Full text of Economic Insights (Federal Reserve Bank of Philadelphia) : Fourth Quarter 2016, Volume 1, Issue 4

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

Fourth Quarter 2016
Volume 1, Issue 4

Just How Important Are New Businesses?
Regional Spotlight
Banking Policy Review
Research Update

INSIDE
ISSN: 0007-7011

FOURTH QUARTER 2016

Economic Insights is published four times

Just How Important Are New Businesses?

1

a year by the Research Department of the
Federal Reserve Bank of Philadelphia. The views
expressed by the authors are not necessarily
those of the Federal Reserve. We welcome your
comments at PHIL BRComments@phil.frb.org.
To receive e-mail notifications of the latest

Economic Insights and other Philadelphia
Fed publications, go to www.philadelphiafed.

New businesses are major job generators, so disappointing trends in firm
formation have raised concern. Thorsten Drautzburg discusses why at least
some of the worry might be misplaced.

Regional Spotlight:
The State of the States

8

org/notifications. Archived articles may be
downloaded at www.philadelphiafed.org/
research-and-data/publications/economicinsights.

The U.S. economy has been expanding for seven years — but don’t tell that to a
handful of states that have suffered recessions recently. Paul R. Flora discusses
how Philadelphia Fed indexes may aid in the tricky business of identifying
recession patterns among the 50 states.

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

Banking Policy Review:
Did Dodd–Frank End ‘Too Big to Fail’?

16

Postcrisis bank reform was intended to end market perceptions that if a big bank
fails, the government will have no choice but to bail it out. Ryan Johnston
examines the evidence from recent studies.

Reserve System. The Philadelphia Fed
serves eastern Pennsylvania, southern New
Jersey, and Delaware.

Research Update

21

Patrick T. Harker

Abstracts of the latest working papers produced by the Federal Reserve Bank of
Philadelphia.

President and Chief Executive Officer
Michael Dotsey
Senior Vice President
and Director of Research
Colleen Gallagher
Research Publications Manager
Dianne Hallowell
Art Director and Manager
On the cover: The art of foiling counterfeiters can be seen by tilting a new $100 bill to reveal
the Liberty Bell inside Ben Franklin’s inkwell. Learn more about currency security features at
https://uscurrency.gov/. Photo by Peter Samulis.

Just How Important Are New Businesses?

New firms are the job engines of the economy, but firm formation has diminished. Should we worry?
BY THORSTEN DRAUTZBURG

New businesses create most of the new jobs in the
U.S. economy each year — not small businesses, as popular wisdom holds. It may thus seem troubling that business
formation has not kept up with overall growth in the U.S.
economy over the last 35 years. And while counting jobs is
just one way to quantify the success of new businesses, their
relative decline matters not only for their owners and employees. That’s because even though many new businesses
fail, some survivors are innovators and grow rapidly, raising
wage growth and productivity across the economy.
But we should be careful not to read too much into
the drop in the headline numbers. The economic theory of
creative destruction suggests that the success of new businesses comes at a cost to existing businesses.1 Also, as I will
show, Americans seem as entrepreneurial today as they were
20 years ago. Much of the fluctuation in the success of new
businesses may actually have been driven by economywide
forces such as demographics or technological opportunities,
and not necessarily vice versa. So, even though it would be
good to reverse the relative trend decline in business formation, it might not be as consequential as some believe.
What do I mean by “new” businesses? And why do they
matter disproportionately for employment? Here I follow
the Census Bureau’s definition and define a new business’s
first, or birth, year to be the year it paid payroll taxes for an
employee for the first time.2 New businesses punch above
their weight in terms of job creation. As Figure 1 illustrates,
if new firms were to disappear and all else equal, employment in the U.S. would have fallen in every five-year period
since 1977. That’s because if the number of jobs created
each year is calculated as a share of all jobs in the economy,
the share created by new firms exceeds the share created by
the U.S. economy as a whole — partly reflecting the fact

FIGURE 1

Share of Jobs at New Businesses Declining
Private nonfarm jobs created by new firms versus by all firms
each year as shares of total jobs.

Sources: Census Bureau and author’s calculations.

that each year many once-new businesses fail and destroy
jobs.3 Startup firms created an average of 3.6 million jobs per
year between 1978 and 2013, but because aging startups and
older firms shed jobs, only 2.1 million jobs a year on average
were created in the economy as a whole during that period.4
Even so, as Figure 1 also makes clear, the share of jobs created by startup firms has been falling since the mid-1980s,
and the decline relative to the whole economy accelerated
again during the Great Recession of 2007–2009.
Before going into details, it is worth emphasizing that
the decline is relative to the
growing U.S. economy. Between
Thorsten Drautzburg is
March 1982 and March 2007,
an economist at the Federal
Reserve Bank of Philadelphia.
just before the Great RecesThe views expressed in this
sion, employment at firms up to
article are not necessarily
those of the Federal Reserve.
three years of age had increased

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 1

17 percent, and the number of firms up to age three had
increased 14 percent. But overall employment had increased
61 percent, and the overall number of firms had increased
47 percent. Startups have failed to keep up.5
Until fairly recently, the role of young businesses in job
creation had gone largely unnoticed, with much emphasis
being placed instead on small businesses.6 In 2013, John
Haltiwanger and his coauthors pioneered the recent wave
of U.S. research on the subject by compiling a data set that
also tabulates the universe of U.S. firms by age. The researchers used the underlying firm-level data to argue “that
once we control for firm age there is no systematic relationship between firm size and growth.” This crucial role of
newly formed businesses is consistent with data from other
countries, such as Germany. Back in 1992, Tito Boeri and
Ulrich Cramer had concluded that the opening of new businesses “is the driving force of trend employment growth.”
Why had previous research focused on small rather
than young firms? Since young firms tend to be small, it
looks as if small firms per se are adding the most jobs unless
one accounts for how long the firms have been in business.7
Crucially, smaller firms do not grow any faster than larger
firms of the same age. But new firms that survive their first
year do grow faster than more established firms do. The
average one-year-old firm increases its workforce by about
15 percent a year. Upon reaching five years of age, firms on
average are adding about 3 percent more workers to their
payrolls, while firms that have been around for more than
10 years are typically growing about 2 percent a year.
The high average growth rates for new businesses

since the late 1970s mask the significant slowdown in new
firm activity that has taken place. Figure 2 illustrates this
slowdown by comparing the contribution to overall employment that new firms made in 1982 versus 2007 — both their
initial share of total jobs in the economy and the growth of
that share over the ensuing five years. Both the initial contribution and the growth were markedly lower in 2007 than
in 1982. Firms that were started in 1982 employed 4.1 percent of private nonfarm workers and increased that share by
an average of 3.2 percentage points over the next five years.
The 2007 cohort, in contrast, initially employed only 2.6
percent of workers and increased that already-smaller share
at the slower rate of 1.8 percentage points per year.
This slowdown has not been limited to the two years
I illustrate here — Figure 3 provides the comprehensive
picture and shows that the two cohorts displayed in Figure 2
are representative of the trend since the early 1980s.
Despite their diminishing contribution, new firms
remain important employers in the U.S. For example,
Figure 3 shows that in 1982, one out of five U.S. workers was
employed at a firm that belonged to the 1977 cohort — and
that was, therefore, up to six years old. By 2012, the ratio
for the corresponding 2007 cohort had fallen to one out of
11, where it stayed in 2013.8 In 2002, about 50 percent of
employees worked at companies that had been started 25
years earlier. In 2013, that number had fallen to 39 percent.9
This smaller role of new businesses is due both to the lower
starting shares evident in Figure 2 (visible as increasingly
lower starting points in Figure 3) and slower growth (visible
in the ever-flatter slopes in Figure 3).

