of
to invest with them. But in that time period, the 1990s
the angel quality, the matching between entrepreneurs
and early 2000s, they seemed to voluntarily stay smaller.
and angels, and whether that can be better facilitated.
What we concluded is that this seems to be an industry
There are online networks like AngelList that are trying
where the quality of the manager, the general partner,
to improve the introduction between investors and entrematters a lot. At the beginning, the high-performing
preneurs, but I think we are still in the process of figuring
managers didn’t say, “Let’s take all the money we can,”
out if this is even possible to do on a digital platform and
which might dilute the marginal performance. We saw in
how scalable that is.
our data that funds that grow very quickly see a reduction
in performance. It seems that manager quality is an asset,
EF: One of your most famous papers documented
a type, an area where it’s tougher to scale up.
persistence among private equity firms: that the
For a mutual fund, once I identify one great investbest-performing funds tend to continue being the
ment strategy, given how big the public market is, it’s
best performers. Can you explain why this was such a
more scalable. If you invest in Google, say, it is possible to
surprising result?
scale your investment — to invest $5 million, $50 million,
or maybe even $500 million. But with venture capital,
Schoar: That paper, from 2005 with Steve Kaplan at the
even if I identify a few really good startups, I can’t invest
University of Chicago, was the first to have large-scale
$500 million. Maybe I can invest $10 million, but then I
18
have to go and find another firm; it’s less scalable.
So that’s what we found. Lots of people found it very
surprising because it shows how different this industry is
from other financial industries.
What then happened was a misinterpretation of our
findings. A lot of investors in venture capital and private
equity said, “Because Kaplan and Schoar find there is
persistence, all you need to do is identify good firms and
then keep on investing no matter what.” But it’s not as
easy as that. Venture capital funds, in particular, might go
through cycles. They were really good and had fantastic
past performance, but that might change if they lose one
of their top managers.
EF: You also found that persistence has declined
recently. What changed, and why does it matter?
Schoar: Here we are 20 years later. In a paper we just
finished with co-authors, Josh Lerner and I look at this
same question using data from State Street, which is one
of the biggest custodians for investors in private equity
and venture capital. We found that this industry has really
transformed, and some of the puzzles we identified in the
1980s and the 1990s have changed.
First, we showed there are big differences between
limited partners — that is, the investors in a private equity
or venture capital fund. Some investors seem to be very
smart about identifying top funds but also about predicting when they will turn south. Other investors don’t have
that skill. We found that foundations, endowments, and
some of the experienced public pension funds are good
at making those decisions, but sovereign wealth funds
and banks that invest money on behalf of their clients are
much worse at it. They stay in a partnership even when
performance goes down.
Also, the way firms set themselves up has changed.
Before, most partnerships would raise money from lots of
limited partners and invest it, and all the investors would
get the same terms. Once you made it into a top fund, your
chances of getting great returns were quite high. And if
you went into a bad fund, everybody got the same returns
— in that case, bad returns.
The puzzle Steve and I originally identified was why
partnerships were willing to give different investors the
same terms. It’s like leaving money on the table, right?
If I’m a fund trying to raise money from one of the top
limited partners, I will be willing to give better terms
to a prestigious limited partner than when I’m trying to
raise money from a no-name investor who doesn’t bring
as much to the table in terms of liquidity or accreditation.
The industry is not stupid. If we as academics could see
and test this, they surely can see it too. So in the paper with
Josh Lerner, we show that firms have started to offer different deals and different investment vehicles to different
investors according to the bargaining power of the investor.
Even the very top funds give top limited partners access to
their best deals, but for less high-powered investors, they
provide investment vehicles that have lower returns.
It matters because as the industry is becoming more
competitive, it leaves much less rent on the table. It also
means the general partners are capturing more of that rent.
I think the shift was caused by a combination of
competition and the fact that some firms have really
manifested their reputation. In the 1980s and 1990s, this
was still a very young industry. Once you have that reputation, you monetize it. Now everybody is bidding like
crazy to get into the top funds, and they can now dictate
the prices to different limited partners.
EF: Switching to consumer finance: How are technology, big data, and fintech changing consumer financial
services?
Schoar: Fintech and big data and machine learning have
really changed the face of many financial services, away
from brick and mortar provision to online, on your cell
phone; it’s much more personalized than in the past.
The credit card industry was really early in this. In the
early 1990s, that was the closest to a machine-learning,
big-data approach one could get, mailing something to
your personal mailbox that’s very much targeted at you.
If I know that you are an educated young woman who is
interested in a certain type of leisure activity, I might send
you a credit card mailer that shows international travel or
going to a museum, things that might appeal to you.
EF: How granular are they getting? At this juncture,
are they profiling someone like you, or are they actually exploiting individual data for you?
Schoar: It’s a bit of both, actually. Machine-learning algorithms are using information about what you just did and
what people who look like you then did after they made a
similar choice. Once you buy a house, what are the other
financial services you might now need? Maybe you will
start renovating your house and therefore might want certain financial products.
On top of that, many financial subfirms are starting to
find that looking at your past financial behavior is very predictive of your future behavior. If you are somebody who
always pays on time and is mindful of your bills, it’s a very
good predictor of whether you will fall late or default in the
future. That’s very valuable information to financial service
companies. To be honest, it’s very valuable information to
you too, because if you are a well-organized, mindful person,
it means a lower cost of financial services.
So there’s definitely a good side to the fact that firms
can target your type better — it reduces the cost of capital
for everyone. I don’t want to lose sight of that. If you think
about the story of American personal finance over the
last 30 years, it’s really that personal finance was able to
expand so much because the banks were becoming better
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
19
at predicting who is a good payer
cards or small business loans
[The housing crisis] wasn’t just a story
and who is a bad payer.
changes.
of banks all making a mistake
We also lose track of the fact
But what we found is that
by lending to subprime borrowers.
that when you look at emerging
for the cards that have all these
The mistake, really, was not caring
market countries, like India or
shrouded features, when the
enough
about what would happen if
Cambodia, people have very litinterest rate goes up, the APR
tle access to personal finance.
doesn’t immediately go up with
collateral values went down.
The financial service industry is
it. Instead, some of the backnot at the same level yet, some of
loaded costs go up — late fees,
it because the banks themselves are not as sophisticated,
over-limit fees, penalty APR, all the things that are hidden
some of it because the infrastructure, such as credit scores,
from less-sophisticated customers. Those customers may
isn’t there. That’s a bad thing; it means a mother whose
not even understand that the cost just went up; they just
child is sick might not be able to get a loan for the treatlook at the low APR and keep on borrowing.
ment even though she would have the ability to pay it back
But this can have a delayed effect on monetary policy.
in the future. Or a small business that has a great idea that
Instead of immediately changing people’s demand for
might be relatively safe can’t get the credit.
credit, it might only change people’s demand for credit
But what my research with a former student, Hong Ru,
once the fees really hit. But then once the fees hit, it might
shows is that there is also a darker side to that personalalso mean that now people are really shocked because the
ization. Now that they can predict whether you’re less
costs are much higher than they thought they were. It
financially literate, in the credit card industry they target
might even create some credit risk for the banks and for
you with offers that are deliberately more complicated and
the people themselves.
more shrouded. Not only are they more complicated in the
What I wanted to highlight in Jackson Hole is that
deal terms, there are more hidden and back-loaded fees.
right now there is only a subset of consumer financial
We even see that the offer itself is more complicated —
products that uses these strategies intensively. But given
there is more distracting material on the first page and more
how much data are becoming available, this will become a
enticing material that shows you the great shopping expebigger channel. It might be something that actually affects
rience you can have. The cost of credit is buried on the last
how the Fed should think about the asymmetric effect
page, and, we show linguistically, using more complicated
that monetary policy can have on people who are finanlanguage when the consumer is less educated.
cially savvy and those who are not.
We believe it’s deliberate. More financially sophisticated people know that somewhere you have to tell me
EF: Innovations in credit also played a role in housing.
what is the cost of credit — so please don’t hide it from
The housing boom and bust was initially interpreted
me because maybe then I will be upset with the offerer.
as primarily a phenomenon centered on subprime
Whereas somebody who is not as financially attuned
borrowers. To what extent has that view held up?
might just think, wow, you are offering me a card with a
zero APR! They don’t think about the fact that the bank
Schoar: Research I did with Manuel Adelino at Duke and
also has a cost, so it must be hidden somewhere.
Felipe Severino at Dartmouth suggests it’s unfair to blame
In our data, we only see the offer side. There is research
it on subprime. It was a broad phenomenon across most
by Sumit Agarwal, John Driscoll, Xavier Gabaix, and
income classes.
David Laibson showing that customers who make misIn dollar value terms, mortgage credit to households
takes in the financial contracts they take up indeed pay
from 2000 to 2007 grew in particular for middle-class
much more for credit and would be better off taking a
and upper-middle-class people. They buy the big houses,
contract with fewer hidden fees. That seems to get worse
and therefore take the big mortgages. It’s stunning: If you
as consumers age. Right now we are working to match the
look at the top 1 percent, you see a drop in leverage, the
offer data with the user data, and then we will see basically
only group for which we don’t see an increase. It’s almost
the entire universe.
like houses were not getting big enough, their income was
growing so quickly over that time period.
