View original document

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




Second Quarter 2019
Volume 4, Issue 2

Exploring the Economic Effects
of the Opioid Epidemic
Regional Spotlight
Implementing Monetary Policy in
a Changing Federal Funds Market

Second Quarter 2019


Volume 4, Issue 2

Exploring the Economic
Effects of the Opioid


Adam Scavette examines how
the opioid crisis affects the labor
market, and why it's been particularly hard on the Third District.


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

Patrick T. Harker
President and
Chief Executive Officer
Michael Dotsey
Executive Vice President and
Director of Research
Adam Steinberg
Managing Editor, Research Publications
Brendan Barry
Data Visualization Manager

ISSN 0007–7011

Implementing Monetary
Policy in a Changing
Federal Funds Market
Benjamin Lester explores how
the Fed set interest rates before,
during, and after the Great
Recession, and how the distribution of reserves might help it
understand when those reserves
are no longer ‘ample.’

Regional Spotlight:
Smart Growth for
Regions of All Sizes
To build a healthy economy, a
region must grow—or so says the
conventional wisdom. Paul R.
Flora examines the data and finds
that growth isn't always the best
way forward.


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

About the Cover
Philadelphia's most famous citizen, Benjamin Franklin, has graced the $100 bill
since the newly created Federal Reserve began issuing “Federal Reserve Notes” in
1914. This particular image is taken from H.B. Hall's engraving of Joseph-Siffred
Duplessis's 1785 portrait of Franklin, which is currently on view at the National
Portrait Gallery in Washington, D.C. In the background are details from the 2009
redesign of the $100 bill, including a reproduction of the Declaration of Independence. Franklin served on the “Committee of Five” that drafted the Declaration
and presented it to the Second Continental Congress, then meeting at the Pennsylvania State House, on July 4, 1776. The State House still stands today, just
two blocks from the Federal Reserve Bank of Philadelphia, and is now known as
Independence Hall.
Photo by Rich Wood.

Connect with Us
We welcome your comments at:


E-mail notifications:


Previous articles:


Exploring the Economic Effects
of the Opioid Epidemic
Hundreds of thousands of Americans have died from opioid
overdoses in recent years. What has this epidemic done to
the economy? And why is the crisis so much worse right here
in the Third District?
Adam Scavette is a senior economic analyst
at the Federal Reserve Bank of Philadelphia.
The views expressed in this article are not
necessarily those of the Federal Reserve.



n a single weekend in July 2018, more than 170
people in Philadelphia overdosed from what investigators said was a single “bad batch” of heroin.
Ten died. Bags containing this particularly harmful
compound were ominously stamped Santa Muerte
(Spanish for “Holy Death”). In the preceding year,
there were more than 70,200 drug overdose deaths
in the United States, over four times the number
of homicide deaths.1 The age-adjusted rate of drug
overdose deaths tripled between 1999 and 2016 and
jumped an additional 10 percent in 2017.2 Nearly 70
percent of those 2017 drug overdose deaths can be
attributed to opioids.3
All three states in the Federal Reserve’s Third
District have been struck particularly hard by this
surge in drug overdose deaths. The age-adjusted drug
overdose rate for the nation was 21.7 deaths per
100,000 people in 2017, while it was 30.0 for New Jersey,
37.0 for Delaware, and 44.3 for Pennsylvania. Only
West Virginia and Ohio had drug overdose death rates
higher than Pennsylvania’s in 2017, at 57.8 and 46.3,
respectively. However, in terms of the absolute number
of drug overdose deaths, Pennsylvania was first in the
country at 5,388. That was 8 percent of the nation’s
drug overdose deaths, even though Pennsylvania had
less than 4 percent of its population. And as is true
in the rest of the United States, the majority of these
Pennsylvanian deaths were opioid related.
This article examines the origins of the crisis, the
nature of the crisis in the Third District, the relationship between the crisis and the labor market, the
costs of the epidemic, and some policy countermeasures designed to alleviate the crisis.

What Is an Opioid?

An opioid is a substance that acts on the opioid
receptors in the nervous system. Among other effects,
opioids relieve pain and, when abused, produce
euphoria. The Centers for Disease Control and
Prevention (CDC) groups opioids into three primary
categories: natural and semisynthetic opioid analgesics that are typically available by prescription (such
as morphine, codeine, oxycodone, and hydrocodone);
synthetic opioid analgesics (such as tramadol and
fentanyl); and heroin.

The Origins of the Opioid Crisis in the
United States

Studies suggest that a large percentage of abusers
began their journey with prescription opioids. In one
study, 80 percent of heroin users admitted to misusing
prescription opioids before turning to heroin. Because
users must obtain these prescription drugs either illicitly through diversion (the illegal transfer of opioids
from the prescribed individual to others) or legally
from a legitimate prescription, it is helpful to examine
the rise in legitimate prescriptions for opioids.
Prior to the 1990s, opioids were prescribed mainly
for cancer patients (or to treat chronic malignant
pain). Beginning in the early 1990s, pharmaceutical
companies encouraged physicians to prescribe
opioids to treat noncancer pain.4 Noncancer patients
tend to require longer-term administration of the
drug than do cancer patients. As a result, practitioners
of pain medicine as well as other medical specialties
were taught to rely more on opioids for general pain
treatment.5 At the time, the public didn’t realize how
risky this would be. Later studies would show that
noncancer pain patients are more likely than cancer

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2


Opioids Hit Third
District Hard

Opioid overdose death
rates per 100,000
population (age-adjusted)
in 2017

U.S. average



Sources: Henry J. Kaiser
Family Foundation and
Centers for Disease Control and Prevention.

Federal Reserve Bank of Philadelphia
Research Department


patients to become dependent on and
eventually abuse opioids.
By decade’s end, 86 percent of patients
using opioids were using them for noncancer pain.6 The national prescribing
rate of opioid medications climbed
throughout the 2000s, peaking in 2012 (see
Figure 2). While the prescribing rate in
Pennsylvania roughly mirrored the nation’s,
Delaware’s rate was typically 20 percent
higher than the nation’s, and New Jersey’s
prescribing rate was 25 percent lower.7
Thanks to increasing awareness and regulatory countermeasures to be discussed
later in this article, by 2017 the rate of
opioid prescriptions had fallen more than
25 percent from its peak in 2012.
As they become more dependent,
abusers of prescription opioids often turn
to heroin, which has experienced a large
supply increase and thus a price decrease
over the past 20 years due to major changes in its production and supply chain.8
According to recent Drug Enforcement
Agency (DEA) estimates, from 1980 to 2016
the average retail price per gram of pure
heroin decreased by more than 70 percent,

from $3,260 (in 2012 dollars) to between
$465 and $1,020.9
An even more recent trend has been
a spike in the availability of fentanyl in the
United States. Fentanyl is a synthetic
painkiller about 50-100 times more potent
than morphine. The drug is commonly
prescribed in the form of transdermal
patches or lozenges. In the United States it
is available as a Schedule II drug, meaning
it is legally available only through a nonrefillable prescription.10 However, drug
cartels began purchasing cheaply produced
fentanyl from Chinese pharmaceutical
labs and shipping it to Mexico to mix with
the heroin supply before it enters the
United States for sale.11 Because small doses
of fentanyl are more likely to be fatal due
to its potency, users are at a much higher
risk of overdosing when abusing fentanyl
or fentanyl-laced heroin. This has escalated
the number of overdose deaths in the
United States for which fentanyl is responsible (see Figure 4).
Natural-opioid overdose deaths, which
can be attributed primarily to prescription
pills, have increased at a steady rate since

1999. But starting in 2010, heroin and
synthetic opioid overdose deaths have
increased much more rapidly. As seen
in the synthetic opioid series in Figure 4,
fentanyl has become responsible for
almost as many overdose deaths as natural
opioids and heroin combined and has
increased nearly tenfold since 2013.

The ‘Synthetic Problem’ in the
Third District

An October 2018 New York Times feature
article about the local and social effects of
the crisis focused on the Philadelphia
neighborhood of Kensington, which has
long been recognized as the highly visible
regional epicenter of the opioid epidemic
due to its open-air drug markets, encampments of drug-addicted homeless users,
and hyper-localized poverty. To address
the crisis, Philadelphia Mayor Jim Kenney
established an opioid task force upon
his inauguration in 2016, and Governor
Tom Wolf signed a statewide disaster
declaration in 2018, an unprecedented
public-health emergency measure in



Opioid Prescriptions Peak in Early 2010s…

…but Vary by County at Their Peak

Prescriptions per 100 people, by state

Third District county-level behavior compared with the nation over peak years, averaged over
2010 through 2012





United States

New Jersey



Prescription Rates
Compared with U.S. rate
Less than 30%
10 to 30% lower
10% lower to 10% higher
10 to 30% higher
More than 30%
Outside the Third District
Source: Henry J. Kaiser Family Foundation.



Source: Centers for Disease Control and Prevention.


Federal Reserve Bank of Philadelphia
Research Department

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

Pennsylvania. We will take a closer look at public policy responses
in a later section, but it is helpful to examine the data on our
District to see how the local crisis became so deadly in the past
few years.
Looking at opioid overdose death rates specifically, we see how
quickly the opioid crisis worsened in our region. Figure 5 depicts
opioid overdose death rates in the United States and the Third
District states. Although Pennsylvania’s and New Jersey’s opioid
overdose death rates were similar to the nation’s in 2015, by 2017
their rates had more than doubled, rising twice as fast as the national rate in two years. Meanwhile, in Delaware the opioid overdose death rate stood at nearly double the national rate in 2017.
Figure 6 breaks down these 2017 overdose death rates by type
of opioid. Nationally, synthetic opioids (e.g., fentanyl) caused
overdose deaths at a higher rate than other types of opioids. In
the Third District, the overdose death rate from synthetic opioids
was even higher.
Although the overdose death rates from natural opioids (e.g.,
prescription pills) in the Third District states are roughly in line
with the national rate, the rate of overdose deaths for synthetic
opioids is roughly double or more in each of our three states.
Fentanyl has been a recent problem in the heroin supply of East
Coast cities,12 which could contribute to the higher overdose
death rates in the Third District—perhaps because East Coast
heroin is sold as a powder and is thus easier to cut with an adulterant like fentanyl,13 whereas West Coast heroin is often sold as
a dark brown paste.14


Third District Overdose Rate Surges Past the Nation’s
Opioid overdose death rates per 100,000 population (age-adjusted), by state


New Jersey


United States




Source: Centers for Disease Control and Prevention.



Fentanyl-Related Deaths Spike After 2013

Death Rate from Synthetics Higher in Third District

Opioid overdose deaths by type for U.S.