FIGURE 2

FIGURE 3

New Firms Used to Have Larger Share of Jobs

New Firms’ Share of Jobs Is Shrinking

Starting share of employment and average growth during
first five years by starting year.

Cumulative share of private nonfarm employment of firms
by starting year, 1977–2013.

Sources: Census Bureau and author’s calculations.

Sources: Census Bureau and author’s calculations.
Note: The data on firm ages top out at 25 years.

2 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

What are the economic implications of this decline in
young companies’ share of total employment? By one estimate, if the U.S. economy had maintained the startup dynamics that had prevailed in the late 1970s and early 1980s,
and if established businesses had still been able to create the
same number of jobs as they did without the added competition, the U.S. would have 15 million to 20 million more
private sector jobs today.10 Amid this trend decline in new

Some of the most prominent new businesses
of the past few decades have become
transformative technology companies such
as Amazon, Google, and Facebook.
firms’ share of employment, the Great Recession accelerated
the decline in firm formation: Thirty percent fewer businesses were created in the recession compared with the previous
peak.11 A decline of this magnitude is unprecedented in the
data, which start in 1977. Worse, according to one study,
those businesses that were created during the recession
were, on average, smaller — and we should expect them to
remain smaller throughout their existence.12
WHY CARE?

While these developments seem disconcerting, they do
not tell us if we should care more about the fate of young
firms than about established ones. After all, what difference
does it make whether a job is created by an established business or a new one? Yet, clearly, startups have gone on to play
an outsize role in today’s economy — not only in terms of
job counts. Some of the most prominent new businesses of
the past few decades have become transformative technology companies such as Amazon, Google, and Facebook.
These companies have gone on to create tens of thousands
of mostly well-paying jobs and have certainly contributed to
a more productive economy.
But it is hard to move beyond anecdotes to establish
whether new businesses in general increase productivity
and employment more than other expanding businesses do.
Looking only at the stars among new businesses is misleading because of survivor bias: Naturally, the top startups were
the successful ones. So we have to look at the job-generating
effects of all the businesses formed within a given period.

But even once we turn to young businesses as a whole, it
becomes hard to tell whether, say, their productivity pushed
overall productivity higher or whether they were pulled
along by a general rise in productivity. And the more important new businesses are for the economy, the more difficult
it is to quantify those benefits because of feedback effects
— whether a productivity boom originated among the new
businesses or was simply adapted by them.
So, to isolate the effects of new businesses, researchers
have to find ways to construct a comparison with a counterfactual model of an otherwise identical economy with fewer
or no new businesses. Consider new businesses in France, as
a starting point. In the French data, new firms tend to have
a productivity rate about 15 percent higher than that of older
firms that are shrinking.13 However, this might be because
new firms use better technology that incumbents could also
invest in. Interpreting the observed higher productivity is,
therefore, hard. One creative study compared U.S. counties
where large factories, called “million dollar plants” in the
study, had chosen to locate with the runner-up counties.14
The new plants made other businesses in the county 3 to
5 percent more productive. But no such increase occurred
among businesses in the runner-up counties. New plants,
like new firms, have access to the latest technologies or can
introduce new product varieties. This difference in the counties’ productivity thus supports the notion that new businesses are both more productive themselves and, unlike older
businesses, make other local businesses more productive.
LIMITS TO THE ROLE OF NEW BUSINESSES

Despite the benefits that new businesses bring, the
headline numbers for employment or productivity may overstate their economic impact for two reasons: First, increases
or decreases in the importance of new businesses might just
reflect other forces at work in the economy. Second, what is
good for new businesses may be bad for old businesses.
One concern is that fluctuations or trends in the number and size of new firms might just be transmitting fluctuations originating elsewhere in the economy. If that were
the case, any remedies would also likely have to address the
underlying cause, and not firm creation, which would merely
be a symptom. For example, one study suggests that supply
shocks from demographic changes largely explain the trend
decline of new businesses.15 Another suggests that changes
in monetary policy barely affect financing conditions for
large firms but have a big impact on the ability of small
firms (which, as we saw, are more likely to be young) to get

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 3

loans — often critical for starting a business and keeping a
young firm going. 16 In these cases, policymakers might want
to address demographics through immigration reform or
credit supply through targeted loan programs.
Are Americans becoming less entrepreneurial and
simply less inclined to start businesses? Even though new
businesses as we have defined them — having an employer
plus at least one employee — have diminished, the same
cannot be said of self-employment in general. Working for
oneself apparently has not declined. On average, around
0.3 percent of Americans reported becoming self-employed
as their primary occupation from 1996 to 2014 (Figure 4).
While the fraction of the newly self-employed fluctuates, it
does so within a fairly narrow range, in contrast with the
trend decline we have seen in the number of new employers.
Figure 4 also shows that a stable fraction of Americans give
up a job to start a business, suggesting that entrepreneurship
is a choice and not due to a lack of jobs.
New technologies also affect the creation and growth of

new businesses. An analysis of different technological eras
from the 1870s to the 1990s that examined the leading new
firms in different sectors found that new firms rose to importance faster during the electrification era in the late 19th
century and in the information technology era of the second
half of the 20th century than during the chemical-pharmaceutical era in the middle of the 20th century.17
The stock market provides a way to quantify the role
of new firms over time. During eras when new firms rose
rapidly, they quickly commanded a large share of the total
stock market valuation. By this metric, today’s startup slump
no longer appears unprecedented. In both the 1890s and
1990s, new firms’ stock market valuation and the growth of
their share were both relatively high — only to be followed
by slowdowns.19 Yet, the slowdown in the mid-20th century
was subsequently reversed with the commercial success of
computers (Figure 5).
FIGURE 4

Americans as Entrepreneurial as in Late 1990s
Share of U.S. adults age 20–64 switching their main occupation
to self-employed, by year.