EF: The implications for consumers are important
Why we think it’s so important is that we also find that
in their own right, but at this year’s annual Jackson
the largest growth in the dollar value of defaults post-2008
Hole central banking conference you also touched on
happened in the middle class and upper middle class.
potential monetary policy implications. Could you
This is really where the big dollars defaulted and also
talk about that?
where the banks were most caught off guard. I did some
research just after the financial crisis where we found that
Schoar: If you look at how monetary policy passes through
in many cases the banks couldn’t even reach the prime custo the consumer, typically what we look at is when the fed
tomers who defaulted. They hadn’t even bothered to take
fund rate changes, how the interest rate offered on credit
a phone number, they were so sure the customer would
20
never need to be reached again. This is the same bank that
is phenomenal at barraging you when you’re even just one
day late on your credit card payment. It was a mindset
that middle-class American customers never default on
their mortgages, so if the collateral is good enough, we
don’t have to worry about the quality or personality of the
borrower.
So it wasn’t just a story of banks all making a mistake
by lending to subprime borrowers. The mistake, really,
was not caring enough about what would happen if collateral values went down. This is important for Fed policy
because regulating misaligned incentives is much easier
than regulating stupidity. The economics profession is
reasonably good at understanding agency problems, but we
are still grappling with what the Fed should do to deflate a
bubble. First, you have to understand what a bubble truly
is, the thing we are worst at as economists. Even if you
could, it’s politically very difficult for a central bank to pull
the brake when everything is going well.
being used to bail out the housing market and being made
unavailable for other services.
EF: Corporate finance is a subfield of economics that
is not particularly known for its diversity. Have your
experiences there led you to believe the economics
profession should do more to improve diversity — and
if so, what?
Schoar: Obviously diversity is a big topic, and I think it’s a
subtle topic. There are some areas in economics I feel have
made bigger progress, in particular labor, public finance,
and development. And then there are areas like finance,
corporate and asset pricing, economic theory, and macroeconomics that really have made very little progress.
Part of it might be that women have interest in some
areas versus not. But finance especially is very broad.
Many of the questions are very close to labor and to
public finance, and so there is no reason why a woman
shouldn’t be interested in corporate finance if she is
interested in labor.
I have never experienced outright sexism in my field.
But in all of us, including women, there are implicit biases,
and I do think that matters. Even well-meaning people
might not be fully aware of the fact that when they listen
to a woman they somehow don’t take everything she says
as seriously, or they are more willing to believe that somebody else made that comment first and therefore attribute
it more to a man than a woman. What I see a lot is when a
man gets hired that somebody doesn’t like, it’s attributed
to a hiring mistake. But if there’s a woman hired whom
they feel didn’t fit in or deserve it, then it goes, “She was
hired because she’s a woman.” I think it’s not very helpful,
and it’s detrimental.
There is some interesting work by Anusha Chari and
Paul Goldsmith-Pinkham that has looked at participation
of economists in large, important industry institutions
like the National Bureau of Economic Research. They
find that the fraction of women who present, especially in
corporate finance and asset pricing, is low relative to the
fraction of women in the field. But they also find that the
average woman who presents at those meetings has a much
more illustrious CV than the average man. So it’s by far not
the case that women are so subsidized that less-qualified
women get to present anywhere, even though that narrative sometimes exists.
There are lots of efforts now to be aware of the biases and
to support women in the hiring process. But seeing female
role models, especially in finance, is still very rare. I feel
what really has to happen is that more women have to study
economics. If the pipeline is rich and is strong, that will
make a difference. But that can’t be the only thing, because
it would take many years before something changes. EF
EF: Have policymakers responded adequately to correcting misaligned incentives in housing finance, or
is there more that should be done?
Schoar: I feel many good things were put in place, and I do
strongly believe that incentives for banks improved.
The place where I’ve been disappointed is around mortgage loans. Fannie Mae and Freddie Mac were nationalized, and the taxpayer absorbed the losses from these
institutions. But we didn’t make any progress on their
regulation and their incentives. We are almost back to a
point where mortgage leverage is very similar to what it
was pre-crisis. The private securitization market has shut
down, which means the banks would not be as caught in
the fire if house prices were to go down, but there might be
losses on Fannie and Freddie and ultimately the taxpayer.
The other thing that really worries me is the mortgage origination sector. The big banks have significantly
reduced their origination activity, so now you have many
fintech lenders that are doing origination and securitizing
these loans via Fannie and Freddie. And while they have
very nice data because of the other services they provide,
we don’t know how good they are at being originators.
They all have one thing in common, which is that they are
extremely thinly capitalized, so they are bearing no true
mortgage risk; it all immediately gets passed on to Fannie
and Freddie.
To me that is worrisome, because Fannie and Freddie,
by the nature of being government institutions, may not
price credit risks correctly and therefore might indeed
feed a bubble again in the housing market. That could
feed a different form of downturn — in this case maybe
not a disruption in the banking industry, but tax money
u
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
21
EconomicHIStory
Founding America’s First Research University
Johns Hopkins put American higher education on the path to world domination
By K a r l R h o d e s
I
n 1872, Daniel Gilman, president of the University
of California, Berkeley, articulated his vision of what
a university should be. During his inaugural address,
he argued that the mission of a university should be “to
advance the arts and sciences of every sort and to train
young men as scholars for all the intellectual callings of life.”
Gilman further stated that universities should elevate
scientific research to the same level as the study of language, history, literature, and art. “Give us more and not
less science,” he demanded. “Encourage the most thorough and prolonged search for the truth which is to be
found in the rocks, the sea, the soil, the air, the sun, and
the stars; in light and heat, and magnetic forces; in plants
and animals, and in the human frame.”
Such ideas were radical in 1872, a time when most
American colleges still focused on teaching Latin, Greek,
and mathematics to undergraduates. The advancement of
knowledge — especially scientific research — was rarely
encouraged.
Near the end of his speech, Gilman imagined what
Berkeley would be like 100 years in the future. “I see a
flourishing University,” he prophesized. “Students are
flocking from east, west, and south, from South America,
and Australia, and India, from Egypt and Asia Minor, with
the ease and rapidity of a locomotion not yet discovered.”
Gilman’s address was eloquent enough but not sufficiently persuasive. He struggled to sell his plan to state
legislators who had their own agendas. He also encountered an aggressive farm lobby that wanted the fledgling
land-grant university to focus primarily on agricultural
and mechanical arts. After two years of slow progress and
growing frustration, he resigned to become the first president of Johns Hopkins University — transplanting his
dream from Berkeley to Baltimore.
Back on the East Coast, funded by the unfettered
bequests of Johns Hopkins — a recently deceased business owner and investor — Gilman and the university’s
trustees established the first research university in the
United States. It was a hybrid of the German model that
emphasized graduate research and the British model
that focused on undergraduate education. The founding
faculty members added the uniquely American features
of greater academic freedom and closer collaboration
between professors and students. They made the laboratory and the seminar the primary centers of learning. They
conducted research and published the results in academic
journals, including several they started themselves.
22
The founding of Johns Hopkins was “perhaps the
single, most decisive event in the history of learning in
the Western Hemisphere,” according to the late Edward
Shils, a University of Chicago sociologist. Shils’ assessment may go a bit too far, responds Jonathan Cole, former
provost of Columbia University and author of The Great
American University. “Nevertheless,” Cole adds, “Gilman’s
molding of Hopkins’ mission represented the beginning
of the great transformation in American higher learning.”
The Economic Impact of Higher Ed
Before the 20th century, American colleges and “universities” were small, and their economic impact was negligible.
Today, American research universities contribute to the
economy in at least three primary ways. First, they have a
direct impact by employing people and purchasing goods
and services. Second, they help students acquire human
capital — knowledge and skills that are useful to employers. And third, they increase productivity by conducting
research that creates new knowledge or applies existing
knowledge in innovative ways.
Johns Hopkins is a good example of the first and second
of these channels. In the university’s 2014 fiscal year, it paid
$3.9 billion in wages and spent $1.5 billion on goods and
services, including construction. During the spring of that
year, more than 20,000 students were enrolled in for-credit
programs at Hopkins. But the third channel — creating and
applying knowledge — is more complicated. Nationally,
Hopkins has ranked first in research spending for 38 consecutive years. In 2016, the university spent more than
$2.4 billion on research and development, according to the
latest survey by the National Science Foundation. No other
university came close to that figure. Globally, basic research
advances knowledge, a public good. Anyone anywhere
might then apply that knowledge in ways that add to economic growth. Locally, however, Hopkins’ high volume of
basic research has not produced a regional concentration of
high-tech spinoffs in Maryland, such as those in California’s
Silicon Valley or North Carolina’s Research Triangle.
For a university to have that level of economic impact
on its home region, a high number of its graduates must
find jobs that keep them there, according to research conducted by Jaison Abel and Richard Deitz, regional economists at the New York Fed. One way to make that happen
is to promote university research that creates spillovers
into the local economy that create more jobs requiring
high levels of human capital.
PHOTOGRAPHY: COURTESY OF JOHNS HOPKINS UNIVERSITY
In recent years, Hopkins has done more to facilitate
such spillovers, but progress has been slow, according to
Stuart Leslie, a professor in the History of Science and
Technology Department who is writing a history of the
university. “Our work to build a biotech industry and
a pharmaceutical industry using faculty research is very
difficult because it runs against a really long-standing prejudice against application,” he says. Traditionally, Hopkins
professors have created knowledge for the sake of creating
knowledge — not for commercial applications. It’s an ethos
that harkens back to Gilman and the founding faculty.