Overdose deaths by type and by state (2017)


Natural and Semisynthetic


l an ath
ura c de
nat eti
re nth
Mo misy

sem Mo
isy re n
nth atu
eti ral
c d an
eat d





in d






← More synthetic deaths

More heroin deaths →









Natural and


Sources: Henry J. Kaiser Family Foundation and Centers for Disease Control and







Sources: Henry J. Kaiser Family Foundation and Centers for Disease Control and Prevention.

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department



Non-Hispanic White Men of Working Age More Likely to Die from Drugs
Drug poisoning mortality in the U.S., 1999–2016, by sex, age, and race/ethnicity
Total Deaths, 2017

Age-Adjusted Death Rate
By race.


Non-Hispanic white

By sex.






80% 100%

All races-all origins
Non-Hispanic black


By age.








20% 25%




2015 2017

Source: Centers for Disease Control and Prevention (

The Opioid Epidemic’s Relationship with the Labor Market

On July 13, 2017, in her semiannual testimony before the Senate Banking Committee,
then–Federal Reserve Chair Janet Yellen
noted the intertwined but complex nature
of the opioid crisis’s relationship with the
labor market and the broader state of the
economy: “I don’t know if [the crisis is]
causal or symptomatic of long-running economic maladies that have affected these
communities and particularly affected
workers who have seen their job opportunities decline.”15
Theoretically, the opioid crisis makes
some workers less likely to search for and
find suitable positions, causing problems
with the labor supply. However, these
same individuals could have been driven
to use drugs as a result of poor health
(e.g., chronic pain due to osteoarthritis
or diabetes)16 or because they were
discouraged due to a long-term decline
in the U.S. demand for low-skilled workers
(e.g., manufacturing jobs), a trend that
is particularly noteworthy for males.17
(See Figure 7.) These labor market symptoms are difficult to discern in large
urban areas such as Philadelphia or New
York, but they become more apparent in
less heavily urbanized regions affected by
the epidemic.18
A 2017 paper19 by Alan Krueger explores
the relationship between a declining labor


Federal Reserve Bank of Philadelphia
Research Department

force participation rate and the opioid
crisis. Krueger notes that “labor force
participation is lower and fell more in the
2000s in areas of the United States that
have a higher volume of opioid medication
prescribed per capita than in other areas.”
He goes on to suggest that 43 percent of
the observed decline in the male labor
force participation rate between 1999 and
2015 could be attributed to the increase
in opioid prescriptions during that time.
However, Krueger notes that it is unclear
whether other factors that result in low
labor force participation (e.g., poor health,
discouraged workers) could have also
resulted in high prescription rates of opioids in certain counties.
Exploring the relationship further, a

2018 NBER paper20 investigated the effect of
opioid prescription rates on employmentto-population ratios at the county level.
Its authors found that the effect is positive
but small for women (suggesting that
higher opioid use in specific counties
allows more women to enter the labor
force), and that there is no relationship for
men. The case for causality is not strong
enough to suggest that opioid prescriptions
lead directly to poor employment outcomes in the above studies.
However, there have been reports of
channels in which this link occurs. One
of those channels is drug testing. A May
2018 Federal Reserve Bank of Cleveland
report noted that, after soliciting input on
how the opioid epidemic was affecting the

TA B L E 1

Accounting for Nonfatal Opioid Costs

Nonfatal costs flow through four main channels.
Health-care sector

Criminal-justice system

Emergency room visits
from overdoses

Police protection

Ambulance rides

Legal hearings and adjudication
Construction, expansion,
and maintenance of
correctional facilities
Property lost due to

Naloxone administration

Disease-related indirect
costs (hepatitis, AIDS,
tuberculosis, etc.)

lost worker productivity

Fewer productive hours
Childcare services due to
as a result of opioid
parental abuse of opioids
abuse and/or dependence
Productivity lost due to

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

on community services

business community, several contacts cited candidates’ inability
to pass drug tests as being a hindrance to finding qualified
employees.21 Thus, while there is not a clear enough causal link
between opioid abuse and poor employment outcomes, the
correlation is sufficient to warrant further study of the link and
optimal policy for assuaging both problems.

Accounting for the Costs of the Opioid Epidemic

The exact costs of this ongoing problem are difficult to measure
precisely, but a number of studies have made an attempt to
reflect the societal costs across a variety of categories. A cost
estimate of the opioid epidemic by the Council of Economic
Advisors (CEA) placed the total cost in 2015 at $504 billion for
the nation, with an expectation for that figure to grow if the
crisis worsened.
In an attempt to quantify the costs of the opioid epidemic
at a more local level, Alex Brill and Scott Ganz used the CEA
estimates along with local variations of health-care costs, criminal
justice services, and worker productivity to arrive at county- and
state-level estimates.22 They found that the total per capita costs
of the opioid epidemic varied from a high of $4,378 in West
Virginia to a low of $394 in Nebraska in 2015, with a median
of $1,672. For the Third District, the per capita costs of the opioid
epidemic in 2015 were $1,907 in New Jersey, $1,945 in Pennsylvania, and $2,530 in Delaware, all above the median. The total
cost of the opioid epidemic across the Third District states in
2015 was nearly $45 billion.
Who bears these costs? How are these costs, fatal and nonfatal,
distributed throughout society? Studies disagree, so rather than
try to estimate the specific numerical costs, we will examine the
channels through which these costs flow.
For fatalities, there are the health-care costs of treating overdosed patients and the losses in future productivity. Much of the
lost productivity is borne by the deceased’s family and the private
sector. Given that the average age for an overdose fatality is 41,
which is considered prime working age,23 the losses in future
productivity are quite high. However, these losses also show up
in federal, state, and local tax receipts.
Aside from these monetized costs, there are numerous
unquantifiable effects on families and communities as a result of
the opioid crisis. Opioid-dependent individuals suffer a substantial decrease in their quality of life. Their families experience
pain and suffering as a result of this dependence and certainly
as a result of any overdoses or overdose fatalities that may occur.
Local communities may suffer any number of problems, including
decreased property values and a loss of community well-being
and safety.

Policy Countermeasures

Since 2010, when opioid prescriptions peaked in the United States,
there have been a number of coordinated policy responses at
the federal, state, and local levels aiming to counteract the opioid
crisis. In 2011, the U.S. Office of National Drug Control Policy
recommended that states have active prescription drug monitoring programs (PDMPs) to counteract overprescribing. A PDMP is

an electronic database that tracks controlled-substance prescriptions in a state.25 These state-level databases track individuals
who obtain prescriptions, informing doctors of patients’ histories
in order to mitigate overprescribing. However, PDMPs are
most effective in preventing overprescribing if the state requires
clinicians to check the state’s PDMP before prescribing a controlled substance. States that have implemented such laws since
2011 have reduced oxycodone prescriptions and oxycodone
deaths.26 Within a year of passing a 2012 law requiring prescribers
to check the state’s PDMP before prescribing opioids, New York
State saw a 75 percent drop in patients seeing multiple prescribers
for the same drugs.
Another major public health response to the proliferation
in opioid overdoses over the past 10 years has been the growing
administration and supply of naloxone. Sold under the brand
name Narcan, naloxone is an opioid antagonist used to temporarily reverse the effects of an opioid overdose.27 State and local
governments have combated overdoses by increasing the supply
of naloxone and training first responders on how to use it. In
2014 the New York Office of the Attorney General provided $1.2
million to supply 20,000 kits to police officers in the state.28 And in
December 2018 the Pennsylvania Department of Health instituted
a program for any Pennsylvania resident to receive naloxone
free from any of 80 locations across the state.29 Although it is
difficult to measure the exact number of overdose deaths that
have been prevented with the drug, in 2017 alone more than
4,000 individuals were administered naloxone by Philadelphia
Emergency Medical Services.30
Improving access to addiction treatment is one of the most
powerful tools for fighting the opioid epidemic, in that it offers
a way out of the cycle of dependence. A number of public
programs treat opioid addiction, including medication-assisted
treatment (MAT), which is a combination of medication, counseling, and behavioral therapy. State and local governments can
set up point-of-contact centers to counsel those seeking recovery
on how to access various treatment options. (Pennsylvania refers
to these as Centers of Excellence.31) States fund these centers
and treatments with Medicaid, a state government insurance
program for limited-income individuals and families, but their
availability varies by state.32

Final Thoughts

As the opioid crisis intensifies in the nation and particularly in our
region, it becomes ever more important to understand its impact
on society. With the rise in the supply and abuse of highly potent
synthetic opioids such as fentanyl, the crisis has entered its
deadliest stage yet. Its costs to society are measured not only in
terms of a diminished labor force and lost productivity but
also in its impact on the health-care sector, the criminal justice
system, and families and communities. Much of the research
on the economic effects of the opioid crisis is still preliminary
and has not captured the last two years’ worth of data, so there
is still much to learn about its continuing effects on society and
the economy.

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department



21 See Fee (2018).

1 See Federal Bureau of Investigation (2018).
22 See Brill and Ganz (2018).
2 The age-adjusted mortality rate standardizes the data by adjusting for
age groups in the population. The age-adjusted rate of drug overdose
deaths rose from 19.8 per 100,000 in 2016 to 21.7 per 100,000 in 2017.

23 See Rhyan (2017).
24 See Florence et al. (2016).

3 Many drug overdose deaths occur from mixing nonopioid drugs with
opioids. For example, the National Institute on Drug Abuse breaks down
overdoses from cocaine and benzodiazepines with and without opioid
involvement. The overdose death rates of these two drugs without opioid
involvement are virtually flat, meaning that opioids, particularly synthetic
opioids, are driving this uptick in overall drug overdose deaths. See Figures
7 and 8 at
4 See Lin et al. (2017).

25 See Centers for Disease Control and Prevention (2017b).
26 See Centers for Disease Control and Prevention (2017a).
27 See U.S. Department of Health & Human Services (2018).
28 See Durando (2014).
29 See Office of the Governor of the Commonwealth of Pennsylvania

5 See Jones et al. (2018).
30 See City of Philadelphia (2018).
6 See Liu et al. (2018).
31 See Pennsylvania Department of Human Services (2019).
7 Opioid prescribing peaked in 2012 for Pennsylvania, 2011 for New
Jersey, and 2010 for Delaware.
8 Prior to the mid-1990s, the heroin market in the United States was
mainly supplied by Asia, via production and transport through countries
such as Afghanistan, Myanmar, and Thailand. In the mid-1990s, traffickers
from Colombia and Mexico flooded the U.S. market with large amounts
of cheap, pure heroin. The supply of heroin from this new source was large
enough to reduce the price per gram of pure heroin throughout the country.
9 See Congressional Research Service (2019).