Tech Startups in History: Not All Gazelles
General Electric’s founding in 1878 represents the start of
the electrification era. It had its breakthrough innovation in
1880, grew rapidly during the electrification era of the late
1800s, incorporated, and went public in 1892. American
Telephone & Telegraph was founded in 1885, had its
breakthrough innovation in 1892, incorporated in 1895,
and had its initial public listing in 1901.
In contrast, major chemical and pharmaceutical companies
were founded in the same era as GE and AT&T but had
their breakthrough innovations and went public at much
later ages. It took Pfizer 51 years to incorporate, in 1900,
and almost 100 years until it achieved its breakthrough
innovation in 1944. Merck progressed a little faster
but still took 43 years to incorporate and 53 years to
reach its breakthrough innovation, also in the chemicalpharmaceutical era of the mid-20th century. These
companies went public in 1944 and 1946, respectively.

Sources: Kauffman Foundation and author’s calculations.

FIGURE 5

Tech Waves? New Public Firms Slow After Roaring ’90s…
Starting share and average growth of stock market capitalization
by firm starting year, 1890–2015.

The information technology era has been characterized
by an even faster rise to prominence by major firms
than during the electrification era. The advent of
computerization is represented by the rapid incorporation,
breakthrough innovation, and initial public offering of
Intel — all within four years of its founding in 1968.
Microsoft reached the same milestones within 11 years of
its incorporation in 1975.18
Sources: Jovanovic and Rousseau, 2002; Compustat*; and author’s calculations.

4 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

rate of startup creation or startup employment shares
per se. For example, competition in the labor market
…But History Shows Big Swings in Value
from new businesses drives up wages so that more or
Cumulative share of U.S. stock market capitalization by firm starting year,
bigger new businesses might lead to fewer or smaller
1890–2000.
existing businesses. Standard economic models20 and
recent empirical estimates suggest that this effect is
sizeable. By one estimate, the crowding-out effect of
increased competition can destroy jobs at established
firms equal to anywhere from one-third to 90 percent
of the jobs created by new firms.21
However, even if the crowding-out were complete
and employment at new businesses came completely at
the cost of old businesses, this reallocation of workers
might still be beneficial for the economy. New firms
are able to crowd out old firms only because they are
more productive. This higher productivity may raise
Source: Jovanovic and Rousseau, 2002.
wages more than employment — my model implies
precisely that the stronger the crowding out, the faster
the wage growth. In the French study mentioned earliFigure 6 shows in more detail how the contribution of
er, even a complete reallocation from old to young businesses
new firms to the U.S. stock market has fluctuated over the
was estimated to raise wages about 10 percent.
course of 110 years. Firms that got started before 1930 grew
rapidly in market value, with cohorts achieving 20 percent
SHOULD WE BE WORRIED?
market capitalization shares within 10 years, reflecting the
rapid growth of firms during the electrification era. During
The pace at which businesses are started matters — but
the pharmaceutical and chemical era of the 1930s, 1940s,
less so than their impressive job creation numbers would
and 1950s, the share of young firms declined markedly. Yet,
suggest. The reason is that ups and downs in the number
it recovered in the subsequent computerization era — before
of new businesses reflect other economic forces such as
slowing again in the 2000s (Figure 5). If history is a good
demographics and technology. New businesses contribute to
guide, we can hope for another rebound.
productivity and employment growth, but partly at the cost
Are publicly traded firms a good indicator of new firms’
of existing businesses. The current slowdown in business
success through history? They might not be. Which firms
go public is not random, and the decisions underlying public
offerings may change for reasons unrelated to startup formaFIGURE 7
tion. However, looking at the census data on all firms, pubFirm Formation Has Fluctuated Greatly over Time
lic as well as privately held, shows that the recent decline in
Changes in the number of new U.S. firms, 1900–2012.
the total number of firms is not unprecedented. True, these
totals do not tell us how much of an observed fall in the
total number of firms is due to fewer startups and how much
is due to more failures of existing firms. Yet, the fluctuations
in the number of firms are consistent with the fluctuations
in stock valuations over time, suggesting that the recent
declines may very well reverse themselves. The decline in
the Great Recession has precedents in the Great Depression
and in the slowdowns in the 1950s and 1960s, all of which
were subsequently reversed (Figure 7). A historical view thus
suggests not reading too much into the decline of new firms
Sources: Census Bureau and author’s calculations.
Notes: Data for 1900–1983 are Census Bureau firm births and deaths statistics. Data
because of technological underpinnings.
for 1978–2012 are Census Bureau Business Dynamics Statistics. Data around World
War I and World War II are averages.
From a macroeconomic view, we do not care about the
FIGURE 6

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 5

formation is therefore serious, but the underlying causes may
well lie outside the realm of policies tailored at nurturing
startups — such as offering new entrepreneurs insurance
against risk or tax incentives. The slowdown in business formation, particularly in the Great Recession, likely reflected
the overall economic slowdown more than it contributed to
it. The experience of the U.S. economy over the 20th century gives reasons to hope that as technology evolves, a new
entrepreneurial boom may well emerge.

That is not to say that policymakers can only stand
by and wait. There is at least limited potential for policy.
French legislation that provided some insurance against
entrepreneurial earnings risk has increased business formation and employment without diminishing the quality of
new firms.22 Even if the employment effects were small,
similar legislation in the U.S. might raise productivity and
wage growth.

REFERENCES
Adelino, Manuel, Antoinette Schoar, and Felipe Severino. “House Prices,
Collateral, and Self-Employment,” Journal of Financial Economics, 117:2
(2015), pp. 288–306.

Haltiwanger, John, Ron S. Jarmin, and Javier Miranda. “Who Creates Jobs?
Small Versus Large Versus Young,” Review of Economics and Statistics, 95:2
(May 2013), pp. 347–361.

Birch, David L. “Who Creates Jobs?” Public Interest, 65 (1981), pp. 3–14.

Hombert, Johan, Antoinette Schoar, David Sraer, and David Thesmar. “Can
Unemployment Insurance Spur Entrepreneurial Activity?” National Bureau of
Economic Research Working Paper 20717 (November 2014).

Boeri, Tito, and Ulrich Cramer. “Employment Growth, Incumbents and
Entrants: Evidence from Germany,” International Journal of Industrial
Organization, 10:4 (December 1992), pp. 545–565.
Decker, Ryan, John Haltiwanger, Ron Jarmin, and Javier Miranda. “The Role
of Entrepreneurship in U.S. Job Creation and Economic Dynamism,” Journal
of Economic Perspectives, 28:3 (Summer 2014), pp. 3–24.
Drautzburg, Thorsten. “Entrepreneurial Tail Risk: Implications for
Employment Dynamics,” Federal Reserve Bank of Philadelphia Working Paper
13–45 (November 2013).
Fairlie, Robert W., Arnobio Morelix, E.J. Reedy, and Joshua Russell. “The
Kauffman Index: Startup Activity, National Trends,” Kauffman Foundation
(June 2015).
Fujita, Shigeru. “Creative Destruction and Aggregate Productivity Growth,”
Federal Reserve Bank of Philadelphia Business Review (Third Quarter 2008).
Gertler, Mark, and Simon Gilchrist. “The Cyclical Behavior of Short-Term
Business Lending: Implications for Financial Propagation Mechanisms,”
European Economic Review, 37:2–3 (April 1993), pp. 623–631.
Greenstone, Michael, Richard Hornbeck, and Enrico Moretti. “Identifying
Agglomeration Spillovers: Evidence from Winners and Losers of Large Plant
Openings,” Journal of Political Economy, 118:3 (2010), pp. 536–598.