But regardless of Hopkins’ regional impact, it has made
a singular contribution to the national economy. The
university created a model of graduate study and research
that has flourished in the United States. This model has
helped develop dozens of top research universities that
have become global leaders in the advancement and dissemination of knowledge in many academic fields of study,
including disciplines that have proved invaluable to economic development both globally and regionally.
“If you are purist on causality, it’s tough to say that
universities generate innovation and economic growth,”
cautions Adam Jaffe, an economist at Brandeis University
who studies the process of technological change and innovation. “But if you are willing to say, ‘It’s only a correlation,
but the correlation is quite robust — it has developed in
a lot of places in a lot of different ways,’ then I think the
story is compelling.”
Promoting Science
Early American colleges, such as Harvard and Yale,
followed the British model of Oxford and Cambridge.
Undergraduate students mostly pursued a classical course
of study.
The University of Virginia, which started classes in
1825, experimented with a more varied and flexible curriculum that stirred debate over the value of the classical
course. At Yale, for example, a trustee resolution suggested that the study of “the dead languages” should be
eliminated. This proposal prompted what is now known as
the “Yale Reports of 1828,” which reaffirmed the conventional wisdom of sticking to the classical course. American
colleges mostly adhered to Yale’s advice for three more
generations — partly because the market demanded it —
but some universities experimented with various electives
and, in some cases, separate schools for scientific studies.
Yale formed such a school in 1847, but its resources were
severely limited.
Gilman graduated from Yale in 1852 and traveled to
Europe, where he visited Noah Porter, a Yale professor
who was studying philosophy at the University of Berlin.
This trip likely was Gilman’s first exposure to the German
model of higher education, and he was impressed by its
emphasis on research and graduate studies. He detailed
his many observations of Berlin and other universities in
letters that he sent back to the United States.
Daniel Gilman, founding president of
Johns Hopkins University
Gilman “accumulated meticulously large stores of
knowledge regarding education, the history of learning
and science, the achievements of great scholars and
scientists, the development of educational institutions
at every level,” noted Abraham Flexner, an expert on
higher education who graduated from Hopkins in 1886
and published a biography of Gilman in 1946. “His ideas
were not original; he sought them here, there, and everywhere, combining and adapting them to American needs
and conditions.”
Gilman returned to Yale to help raise money for its
scientific school. “His task was, in essence, to win adherents to the teaching of science,” Flexner wrote, but “the
classicists fought hard to maintain their monopoly.” In
1856, Gilman published a pamphlet calling for greater
opportunities for Americans to study basic and applied
science. Because these opportunities were lacking, he
lamented, the United States was “far behind European
nations in many important branches of industry.”
In 1866, Gilman became essentially the chief operating officer of the scientific school, which was making
good progress with increased funding and a new corporate structure that further separated it from Yale.
“While Yale College continued to operate on traditional
lines, the Scientific School embraced modern subjects,”
Flexner wrote. Forward-looking “Yale men,” he noted,
started to hope that Gilman might become the university’s next president, but when the job went to Porter,
Gilman went to Berkeley.
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
23
Founding Johns Hopkins
While Gilman was accumulating ideas in Europe and
experience at Yale, Johns Hopkins (the man) was amassing a fortune in Baltimore as a wholesaler of groceries, a
financier of various enterprises, and a major shareholder
in the Baltimore and Ohio Railroad. Hopkins died in 1873,
leaving $3.5 million to start a university and $3.5 million
to establish an affiliated hospital. There were few strings
attached to these gifts, but he advised the trustees to never
sell the B&O stock, a recommendation they followed and
would later regret.
After Hopkins’ death, the trustees began researching
various models of American higher education. They asked
the presidents of three universities — Charles Eliot of
Harvard, Andrew White of Cornell, and James Angell of
Michigan — to advise them. “White made encouraging
suggestions from his experience at Cornell, Angell was
skeptical, and Eliot could not imagine more than a fledgling
regional college in Baltimore,” wrote Roger Geiger, a Penn
State education professor who in 2015 published The History
of American Higher Education. “But all agreed that the best
person to lead such a venture was Gilman.”
Reverdy Johnson Jr., chairman of the trustees’ executive
committee, offered the job to Gilman in an 1874 letter.
“The Institution which I represent,” Johnson began, “is the
recipient of a fund of some three and a half millions of dollars — with no shackles of state or political influence, and
with no restriction but the wisdom and sound judgment of
the Board of Trustees.” The institution, he added, would be
“entirely plastic in the hands of those to whom its founder
has entrusted its organization and management.” In short,
Hopkins would not suffer from any of the impediments
that Gilman was struggling against at Berkeley.
“The trustees of the university believed in Gilman from
the start: He had no opposition to overcome, no vested
interest to combat, no tradition to defy,” Flexner wrote.
Gilman took the job and opened the university in 1876
with 54 graduate students, 12 matriculates, and 23 special
students.
In Hopkins’ third annual report, Gilman cannily
attributed the university’s emphasis on graduate education not to himself but to the trustees. He said the trustees
found strong demand “for opportunities to study beyond
the ordinary courses of a college or scientific school.”
The best evidence of this demand was “the increasing
attendance of American students upon the lectures of the
German Universities.”
To attract such students, Gilman sought professors
who were devoted to specific disciplines and eminent
in their fields with “power to pursue independent and
original investigation, and to inspire the young with
enthusiasm for study and research.” He hired three professors immersed in science: Henry Rowland (physics), Ira
Remsen (chemistry), and H. Newell Martin (biology). He
also hired three professors steeped in classical instruction:
J.J. Sylvester (mathematics), Basil Gildersleeve (Greek),
24
and Charles Morris (Latin and Greek).
“Sylvester and Gildersleeve were the elder statesmen.
The rest of the faculty members were about 30, which is
astonishing when assembling what was supposed to be a
world-class faculty,” Leslie says. “How do you get a great
university without a great faculty? And how do you get a
great faculty without a great university? You have to think
about it differently. You have to think about assembling a
future great faculty.”
Many young professors from the early years at Hopkins
never became outstanding scholars, but they trained hundreds of Ph.D.s who spread Hopkins’ research-centric
model of learning to many other American universities.
“The numbers are staggering by today’s standards,”
Leslie says. Rowland died at 52 after training more than
100 Ph.D.s, including more than 30 who went on to chair
departments at other universities. “Remsen was not a
great scholar at all, but as a trainer of people who would
train other graduate students at other universities, he was
unmatched,” Leslie says. “This was also true of William
Welch in medicine. He set up an environment in which
great medical researchers flourished and went on to do
tremendous things all over the world.”
In economics, Hopkins awarded its first Ph.D. in
1878 to Henry Adams, who later became a co-founder of
the American Economic Association along with Richard
Ely, who was among Hopkins’ first professors of political
economy.
Imitation and Acclaim
Gilman corresponded frequently with Eliot, the president
of Harvard. Eliot understood the importance of science.
He was familiar with the German model of graduate education, but he also was bound by the ancient traditions
of America’s oldest college. He famously stated that the
German approach would suit Harvard freshmen “about as
well as a barnyard would suit a whale.”
Harvard was not alone in its complacency. “By 1890,
the German ideal of advanced scholarship, professors as
experts, doctoral programs with graduate students, and a
hierarchy of study had few adherents in the United States
outside of Johns Hopkins,” wrote University of Kentucky
professor John Thelin in his 2004 book, A History of
American Higher Education.
Hopkins took full advantage of this head start. It graduated its first Ph.D.s in 1878, and by 1889, it had produced
a total of 151 — more than Harvard and Yale combined,
according to Geiger. “Hopkins’ Ph.D.s were soon sought
by ambitious universities throughout the country.”
The university’s first obvious imitator was Clark
University, which opened in Worcester, Mass., in 1889
with G. Stanley Hall (a former Hopkins professor of
psychology) as its founding president. Clark was the first
all-graduate studies institution in the United States. It
focused on mathematics, physics, chemistry, biology, and
most importantly, psychology.
The University of Chicago was not a Hopkins imitator, according to Cole, but its first president, William
Harper, came from the same school of thought as
Gilman. Beginning in 1890, and 14 years after the
founding of Hopkins, Harper created “a small undergraduate body and a much more elaborate and important research enterprise,” Cole says. “He recruited
extraordinarily able faculty members by raiding a lot
of the prestigious eastern universities. He essentially
killed Clark University by stealing almost all of their
very good psychologists.”
Hopkins’ emphasis on research also attracted talented
professors. “Harvard was forced to adopt the model
because Hopkins began to raid some of Harvard’s faculty
who were interested in doing research,” Cole says. “They
revered the German universities and the opportunity to
produce new knowledge rather than simply to transmit
existing knowledge.”
Harvard’s Eliot was not an early adopter of the Hopkins
model, Cole says, “but when he got into it, he got into it
in a big way. He not only adopted it, he quintessentially
adopted it.”
As Harvard surged ahead, Hopkins lost momentum.