32 See Grogan et al. (2016).

Brill, Alex and Scott Ganz. “The Geographic Variation in the Cost of the
Opioid Crisis,” AEI Economics Working Paper Series 2018-03 (2018).
Centers for Disease Control and Prevention. “State Successes” (2017a).

10 See O’Connor (2017).
Centers for Disease Control and Prevention. “What States Need to Know
About PDMPs” (2017b).

11 See Ciccarone (2017).
12 See Mars et al. (2017).

Ciccarone, Daniel. “Fentanyl in the U.S. Heroin Supply: A Rapidly Changing
Risk Environment,” International Journal of Drug Policy, 46 (August 2017),
pp. 107–111.

13 See Vestal (2019).
14 See O’Brien (2018).

City of Philadelphia. “Opioid Misuse and Overdose Report,” Department
of Public Health (November 29, 2018).

15 See Lovelace (2017).
16 See Semuels (2017).
17 See Michaels (2017).

Congressional Research Service. “Heroin Trafficking in the United States,”
CRS Report R44599 (2019).

18 Because the Centers for Disease Control and Prevention does not
break out opioid data by sex or race, this relationship must remain
theoretical for now.

Currie, Janet, Jonas Y. Jin, and Molly Schnell. “U.S. Employment and
Opioids: Is There a Connection?” NBER Working Paper 24440 (2018).

19 See Krueger (2018).

Durando, Jessica. “NYPD Officers to Carry Heroin Antidote,” USA Today,
May 27, 2014.

20 See Currie et al. (2018).


Federal Reserve Bank of Philadelphia
Research Department

Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

Federal Bureau of Investigation. “Crime in the United States: 2017—Murder”

pp. 7–16.

Fee, Kyle. “The Opioid Epidemic and Its Effects: A Perspective on What
We Know from the Federal Reserve Bank of Cleveland,” Federal Reserve
Bank of Cleveland Community Development Briefs (May 31, 2018).

O’Brien, Michael. “Fentanyl Changed the Opioid Epidemic. Now It’s
Getting Worse,” Rolling Stone, August 31, 2018. https://www.rollingstone.

Florence, Curtis, Chao Zhou, Feijun Luo, et al. “The Economic Burden
of Prescription Opioid Overdose, Abuse, and Dependence in the United
States, 2013,” Medical Care, 54:10 (October 2016), pp. 901–906.
Governor of the Commonwealth of Pennsylvania, Office of. “Gov. Wolf
Receives Naloxone Kit for Stop Overdoses in PA: Get Help Now Week”
(December 10, 2018).
Grogan, Colleen, Christina Andrews, Amanda Abraham, et al. “Survey
Highlights Differences in Medicaid Coverage for Substance Use Treatment
and Opioid Use Disorder Medications,” Health Affairs, 35:12 (December
2016), pp. 2289–2296.
Jones, Mark R., Omar Viswanath, Jacquelin Peck, et al. “A Brief History of
the Opioid Epidemic and Strategies for Pain Medicine,” Pain and Therapy,
7:1 (June 2018), pp. 13–21.
Krueger, Alan. “Where Have All the Workers Gone? An Inquiry into the
Decline of the U.S. Labor Force Participation Rate,” Brookings Papers on
Economic Activity (2018).
Lin, Dora H., Eleanor Lucas, Irene B. Murimi, et al. “Financial Conflicts
of Interest and the Centers for Disease Control and Prevention’s 2016
Guideline for Prescribing Opioids for Chronic Pain.” JAMA Internal Medicine,
177:3 (March 2017), pp. 427–428.

O’Connor, Sean. “Fentanyl: China’s Deadly Export to the United States,” USChina Economic and Security Review Commission Staff Research Report
(February 1, 2017).’sdeadly-export-united-states.
Pennsylvania Department of Human Services. “Centers of Excellence”
Rhyan, Corwin. “The Potential Societal Benefit of Eliminating Opioid
Overdoses, Deaths, and Substance Use Disorders Exceeds $95 Billion Per
Year,” Altarum Research Brief (November 16, 2017).
Semuels, Alana. “Maybe the Economy Isn’t the Reason Why So Many
American Men Aren’t Working,” The Atlantic, March 22, 2017. https://
U.S. Department of Health & Human Services. “Surgeon General’s Advisory
on Naloxone and Opioid Overdose” (2018). https://www.surgeongeneral.
Vestal, Christine. “So Far, the Deadly Fentanyl Epidemic Hasn’t Hit
California. Here’s Why,” Washington Post, February 4, 2019. https://www.

Liu, Lindsy, Diana N. Pei, and Pela Soto. “History of the Opioid Epidemic:
How Did We Get Here?” Poison Control: National Capital Poison Center
Lovelace, Jr., Berkeley. “The Head of the Federal Reserve Believes Opioid
Abuse Could Be Holding Back the U.S. Economy,” CNBC, July 13, 2017.
Mars, Sarah, Jeff Ondocsin, and Daniel Ciccarone. “Sold as Heroin:
Perceptions and Use of an Evolving Drug in Baltimore, MD,” Journal of
Psychoactive Drugs, 50:2 (December 6, 2017), pp. 167–176.
Michaels, Ryan. “Why Are Men Working Less These Days?” Federal
Reserve Bank of Philadelphia Economic Insights (Fourth Quarter 2017),
Exploring the Economic Effects of the Opioid Epidemic
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department


Regional Spotlight

Smart Growth for
Regions of All Sizes
Pittsburgh’s population has shrunk by 400,000 since 1969,
making it the poster child for urban shrinkage. So why is it
doing so well? Sometimes, smaller really is better.

Paul R. Flora is
the manager of
regional economic analysis
in the Research
Department of
the Federal
Reserve Bank of
Philadelphia. The
views expressed
in this article are
not necessarily
those of the
Federal Reserve.



mong the truisms to which regional policymakers frequently adhere, the most pervasive may be that a region
must grow to be successful. However, population growth
and job growth are not preconditions for a region to become
economically healthy. Rather, the composition and characteristics
of jobs required to meet the demands of a region’s growing (or
changing) industrial structure typically determine the health of a
region’s economy. High-productivity industries and progressively
managed firms generate high-skill, high-wage jobs that raise
a region’s per capita income.
Indeed, for a region that has lost its prior locational advantage
or is unable to attract high-productivity industries, growth may
falter or reverse, but its economic health need not decline if
policies recognize the transition; assist those most impacted; and
generally lower the cost of living, including adjustments to the
scope and cost of public infrastructure. Too often, shrinking
regions refuse to accept the reality that their population will not
resume a growth path.
Aside from the apparent fallacy of the growth paradigm, the
policies that emanate from such a belief typically maintain unrealistic expectations given a region’s economic structure, often
ignore stronger countervailing market forces, and routinely waste
resources in pursuit of misplaced goals. Pittsburgh’s regional
economy offers a persuasive counterexample to the growth paradigm even though its policymakers also often chased economic
growth rather than economic health.
Pittsburgh has arguably held the crown for population decline
among the nation’s largest metropolitan regions for more than
half a century. The region’s population began falling in the 1960s.
From 1969 to 2017, the Pittsburgh region’s population fell further,
and more consistently year to year, than any other major metropolitan region in the country, including Detroit.1 Yet no sooner
had most of Pittsburgh’s steel mills closed for good in the early
1980s than Pittsburgh began garnering accolades as the “most
livable city” and generating a stream of positive media coverage
describing its economic comeback.


Federal Reserve Bank of Philadelphia
Research Department

Moreover, consider that in 1969 the Pittsburgh metro area was
the ninth largest in the nation.2 Pittsburgh has since lost population in 41 of the 48 years through 2017, amounting to a net loss
of more than 400,000 people.3 As Pittsburgh shrank, 17 other
metro areas surpassed it in population (Figure 1).4
While Pittsburgh’s net population was falling at a pace of 0.3
percent per year, Orlando, FL, added nearly 2 million people (3.4
percent annualized growth). Las Vegas, which will likely surpass
Pittsburgh within the next five years, has grown at a 4.5 percent
annualized rate since 1969. Has rapid growth in Orlando and Las
Vegas made their economies healthier than Pittsburgh’s?
In 2016, Pittsburgh’s real per capita income was $60,797—nearly
$12,000 more than in Las Vegas and over $15,000 more than
in Orlando.5 In 2017, Pittsburgh’s poverty rate was 11.8 percent
(a five-year-average estimate); the poverty rate was 14.6 percent in
Las Vegas and 15.4 percent in Orlando. In Las Vegas and Orlando,
population and job growth have produced neither greater overall
incomes per capita nor a more equitable distribution of income.
Of the 17 metro areas that have surpassed Pittsburgh in
population since 1969, only three (Minneapolis, St. Louis, and
Seattle) had higher real per capita incomes
than Pittsburgh as of 2016. And only four
See Pittsburgh’s
(Baltimore, Denver, Minneapolis, and
Path to EconomSeattle) had lower poverty rates as of 2017.
ic Renewal BeDespite over five decades of population
gan Long Before
decline, Pittsburgh remains an economically Its Decline
better place to live than most of its now
larger peers.6
How does a region’s economy improve while declining in population? What does regional economic growth mean, and how
should we measure it? Most important, how can we ensure that
policymakers are governed by a realistic appraisal of a region’s
prospects and develop strategies to grow better, not just bigger?
These questions are especially important in Pennsylvania. In
seven of the state’s 15 more mature metropolitan statistical areas,
population has been shrinking for most of the past 50 years.