Jovanovic, Boyan, and Peter L. Rousseau. “Stock Markets in the New
Economy,” in Chong-En Bai and Chi-Wa Yuen, eds., Technology and the New
Economy, Cambridge MA: MIT Press, 2002.
Karahan, Fatih, Ben Pugsley, and Aysegul Sahin. “Understanding the 30-Year
Decline in the Startup Rate: A General Equilibrium Approach,” unpublished
manuscript (2015).
Pugsley, Benjamin Wild, and Aysegul Sahin. “Grown-Up Business Cycles,”
unpublished manuscript (2015).
Schmalz, Martin C., David A. Sraer, and David Thesmar. “Housing Collateral
and Entrepreneurship,” National Bureau of Economic Research Working
Paper 19680 (November 2013).
Sedlacek, Petr, and Vincent Sterk. “The Growth Potential of Startups over
the Business Cycle,” Centre for Macroeconomics Discussion Paper 1403
(February 2014).
Siemer, Michael. “Firm Entry and Employment Dynamics in the Great
Recession,” Federal Reserve Board Finance and Economics Discussion Series
2014–56 (2014).

6 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

NOTES
Shigeru Fujita’s Business Review article provides an overview of studies
quantifying economist Joseph Schumpeter’s famous insight that the
continual churn of firm formations and failures is the “essential fact about
capitalism.”
1

11

See the discussion paper by Michael Siemer.

12

See the paper by Petr Sedlacek and Vincent Sterk.

See the paper by Johan Hombert and others. Ideally, one should compare
entering with exiting firms. Because we do not observe the hypothetical
productivity of firms that exited, Hombert and his coauthors instead compare
the productivity of new and existing shrinking firms.
13

Formally, a new business has been in existence for no more than a year,
has at least one paid employee, and is not owned by another business.
Excluded are the self-employed who have no employees; private households
that employ domestic help; and railroads, agricultural producers, and most
government entities.
2

The Business Dynamics Statistics data set assembled by John Haltiwanger
and his coauthors and provided by the U.S. Census Bureau underlies this
article.
3

This difference partly reflects how business-level job creation is calculated:
as the change from the size of the firm’s workforce in the prior year. Since
by definition a new firm has no prior year, it can only add jobs, while an
older firm can shed them. For details on how the Bureau of Labor Statistics
measures net changes in employment at the business level, see its Business
Employment Dynamics FAQs, in particular question No. 9: http://www.bls.
gov/bdm/bdmfaq.htm#9. To calculate these annual averages, I adjusted
for changes in the working-age population by dividing by the ratio of the
working-age population in a given year relative to 2013. I dropped 1977,
which was a (positive) outlier. Note that gross job creation averaged 19.63
million jobs per year, adjusted for changes in the working-age population.

14
The article by Michael Greenstone and his coauthors details the
comparisons. Note that winning counties could also just have better
productivity to start with than losing counties, but Greenstone and his
coauthors find that “compared to losing counties in the years before the
opening of the new plant, winning counties have similar trends in most
economic variables,” (p. 539).

4

The Census Bureau’s Business Dynamics Statistics also charts the decline in
absolute terms: http://www.census.gov/ces/dataproducts/bds/.

Fatih Karahan and his colleagues argue that the trend decline of new
businesses that Pugsley and Sahin documented is, in fact, largely attributable
to supply shocks arising from demographics.
15

See Mark Gertler and Simon Gilchrist’s study of monetary policy’s effects
on financing conditions for large versus small firms. Martin Schmalz and
his coauthors and Manuel Adelino and his coauthors argue that because
collateralized loans matter for entrepreneurs, startups transmit events in the
housing market. They claim that 15 to 25 percent of the employment growth
between 2002 and 2007 can be attributed to the U.S. housing boom’s benefit
to entrepreneurs.
16

5

6
David Birch’s 1981 work actually emphasizes both the role of being a young
firm and of being a small firm but does not address the correlation of young
with small.

Boyan Jovanovic and Peter Rousseau’s account of U.S. history shows how
firm formation has reflected technological opportunities.
17

18

Jovanovic and Rousseau, 2002.

Because the data in Jovanovic and Rousseau end in 2000, I supplement
calculations based on their data with data from Compustat in Figure 5.
When the two data sets overlap in the 1990s, the implied starting share and
growth rate are very similar.

19
7

Haltiwanger and his coauthors pointed out this misperception.

8

Data for 2013 are not shown.

9
This trend holds true within industries and across regions in the U.S. See
also the research by Ryan Decker and his colleagues and by Benjamin Pugsley
and Aysegul Sahin.

20

For example, a “span of control” model as calibrated in my 2013 paper.

21

See the paper by Johan Hombert and his coauthors.

10
Pugsley and Sahin, 2015. Such a high number exceeds the number of
people unemployed in the U.S., which peaked at 15.2 million in 2009, and
would thus imply an increase in labor force participation.

22

Hombert and coauthors.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 7

REGIONAL SPOTLIGHT
The State of the States
Even if the U.S. economy is thriving, some states can be in recession, and vice versa.
But identifying state cycles is not so easy.
BY PAUL R. FLORA

Of the five U.S. recessions since 1979, Florida’s economy continued to expand throughout three of them. In
contrast, Alaska has had eight recessions since 1979, but
only three of them occurred during a national recession.
In fact, over the past 37 years, only eight states have been
in recession during — and only during — all five of those
U.S. recessions.1 Whether a state’s economy hews closely
to the expansions and contractions of the U.S. business
cycle depends on a variety of factors, including the state’s
industry mix and demographic trends. Florida’s economy, for
instance, has been propelled by rapid population growth as
one of the main Sun Belt destinations for domestic migration and as a gateway state for tens of thousands of Latin
American immigrants each year. Energy price shocks have
frequently buffeted Alaska’s economy, which relies heavily
on the volatile and risk-prone oil industry.
Understanding a state’s unique trends as well as the
geographic distribution of state recessions is of great interest to households, firms, and policymakers. Tracking state
cycles helps clarify the underlying causes of national recessions,2 informs policymakers regarding appropriate monetary
policy,3 and aids in recognizing in real time an emerging
national recession.4
However, as this article will show, the greater volatility of state data and other complications make determining
business cycles for an individual state more difficult than for
the U.S. economy. Since 2005, the Federal Reserve Bank of
Philadelphia has facilitated state business cycle research by
producing coincident indexes of economic activity for all 50
states and the nation. Over the past decade, researchers have
used the indexes to identify individual state business cycles.
With an additional 11 years of data since the indexes

were first published, and with the Great Recession behind
us, I explore a method for using our indexes to pinpoint the
onset and end dates of state business cycles and assess its results: What do the state coincident indexes now tell us about
state cycles? And have any states entered a recession lately?
HOW ARE BUSINESS CYCLES DETERMINED?