The university struggled financially after its endowment
— almost all of it still in B&O Railroad stock — stopped
generating cash in 1887. Also, the university started to
lose some of its most distinguished faculty to retirement,
death, and academic free agency. By the time Gilman
retired from Hopkins in 1901, Eliot and Harper were
beating him at his own game. But at the university’s 25th
anniversary celebration, they gave him credit for much of
their success.
“We are celebrating the close of the first period of
University Education in these United States,” Harper said.
“During this first period, the Johns Hopkins University
has been the most conspicuous figure in the American
University world, and, to its achievements we are largely
indebted for the fact that we may now enter upon a higher
mission.”
Eliot’s tribute went even further. Eating crow from his
infamous barnyard quote, he said, “I want to testify that
the graduate school of Harvard University, started feebly
in 1870 and 1871, did not thrive, until the example of Johns
Hopkins forced our Faculty to put their strength into the
development of our instruction for graduates. And what
was true of Harvard was true of every other university in
the land which aspired to create an advanced school of
arts and sciences.”
American Dominance
In 1910, Johns Hopkins still was struggling financially, but
its reputation was intact. It appeared in Edwin Slosson’s
book of 14 Great American Universities. “Slosson chose
universities with the largest instructional budgets,” Geiger
noted. “Johns Hopkins, with a slightly smaller budget,
was grandfathered in” at the expense of MIT. In another
1910 tome, written by Flexner under the auspices of
the Carnegie Foundation, Hopkins’ medical school was
deemed the best in the nation, an excellent model for
others to emulate.
In the early 20th century, European research universities were still considered better than American research
universities on average. Nobel prizes, for example, mostly
went to professors in Germany, France, and Great Britain.
But American research universities were improving rapidly under the Hopkins model. By then, they had developed strong ties to their European colleagues — including
those in Germany.
When Hitler came to power in 1933, American universities were well-positioned to garner the lion’s share
of academic talent flowing out of Germany. The United
States welcomed these refugees “with open arms as well as
with university appointments, research fellowships, and a
level of academic freedom they quickly learned to cherish,”
Cole says. “It’s hard to say what might have happened if our
system had not been receptive to the German researchers.”
Following World War II, the United States continued to invest heavily in higher education, producing
armies of Ph.D.s. who facilitated the rapid expansion
of American higher education. Gilman’s international
vision for Berkeley came true at Johns Hopkins, at
Berkeley, and at dozens more American universities that
have become magnets for the brightest scholars from
all over the world. Hopkins currently serves more than
4,000 of the 1 million international students who are
enrolled in American universities. The United States
hosts almost twice as many international students as
any other country, according to the U.S. Department
of Commerce. America also has the largest number of
colleges and universities — even more than China, whose
population is four times larger.
Economists may debate the degree to which American
higher education has driven U.S. innovation and economic growth, but there’s no question that America’s top
research universities are the envy of the world — thanks in
part to the intellectual entrepreneurship of Daniel Gilman
at a new university in Baltimore.
EF
Readings
Cole, Jonathan R. The Great American University. New York:
PublicAffairs, 2009.
Geiger, Roger L. The History of American Higher Education.
Princeton, N.J.: Princeton University Press, 2015.
Flexner, Abraham. Daniel Coit Gilman: Creator of the American Type
of University. New York: Harcourt, Brace and Company, 1946.
Thelin, John R. A History of American Higher Education. Baltimore:
Johns Hopkins University Press, 2004
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
25
POLICYUPDATE
Tailoring Bank Regulations
B y T i m S a bl i k
O
n Oct. 2, 2018, Fed Governor and Vice Chairman
for Supervision Randal Quarles appeared before
the Senate Committee on Banking, Housing, and
Urban Affairs to discuss the Fed’s progress in implementing new reforms to bank regulation.
The reforms are a result of the Economic Growth,
Regulatory Relief, and Consumer Protection Act, or
EGRRCPA, signed into law in May 2018. In addition to
introducing protections for consumers and relaxing mortgage lending rules, EGRRCPA amended several banking
regulations put in place by the Dodd-Frank Act of 2010.
Dodd-Frank introduced new rules for financial firms,
such as stronger capital and liquidity requirements and
stress tests to assess institutions’ resiliency to future crises.
While Dodd-Frank set easier requirements for smaller
entities than for large institutions whose failure could
create spillover effects for the overall economy, leaders of
community banks argued that the law’s new requirements
were nevertheless disproportionately burdensome for their
smaller staff.
There is some evidence that the cost of regulatory compliance can be significant for smaller financial institutions.
A 2013 Minneapolis Fed study found that for a third of
community banks, hiring just two additional employees
to manage regulatory compliance would be enough to
make them unprofitable. Some have suggested this burden
may have contributed to the dearth of new community
banks formed in recent years. (See “Who Wants to Start a
Bank?” Econ Focus, First Quarter 2016.)
To address these concerns, EGRRCPA makes a number of changes to how Dodd-Frank applies to community
banks. Existing rules require all banks to maintain capital at
no less than 4 percent of total assets. Banks are also subject
to additional risk-weighted capital requirements. Under
EGRRCPA, banks with less than $10 billion in assets
have the option to instead meet a new Community Bank
Leverage Ratio, which requires them to maintain capital at
between 8 percent and 10 percent of total assets unweighted
by risk. While this new leverage ratio is higher than the
old one, banks that meet this requirement will be exempt
from any additional capital requirements. In his October
testimony, Quarles noted that the Fed and other banking
regulators plan to issue a proposal for implementing the
Community Bank Leverage Ratio “in the very near future.”
Under EGRRCPA, banks with less than $10 billion in
assets are also exempt from the so-called “Volcker Rule,”
which prohibits proprietary trading by banks. (See “Rolling
Out the Volcker Rule,” Econ Focus, First Quarter 2014.)
Finally, EGRRCPA seeks to ease some of the reporting
costs for smaller banks. Banks must file quarterly “call
26
reports” collecting a variety of information on their operations for regulators. Under the new law, banks with less
than $5 billion in assets can file more simplified reports
for the first and third quarters of the year. Additionally,
EGRRCPA allows banks with less than $3 billion in assets
to reduce the frequency of bank examinations from yearly
to once every 18 months. Under Dodd-Frank, this was only
available to banks with less than $1 billion in assets.
EGRRCPA makes a number of changes to regulations
for larger banks as well. Under Dodd-Frank, all banks with
more than $50 billion in assets were subject to additional
requirements and regulatory scrutiny. They were subject to periodic stress tests, asked to provide living wills
detailing how regulators could unwind them in the event
of failure, and required to maintain a certain threshold
of assets that could easily be liquidated in a crisis, among
other measures. EGRRCPA raised the asset threshold for
applying these requirements to $100 billion and gave the
Fed greater discretion to tailor requirements for banks
with more than $100 billion in assets. In an Aug. 17 letter,
a group of senators urged the Fed to use that discretion
to reduce regulations for large banks that do not pose a
systemic risk to the economy.
On Oct. 31, the Fed’s Board of Governors released a
draft framework for implementing these changes. The
proposal creates four categories for large banks based on
asset size and risk profile. The first category applies to globally systemically important banks and their subsidiaries.
Regulators have determined that these institutions pose
the greatest risk for the financial system, and under the new
framework they would remain subject to the most stringent
requirements introduced by Dodd-Frank and the international Basel Committee on Banking Supervision.
The second category of regulations would apply to
banks with $700 billion or more in assets as well as those
firms engaged in significant international activity. These
firms would be subject to many of the same enhanced
requirements as the firms in the first category. Firms with
$250 billion or more in assets that do not meet the criteria
for the first two categories would also face similar rules but
less demanding liquidity requirements.
The biggest changes under the proposed framework
would be for firms with between $100 billion and $250
billion in assets. Those firms would no longer be subject to
certain liquidity requirements and would need to submit
to stress tests only every other year rather than annually.
In a statement accompanying the new framework,
Quarles said that he was “hopeful that firms will see
reduced regulatory complexity and easier compliance with
no decline in the resiliency of the U.S. banking system.” EF
BOOKREVIEW
Updating the Schoolhouse
INFORMATION, INCENTIVES, AND
EDUCATION POLICY
BY DEREK A. NEAL, CAMBRIDGE,
MASS.: HARVARD UNIVERSITY PRESS,
2018, 224 PAGES
REVIEWED BY AARON STEELMAN
H
uman capital — the collection of traits that
improve people’s productive capacities, such
as health, skills, knowledge, and habits — is a
primary driver of economic growth at the aggregate level
and earnings potential at an individual level. But how do
people acquire human capital — and what can policymakers
do to facilitate that acquisition? One of the most significant
ways is formal education.
In Information, Incentives, and Education Policy, Derek
Neal, an economist at the University of Chicago, explores
how K-12 education might be provided most efficiently.
To address that issue, Neal first examines the performance of existing public schools, which he often finds
wanting due to inefficient personnel policies and waste,
and then assesses three widely proposed reforms.
He first discusses assessment-based incentive (ABI)
systems, which reward or penalize educators based on
how their students perform on standardized tests. He
notes that the “teacher effects” literature has identified
three facts about the determinants of test scores. First,
among a population of students with similar records of
past achievement who attend schools with comparable
resources, expected test scores differ systematically across
classrooms. Second, much of those differences can be
predicted by the performance of students taught previously by the teachers assigned to each classroom. Third,
researchers have not been able to identify the characteristics of new teachers that will foster higher achievement
among their students.