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2


Since 1969, 17 Metro Areas Have Surpassed
Pittsburgh’s Population

Millions of residents; excluding New York, Los Angeles, and Chicago

Pittsburgh lost over


areas have
surpassed Pittsburgh
17 metro

people, 1969–2017





Of the 17 metro areas,
compared to Pittsburgh,
only these cities have:
Lower poverty rates
Higher real per capita income




San Diego
St. Louis
San Antonio



In 1969, Pittsburgh was the ninth largest metropolitan region in the
country, but its population had been falling since 1960, in step
with the decline of the nation’s integrated steel industry. Other than
auto manufacturing in Detroit, arguably no other industry had
dominated a region more thoroughly than steel had dominated
Pittsburgh.16 During the double-dip recession of the early 1980s,
most of Pittsburgh’s large integrated steel mills closed their doors
for good.17
Meanwhile, the advent of air conditioning ushered in the migration
from the Rust Belt to the Sun Belt and propelled numerous southern regions into periods of rapid growth. As of 2017, 17 metropolitan
statistical areas (MSAs) had surpassed Pittsburgh in population.
Miami, Dallas, Phoenix, and Orlando shot past on their own rapid
trajectories. Others, such as St. Louis and Baltimore, edged past
by rising slowly.
The seeds for Pittsburgh’s revival had been sown over many prior
decades, beginning perhaps with the writings of the 1930s business
economist Glenn McLaughlin, who warned city leaders that steel
was a mature industry and the region should diversify to prepare
for steel’s gradual decline.18 Pittsburgh engineered numerous critical
changes, including strict county air pollution regulations, flood
control projects, and downtown renewals such as Renaissance
One, Renaissance Two, and several successors.
While later researchers have pointed to myriad current explanations
for Pittsburgh’s economic revival, the decades of prior city renewal
efforts enabled Pittsburgh to reinvent itself after the decline of
steel. From the 1990s on, firms like Google and RAND have been
attracted to Pittsburgh as Carnegie Mellon University and the University of Pittsburgh led advances in higher education, the medical
sector, life sciences, computer science, and robotics. Meanwhile,
PNC Bank supplanted Mellon Bank as the region’s major financial
employer and has been a significant downtown developer since the
1990s. Thus the stability of Pittsburgh’s economic health during
the past 35 years has resulted from the emergence of high-skill,
high-paying professional and technical jobs, which eased the sting
of losing thousands of lower-skill, well-paying steelworker jobs.
Yet the region’s population continued to shrink. Following the
demise of steel, economic policymakers persisted with the assumption that the region would begin to grow next year and they
planned accordingly—against the advice of regional economists
that ongoing decline was more realistic.19 Pittsburgh likely would
have fared better (with greater gains or a smoother transition) had
policymakers continued to heed the advice of regional economists,
as in prior decades.



Pittsburgh’s Path to Economic Renewal
Began Long Before Its Decline



Source: Bureau of Economic Analysis.

Note: The chart was constructed for the 53 msas in the U.S. with populations greater than 1 million people in 2017.

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department


Moreover, 12 of 20 smaller micropolitan
statistical areas, such as Oil City, PA, have
seen declining populations since at least
2000. Only seven micro areas have grown
during that period, and only four grew
sufficiently to earn a “promotion” to metro
area status.

The Case for Regions to Pursue
Per Capita Income Growth
Rather Than Jobs

Regions are too often compared on sizebased measures such as population
growth, employment growth, GDP, and
total personal income. Less often, or as
a secondary measure, regions are compared on a per capita basis, such as per
capita income. Let’s call these economichealth-based measures.7
Whereas GDP measures a region’s total
economy, GDP per capita represents a
region’s average potential spending power,
which falls when population rises faster
than GDP. Thus, growth in per capita
measures is more vital to the well-being of
a region’s people. Individuals and families
are better off in regions with higher
per capita income, or where per capita
incomes are rising faster.8
In fact, a region’s population growth
has a clear, strong correlation with the
growth of its total real personal income.
Since 1969, Austin, TX, Las Vegas, Orlando,
Phoenix, and Raleigh, NC, have been the
five fastest-growing regions for both population and total real personal income. By
contrast, Cleveland, Detroit, Pittsburgh,
and Buffalo, NY—the only four regions in
which population has fallen since 1969—
bring up the rear with the slowest growth
of total real personal income (Figure 2).9
However, the strong positive correlation
dissipates when real per capita income
is examined in place of total income.
Instead, a weak negative correlation appears. Austin and Raleigh still show strong
growth rates for real per capita income,
but Las Vegas, Orlando, and Phoenix
show relatively weak growth rates. Meanwhile, Buffalo, Detroit, and Cleveland
show below-average growth in real per
capita income, while Pittsburgh’s growth
is above average (Figure 3).
Why do size comparisons persist if per
capita comparisons matter more for individual welfare? For one, size comparisons


Federal Reserve Bank of Philadelphia
Research Department

were once much simpler to make, and
their use continues through inertia.10
Moreover, aggregate economic growth
is generally perceived as a desirable
outcome that benefits firms as well as
state and local governments. Because
local businesses and local politicians
typically frame the economic development
conversation, size comparisons tend
to dominate the analysis. To be sure,
population growth leads to new home
construction, rising retail sales, and
a larger tax base.
Thus, firms seeking to increase revenues or market share will expand
into growing suburban areas and in the

Sun Belt, where population and consumer
demand are growing most rapidly. At least
three unfortunate events await many of
these firms.
First, retailers have often extended
their suburban reach too far, only to
be reminded that the urban center retains
a locational advantage for well-heeled
customers. Second, over time in regions
with a rising percentage of lower-income
households, local firms will find their
profit margins squeezed by price-sensitive
consumers. Finally, when growth does
stop, local businesses will be left servicing
less-profitable customers than their counterparts in high per capita regions.


Total Personal Income Growth Is Highly Correlated with Population Growth
Austin, TX, and Las Vegas are the latest U.S. boom towns.

Annualized change in aggregate real personal income vs. annualized percent change in pop., 1969 to 2017
Aggregate real personal income change

Las Vegas



Dallas Riverside




San Antonio
San Diego



St. Louis



Population change




Source: Bureau of Economic Analysis.

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2



Taxing entities also prefer population
growth, which often masks an unsustainable fiscal structure. Florida is able to provide state and local government services
without an income tax by relying on
impact fees for new development, a high
sales tax that generates a large proportion
of its revenues from tourists, and a “welcome stranger” property tax that caps
increases for homestead property owners
and shifts a disproportionate burden to
new homebuyers.11
However, when growth stops, the cost
of providing and maintaining infrastructure and delivering services grows
faster than tax revenues. In Florida, the

Great Recession revealed an unpleasant,
surprise feature of its tax cap for existing
homeowners: As their property values
plummeted, their assessed values continued to rise, which resulted in higher
property taxes.12
If a region’s population declines permanently, so too does its tax base, often
forcing state and local governments
to raise taxes on a population with lower
incomes and less wealth. The resulting
fiscal stress often prompts local officials to
pursue growth strategies—unsuccessfully
in the face of much stronger countervailing economic forces. Within a region, this
problem may be further magnified by


But Per Capita Income Growth Is Not Correlated with Population Growth

Austin, TX, and Las Vegas have widely disparate rates of per capita income.

Annualized change in real personal income per capita vs. annualized percent change in pop., 1969 to 2017
Per capita real personal income change


St. Louis


San Antonio




Portland Tampa



Las Vegas


Population change





Source: Bureau of Economic Analysis.

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2


local government fragmentation, which
can accelerate a migration of households
with means from fiscally distressed
cities to surrounding suburban and rural
Understandably, firms and regions are
inclined to try to stay on the easier path
by pursuing continuous growth, but it is
unrealistic to expect to be forever immune
from economic shocks that cause growth
to slow, stop, or reverse.

Industrial Structure Remains
Key to Assessing a Region’s

Since François Quesnay published his
Tableau économique in 1758, economists
have studied nations, and then regions,
in terms of the industries that compose
the economy to better understand what
drives economic growth. Quesnay believed that agricultural surpluses were the
prime mover.
Today, economists speak of a region’s
export base (or economic base) as those
sectors associated with the region’s production of goods or services in excess of
local demand. The auto industry remains
a significant part of Detroit’s economic
base, film studios of Los Angeles’s economic base, and finance of New York’s
economic base.
A region’s export base affects its economy in two key ways. First, employment
typically rises (or falls) as industries
present within a region’s export base grow
(or decline). Second, per capita income is
greater in regions whose export-base
sectors utilize highly skilled, high-paid
In turn, the usual multiplier effects
that generate local jobs (e.g., carpenters,
teachers, clerks, and wait staff ) will be
stronger in regions in which higher-paid
export-base workers will consume highvalue goods and services.
To better understand how a region
attains high real per capita incomes with
or without population growth, I compare
four regional economies that represent
four extremes of the distribution and
examine their industry mix: Austin, with
high population growth and high per
capita income growth; Las Vegas, with
high population growth and low per
capita income growth; Cleveland, with
Federal Reserve Bank of Philadelphia
Research Department


low population growth and low per capita income growth; and
Pittsburgh, with low population growth and high per capita
income growth.
Not surprisingly, the industry mix differs substantially among
these four regions. A good sense of the differences can be gained
by identifying within each region the top five industrial sectors
(by location quotient, a measure of an industry's concentration
in an area) for which the sector employs at least 5,000 workers
(Figure 4).13
Austin’s top five sectors are representative of an economy with
a large concentration of high-paying, high-tech jobs at firms that
design and produce computer hardware and software. Pittsburgh
is still characterized by its steel industry legacy plus its education
and health sectors. Cleveland retains concentrations in many
small, diversified manufacturing industries (nine of its 10 largest
location quotients were manufacturing sectors) but has also
experienced a shift to health care. In contrast, the Las Vegas
economy is heavily concentrated in tourism sectors that do not
pay very high wages.
To provide a more comprehensive comparison, I computed
a weighted average wage for employees who represent the export
base of each region.14 The export-base employment of Austin
generated an average weekly wage of $1,841 in 2016—significantly
higher than Pittsburgh’s $1,346 and Cleveland’s $1,245. Exportbase workers in Las Vegas averaged only $793 per week.
How a region grows, whether in high-skill, high-wage sectors
or in low-skill, low-wage sectors, has important implications for
the overall long-term health of the region’s households and of the
region itself. Were growth alone responsible for lifting a regional
economy and all of its participants, then 46 years of rapid
growth should have turned Las Vegas into one of the healthiest
economies in the nation. If instead population growth is driven

Industry Matters More
Than Population Growth
2017 annual average salary;
2017 average weekly wages; sectors
employing at least 5,000 workers

Real per capita
income growth

Four Examples of Extreme Distribution
Population change



Las Vegas

primarily by low-wage jobs in sectors such as call centers,
construction, tourism, and warehousing, then a region may grow
poorer while its population is growing larger.
The distribution of 2016 real per capita incomes adjusted by
regional price parities further demonstrates a lack of correlation
with population growth and reflects instead the industrial
structure (Figure 5). In every income bracket, one can find fastgrowth and slow-growth regions. Despite very slow growth (or
no growth), Cleveland, Philadelphia, Pittsburgh, and St. Louis
enjoy per capita income levels equal to Denver, Houston, and
Nashville, TN, which have all grown at twice the national rate. Of
note, high-income outliers (San Jose, CA, San Francisco, and
Boston) have significantly lower population growth rates than the
low-income outliers (Orlando, Tucson, AZ, Las Vegas, Phoenix,
and Tampa, FL).
Finally, 2017 poverty rates show a distribution that is likewise
uncorrelated with population growth. Austin, Cleveland, Las
Vegas, and Pittsburgh are once again located in separate quadrants of the scatter plot (Figure 6).15
Despite enduring population losses for over 50 years, the
Pittsburgh region maintains higher real per capita income and
a lower poverty rate than most of its peers. This outcome suggests
that regional policymakers should not simply seek job growth
but should pursue development strategies that emphasize the
quality of jobs and the needs of the resident population.