Unfortunately, no official entity exists for dating the
peaks and troughs of economic activity for each of the 50
states. For the overall U.S. economy, however, the National
Bureau of Economic Research (NBER), a private organization, began publishing its determination of the timing of
peaks and troughs in economic activity in 1929, becoming
the unofficial but widely accepted arbiter of the nation’s
business cycles.
Within the NBER framework of alternating peaks
and troughs in economic activity, “a recession is a period
between a peak and a trough, and an expansion is a period
between a trough and a peak.” A recession is marked by a
“significant decline in economic activity” lasting at least a
few months, while an expansion is a typically longer period
of increasing economic activity.5
Using judgment rather than a rule, the NBER’s Business
Cycle Dating Committee decides when the last turning
point in a cycle occurred by
Paul R. Flora is a research
and policy support manager
examining an assortment of
and senior economic analyst
quarterly and monthly data,
in the Research Department
of the Federal Reserve Bank of
but only after waiting until
Philadelphia. The views expressed
the risk of significant data
in this article are not necessarily
revisions has abated. The
those of the Federal Reserve.

8 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

NBER waited 15 months before pronouncing that June 2009
was the trough month in which the Great Recession ended.6
For the states, a lack of comparable data represents the
greatest challenge for determining individual state business
cycles. Most critically, quarterly state GDP has been available
only since 2015 and is still considered an experimental
measure. When it is released, state GDP lags the comparable
national data by three months. Of the 10 monthly indicators
recently used by the NBER, only three are available for
the states on a monthly basis: employment as measured by
Bureau of Labor Statistics payroll and household surveys, and
aggregate hours worked. All three are employment-related,
so potential signals from other economic factors that are
typically included in national aggregates of economic activity
such as corporate profits are missed.
Our state coincident indexes were designed to compensate for the lack of comparable data by modeling the overall
underlying growth of a state’s economy using available data.
Three monthly variables (nonfarm payroll employment,
average hours worked in manufacturing, and the unemployment rate) plus one quarterly variable (real wages and salaries) are used to estimate an underlying (sometimes called a
hidden) fifth variable that represents a state’s gross domestic
product.7 However, divining state business cycles is further
complicated by two additional challenges.
First, the smaller size of state economies and the smaller
sample sizes used to estimate state economic indicators
generate greater data volatility and noisier trends, making it
more difficult to discern true peaks and troughs. The second problem results from the longer lags in reporting state
variables and the greater revisions to state estimates, which
allow any false signals to persist until annual revisions are
conducted to update the data. Thus, just as the NBER does
in declaring national cycle dates, it is better to wait before
pronouncing state peaks and troughs. Still, studies have
demonstrated that examining state business cycles in real
time is a potential — though not risk-free — way to assess
the probability that the nation is currently in recession —
an assessment that is beyond the scope of this article.8
BUT HOW TO DETERMINE A STATE CYCLE?

Undertaking the task of identifying peaks and troughs
for 50 individual states over a 37-year period calls for establishing a set of simple, straightforward criteria that capture
the spirit of the NBER dating committee.
Criteria for the states are established by first examining
how our national coincident index has performed relative to

FIGURE 1

U.S. Index Aligns Well with NBER Recessions
Pennsylvania’s as well, but state indexes are inherently more volatile.

Sources: Federal Reserve Bank of Philadelphia; National Bureau of Economic Research.

NBER-determined cycles.9 Our national coincident index,
which was created at the same time as the state indexes for
comparison purposes, is relatively well behaved, capturing
all five NBER recessions as uninterrupted declines in activity, interspersed with uninterrupted increases in activity,
or expansions (Figure 1).10 The durations of the declines
range in length from four months in the 1980 recession to
18 months in the Great Recession. The depths of the recessions (calculated as the simple sum of the monthly percent
changes during each recession period) ranged from -0.24 in
the 1980 recession to -4.55 in the Great Recession.11
As the 1980 recession was the shortest and shallowest
national recession since 1979, its characteristics were used
as the minimum criteria for determining state recessions: a
minimum duration of four months and a minimum decline
equal to or exceeding a simple variance measure computed
for each state. Brief, one-time economic shocks that may result from a labor strike, factory closing, or natural disaster are
less likely to be labeled a recession because a duration threshold is applied. Similarly, longer patches of slight declines
avoid a recession label by virtue of a variance threshold.
For the nation, the average absolute value of the
monthly percent changes in the national index was 0.24, the
same as the aggregate change during the nation’s smallest
recession. Thus, the minimum decline for a state recession
and minimum increase for a state expansion are established as the average absolute value of the monthly percent
changes in each state index. Using a state-specific variance
acknowledges the potential for state business cycles to have
smaller or greater amplitudes than the nation’s cycle.12 (See
the accompanying notes on Determining State Peaks and
Troughs for examples of how the criteria are applied.)

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 9

Determining State Peaks and Troughs
Criteria

A state business cycle peak is determined as the last month in which the index has a positive monthly change prior to a
period of at least four months in which the sum of the monthly changes is negative and its absolute value equals or exceeds
the simple variance in that state’s coincident index.
A state business cycle trough is determined as the last month of a qualifying recession (and one with a negative monthly
change) prior to a period of at least four months in which the sum of the monthly changes is positive and its absolute value
equals or exceeds the simple variance.
A period with offsetting monthly changes (a net change of zero for two or more months) at the start of a qualifying
recession is treated as part of the prior expansion. Likewise, a period of two or more months of no net change at the end of a
qualifying recession is treated as part of the subsequent expansion.
Examples

The very different experiences of five states and the U.S. during the double-dip U.S. recessions of the early 1980s are
representative.
•

Connecticut avoided both recessions. It did experience a seven-month decline (shaded yellow) during the second
U.S. recession that was too shallow to qualify as a recession.

•

Florida avoided both recessions. Although its growth rate was well below its norm, the state economy continued
to expand.

•

Illinois experienced one long recession. While the U.S. enjoyed a brief intervening expansion, Illinois was one of
two states that declined throughout. Three other states escaped that fate by virtue of a bare minimum fourmonth expansion.

•

New Hampshire avoided the first recession because of an insufficient duration, although it had a sufficiently deep
decline (shaded yellow). Eight other states avoided the first recession with little or no decline, but not the second,
while Alaska experienced the first and avoided the second.

•

Pennsylvania followed the nation into and out of both recessions — one of 36 states to do so.