Neal argues that ABI systems, in principle, show promise but can fail to deliver for several reasons. There have
been numerous cases of corruption, in which students’ test
scores have been manipulated by teachers and principals.
Also, ABI systems can encourage instructors to teach to
the test rather than devote class time to more productive
uses. In addition, teachers may neglect students who
they perceive as having little chance of passing the test
in favor of other students. Neal and Gadi Barlevy of the
Chicago Fed have proposed an ABI scheme called Pay
for Percentile (PFP), which defies succinct explanation,
but which they argue would avoid those pitfalls. Even if
PFP were adopted, though, it still would not necessarily
foster other important skills and character traits, such as
acting more maturely and responsibly. Some have argued
that school choice proposals, including charter schools and
voucher programs, would allow parents to choose schools
that seem to perform well along those dimensions.
Charter schools, the second reform he considers, are
schools that the government authorizes and pays for but
does not operate. There are a large number of charter
schools across the country, which have been the subject
of much research. Neal argues that performance among
charter schools can be quite variable and what has been
described as the “No Excuses” model has often yielded
significant returns, particularly in urban schools with significant minority populations. He cites work by Roland
Fryer Jr., an economist at Harvard University, who has
written that the No Excuses approach follows five broad
tenets: (1) an increase in instructional time, including longer
school days, longer school years, and classes on Saturdays;
(2) changes in the human capital in schools, including rigorous screening of school principals, resources dedicated
to teacher training, and frequent feedback given to teachers on the quality of their instruction; (3) significant time
devoted to tutoring; (4) regular assessments of students’
progress and updated performance goals based on those
assessments; and (5) a culture of high expectations, including clearly stated objectives, with both school administrators and students’ parents signing agreements to honor the
policies designed to ensure that students succeed.
The third reform is voucher programs that provide parents with funds they could use to send their children to any
school they choose, public or private, and perhaps would
allow parents to provide additional personal funds to meet
any shortfall between the amount of the voucher and the
tuition rate. Neal has less to say about voucher programs
than he does about ABI systems or charter schools, in
part because there is less evidence to draw on. He is generally favorable toward them, though he worries about the
inequality of educational outcomes they may produce and
argues that their adoption faces large political barriers.
In the end, Neal argues that optimal education policies
might draw from all three reforms discussed, producing
a “system of regulated competition among sets of education providers that education authorities deem eligible to
receive public funds.” He sees a parallel with the Medicare
system, which allows beneficiaries to receive care at public
or private hospitals, but the providers are licensed and
treatments deemed wasteful are not covered. It may seem
like a less than satisfying proposal to some — including both
those who see such reforms as undermining a system they
believe works reasonably well now and those who would like
to see schooling separated from the state altogether — but
solutions to issues this complex often are.
EF
E CO N F O C U S | T H I R D Q U A RT E R | 2 0 1 8
27
DISTRICTDIGEST
Economic Trends Across the Region
Definitions Matter: The Rural-Urban Dichotomy
B y J o s ep h Me n g e d o t h
E
conomic disparities between urban and rural areas
have been discussed widely in recent years, with
larger metro areas seeing remarkably stronger
growth, on average, than their smaller or more rural
counterparts. The Richmond Fed’s district, the Fifth
Federal Reserve District, encompasses many points
along this spectrum, from the Appalachian region of
West Virginia to the Lowcountry of South Carolina, and
from large metro areas such as Washington, D.C., and
Charlotte to the sparsely populated Highland County in
the western mountain region of Virginia.
What is the nature of the disparities across those
regions? Do we see what is commonly called an urban-rural
divide? And how is it influenced by the definition of urban
or rural? This article will take a look at some commonly
used ways to define urban and rural areas from agencies
such as the U.S. Department of Agriculture (USDA), the
National Center for Health Statistics (NCHS), and the
Census Bureau to see what the current data tell us about
disparities across measures of demography, education,
employment, and poverty.
Defining Rural and Urban Areas
The first step is deciding how to define urban and rural. One
crude approach is to rely on the Office of Management and
Budget’s (OMB) categorization of economically integrated
Share of Population Living in Urban Areas
by Census Tract
0% - 5%
5% - 10%
10% - 20%
20% - 50%
50% - 100%
Baltimore, MD •
Washington, DC •
• Charleston, WV
Richmond, VA •
Virginia Beach, VA •
• Raleigh, NC
Charlotte, NC •
• Columbia, SC
Source: Census Bureau,
2010 Decennial Census
28
•
Charleston, SC
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
areas: Any county that is within a metropolitan or micropolitan statistical area (metropolitans have at least one
urbanized area with more than 50,000 inhabitants while
micropolitans have at least one with a population between
10,000 and 49,999) could be treated as “urban” and the rest
could be treated as “rural.” However, this method would
classify many counties as urban that, to most people, would
seem less urban than some counties classified as rural. For
example, Goochland County, Va., which is part of the
Richmond MSA, has a population density of about 77 people per square mile, which is more similar to the density of
Accomack County on Virginia’s Eastern Shore (which is
not in a metro area) than to Richmond City’s approximately
3,400 people per square mile. Thus, it may be helpful to
have ways of categorizing areas as urban or rural beyond the
basic metro and nonmetro definitions.
Census Bureau data are a commonly used alternative.
The Census Bureau designates rural and urban areas at the
Census tract or block level with each decennial Census.
A block is considered urban if the density is greater than
1,000 people per square mile or if it has a density of 500 to
1,000 people per square mile plus a mix of residential and
certain nonresidential land uses. The latter may include
such uses as parks, schools, office buildings, or retail. Rural
areas are simply those that do not meet the criteria to be
considered urban. Using this definition, 81.2 percent of the
U.S. population and 72.3 percent of the Fifth District population lived in urban areas in 2010. Of course, urban areas
are denser, accounting for only 3.1 percent of land area in
the United States in 2010 and 8.1 percent of land area in the
Fifth District. (See map.)
The USDA has another method of categorizing areas
as urban or rural. The USDA uses the OMB determinations of core-based statistical areas as a starting point to
create a county-level classification system that it calls the
Rural-Urban Continuum. This system divides counties
into nine classifications. The first three include all counties within an OMB-defined metropolitan statistical area,
which are then separated by size of the metro area population. For the remaining six classifications, the USDA
ignores the OMB’s designations of a micropolitan statistical area and instead separates counties based on the size of
the total population that live in urban areas of each county
(based on data from the decennial census) and then by
their adjacency to a metro area. The USDA system assigns
a code that ranges from one (counties in metro areas with
1 million or more in population) to nine (counties with
less than 2,500 urban population that are nonadjacent to
metro areas).
The largest number of counties in the Fifth District
fall into the first category, large metro areas, and include
those that are a part of the Washington, D.C., Virginia
Beach, Richmond, Baltimore, or Charlotte MSAs. (See
map.) These counties contained almost half of the 2016
population of the Fifth District. Nationally, this category
contained 55 percent of the total population. The second-largest number of counties fall in the sixth category
— counties not part of but adjacent to an MSA that have
between 2,500 and 19,999 people living in urban areas
of the county. This includes counties such as Logan and
Wyoming in West Virginia that are near but not part of
the Charleston and Beckley MSAs. Although a large number of counties fit in this category, they contained only
5.8 percent of the District’s population in 2016, slightly
less than the national share of 8.4 percent.
Another useful scheme for categorizing counties as
more rural or less rural is that of the NCHS, which, similar
to the USDA, uses the core-based statistical areas developed by the OMB as a basis. This system has six categories.
The first four include counties in metropolitan statistical
areas, and the last two include counties in micropolitan
statistical areas and counties that are neither in a metro or
micro area, which are labeled “noncore.” (See table.)
One of the major benefits of using the NCHS system is
that it separates the counties in large metro areas into one
central county and fringe counties. As a result, the 78 Fifth
District counties that were in the most urban category of
the USDA rural-urban continuum are separated into two
NCHS categories. Nine of these counties are in the most
urban category of the NCHS system, while 69 are in what
the NCHS deems “fringe” counties. It is easy to see why
separating metropolitan counties might be important; in
many metro areas, it is the suburbs of the cities that tend to
be in the strongest economic position.
Meanwhile, on the rural end of the spectrum, the
NCHS system groups all counties that are not part of any
MSA or micropolitan statistical area into one “rural” category. In the Fifth District, then, 112 counties are considered to be rural, or “noncore,” and there is no distinction
by proximity to metro areas. (See map on next page.) And
Fifth District Counties by Rural-Urban Continuum Area
Nonmetro: 2,500 – 19,999, Adjacent to Metro
Nonmetro: 2,500 – 19,999, Not Adjacent to Metro
Rural: Adjacent to Metro
Rural: Not Adjacent to Metro
Metro > 1 Million
Metro: 250,000 – 1 Million
Metro: < 250,000
Nonmetro: > 20,000,
Adjacent to Metro
Nonmetro: > 20,000,
Not Adjacent to
Metro
Baltimore, MD •
Washington, DC •
• Charleston, WV
Richmond, VA •
Virginia Beach, VA •
• Raleigh, NC
Charlotte, NC •
• Columbia, SC
NOTE: Population
figures in the legend refer
to the urban population.
•
Source: U.S. Department
of Agriculture
Charleston, SC
counties that are adjacent to metro areas might see different economic outcomes than those that are more distant
from an urban core.