How to Grow a Healthier Economy—Without
(Necessarily) Growing More Populous

Regions face the same basic challenges, whether they are anchored by a large, mature, slow-growing city; a midsize, youthful,
rapidly growing city; or a small, declining city contemplating the

Salaries for Sectors with Five Highest Location Quotients in Each Region
Annual average salary




Computer systems design

Data processing, hosting
Nonbank lenders


Other machinery mfg.




Taxis & limos





Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2



Steel mills

Machine shops
Note: Location quotient quantifies how
Mental health
Other support services
concentrated an industry is compared to the U.S.
facilities Bars
average. A location quotient of 10 indicates
Nursing Business support services
jobs in a specified industry are 10 times more
concentrated in a specified region as in the
School transport
nation as a whole.
Source: Quarterly Census of Employment and
Wages, Bureau of Labor Statistics.
Location quotient

Federal Reserve Bank of Philadelphia
Research Department


Semiconductor & electronics mfg.

Business mgmt.


Computer & peripherals mfg.


health care


Average Weekly Wage of
Export-Related Employment


Las Vegas

promise from the latest resource boom.
Regions typically strive to deliver public
services and to enable the provision of
amenities to meet the needs of residents
and firms within a fiscally sound, longrange budget constraint. A region’s industrial structure is a major determinant of
the budget constraint. First, policymakers
should approach economic development
with a realistic understanding of their
region’s place and prospects in the world
economy. A comprehensive economic
base analysis provides a good start to avoid
setting unattainable goals and wasting
resources on empty strategies. This
analysis should undergird any objective
assumptions about future population or
employment trends.
Next, regions should develop an infrastructure plan and pragmatic policy
solutions for addressing the economic
needs of existing and future residents,
including a sustainable fiscal path for
the region’s local governments. Ideally,
each region would produce a multijurisdictional fiscal impact analysis of its

long-range comprehensive plan to ensure
its efficiency and feasibility.
In addition, all regions, but especially
those with a high proportion of low-wage
jobs, may find that they need to budget for
strategies that reduce the cost of living
for households living on those minimumwage jobs. Policymakers may need to reduce barriers to education and labor force
participation by, for example, assisting
with day care, health care, job training,
and transportation needs for marginally
attached workers, and by encouraging
the provision of affordable housing near
jobs and in transit-oriented locations.
Finally, and especially for regions that
are stagnant or declining in population,
policymakers should consider developing
people-oriented policies that help persons
relocate to regions with greater job opportunities and consider rationalizing public
infrastructure with incentives for the region’s residents to consolidate into a more
compact urban form.
These recommendations do not include
policies to target and attract particular

businesses. Better perhaps is a policy to
not target or subsidize any business that
does not utilize high-value occupations, or
that does not make immediate use of a
region’s existing underutilized labor force.
A final key point is that all of the above
policy prescriptions are (optimally)
regional. Cities, their suburbs, and their
hinterlands will realize their greatest
economic success by working as one.
Ideally, states would encourage municipal
consolidations that expand the political/
fiscal base to match a region’s economic
footprint. Since municipalities are
creatures of the state, the burden of failing
to relieve the inefficiencies of local government fragmentation will fall to the state
to address.



Population Gains Do Not Determine Levels of Income

…Nor Poverty Rates

Real per capita income (2016) adjusted with regional price parities (2018 dollars);
annualized percent change pop.: 1969 to 2017

Five-year poverty estimates (2017); annualized percent change pop.: 1969 to 2017

Largest U.S. metropolitan areas' growth and income.
Real per capita income

Largest U.S. metropolitan areas' growth and poverty.
Poverty rate


60,000 Pittsburgh


St. Louis


Baltimore Charlotte
San Diego
San Antonio
Las Vegas



Las Vegas


San Anton


Population change






Population change




Note: Income has been adjusted for regional cost-of-living differences.

Note: Poverty rates are not adjusted for regional cost of living.

Source: Bureau of Economic Analysis.

Sources: Bureau of Economic Analysis and U.S. Census Bureau.

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department


San Diego




St. Louis




San Diego




San Antonio







1 Cleveland’s population fell during the same period at a slower rate than
did Pittsburgh’s. Detroit’s population fell at a much slower rate and rose
in more years than it fell. Buffalo’s population declined at a similar rate,
but less consistently, and Buffalo was, and is, half the size of Pittsburgh.
2 Unless otherwise noted, region and metro area refer to official metropolitan statistical area (MSA). Analysis in this article is based on data
for each MSA as delineated in the Office of Management and Budget
Bulletin 18-03 issued April 10, 2018. This article truncates these official
names to the names of their largest principal cities.

legislated increase even when the market value has fallen, as long as the
assessed value is below market value.
13 This selection was necessarily an arbitrary one that misses sectors
with smaller location quotients, which may employ significantly more
workers or pay significantly higher wages. Moreover, key sectors with
higher location quotients may have been suppressed in the Quarterly
Census of Employment and Wages data set based on nondisclosure
rules of the Bureau of Labor Statistics.

3 The brief interludes of population growth occurred in years that
followed economic recessions, suggesting that some of the Pittsburgh
diaspora returned home after losing jobs in other regions. (They may
have felt that it is better to be unemployed near family and friends than
in a relatively strange place.)

14 The average weekly wage was calculated on a weighted basis across
all sectors within each region with a location quotient of 1.2 or higher.
The average weekly wage in each of these sectors was multiplied by the
number of workers in each sector in excess of the number required to
reach a location quotient of 1.2. The employment and wage data are 2017
annual averages for all sectors in a region except those sectors for which
the BLS suppressed data because of nondisclosure rules.

4 Recently released census estimates of 2018 population indicate that
Sacramento, CA, became the 18th metro area to surpass Pittsburgh’s

15 Some shifting and some compression would occur in this scatter plot
if poverty rates could be adjusted for regional price parities, as was done
with per capita income.

5 Per capita income estimates have been adjusted for cost-of-living
differences and are expressed in 2018 dollars.

16 See Chinitz (1961).
17 See Hoerr (1988).

6 The 17 metro areas that have surpassed Pittsburgh in population since
1969, with the year in which they reached that milestone, are Miami (1974),
Dallas and Houston (1976), Atlanta (1985), St. Louis (1986), Minneapolis
and Seattle (1989), Riverside, CA, and San Diego (1990), Phoenix (1993),
Baltimore (1995), Tampa, FL (2001), Denver (2006), Charlotte, NC (2014),
and Orlando, FL, Portland, OR, and San Antonio (2015).
7 Other economic-health-based measures include poverty rates,
unemployment rates, and comparative cost-of-living measures.
8 This statement assumes that other variables are the same, including
potential income inequality.
9 The article analyzes the 53 metro areas in the United States with
populations greater than 1 million in 2017. The United States as a whole
is also represented. The variables are based on data from the U.S.
Census Bureau, the Bureau of Labor Statistics, and the Bureau of
Economic Analysis.
10 Adjusting economic data to eliminate potentially distorting underlying
factors—such as population growth or the presence of an unusual
number of college students, retirees, migrants, or prisoners—was difficult
before computers and remains complicated today.

18 See McLaughlin (1938).
19 See Giarratani and Houston (1989).

Chinitz, Benjamin. “Contrasts in Agglomeration: New York and Pittsburgh,” The American Economic Review, 51:2 (May 1961), pp. 279–289.
Giarratani, Frank, and David B. Houston. “Structural Change and Economic Policy in a Declining Metropolitan Region: Implications of the
Pittsburgh Experience,” Urban Studies, 26:6 (1989), pp. 549–558.
Hoerr, John. And the Wolf Finally Came: The Decline of the American Steel
Industry. Pittsburgh: University of Pittsburgh Press, 1988.
McLaughlin, Glenn E. Growth of American Manufacturing Areas: A
Comparative Analysis with Special Emphasis on Trends in the Pittsburgh
District. Pittsburgh: University of Pittsburgh, Bureau of Business Research, 1938.

11 Florida’s constitution limits the annual increase in assessed value of
properties with a homestead exemption to 3 percent or the change in
the Consumer Price Index, whichever is lower. New homebuyers can
face tax bills that are several times greater than their long-tenured neighbors in comparable properties.
12 The recapture rule of the Florida law requires that homestead properties with an assessed value below market value must be assessed the


Federal Reserve Bank of Philadelphia
Research Department

Regional Spotlight: Smart Growth for Regions of All Sizes
2019 Q2

Implementing Monetary
Policy in a Changing
Federal Funds Market
As the Fed normalizes its balance sheet, it helps to understand how the federal funds market used to operate, how it
changed in the wake of the crisis, and what comes next.

Benjamin Lester is
a senior economic
advisor and economist at the Federal
Reserve Bank of
Philadelphia. The
views expressed in
this article are not
necessarily those of
the Federal Reserve.



very six weeks or so, the financial world watches as the
Federal Open Market Committee (FOMC) decides on
a target interest rate in the federal funds market. But what
happens next? How do policymakers make sure that interest
rates in the fed funds market trade within the target range?
Though not widely discussed, the framework that the FOMC
uses to implement monetary policy has changed over the last
decade and continues to evolve today. Before the financial
crisis—when reserves were scarce—policymakers used one set
of instruments to achieve the target rate. However, several
important policy interventions introduced soon after the crisis
drastically altered the landscape of the fed funds market. This
new environment—with ample reserves—necessitated a new set
of instruments for monetary policy implementation. Now, as the
FOMC begins to unwind the effects of these policy interventions,
the question arises: What happens next as the fed funds market
converges to a “new normal”?

Implementing Monetary Policy Before the Crisis

Banks hold reserves in an account at the Fed and are required to
maintain a balance above a certain fraction of their deposits—
so-called required reserves.1 Prior to the onset of the Great
Recession in December 2007, a defining feature of the fed funds
market was that reserves were scarce. As a result, throughout
the day a bank’s reserves would fluctuate as payments were
made or received, and some banks would find themselves short
of their reserve requirements at the end of the day. In order
to avoid borrowing at the Fed’s discount window, these banks
would look to borrow from other banks in the fed funds market.2
At the same time, some other banks would find themselves
holding excess reserves at the end of the day. Since the Fed
didn’t pay interest on excess reserves deposited overnight, these
banks would look to lend in the federal funds market to earn
a positive rate of return. As there were a significant number of
banks on both sides of the market—some looking to borrow and

others looking to lend—trading volume in the fed funds market
was substantial, and interbank trades dominated market activity.
For instance, Afonso, Entz, and LeSueur estimate an average
daily trading volume of approximately $200 billion in the fourth
quarter of 2006, of which approximately 60 percent was accounted for by bank-to-bank lending.
In this environment of scarce reserves, monetary policy implementation was fairly straightforward. The Open Market Trading
Desk (the Desk) at the Federal Reserve Bank of New York would
implement the desired target for the effective federal funds rate
(EFFR) by adjusting the supply of reserves via open market operations.3 For example, if the Desk wanted to increase market rates,
it would sell securities (such as Treasury bills) in the market, thereby decreasing the supply of cash held by banks. As banks’ supply
of cash became scarcer, the rate at which they would be willing to
lend would rise. Hence, as in the usual model of supply and
demand, a reduction in the supply of reserves in the market
would lead to an increase in the fed funds rate. (See Figures 1 and
2.) As the fed funds rate rose, market rates would rise as well.