It is important to note that peaks also represent the maximum for that cycle. For example, June 1981 was a peak month
for Pennsylvania, with a subsequent trough in February 1983. June 1981 is the cycle maximum, not February or April,
because the cumulative change from March 1981 through June 1981 is positive. Likewise, troughs represent a minimum
for that cycle.
There were seven instances in which the depth was sufficient to qualify as a recession, but the duration was too short.
Only the New Hampshire episode fell within a national recession. In addition, a 2006 bank merger in Delaware generated
a three-month decline, a 1998 General Motors strike in Michigan caused a deep, two-month decline, and Florida’s index
declined sharply for one month following 9/11. The remaining three cases involved the energy states of Alaska, South
Dakota, and West Virginia.
A spreadsheet showing onsets and end dates of all recessions since 1979 for all 50 states can be viewed at: https://www.
philadelphiafed.org/-/media/research-and-data/publications/regional-spotlight/2016/Q4-state-peaks-and-troughs.xlsx.

10 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

TABLE 1

Results
Monthly percent change in each
coincident index

Monthly percent change in each
coincident index

CT

FL

IL

NH

PA

US

State Absolute
Average

0.29

0.36

0.29

0.35

0.25

0.24

Feb-79

0.39

0.57

0.06

0.54

0.20

Mar-79

0.40

0.59

0.18

0.61

Apr-79

0.41

0.60

0.39

0.55

May-79

0.42

0.62

0.04

0.39

0.09

0.30

Oct-81

0.00

0.07

(0.43)

Jun-79

0.41

0.47

0.14

0.37

0.09

0.28

Nov-81

(0.02)

0.05

(0.26)

Jul-79

0.41

0.65

(0.04)

0.19

0.02

0.25

Dec-81

(0.04)

0.04

(0.50)

Aug-79

0.39

0.50

(0.13)

0.32

0.04

0.23

Jan-82

(0.05)

0.03

Sep-79

0.37

0.68

(0.42)

0.36

0.08

0.21

Feb-82

(0.05)

Oct-79

0.35

0.68

(0.18)

0.44

(0.01)

0.19

Mar-82

Nov-79

0.33

0.69

(0.42)

0.54

0.04

0.17

Dec-79

0.29

0.70

(0.18)

0.41

(0.06)

0.15

NBER Peak
Jan-80

0.24

0.57

(0.30)

0.33

(0.11)

0.12

Feb-80

0.18

0.61

(0.55)

0.30

(0.36)

0.06

Mar-80

0.12

0.30

(0.53)

0.15

(0.52)

(0.00)

Apr-80

0.07

0.32

(0.85)

(0.16)

(0.69)

(0.07)

May-80

0.04

0.31

(0.55)

(0.22)

(0.64)

(0.10)

Jun-80

0.03

0.47

(0.67)

(0.02)

(0.55)

(0.07)

NBER Trough
Jul-80

0.05

0.30

(0.59)

0.09

(0.50)

0.00

Aug-80

0.09

0.60

(0.25)

0.35

0.11

0.09

Sep-80

0.13

0.56

(0.30)

0.45

0.02

0.16

Oct-80

0.17

0.56

(0.25)

0.60

0.48

0.22

Nov-80

0.20

0.55

(0.28)

0.51

0.20

0.25

Dec-80

0.21

0.54

(0.10)

0.55

0.38

0.24

Jan-81

0.22

0.52

(0.18)

0.32

0.03

0.22

Feb-81

0.21

0.51

0.04

0.37

0.11

0.23

Mar-81

0.19

0.48

(0.12)

0.40

(0.06)

0.24

Apr-81

0.18

0.46

(0.06)

0.40

0.06

0.24

May-81

0.16

0.43

0.01

0.43

(0.11)

0.21

CT

FL

IL

NH

PA

US

Jun-81

0.13

0.41

(0.04)

0.44

0.20

0.16

0.33

NBER Peak
Jul-81

0.11

0.23

(0.22)

0.33

(0.25)

0.11

0.17

0.32

Aug-81

0.07

0.23

(0.15)

0.31

(0.12)

0.04

0.17

0.31

Sep-81

0.04

0.08

(0.23)

0.15

(0.48)

(0.00)

0.12

(0.35)

(0.05)

0.04

(0.56)

(0.10)

(0.09)

(0.58)

(0.13)

(0.46)

(0.04)

(0.47)

(0.14)

0.02

(0.62)

(0.18)

(0.37)

(0.13)

(0.04)

(0.00)

(0.58)

(0.07)

(0.44)

(0.13)

Apr-82

(0.03)

0.13

(0.60)

(0.01)

(0.40)

(0.11)

May-82

(0.01)

0.11

(0.59)

0.11

(0.42)

(0.10)

Jun-82

0.01

0.13

(0.63)

0.23

(0.50)

(0.11)

Jul-82

0.04

0.16

(0.48)

0.26

(0.53)

(0.13)

Aug-82

0.06

0.04

(0.47)

0.20

(0.51)

(0.13)

Sep-82

0.08

0.09

(0.49)

0.13

(0.52)

(0.12)

Oct-82

0.12

0.12

(0.34)

0.04

(0.68)

(0.07)

NBER Trough
Nov-82

0.18

0.16

(0.35)

0.11

(0.32)

(0.01)

Dec-82

0.25

0.18

(0.18)

0.29

(0.22)

0.07

Jan-83

0.33

0.31

(0.05)

0.46

(0.11)

0.14

Feb-83

0.42

0.61

0.12

0.67

(0.05)

0.21

Mar-83

0.52

0.61

0.24

0.83

0.35

0.28

Apr-83

0.60

0.64

0.43

1.05

0.44

0.34

May-83

0.68

0.78

0.43

1.12

0.57

0.40

Jun-83

0.74

0.93

0.61

1.01

0.44

0.45

Sources: Federal Reserve Bank of Philadelphia; National Bureau of Economic Research.
Notes: Declines are shown in parentheses.
The NBER indicates the months in which peaks and troughs occur and the duration (in
months) of recessions and expansions. It makes no determination of exactly when during the month a recession or expansion starts or ends.
A spreadsheet showing onsets and end dates of all recessions since 1979 for all 50
states can be viewed at: https://www.philadelphiafed.org/-/media/research-and-data/
publications/regional-spotlight/2016/Q4-state-peaks-and-troughs.xlsx.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 11

ASSESSING STATE CYCLES SINCE 1979

While the 1990–1991 recession was much shorter, the
distribution of its impact among the states was much more
uneven. Of the 31 states in recession, Alaska and Wisconsin
began to recover after just six months, while Connecticut
and New Jersey endured 37 months of contraction. Sometimes referred to as the bicoastal recession, the 1990–1991