Demographics of the Fifth District’s Rural Areas
So what do the data tell us about differences between
urban and rural areas? For one thing, no matter which
definition of urban and rural we use — Census, USDA, or
NCHS — rural populations are older. Using the Census
definition, about 17.5 percent of the 2016 U.S. population
in rural areas was over the age of 65 compared to only
about 14 percent of the urban population. Most of the
states in the Fifth District had similar shares and similar
discrepancies between urban and rural areas. Virginia had
the largest urban/rural divide with respect to the older
NCHS 2013 Rural-Urban Classifications
Code
Name
Description
Metropolitan
1
Large central metro NCHS-defined “central” counties of MSAs of 1 million or more population
2
Large fringe metro
NCHS-defined “fringe” counties of MSAs of 1 million or more population
3
Medium metro
Counties within MSAs of 250,000 – 999,999 population
4
Small metro
Counties within MSAS of 50,000 – 249,999 population
Nonmetropolitan
5
Micropolitan
Counties in micropolitan statistical areas areas (urban core population 10,000 – 49,999)
6
Noncore
Counties not within micropolitan statistical areas
Source: National Center for Health Statistics, U.S. Dept. of Health and Human Services
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
29
Fifth District Counties by NCHS Rural-Urban Classification
Large Central Metro
Large Fringe Metro
Medium Metro
Small Metro
Micropolitan
Noncore
Baltimore, MD •
Washington, DC •
• Charleston, WV
Richmond, VA •
Virginia Beach, VA •
• Raleigh, NC
Charlotte, NC •
• Columbia, SC
•
Source: National Center
for Health Statistics
Charleston, SC
population with just under 13 percent of the urban population over 65 compared to around 20 percent of the population in rural areas. Moreover, while the rural population
in West Virginia is older than its urban population, West
Virginia’s population is also just generally older. In fact,
the share of the population over 65 in urban areas of West
Virginia was higher than that in rural areas of the nation.
Applying the USDA and NCHS definitions to
county-level data corroborates this pattern in age demographics, with the lowest shares of the over-65 population occurring in the most urban categories and the
highest shares in the most rural. In particular, applying
the USDA definitions to the Fifth District and aggregating counties within each category gave a range from
about 18 percent of the population over age 65 in large
MSAs to almost 31 percent in the most rural counties.
And under the NCHS system’s separation of large MSAs
into central and fringe counties, the central counties had
an even lower share, about 16 percent.
Rural areas also tend to be less educated. According to
the Census, each state in the Fifth District had a smaller
share with a bachelor’s degree (or higher) in rural areas
than in urban areas. The District of Columbia, which is
completely urban, had the highest share of those with at
least a bachelor’s degree, almost 57 percent; on the other
hand, among the rural population of West Virginia, only
about 16 percent of the population had a college degree.
(West Virginia’s urban population also had a relatively
low share at about 26 percent.) Of course, this is only
one measure of human capital accumulation — if one is
interested in economic divides between rural and urban
areas, it is important to also understand high school
graduation rates and access to vocational or technical
30
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
schools, among other human capital measures.
As with the Census definitions, statistics based on the
USDA definitions show that college attainment was highest in the most urban areas and lowest in the most rural.
But the data also show, perhaps unsurprisingly, that the
counties that were adjacent to metro areas generally had
higher shares of college graduates than those that were
nonadjacent. In the Fifth District, this was particularly the
case for Virginia. Further, the NCHS categories showed
a substantial difference in college attainment between
central and fringe counties of large urban areas, with about
45 percent of the population in central counties having a
bachelor’s degree or higher compared to 39 percent in the
fringe counties. But as always, or almost always, there are
exceptions; for example, fringe counties of Baltimore City
had a higher share of the college educated than the central
county of Baltimore City.
Labor Market Outcomes and Poverty
in the Rural Fifth District
There are many measures of labor market outcomes; here,
we explore labor force participation, unemployment, and
wages to get some sense of labor market activity across
areas. In general, labor force participation rates were
higher in urban areas than rural areas. This is true at the
state level according to the Census data, but the county-level
systems offer further insight. Although generally the highest participation rates were in central counties of large
urban areas (using the NCHS system), this was not always
the case. For example, in Maryland, fringe counties had
the highest participation rates.
Again, the USDA system offers insight for rural counties
by considering adjacency to metro areas. For example, some
of the lowest participation rates in the Fifth District of
35.1 percent and 36.5 percent occurred in Virginia’s neighboring counties of Buchanan and Dickenson, respectively.
Both counties are in the southwest region of the commonwealth and are designated as rural, not adjacent to
a metro area by the USDA. Meanwhile, in the nearby
county of Wythe, Va., which is also rural but adjacent to
the Blacksburg MSA, the participation rate was more than
20 percentage points higher at 57.1 percent.
Such differences in the data may lead a policymaker to
ask: What drives the lower participation rates in more rural
counties? Is it the fact that those areas tend to be older, or
are there potential workers on the sidelines in rural areas
who can be brought into the fold, especially in this period
of very low unemployment?
Using data from 2016, one can aggregate the number of
unemployed and the labor force for all the counties in each
rural-urban continuum group and calculate an unemployment rate. Doing so reveals that unemployment was lowest
in large metro areas and generally increased with rurality,
with a notable exception of counties considered rural but
adjacent to a metro area. Fifth District rural, metro-adjacent counties had a combined unemployment rate of 5.2
percent. That was lower than nonadjacent rural areas, which
had a rate of 6.7 percent.
Meanwhile, under the NCHS scheme, unemployment
was lowest in fringe counties of large MSAs in the Fifth
District, which had an aggregate rate of 4.1 percent compared to central counties and medium-sized cities with
4.8 percent each. And the highest aggregate rate of 6.1 percent occurred in the category for micropolitan areas. Some
of the highest county-level unemployment rates were in
rural counties, but this category is large and included many
counties with very low unemployment rates.
Wage data aggregated across the USDA’s urban-rural
continuum likewise show that wages were higher the
more urban the area. In fact, in 2016, wages were almost
40 percent higher in large metro areas than in mid-sized
and smaller MSAs. But unlike with educational attainment and labor force participation, there was very little
difference between adjacent and nonadjacent rural areas.
Using the NCHS classification further showed the wage
premium earned in large cities. The average employee
working in a central county of one of the Fifth District’s
large MSAs earned a wage about 20 percent higher than
the average employee working in a fringe county. Wage
data are reported by place of work, so a person earning
those higher wages might live in the fringe county and
commute to the central county.
Although the data above show generally lower unemployment, higher wages, and higher labor force participation in urban areas (with some key exceptions), those areas
also tend to have higher levels of poverty. Using Census
definitions, a higher share of the urban population lived
below the poverty line than the rural population in the
United States and in most Fifth District jurisdictions (the
exceptions being Virginia and South Carolina). This, however, is where the classification system matters. Using the
USDA classification and creating an aggregate measure of
the share of the population living below the poverty line
for each urban-rural category, we observe that counties in
large metro areas actually tended to have the lowest poverty
rate at around 11 percent, followed by mid-sized and small
MSAs. Among the nonmetro categories, poverty rates were
between 20 percent and 22 percent, with the exception
of rural areas that are adjacent to a metro area, where the
aggregate measure was about 18 percent.
There is also a considerable amount of variation within
categories. For example, poverty rates among large metro
areas ranged from 2.7 percent in Falls Church City, Va.
(part of the Washington, D.C., MSA) to 29.4 percent in
Petersburg City, Va. (part of the Richmond MSA). In fact,
the poverty rate in Petersburg is higher than that of any
completely rural county, adjacent to a metro area or not.
Similarly, there was a large variation in poverty rates in
nonmetro counties with urban populations between 2,500
and 19,999, which ranged from 5.6 percent to 33.0 percent.
Meanwhile, using the NCHS system showed that fringe
counties of large MSAs had the lowest poverty rate
(9 percent), while central counties of large MSAs actually had similar poverty rates to medium and small sized
MSAs. The category with the highest poverty rate was
micropolitan areas, at just over 20 percent. Although this
classification system showed that central counties had
higher poverty rates than fringe counties, it is important to
remember that aggregating to the county level might mask
large differences within counties, such as the difference
between poverty inside the city and outside of the city.
For example, the central county of the Richmond, Va.,
metro area is the independent city of Richmond, where the
poverty rate was 25.4 percent in 2016. Meanwhile, the more
suburban neighboring counties of Henrico and Chesterfield
had poverty rates of just 10.6 percent and 7.4 percent,
respectively. Comparatively, in Mecklenburg County, N.C.
(the central county of the Charlotte MSA), the poverty rate
was 14.2 percent. But although Mecklenburg’s 524 square
miles contain the center city of Charlotte, the county also
contains a large portion of the city’s suburban areas. This
could mask differences in economic outcomes between the
city of Charlotte and its suburbs.
Definitions Matter, but so Does Geography
No matter the classification system, the data indicate that
more rural areas tend to be both older and less educated. It
is no coincidence, then, that measures such as labor force
participation and wages tend to be lower. Nonetheless,
analyzing the classification systems for urban and rural
areas shows the importance of close attention to how data
are aggregated.