Three Important Changes

The landscape of the fed funds market was altered dramatically
following the financial crisis. First, and most important, the Fed’s
large-scale asset purchase programs left depository institutions
awash with reserves. Over three rounds of “quantitative easing”
in 2008, 2010, and 2012, the Fed purchased assets such as U.S.
Treasury debt and agency mortgage-backed securities.4 As the
Fed bought these assets, the banks that sold them saw their
reserve balances soar. As a result, excess reserves held by depository institutions reached nearly $2.7 trillion by August 2014.
To put that in perspective, in the precrisis years, excess reserves
typically hovered between just $1 and $2 billion.
Second, changes in the assessment of FDIC fees made borrowing in the interbank market more expensive for domestic banks.
In response to the Dodd–Frank Act of 2010, the Federal Deposit

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department



How Interbank Lending in the Fed Funds Market Worked Before the Crisis
Until 2008, the Fed did not pay
interest on excess reserves
and so there was no incentive
to hold onto them.

Amount required
Banks’ deposits
determine the
amount of
reserves they
are required to
hold at the
Federal Reserve.

If Bank A cannot find a lender,
it can borrow from the Fed
directly at the discount
window rate, which is higher
than the rates at which
reserves were typically traded
in the federal funds market.

1 Bank B
lends its
to cover
Bank A’s
A Banks B


Until 2015, the effective federal funds rate
was calculated as the volume-weighted
average of overnight loans in the federal
funds market.


Bank A can now
meet reserve
requirements and
will repay Bank B
its loan with


How the FOMC Raised Rates

The Federal Open Market Committee (fomc) did not achieve the desired rate directly.
Instead it used supply and demand for reserves to achieve a rate within the target range.
Effective federal
funds rate

Federal Funds Market Open Market Operations
Treasury securities
held by the Fed

Postsale rate

To raise rates, the
Fed sold Treasury
securities to
decrease the
supply of reserves
within the federal
funds market.

Presale rate

an s
m rve
De ese


re upp
se ly
rv o
es f


Postcrisis Implications

to banks


Federal Funds Trade Volume
Federal funds lending, usd, bn

2006 Q4

2012 Q4

Source: Federal Reserve Bank of New York (


Federal Reserve Bank of Philadelphia
Research Department

Insurance Corporation (FDIC) changed the
basis for its fees from a bank’s deposits
to its assets. Since a bank’s reserves are
included in the calculation of its assets,
this policy change increased FDIC fees and,
hence, the cost of borrowing reserves on
the interbank market. Economists estimate
that these policy changes implied an
additional cost between 4 and 7 basis
points for each extra dollar of cash on
a bank’s balance sheet.5 However, FDIC fees
are imposed only on banks with U.S.
deposits, and branches of foreign banks
typically don’t hold U.S. deposits, so
this policy change raised the cost of borrowing for domestic banks while leaving
foreign banks with U.S. subsidiaries largely
Third, in October 2008, in the hope of
putting a “floor” beneath market rates, the
Fed started paying an interest rate of 25
basis points on overnight reserves deposited by banks.6 However, this overnight rate
was not made available to other financial
institutions, including governmentsponsored entities like the Federal Home
Loan Banks (fhlbs) as well as money
market funds. As a result, the introduction
of interest on reserves (IOR), with eligibility
restrictions, created a gap between the
interest rates available to different types
of financial institutions.

These changes altered the fed funds market
in a number of important ways, including
the types of financial institutions that
were trading, the rates at which they were
borrowing and lending, and the tools
available to the FOMC that could effectively
influence these market rates.
Because banks were awash with
reserves, their desire to borrow effectively
vanished, and bank-to-bank lending largely
disappeared. However, once the Fed
started paying interest on reserves to
some (but not all) financial institutions,
a new lending opportunity emerged.
To understand this opportunity better,
consider a financial institution ineligible to
receive interest on reserves at the Fed,
such as an FHLB.7 At the end of the day, it
likely holds some amount of cash, but the
highest overnight interest rate it could
receive—what economists call its “outside
option”—was a zero percent net return.

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

Eligible financial institutions, however,
had a better outside option, since they
could deposit money at the Fed and
earn the IOR rate (initially set at 25 basis
points), less any costs associated with
expanding their balance sheet. Because
only domestic banks incurred FDIC fees
from increasing their asset position,
foreign banks faced smaller costs and
thus had an advantage in borrowing.
Hence, an opportunity for arbitrage
emerged: The FHLB could lend to an
eligible bank at a rate above zero (its outside option) but less
than the IOR rate,
See Arbitrage
and the eligible bank in the Fed Funds
could lend those
reserves to the Fed
at the IOR rate (its outside option). “Arbitrage in the Fed Funds Market” describes
in greater detail the arbitrage opportunity
that emerged because of differing outside
options, the effects of borrowing costs
like FDIC fees, and the determination of
a mutually agreeable interest rate.
As a result of the many changes in
the immediate aftermath of the crisis, the

majority of trading in the fed funds
market was occurring between ineligible
financial institutions, like FHLBs, and
eligible financial institutions with low costs
of borrowing, like U.S. branches of foreign
banks, at rates below the IOR rate being
offered at the Fed. Moreover, with no
bank-to-bank lending, the overall market
volume dropped precipitously, to $80
billion or less per day. (See Figure 3.)

The Fed had been controlling the outside option of eligible banks via the IOR
rate since October 2008. However, if the
Fed adjusted this rate alone, the gap
between the two outside options would
widen as the IOR increased and, as
a result, market rates might not rise in
sync with the IOR. So in September 2013
the FOMC introduced an instrument to
adjust the outside option of ineligible
institutions, too, via the overnight reverse
repurchase agreement facility, or ON RRP.
In a reverse repurchase, the Desk sells
a security to an eligible counterparty with
an agreement to buy the security back
at a specified date and price, with the
interest rate computed from the difference
between the original purchase price and
the (higher) repurchase price. Importantly,
the FOMC included a wide range of market
participants as eligible counterparties at
the ON RRP facility, including FHLBs and
key money market funds.9 By adjusting the
rate being offered at the ON RRP facility,
the FOMC was thus adjusting the outside
option of essentially all major financial
institutions ineligible to earn IOR at the Fed.

Implementing Monetary Policy
After the Crisis

These changes to the fed funds market
required policymakers to devise a new
system for implementing monetary policy.
Since the market rate was no longer
primarily determined by banks’ supply
and demand for reserves, typical open
market operations would have essentially
no effect on market rates.8 Instead, when
the FOMC decided to raise interest rates
after a long period at zero, it did so by adjusting the outside options of the lenders
and the borrowers in this market via
administered rates.

Arbitrage in the Fed Funds Market

A number of factors could determine the bargaining power of a bank
or an FHLB. For example, an FHLB that can quickly and easily find
an alternative bank to trade with would be in a relatively strong bargaining position. However, a bank that was desperate to borrow
to avoid violating reserve requirements would be in a relatively weak
bargaining position.

Before the
introduction of
the on rrp
facility, fhlbs
earned no
interest on
overnight cash


0 bp

This gap creates an
opportunity for arbitrage.

25 bp

Bank has more
bargaining power.
25 bp
20 bp


What determines the interest rate at which they actually trade? In
bilateral transactions like this, we often assume that the two parties
negotiate or “bargain.” Moreover, we assume that the interest rate
at which they agree to trade depends on each party’s relative negotiating skill or “bargaining power.” If the bank has more bargaining
power, it negotiates an interest rate r closer to zero so that its profit,
20−r, is relatively large. If the FHLB has more bargaining power, it
negotiates an interest rate closer to 20 so that it earns more profit
on its overnight loan.

Banks could
earn the ior
rate of 25 bp.

0 bp

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

1 fhlbs can loan
cash holdings to a
bank, and the two
can split the gains.

fhlb has more
bargaining power.

Outside option
for banks

Gains from trade

Between October 2011 and September 2013, an FHLB could earn a zero
net return on any cash it held at the end of the day. However, it could
lend that money to a bank eligible to earn the IOR rate, 25 basis
points, less any costs associated with expanding its balance sheet.
Suppose these costs were 5 basis points, so there were “gains from
trade” between the FHLB and the bank of 25−5=20 basis points.
This means the two parties would agree to trade at any interest rate
between 0 and 20 basis points.

profits for
the bank

More profits
for the fhlb

Outside option
for fhlbs

Federal Reserve Bank of Philadelphia
Research Department


Since the FOMC began raising the
target rate in December 2015, it has used
these two instruments—the IOR and
ON RRP rates—to raise and control the fed
funds rate in a market characterized by
ample excess reserves. In particular, as
Armenter and Lester (2017) describe, the
FOMC has raised rates by increasing both
the ON RRP and IOR rates at the same
time, while it has adjusted where the fed
funds rate falls within the target range by
adjusting the IOR rate.
The top panel of Figure 4 illustrates the
relationship between the ON RRP rate,
the IOR rate, and the fed funds rate between December 2015 and September 2018.
The bottom panel of Figure 4 plots the

spread between the IOR and ON RRP rates
between June 2017 and September 2018,
and it plots where the EFFR rate falls within
this spread (the red line).
From the time it “lifted off” from zero
until 2018, the FOMC raised the IOR and
ON RRP rates in tandem, with a 25 basis
point spread between the two. The EFFR
followed suit, staying safely within the target range until the second quarter of 2018.
At that time, however, the outside option
of ineligible financial institutions began
rising, putting upward pressure on the
EFFR. In response, when the FOMC raised
the target range in June 2018, it
increased the ON RRP rate by 25 basis
points but the IOR rate by only 20 basis


How the Fed Changes Rates Post-Great Recession
The Fed uses the IOR and ON RRP rates to adjust the EFFR.



on rrp

See below for the effr within
the ior–on rrp spread


1 Dec 2015


3 Sep 2018

Vertical lines represent
dates when the fomc
changed the target rate


effr within
the ior–on
rrp spread


ior–on rrp


1 Jun 2017

points. Decreasing the spread between the
IOR and ON RRP rates puts downward
pressure on the fed funds rate, helping to
keep it within the target range.