Using these criteria, I determined the peaks and
troughs for all 50 states, five of which are highlighted, along
with the United States, in Determining State Peaks and
Troughs and all of which are viewable through the accompanying link. The most notable finding is that the Great Recession was so
FIGURE 2
severe that no state economy avoided a
No State Avoided the Great Recession of 2007–2009
recession. The all-encompassing nature
Length of each state’s recession, in months.
of that downturn stands in contrast to
the prior four national recessions. In
particular, 19 states avoided a contraction during the 1990–1991 recession
(Figures 2 and 3). During the double-dip
recessions, 11 states avoided the brief
1980 recession, while only three states
avoided the deeper, longer recession that
followed in 1981–1982. Connecticut
and Florida avoided both, while Alaska
avoided the second. Eight states avoided
the 2001 national recession.
The national economy endured the
Great Recession for 18 months, according to the NBER. Our national index
Source: Federal Reserve Bank of Philadelphia.
also indicated an 18-month duration.
Note: The duration of a recession is the number of months from the peak to the trough. The Great Recession was
18 months long for the nation as a whole.
However, the peak and trough indicated
by our index lag the NBER’s dates by
four months. For the other four recesFIGURE 3
sions, all peaks and troughs for the U.S.
19 States Avoided the 1990–1991 Recession
Length of each state’s recession, in months.
economy had been indicated within two
months or less of the NBER determinations.
The durations of those state recessions that accompanied the Great
Recession ranged from five months in
North Dakota to 64 months in Michigan. However, the latter was mired in
a long-term structural change (more
on that later). The more representative
extreme during the Great Recession was
Nevada, which endured 52 months of
economic decline as its housing market
collapsed. On average, recessions lasted
a full year longer in the sand states
of Arizona, California, Florida, and
Source: Federal Reserve Bank of Philadelphia.
Nevada than in the other 46 states: 36
Note: The duration of a recession is the number of months from the peak to the trough. The 1990–1991 recession was
months as opposed to 24 months.
eight months long for the nation as a whole.

12 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

recession hit New England and the Mid-Atlantic states especially hard. The average duration of recessions in the nine
states in those two regions was 30 months; the average in the
other 22 states was just 12 months.
Many of the 19 states that avoided the 1990–1991
national recession had hit bottom just a few years earlier as
part of a series of mid-1980s state recessions that struck 14
farm and energy states, predominately located in the nation’s heartland. The farm states suffered early in the 1980s
as increased planting and greater yields collided with trade
disruptions and a stronger dollar. Farmland values followed agricultural prices and profits in a downward spiral,
and many farms went bankrupt. Rolling recessions became a
popular descriptor, as 10 of those 14 states would later avoid
the 1990–1991 U.S. recession, while Alaska, Mississippi,
Montana, and West Virginia would succumb a second time.
The timing and duration of the farm and energy state
recessions were somewhat idiosyncratic. Farm states tended
to be hit earlier but adjust more quickly, such as Iowa, with a

Hawaii and Michigan have had
recessionary periods lasting in excess
of five years that may be more accurately
described as secular declines due to
long-term structural change.
July 1984 peak and a February 1985 trough. With a dependence on agriculture, metal mining, and energy extraction,
Montana was the first state to enter a recession during this
period, with a February 1984 peak, and it was the last to
emerge, with a September 1987 trough.
The sense many people had of a “jobless” recovery following the eight-month 2001 national recession gains credence after examining state recessions rather than just the
U.S. Of the 42 states that experienced a recession, only 15
had a single, relatively brief recession like the national one.
Recessions extended 12 to 18 months longer in 14 states.
During that same postrecession period, a dozen more states
experienced a second recession following a brief interlude of
expansion. Often the anomaly, West Virginia did not begin
its 18-month recession until the national recession had ended.
When is a recession not a recession? Following our
criteria, Hawaii and Michigan have had recessionary periods

lasting in excess of five years that may be more accurately described as secular declines due to long-term structural change.
Hawaii, which avoided the 1990–1991 recession, peaked
instead in December 1991. An 81-month recession ensued
that corresponded to the massive asset bubble burst and recession that enveloped Japan. The nearly seven years it took
for Hawaii to hit bottom represents the painful adjustment as
business and personal investment from Japan dropped sharply.
While the nation underwent the relatively shallow
eight-month recession of 2001, Michigan was in the midst
of a much deeper 21-month recession. Michigan’s economic activity had peaked in April 2000 and hit bottom
in January 2002. Like many other states during the jobless
recovery, Michigan experienced a short, shallow expansion
of seven months then entered another 11-month recession
— hitting a second trough in July 2003. However, unlike
other states, Michigan’s next expansion was equally short
and shallow, again lasting just seven months and peaking
in February 2004. Michigan did not hit bottom again until
June 2009, when the Great Recession ended. Essentially,
Michigan gained little from the six-year national expansion. Rather, the state suffered significant employment
losses as its manufacturing sector restructured and retooled
over more than a decade.
HOW HAVE STATES FARED SINCE THE GREAT RECESSION?

Aside from a few late echoes following the Great Recession — as in the jobless recovery in the wake of the 2001
recession — six energy states were in recession for at least
part of 2015: Alaska, Louisiana, North Dakota, Oklahoma,
West Virginia, and Wyoming. For Alaska and West Virginia, these were their second recessions since the Great
Recession. Most of these state economies have been severely
hurt by the fall in oil prices. West Texas crude dropped 75
percent (annualized) from $105.80 per barrel in June 2014 to
$47.50 per barrel in January 2015. West Virginia’s economy,
which expanded again in the latter half of 2015, has suffered
due to coal industry conditions.
These six states are among the top eight states in terms
of the share of total wages attributable to the natural resources and mining sectors. Wyoming leads the pack, with
Louisiana eighth. New Mexico and Texas are sixth and
seventh (Table 2).
The current energy state downturn resembles the previously discussed farm and energy slump that sent 14 states
into recession at some point from 1984 to 1987 (Figure 4).
Back then, West Texas crude oil had dropped 93 percent

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 13

TABLE 2

Recession States in 2015 Highly Dependent on Energy
Location quotients* for state natural resources and mining sectors.
Recession states are shaded.
Total annual
wages

Annual average
employment

Wyoming

10.53

6.52

Alaska

8.17

3.84

North Dakota

7.26

4.41

Oklahoma

5.26

2.82

West Virginia

4.41

2.69

New Mexico

4.11

3.19

Texas

3.58

1.97

Louisiana

3.31

1.97

Montana

3.04

2.02

Idaho

2.35

2.80

U.S.

1.00

1.00

Source: Bureau of Labor Statistics Quarterly Census of Employment and Wages.
*A location quotient represents the proportionate contribution that wages or employment from a given economic sector makes to a state’s total economy relative to that
sector’s contribution within the nation’s economy.

FIGURE 4

Latest Energy State Recession Less Widespread
Instances of state recession, by recession period.