The USDA and NCHS county-based systems enable
distinctions among urban areas by population size and
between central and fringe counties of large MSAs, which
can highlight important differences. Likewise, separating
rural areas by adjacency to metro areas shows the potential
importance of proximity to a city.
With each method comes costs and benefits. The Census’
block-level definition, while the most comprehensive, is
also the most difficult to use with other data, since data at
the block or tract level are usually collected at best every
few years and at worst every decade. What is more, since
counties are a common geographic category, classification
schemes like those of the USDA or the NCHS enable a
researcher to use much more data to characterize an area
that most people can relate to. On the other hand, the
USDA classification system, with its nine categories, allows
for a richer comparison across the rural and urban continuum but obfuscates some differences between central and
fringe counties of large MSAs. The NCHS system’s fewer
categories do distinguish between central and fringe counties, but it groups a larger set of rural areas together, making
it difficult to understand how adjacency to metro areas
might influence outcomes. Recognizing and understanding
these limitations is important for researchers and policymakers when trying to understand geographic disparities
broadly and the urban-rural divide more specifically. EF
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
31
State Data, Q1:18
DC MD NC
SC VA
WV
Nonfarm Employment (000s)
794.1
2,731.5
4,466.5
2,117.4
3,979.0
751.0
Q/Q Percent Change
0.0
0.3
0.6
0.6
0.6
0.5
Y/Y Percent Change
0.7
0.3
1.6
1.5
0.8
0.7
Manufacturing Employment (000s)
1.3
108.2
469.8
243.5
238.3
47.3
Q/Q Percent Change
0.0
0.9
0.5
0.7
1.5
0.5
Y/Y Percent Change
8.3
1.6
0.8
1.9
2.0
1.1
Professional/Business Services Employment (000s) 166.9
Q/Q Percent Change
0.0
Y/Y Percent Change
0.5
447.1
0.8
0.6
632.5
0.7
2.9
274.5
-2.3
0.7
736.6
0.6
1.8
66.1
0.1
-0.3
Government Employment (000s)
238.4
503.5
735.8
367.5
715.9
154.1
Q/Q Percent Change
-0.2
0.0
0.2
0.2
0.0
0.3
Y/Y Percent Change
-1.4
-0.3
0.9
0.8
-0.1
-1.0
Civilian Labor Force (000s)
402.7
3,224.7
4,974.0
2,324.5
4,321.1
784.4
Q/Q Percent Change
0.3
0.1
0.1
0.2
0.0
0.3
Y/Y Percent Change
0.8
0.4
1.3
0.8
0.7
0.9
Unemployment Rate (%)
5.7
4.2
4.5
4.4
3.5
5.4
Q4:17
5.9
4.1
4.5
4.2
3.6
5.4
Q1:17
6.1
4.3
4.8
4.4
4.0
5.1
Real Personal Income ($Bil)
52.8
350.1
434.9
199.5
445.4
66.4
Q/Q Percent Change
0.5
0.3
0.7
0.8
0.6
0.5
Y/Y Percent Change
1.4
1.6
2.4
1.7
2.0
1.1
New Housing Units
721
4,405
18,075
8,756
8,399
619
Q/Q Percent Change
-69.3
56.4
10.4
11.6
7.8
2.7
Y/Y Percent Change
6.5
16.3
14.1
5.6
15.9
-8.3
House Price Index (1980=100)
878.2
473.0
371.3
381.6
458.0
235.0
Q/Q Percent Change
1.8
0.9
1.6
2.0
1.0
-0.3
Y/Y Percent Change
7.4
5.0
7.2
7.6
4.9
3.3
Notes:
Sources:
1) FRB-Richmond survey indexes are diffusion indexes representing the percentage of responding firms
reporting increase minus the percentage reporting decrease.
The manufacturing composite index is a weighted average of the shipments, new orders, and
employment indexes.
2) Building permits and house prices are not seasonally adjusted; all other series are seasonally adjusted.
3) Manufacturing employment for DC is not seasonally adjusted
Real Personal Income: Bureau of Economic Analysis/Haver Analytics.
Unemployment Rate: LAUS Program, Bureau of Labor Statistics, U.S. Department of Labor/Haver
Analytics
Employment: CES Survey, Bureau of Labor Statistics, U.S. Department of Labor/Haver Analytics
Building Permits: U.S. Census Bureau/Haver Analytics
House Prices: Federal Housing Finance Agency/Haver Analytics
For more information, contact Akbar Naqvi at (804) 697-8437 or e-mail Akbar.Naqvi@rich.frb.org
32
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
Nonfarm Employment
Unemployment Rate
Real Personal Income
Change From Prior Year
First Quarter 2007 – First Quarter 2018
Change From Prior Year
First Quarter 2007 – First Quarter 2018
4%
3%
2%
1%
0%
-1%
-2%
-3%
-4%
-5%
-6%
07 08 09 10
First Quarter 2007 – First Quarter 2018
10%
9%
8%
7%
6%
5%
4%
11
12
13
14
15
16
17
18
3%
07 08 09 10
11
12
13
14
15
16
Fifth District
17
18
8%
7%
6%
5%
4%
3%
2%
1%
0%
-1%
-2%
-3%
-4%
-5%
07 08 09 10
11
12
13
14
15
16
17
18
14
15
16
17
18
17
18
United States
Nonfarm Employment
Major Metro Areas
Unemployment Rate
Major Metro Areas
New Housing Units
Change From Prior Year
First Quarter 2007 – First Quarter 2018
First Quarter 2007 – First Quarter 2018
Change From Prior Year
First Quarter 2007 – First Quarter 2018
7%
6%
5%
4%
3%
2%
1%
0%
-1%
-2%
-3%
-4%
-5%
-6%
-7%
-8%
07 08 09 10
11
Charlotte
12
13
14
Baltimore
15
16
17
18
13%
12%
11%
10%
9%
8%
7%
6%
5%
4%
3%
2%
1%
07 08 09 10
Washington
40%
30%
20%
10%
0%
-10%
-20%
-30%
-40%
11
Charlotte
12
13
14
Baltimore
15
16
FRB—Richmond
Manufacturing Composite Index
First Quarter 2007 – First Quarter 2018
First Quarter 2007 – First Quarter 2018
30
30
20
20
10
-30
-20
-40
11
12
13
14
15
16
17
18
-50
07 08 09 10
11
12
13
14
15
16
12
13
United States
Change From Prior Year
First Quarter 2007 – First Quarter 2018
-20
-10
11
House Prices
-10
0
-50%
07 08 09 10
Fifth District
0
10
-30
07 08 09 10
18
Washington
FRB—Richmond
Services Revenues Index
40
17
17
18
10%
8%
6%
4%
2%
0%
-2%
-4%
-6%
-8%
-10%
-12%
-14%
07 08 09 10
11
Fifth District
12
13
14
15
16
United States
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
33
Metropolitan Area Data, Q1:18
Washington, DC
Baltimore, MD Hagerstown-Martinsburg, MD-WV
Nonfarm Employment (000s)
2,683.8
1,392.0
103.9
Q/Q Percent Change
-0.9
-1.4
-3.4
Y/Y Percent Change
1.3
1.4
0.2
Unemployment Rate (%)
3.6
4.4
4.4
Q4:17
3.6
4.2
4.3
Q1:17
3.8
4.4
4.2
New Housing Units
6,443
2,144
257
Q/Q Percent Change
-5.2
86.6
-13.8
Y/Y Percent Change
27.1
48.2
-1.5
Asheville, NC Charlotte, NC
Durham, NC
Nonfarm Employment (000s)
191.0
1,192.4
310.5
Q/Q Percent Change
-1.7
-1.2
-1.1
Y/Y Percent Change
1.7
2.8
1.2
Unemployment Rate (%)
3.4
4.2
3.9
Q4:17
3.7
4.2
4.0
Q1:17
3.8
4.5
4.3
New Housing Units
796
7,107
1,133
Q/Q Percent Change
15.5
25.6
0.8
Y/Y Percent Change
68.3
42.8
3.1
Greensboro-High Point, NC
Raleigh, NC
Wilmington, NC
Nonfarm Employment (000s)
358.6
620.7
124.2
Q/Q Percent Change
-1.2
-0.9
-1.9
Y/Y Percent Change
0.6
2.7
1.0
Unemployment Rate (%)
4.6
3.9
4.1
Q4:17
4.8
4.0
4.3
Q1:17
5.0
4.2
4.4
New Housing Units
569
4,363
526
Q/Q Percent Change
-3.7
26.2
-23.9
Y/Y Percent Change
-34.6
12.7
21.8
Note: Nonfarm employment and new housing units are not seasonally adjusted. Unemployment rates are seasonally adjusted.