In the summer of 2017 the FOMC announced its intention to stop reinvesting
the proceeds from maturing assets (such
as mortgage-backed securities) on its
balance sheet. This decision marked the
beginning of the Fed unwinding or “normalizing” its balance sheet. As the Fed’s
balance sheet shrinks, excess reserves
in the banking sector decline. However,
at the time, the FOMC did not provide an
explicit endpoint for this process.10
More recently, in January 2019 the
FOMC announced how it planned to hold
“no more securities than necessary to
implement monetary policy efficiently and
effectively”: by using a “regime in which
an ample supply of reserves ensures that
control over the level of the federal funds
rate and other short-term interest rates is
exercised primarily through the setting
of the Federal Reserve’s administered rates,
and in which active management of the
supply of reserves is not required.”11 In
other words, the FOMC decided to shrink
the balance sheet until reaching the
minimal size still consistent with “ample”
excess reserves, and to use the ON RRP
and IOR rates to achieve the target fed
funds rate.
This decline in aggregate excess
reserves changes the individual behavior
of market participants, and this in turn
affects overall market conditions in the fed
funds market, including interest rates and
trading volume. In particular, if total excess reserves decline enough, the market
will transition from the ample-reserve
regime—in which open market operations
have little effect—to the precrisis scarcereserve regime. However, it is difficult to
forecast when this transition will occur
because it depends not only on the level
of excess reserves in the market but also
on the distribution of these reserves
across banks, which is hard to predict.

3 Sep 2018

Source: FRED, Federal Reserve Bank of St. Louis.


Federal Reserve Bank of Philadelphia
Research Department

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

Who Trades with Whom, and at What Price?

In the fed funds market, a bank can try to find a counterparty to
borrow from (either an ineligible financial institution, like an
FHLB, or another bank), it can try to find a counterparty to lend
to (another bank), or it can remain idle. When all banks are
awash with reserves, there is no motive to lend, since nobody in
the market is willing to pay more than the IOR rate. Hence, when
reserves are ample, banks with sufficiently low balance-sheet
costs (such as banks not subject to FDIC fees) will borrow from
institutions such as FHLBs at a rate between the ON RRP rate and
the IOR rate, and the remainder of banks (with higher costs from
expanding their balance sheets) will remain idle.
However, as total reserves decline, some banks will find themselves close to their reserve requirement. To avoid coming up
short of required reserves—and being forced to borrow at the
discount window, where rates are typically 50 basis points higher
than the IOR rate—these “desperate” banks will look to borrow
from either an FHLB or another bank. If there are only a few
desperate banks looking to borrow, they can likely satisfy their
reserve requirements by borrowing from FHLBs at a rate below
the IOR rate. But as total reserves decline further, there will be
more and more desperate banks looking to borrow.
When this occurs, banks that are far from their reserve
requirements will face a choice. These “nondesperate” banks can
continue looking to borrow from an FHLB at a rate below the IOR
rate, pocketing the difference (less any balance-sheet costs),
or they can try to lend to desperate banks at a rate above the IOR
rate. As the Fed’s balance sheet shrinks and reserves become
increasingly scarce, the demand for reserves from desperate
banks will grow, the supply of reserves from nondesperate banks
will shrink, and lending to desperate banks will become more
attractive. At some point, nondesperate banks will once again
find themselves lending in the fed funds market, and they will do
so at rates above the IOR rate.
This shift in the behavior of individual market participants has
several important implications for the fed funds market as
a whole. First, the fed funds rate, which is an average of all rates
in the fed funds market, will no longer reside within the corridor
formed by the ON RRP and IOR rates. It will instead lie within the
corridor formed by the IOR and discount-window rates. Second,
as bank-to-bank lending resumes alongside trades between
FHLBs and banks, trading volume should also increase. Lastly,
since the market rate will be determined by supply and demand
once again, the fed funds rate will be sensitive to relatively small
changes in the supply of reserves.

When Are Reserves No Longer ‘Ample’?

How much must total reserves shrink before we see these
changes? Because the logic above suggests that the fed funds rate
should move from one corridor to another when enough banks
find themselves with scarce reserves, it is not sufficient to know
the total level of reserves. In addition, we need to know the
distribution of those reserves across banks! To see why, consider
what would happen if the total amount of excess reserves
declined by $100 billion and the entirety of this decline came off
the balance sheets of banks already close to their reserve requirements. This would immediately force a number of banks to enter
the fed funds market as borrowers, prompting other banks to
lend above the IOR, thus raising rates. However, if this decline in
reserves came off the balance sheets of banks far from their
reserve requirements, it would have little effect; all banks would
continue to borrow from FHLBs at rates below the IOR.
Hence, to forecast the level of reserves at which the market
transitions from ample to scarce reserves, we need to predict the
distribution of reserves across banks as the Fed’s balance sheet
shrinks. Several factors determine this distribution, including each
bank’s size and the regulatory costs they face. In a recent paper
with Afonso and Armenter, we estimate the total quantity of
reserves consistent with the fed funds rate returning to a corridor
between the IOR and discount-window rates. Our benchmark
model suggests an answer of approximately $900 billion. However,
we find that our estimates are quite sensitive to what we assume
about the evolution of the distribution of reserves. In particular,
assuming that the majority of the decline in aggregate reserves is
absorbed by the smallest or largest banks, respectively, produces
estimates as large as $1.1 trillion and as small as $500 billion.


In response to the financial crisis, the Federal Reserve introduced
new programs and policies to stabilize markets, restore liquidity,
and spur economic activity. However, a byproduct of these
changes was that the fed funds market was dramatically altered,
necessitating a new framework for monetary policy implementation. More recently, as the Fed began to unwind some of these
programs, it was forced to reassess the long-run size of its balance
sheet—and the tools it intended to use for monetary policy implementation—given the current economic and regulatory
environment. It has chosen to maintain a balance sheet that is
sufficiently large to support a market with ample reserves, and to
use the administered (IOR and ON RRP) rates to achieve the target
range. A lingering challenge is identifying the minimum balancesheet size consistent with these goals, as this requires forecasting
the evolution of the distribution of reserves across banks.

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department




1 Although not all banks are depository institutions, and not all depository
institutions are banks, we will use “bank” to refer to depository institutions trading in the fed funds market, including bank holding companies,
standalone commercial banks, and thrifts. However, institutions other
than banks also trade in the federal funds market. Under current regulation, once deposits exceed a minimal threshold, these banks are required
to hold at least 10 percent of any additional deposits as reserves at the Fed.

Afonso, Gara, Roc Armenter, and Benjamin R. Lester. “A Model of the
Federal Funds Market: Yesterday, Today, and Tomorrow,” Federal Reserve
Bank of Philadelphia Working Paper 18-10 (2018).

2 Banks would try to avoid borrowing at the discount window because
the rate was higher than the typical rate being offered in the fed funds
market, and because there was a stigma associated with borrowing at
the discount window. See Ennis and Weinberg (2013).
3 For a more detailed description of open market operations, see
4 For more details on quantitative easing, see Yu (2018).
5 A basis point equals one hundredth of 1 percent. McCauley and McGuire
(2014) estimate a cost of 4 basis points, while Banegas and Tase (2016)
find a cost of 7 basis points.
6 This policy change was made possible when Congress passed the
Financial Services Regulatory Relief Act in 2006, clearing the way for the
Federal Reserve to start paying interest on reserves to eligible depository
institutions effective October 1, 2011. This date was later moved up to
October 1, 2008, as a result of the Emergency Economic Stabilization Act
of 2008.
7 The Federal Home Loan Banks provide funds to depository institutions
in the form of loans collateralized by real estate. They were initially set
up to provide liquidity to savings and loans but are now a source of funds
for all banks.
8 If the Fed tried to conduct policy on precrisis terms, it would have had
to execute very large open market operations to drain reserves in relatively
short order. Selling large quantities of certain assets in a very short
period would have negative side effects, as prices in these markets would
likely experience sudden declines.

Afonso, Gara, Alex Entz, and Eric LeSueur. “Who’s Lending in the Fed
Funds Market?” Federal Reserve Bank of New York Liberty Street
Economics Blog (December 2, 2013).
Afonso, Gara, and Ricardo Lagos. “Trade Dynamics in the Market for
Federal Funds,” Econometrica, 83:1 (January 2015), pp. 263–313.
Armenter, Roc, and Benjamin Lester. “Excess Reserves and Monetary
Policy Implementation,” Review of Economic Dynamics, 23 (2017), pp.
Banegas, Ayelen, and Manjola Tase. “Reserve Balances, the Federal
Funds Market and Arbitrage in the New Regulatory Framework,” Board
of Governors of the Federal Reserve System Finance and Economics
Discussion Series 2016-079 (September 2016).
Ennis, Huberto M., and John A. Weinberg. “Over-the-Counter Loans,
Adverse Selection, and Stigma in the Interbank Market.” Review of
Economic Dynamics 16:4 (2013): pp. 601–616.
McCauley, Robert, and Patrick McGuire. “Non-U.S. Banks’ Claims on the
Federal Reserve,” BIS Quarterly Review (March 2014), pp. 89–97. https://
Yu, Edison. “Did Quantitative Easing Work?” Federal Reserve Bank of
Philadelphia Economic Insights (First Quarter 2016), pp. 5–13. https://

9 For more information about eligible counterparties at the ON RRP facility,
10 In its June 14, 2017, statement, the FOMC announced only that “the
Federal Reserve’s securities holdings will continue to decline in a gradual
and predictable manner until the Committee judges that the Federal
Reserve is holding no more securities than necessary to implement monetary policy efficiently and effectively.”
11 See


Federal Reserve Bank of Philadelphia
Research Department

Implementing Monetary Policy in a Changing Federal Funds Market
2019 Q2

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

Beautiful City: Leisure Amenities and Urban Growth
Modern urban economic theory and policymakers are coming to see
the provision of consumer-leisure amenities as a way to attract
population, especially the highly skilled and their employers. However,
past studies have arguably only provided indirect evidence of the
importance of leisure amenities for urban development. In this paper,
we propose and validate the number of tourist trips and the number of
crowdsourced picturesque locations as measures of consumer
revealed preferences for local lifestyle amenities. Urban population
growth in the 1990–2010 period was about 10 percentage points
(about one standard deviation) higher in a metro area that was perceived as twice more picturesque. This measure ties with low taxes as
the most important predictor of urban population growth. “Beautiful
cities” disproportionally attracted highly educated individuals and
experienced faster housing price appreciation, especially in supplyinelastic markets. In contrast to the generally declining trend of the
American central city, neighborhoods that were close to central
recreational districts have experienced economic growth, albeit at the
cost of minority displacement.
Supersedes Working Paper 08-22.
Working Paper 19-16. Gerald A. Carlino, Federal Reserve Bank of
Philadelphia; Albert Saiz, Massachusetts Institute of Technology.