34
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
Winston-Salem, NC Charleston, SC Columbia, SC
Nonfarm Employment (000s)
264.0
352.7
394.4
Q/Q Percent Change
-0.9
-1.0
-0.8
Y/Y Percent Change
1.0
1.4
-0.3
Unemployment Rate (%)
4.2
3.8
4.4
Q4:17
4.3
3.6
4.3
Q1:17
4.6
3.9
4.2
New Housing Units
593
1,501
1,218
Q/Q Percent Change
-6.6
-6.8
18.7
Y/Y Percent Change
30.0
-13.0
2.4
Greenville, SC
Richmond, VA
Roanoke, VA
Nonfarm Employment (000s)
418.6
666.9
158.6
Q/Q Percent Change
-0.7
-1.3
-1.4
Y/Y Percent Change
1.9
0.7
-0.3
Unemployment Rate (%)
4.0
3.4
3.3
Q4:17
3.8
3.8
3.7
Q1:17
3.9
4.1
3.9
New Housing Units
1,434
1,667 N/A
Q/Q Percent Change
20.3
17.7 N/A
Y/Y Percent Change
20.5
-0.2 N/A
Virginia Beach-Norfolk, VA Charleston, WV Huntington, WV
Nonfarm Employment (000s)
768.1
115.5
135.4
Q/Q Percent Change
-1.7
-1.7
-3.5
Y/Y Percent Change
0.1
-0.5
-1.2
Unemployment Rate (%)
3.5
5.4
5.4
Q4:17
4.0
5.5
5.6
Q1:17
4.4
5.1
5.8
New Housing Units
1,441
54
38
Q/Q Percent Change
5.9
0.0
0.0
Y/Y Percent Change
-11.5
0.0
0.0
For
more information, contact Akbar Naqvi at (804) 697-8437 or e-mail Akbar.Naqvi@rich.frb.org
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
35
Opinion
What Have We Learned since the Financial Crisis?
By K a rt i k At h r e ya
T
he financial crisis of 2007-2008 and the ensuing
recession raised important questions for policymakers and researchers alike. In the 10 years since,
research by economists has helped improve our understanding of financial markets, labor markets, economic
shocks, and policy responses and the interactions among
them. (Many of these developments are nicely summarized in the Summer 2018 issue of the Journal of Economic
Perspectives.) I’d like to share with you some of the ways
that researchers at the Richmond Fed and elsewhere in
the Federal Reserve System are contributing to this work.
One feature of the financial crisis was the sudden largescale withdrawal of funds from large financial institutions,
most famously Lehman Brothers. Some observers have
likened these withdrawals to the classic bank runs of the
19th and early 20th century. Research on how to prevent
such rapid liquidations can help make both the financial
system and the real economy more stable.
To that end, Bruno Sultanum of the Richmond Fed,
David Andolfatto of the St. Louis Fed, and Ed Nosal of the
Atlanta Fed have proposed creating a new type of financial
instrument to help detect runs. This instrument creates
an incentive for investors to signal to financial institutions
when they believe a run is imminent. In theory, this new
instrument would give institutions the necessary information to take actions such as temporarily suspending payments and, ideally, prevent runs from happening.
Understanding the labor market — the market most
important to most of us — was also a focus for policymakers
and economists in the immediate aftermath of the crisis.
Millions of workers who lost their jobs during the recession sought new positions, but the labor market recovery
took a long time compared to past downturns. Research
by Andreas Hornstein of the Richmond Fed, Marianna
Kudlyak of the San Francisco Fed, and Fabian Lange of
McGill University provides an account of this fact: Not all
job seekers transition from unemployment to employment
at the same rate. Depending on their circumstances, job
seekers face different probabilities of returning to work.
Using this information, Hornstein, Kudlyak, and Lange
constructed an alternative measure of unemployment called
the Non-Employment Index. Tools like this help policymakers better understand phenomena observed during the
Great Recession, such as the elevated long-term unemployment rates and slow labor market recovery.
Another key aspect of the Great Recession was the
apparent transmission of shocks in some sectors of the
economy (mortgage finance and housing) to the economy as a whole. To understand how such transmission
occurs, Pierre-Daniel Sarte of the Richmond Fed along
36
E co n F o c u s | T h i r d Q u a rt e r | 2 0 1 8
with Lorenzo Caliendo of Yale University, Esteban RossiHansberg of Princeton University, and Fernando Parro of
Johns Hopkins University modeled these types of linkages
across the United States. Sectors tend to be located across
different regions; Sarte, Caliendo, Rossi-Hansberg, and
Parro showed that understanding the regional and sectoral
characteristics of the economy is essential to understanding
how shocks affect the overall economy.
Fed researchers have also improved the tools available
to policymakers and have advanced our understanding of
the effects of policy. This work is crucial because the fiscal
and monetary policy responses to the financial crisis and
subsequent recession were large but neither their costs
nor their benefits are fully understood. Richmond Fed
economist Christian Matthes and Fabio Canova of the BI
Norwegian Business School have presented new solutions
to problems arising in macroeconomic models used to
formulate advice for policymakers. This helps to ensure
that researchers and policymakers alike are working with
the best available model of the economy.
As for understanding the effect of policy, in work with
Regis Barnichon of the San Francisco Fed, Matthes found
that monetary policy has a larger effect on unemployment
when it is contractionary than when it is expansionary,
and the same is true for fiscal policy. Policymakers aware
of these potential asymmetries can make more informed
decisions about how best to respond to downturns such as
the Great Recession.
A final strand of work I want to mention is also aimed at
improving policy evaluation, this time by allowing for much
greater differences across households in economic models,
especially in their income and wealth. These new models
are known as Heterogeneous Agent New Keynesian, or
“HANK.” Along with other leading macroeconomists,
Richmond Fed economists Felipe Schwartzman and Marios
Karabarbounis and their co-authors have analyzed and used
HANK models to help us better understand how both fiscal and monetary policy work by allowing us to selectively
incorporate real-world features such as credit constraints,
illiquid wealth, and uninsurable risks.
These are just a few examples of how work by Richmond
Fed economists has helped improve our understanding of
the economy over the last decade. As we get further away
from the financial crisis of 2007-2008, it is important to
continue to push the research frontier using existing tools
and also improve our toolkit if we want to be prepared to
face the next crisis — or, ideally, avoid it altogether. EF
Kartik Athreya is executive vice president and director
of research at the Federal Reserve Bank of Richmond.
NextIsSue
Energy Booms and Busts
Fracking has led to a boom in natural gas exporting in West Virginia
and other states, but the energy sector is also prone to busts due
to the variability of energy prices. How do these booms and busts
affect how local workers manage their labor market risks?
Executive Pay
How do you compensate CEOs so they have the right incentives
to maximize value to shareholders? It’s a question that economists
(and boards of directors) have long tried to answer. Some research
suggests the average CEO is overpaid, while other research argues
that the value of high-performing CEOs has increased — justifying
bigger pay plans.
Federal Reserve
The Fed’s long-term policy goal is to steer interest rates toward
their “neutral” level, where the economy is stable and inflation
is neither rising nor falling. The trouble is there is no way to
directly observe this variable, referred to as R-star by economists
and policymakers. Is R-star a useful tool for monetary policy,
despite its elusive nature?
Interview
Preston McAfee, former chief economist
of Microsoft, director of Google strategic
technologies, and professor at the California
Institute of Technology, on designing
markets, entrepreneurship within large
tech companies, and what machine learning
means for competition and antitrust.
Economic History
Motor City … Maryland? In the early 1900s,
there were around 200 auto manufacturers
in the United States, including in Maryland,
Virginia, and North Carolina. By the 1940s,
there were only eight — and nearly all of
them were located in Detroit. Why didn’t the
auto industry take hold in the Fifth District?
At the Richmond Fed
For Richmond Fed economist Toan Phan,
seeing the effects of Japan’s housing bubble
in the 1990s sparked an interest in why asset
bubbles form, what the economic tradeoffs of a bubble are, and how the bursting
of a bubble affects the economy.
Visit us online:
www.richmondfed.org
• To view each issue’s articles
and Web-exclusive content
• To view related Web links of
additional readings and
references
• To subscribe to our magazine
• To request an email alert of
our online issue postings
Federal Reserve Bank
of Richmond
P.O. Box 27622
Richmond, VA 23261
Change Service Requested
To subscribe or make subscription changes, please email us at research.publications@rich.frb.org or call 800-322-0565.
Economic Quarterly is published by the Research Department
of the Federal Reserve Bank of Richmond. The journal contains articles written
by our staff economists and visiting scholars and includes economic analysis
pertinent to Federal Reserve monetary and banking policy.
Third Quarter 2018
First Quarter 2018
The Lifetime Medical Spending of Retirees
Cyclical Properties of Bank Margins:
Small versus Large Banks
John Bailey Jones, Mariacristina De Nardi, Eric French,
Rory McGee, and Justin Kirschner
Inefficiency in a Simple Model of Production
and Bilateral Trade
Zachary Bethune, Bruno Sultanum, and Nicholas Trachter
Second Quarter 2018
The Decline in Currency Use at a National
Retail Chain
Zhu Wang and Alexander L. Wolman
Idiosyncratic Sectoral Growth, Balanced Growth,
and Sectoral Linkages
Andrew Foerster, Eric LaRose, and Pierre-Daniel G. Sarte
Borys Grochulski, Daniel Schwam, and Yuzhe Zhang
Self-Insurance and the Risk-Sharing Role of Money
Russell Wong
First-Fourth Quarter 2017
Inequality Across and Within US Cities around
the Turn of the Twenty-First Century
Felipe F. Schwartzman
The Fed’s Discount Window: An Overview
of Recent Data
Felix P. Ackon and Huberto M. Ennis
Using the Richmond Fed Manufacturing Survey to
Gauge National and Regional Economic Conditions
Nika Lazaryan and Santiago Pinto
To access Economic Quarterly visit: https://www.richmondfed.org/publications/research/economic_quarterly
@FedFRASER