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

Building Credit History with Heterogeneously
Informed Lenders
This paper examines a novel mechanism of credit-history building as
a way of aggregating information across multiple lenders. We build a
dynamic model with multiple competing lenders, who have heterogeneous private information about a consumer’s creditworthiness,
and extend credit over multiple stages. Acquiring a loan at an early
stage serves as a positive signal—it allows the borrower to convey to
other lenders the existence of a positively informed lender (advancing
that early loan)—thereby convincing other lenders to extend further
credit in future stages. This signaling may be costly to the least risky
borrowers for two reasons. First, taking on an early loan may involve
cross-subsidization from the least risky borrowers to more risky
borrowers. Second, the least risky borrowers may take inefficiently
large loans relative to the symmetric-information benchmark. We
demonstrate that, despite these two possible costs, the least risky
borrowers often prefer these equilibria to those without information
aggregation. Our analysis offers an interesting and novel insight
into debt dilution. Contrary to the conventional wisdom, repayment
of the early loan is more likely when a borrower subsequently takes
on a larger rather than a smaller additional loan. This result hinges on
a selection effect: Larger subsequent loans are only given to the least
risky borrowers.
Working Paper 19-17. Natalia Kovrijnykh, Arizona State University;
Igor Livshits, Federal Reserve Bank of Philadelphia; Ariel Zetlin-Jones,
Carnegie Mellon University.

Research Update
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department


Capitalization as a Two-Part Tariff: The Role of

The Firm Size and Leverage Relationship and Its
Implications for Entry and Concentration in a Low
Interest Rate World
Larger firms (by sales or employment) have higher leverage. This
pattern is explained using a model in which firms produce multiple
varieties and borrow with the option to default against their future
cash flow. A variety can die with a constant probability, implying that
bigger firms (those with more varieties) have lower coefficient of
variation of sales and higher leverage. A lower risk-free rate benefits
bigger firms more as they are able to lever more and existing firms
buy more of the new varieties arriving into the economy. This leads to
lower startup rates and greater concentration of sales.
Working Paper 19-18. Satyajit Chatterjee, Federal Reserve Bank of
Philadelphia; Burcu Eyigungor, Federal Reserve Bank of Philadelphia.

Mortgage Loss Severities: What Keeps Them So
Mortgage loss-given-default (LGD) increased significantly when house
prices plummeted and delinquencies rose during the financial crisis,
but it has remained over 40 percent in recent years despite a strong
housing recovery. Our results indicate that the sustained high LGDs
postcrisis are due to a combination of an overhang of crisis-era foreclosures and prolonged foreclosure timelines, which have offset higher
sales recoveries. Simulations show that cutting foreclosure timelines
by one year would cause LGD to decrease by 5–8 percentage points,
depending on the trade-off between lower liquidation expenses and
lower sales recoveries. Using difference-in-differences tests, we also
find that recent consumer protection programs have extended foreclosure timelines and increased loss severities in spite of their benefits
of increasing loan modifications and enhancing consumer protections.
Supersedes Working Paper 17-08.
Working Paper 19-19. Xudong An, Federal Reserve Bank of Philadelphia;
Larry Cordell, Federal Reserve Bank of Philadelphia.

This paper shows that the capitalization of local amenities is effectively
priced into land via a two-part pricing formula: a “ticket” price paid
regardless of the amount of housing service consumed and a “slope”
price paid per unit of services. We first show theoretically how tickets
arise as an extensive margin price when there are binding constraints
on the number of households admitted to a neighborhood. We use
a large national dataset of housing transactions, property characteristics, and neighborhood attributes to measure the extent to which
local amenities are capitalized in ticket prices vis-à-vis slopes. We find
that in most U.S. cities, the majority of neighborhood variation in
pricing occurs via tickets, although the importance of tickets rises
sharply in the stringency of land development regulations, as predicted
by theory. We discuss implications of two-part pricing for effciency and
equity in neighborhood sorting equilibria and for empirical estimates
of willingness to pay for nonmarketed amenities, which generally
assume proportional pricing only.
Working Paper 19-20. H. Spencer Banzhaf, Georgia State University;
Kyle Mangum, Federal Reserve Bank of Philadelphia.

Demographic Aging, Industrial Policy, and Chinese
Economic Growth
We examine the role of demographics and changing industrial policies
in accounting for the rapid rise in household savings and in per capita
output growth in China since the mid-1970s. The demographic
changes come from reductions in the fertility rate and increases in the
life expectancy, while the industrial policies take many forms. These
policies cause important structural changes; first benefiting private
labor-intensive firms by incentivizing them to increase their share of
employment, and later on benefiting capital-intensive firms resulting
in an increasing share of capital devoted to heavy industries. We
conduct our analysis in a general equilibrium economy that also
features endogenous human capital investment. We calibrate the
model to match key economic variables of the Chinese economy and
show that demographic changes and industrial policies both contributed to increases in savings and output growth but with differing
intensities and at different horizons. We further demonstrate the
importance of endogenous human capital investment in accounting
for the economic growth in China.
Working Paper 19-21. Michael Dotsey, Federal Reserve Bank of
Philadelphia; Wenli Li, Federal Reserve Bank of Philadelphia; Fang
Yang, Louisiana State University.


Federal Reserve Bank of Philadelphia
Research Department

Research Update
2019 Q2

Commuting, Labor, and Housing Market Effects of
Mass Transportation: Welfare and Identification

Consumer Lending Efficiency: Commercial Banks
Versus a Fintech Lender

Using a panel of tract-level bilateral commuting flows, I estimate the
causal effect of Los Angeles Metro Rail on commuting between
connected locations. Unique data, in conjunction with a spatial general
equilibrium model, isolate commuting benefits from other channels.
A novel strategy interacts local innovations with intraurban geography
to identify all model parameters (local housing and labor elasticities).
Metro Rail connections increase commuting between locations
containing (adjacent to) stations by 15 percent (10 percent), relative to
control routes selected using proposed and historical rail networks.
Other margins are not affected. Elasticity estimates suggest relatively
inelastic mobility and housing supply. Metro Rail increases welfare
$146 million annually by 2000, less than both operational subsidies
and the annual cost of capital. More recent data show some additional
commuting growth.

We compare the performance of unsecured personal installment
loans made by traditional bank lenders with that of LendingClub,
using a stochastic frontier estimation technique to decompose the
observed nonperforming loans into three components. The first is
the best-practice minimum ratio that a lender could achieve if it were
fully efficient at credit-risk evaluation and loan management. The
second is a ratio that reflects the difference between the observed
ratio (adjusted for noise) and the minimum ratio that gauges the
lender’s relative proficiency at credit analysis and loan monitoring.
The third is statistical noise. In 2013 and 2016, the largest bank
lenders experienced the highest ratio of nonperformance, the highest
inherent credit risk, and the highest lending efficiency, indicating that
their high ratio of nonperformance is driven by inherent credit risk,
rather than by lending inefficiency. LendingClub’s performance was
similar to small bank lenders as of 2013. As of 2016, LendingClub’s
performance resembled the largest bank lenders—the highest ratio of
nonperforming loans, inherent credit risk, and lending efficiency—although its loan volume was smaller. Our findings are consistent with
a previous study that suggests LendingClub became more effective in
risk identification and pricing starting in 2015. Caveat: We note that
this conclusion may not be applicable to fintech lenders in general,
and the results may not hold under different economic conditions
such as a downturn.

Working Paper 18-14 Revised. Christopher Severen, Federal Reserve
Bank of Philadelphia.

Elasticities of Labor Supply and Labor Force
Participation Flows
Using a representative-household search and matching model with
endogenous labor force participation, we study the interactions
between extensive-margin labor supply elasticities and the cyclicality
of labor force participation flows. Our model successfully replicates
salient business-cycle features of all transition rates between three
labor market states, the unemployment rate, and the labor force
participation rate, while using values of elasticities consistent with
micro evidence. Our results underscore the importance of the procyclical opportunity cost of employment, together with wage rigidity,
in understanding the cyclicality of labor market flows and stocks.

Working Paper 19-22. Joseph P. Hughes, Rutgers University; Julapa
Jagtiani, Federal Reserve Bank of Philadelphia; Choon-Geol Moon,
Hanyang University.

Working Paper 19-03 Revised. Isabel Cairó, Board of Governors of
the Federal Reserve System; Shigeru Fujita, Federal Reserve Bank
of Philadelphia; Camilo Morales-Jiménez, Board of Governors of the
Federal Reserve System.

Research Update
2019 Q2

Federal Reserve Bank of Philadelphia
Research Department


A Generalized Factor Model with Local Factors
I extend the theory on factor models by incorporating local factors into
the model. Local factors only affect an unknown subset of the observed
variables. This implies a continuum of eigenvalues of the covariance
matrix, as is commonly observed in applications. I derive which
factors are pervasive enough to be economically important and which
factors are pervasive enough to be estimable using the common
principal component estimator. I then introduce a new class of
estimators to determine the number of those relevant factors. Unlike
existing estimators, my estimators use not only the eigenvalues of
the covariance matrix, but also its eigenvectors. I find strong evidence
of local factors in a large panel of U.S. macroeconomic indicators.
Working Paper 19-23. Simon Freyaldenhoven, Federal Reserve Bank
of Philadelphia.

Institution, Major, and Firm-Specific Premia:
Evidence from Administrative Data
We examine how a student’s major and the institution attended
contribute to the labor market outcomes of young graduates. Administrative panel data that combine student transcripts with matched
employer-employee records allow us to provide the first decomposition
of premia into individual and firm-specific components. We find that
both major and institutional premia are more strongly related to
the firm-specific component of wages than the individual-specific
component of wages. On average, a student’s major is a more
important predictor of future wages than the selectivity of the
institution attended, but major premia (and their relative ranking) can
differ substantially across institutions, suggesting the importance of
program-level data for prospective students and their parents.
Working Paper 19-24. Ben Ost, University of Illinois–Chicago; Weixiang
Pan, Georgia State University; Douglas Webber, Temple University.


Federal Reserve Bank of Philadelphia
Research Department

Research Update
2019 Q2


Is the Phillips Curve Dead?
Collateral Damage: House
Prices and Consumption
During the Great Recession
Banking Trends:

Have Foreign Banks
Changed How They
Connect with Us
Read previous articles
Economic Insights

Business Review

Sign up for e-mail notifications


Federal Reserve Bank of Philadelphia


Ten Independence Mall
Philadelphia, Pennsylvania 19106-1574

You can find Economic
Insights via the
Research Publications
part of our website.