(annualized) from $30.80 in November 1985 to $12.60 in
March 1986. Besides the current six, Colorado, Idaho, Iowa,
Mississippi, Montana, Nebraska, New Mexico, and Texas
had also been in recession.
As with the nation’s mid-1980s experience with an
energy recession, the current state recessions in six energy
states do not indicate a nationwide problem. The misfortunes of businesses and households from those six states are
linked to significantly lower energy prices, which represent a
substantial consumer benefit for everyone else. Thus, the nation’s economy typically grows faster, even as regions tied to
energy production retrench. Similarly, we can draw distinctions within our Third District between those manufacturing firms that supply the energy sector and those that supply
consumers, either directly or indirectly. Producers of food
products and building materials, such as windows for new
homes, have enjoyed lower input prices and lower production costs. Conversely, producers of heavy industrial equipment used by shale gas firms in Pennsylvania and by energy
firms worldwide have suffered a sharp decline in orders.
FINAL OBSERVATIONS

Based on my analysis of the 50 state coincident indexes,
just six energy states were in recession during 2015, and as
was the case in the mid-1980s, this energy state recession posed no risk to the
national expansion.
However, as new data continually
arrive and previous data are revised,
our indexes may reveal somewhat
different trends. Nevertheless, the
economic data we’ve seen through
most of 2016, and our knowledge of the
direction and extent of potential data
revisions, do not alter the conclusion
that the nation’s economic expansion
continues unabated. And most states
are following along.

Source: Federal Reserve Bank of Philadelphia.

14 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

NOTES
1
These states are Georgia, Kansas, Missouri, New Jersey, Ohio, South
Carolina, Vermont, and Virginia.

2

See the research by Michael Owyang and his colleagues.

3

See the article by Gerald Carlino and Robert DeFina.

4

Ted Crone’s 2006 Business Review article goes into detail.

sales, the index of industrial production, real personal income less transfer
payments, aggregate hours of work in the total economy, payroll survey
employment, and household survey employment.
7
For more details on the construction of the state coincident indexes, see Ted
Crone’s 2006 paper or our website at: www.philadelphiafed.org/researchand-data/regional-economy/indexes/coincident.

8

The NBER’s Business Cycle Dating Committee defines a recession as a
“significant decline in economic activity” lasting a few months to more than
a year, but it uses a variety of indicators as well as its members’ judgment to
decide what constitutes significant: “The Committee does not have a fixed
definition of economic activity. It examines and compares the behavior of
various measures of broad activity: real GDP measured on the product and
income sides, economy-wide employment, and real income. The Committee
also may consider indicators that do not cover the entire economy, such as
real sales and the Federal Reserve’s index of industrial production (IP). The
Committee’s use of these indicators in conjunction with the broad measures
recognizes the issue of double-counting of sectors included in both those
indicators and the broad measures. Still, a well-defined peak or trough in
real sales or IP might help to determine the overall peak or trough dates,
particularly if the economy-wide indicators are in conflict or do not have
well-defined peaks or troughs.” For more details on the NBER’s approach
to determining national business cycles, see its Business Cycle Dating
Committee website, including its frequently asked questions at www.nber.
org/cycles/recessions_faq.html.

See the 2006 article by Ted Crone and the 2008 report by Jason Novak.

5

Although we used the state coincident indexes as of June 2016, we did not
consider the data beyond December 2015 for the purpose of determining
business cycles. In the June 2016 vintage, state employment data after
September 2015 are subject to significant potential revisions. However,
this vintage also includes first quarter personal income data, which itself
incorporates some of the employment data revisions through December 2015.
9

10
As such, peaks and troughs from the national index are easily determined.
A peak occurs in the last month of growth prior to a sequence of declines
in the index, and a trough occurs in the last month of decline prior to a
sequence of growth in the index.

For a spreadsheet of the underlying data of these results for all 50 states
and the nation over the entire 37 years, see https://www.philadelphiafed.
org/-/media/research-and-data/publications/regional-spotlight/2016/Q4state-peaks-and-troughs.xlsx.
11

Using a state-specific variance as a threshold rather than the nation’s
variance is the main conceptual change from the approach used in Ted
Crone’s 1994 and 2006 articles. This change also accommodates the fact
that our state coincident index approach can introduce greater variance. In
particular, our process standardizes the input variables to have a mean of
0 and a standard deviation of 1. After estimating, we retrend the result to
match the growth of state GDP. However, we do not revariance the indexes;
thus, they may fluctuate more or less than their underlying data.
12

To establish the June 2009 recession trough, the NBER reviewed quarterly
estimates of real gross domestic product (GDP) and real gross domestic
income (GDI) issued by the Bureau of Economic Analysis to determine the
quarter. Then they examined 10 monthly indicators to set the month. These
included: Macroeconomic Advisers’ monthly GDP, the Stock-Watson index of
monthly GDP, the Stock-Watson index of monthly GDI, the average of StockWatson indexes of monthly GDP and GDI, real manufacturing and trade
6

REFERENCES
Carlino, Gerald A., and Robert H. DeFina. “The Differential Regional Effects
of Monetary Policy,” Review of Economics and Statistics, 80:4 (November
1998), pp. 572–587.

Federal Reserve Bank of Philadelphia. “Predicting Benchmark Revisions of
Payroll Employment Estimates in Third District States” Regional Economic
Analysis (April 23, 2014).

Crone, Theodore M. “New Indexes Track the State of the States,” Federal
Reserve Bank of Philadelphia Business Review (January/February 1994).

Henderson, Jason. “Is This Farm Boom Different?” Federal Reserve Bank of
Kansas City Main Street Economist, 5 (2011).

Crone, Theodore M. “What a New Set of Indexes Tells Us About State and
National Business Cycles,” Federal Reserve Bank of Philadelphia Business
Review (First Quarter 2006).

Matthews, Steve. “The U.S. States Where Recession Is Already a Reality.”
Bloomberg (February 22, 2016).

Crone, Theodore M., and Alan Clayton Matthews. “Consistent Economic
Indexes for the 50 States,” Review of Economics and Statistics, 87:4,
(November 2005) pp. 593–603.
Federal Reserve Bank of Dallas. “After the Boom: Texas Economy Downshifts
in Energy Bust,” Annual Report 2015 (April 2016).

National Bureau of Economic Research. “The NBER’s Business Cycle Dating
Committee,” www.nber.org/cycles/recessions.html.
Novak, Jason. “Marking NBER Recessions with State Data,” Federal Reserve
Bank of Philadelphia Research Rap Special Report, April 2008.
Owyang, Michael, Jeremy Piger, and Howard Wall. “Business Cycle Phases
in U.S. States,” Review of Economics and Statistics, 87:4 (November 2005),
pp. 604–616.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 15

BANKING POLICY REVIEW
Did Dodd–Frank End ‘Too Big to Fail’?
Despite reforms, do big banks still benefit from market perceptions that the government
will bail them out if they falter?
BY RYAN JOHNSTON

During the financial crisis in 2008, the U.S. government bailed out some very large banks for fear the collapse of any bank that large would profoundly harm the
U.S. economy and destabilize the global financial system.1
That is, they were too big to be allowed to fail. Passage of
the Dodd–Frank Act two years later was intended to rule
out future bailouts through tighter safety-and-soundness
requirements, among other measures. Yet, some worry that
investors may still view certain banks as “too big to fail,” a
perception that would confer an arguably unfair and potentially risky funding advantage over smaller banks. If a bank’s
uninsured depositors or bondholders expect to be protected
against losses, they will accept lower interest rates. So, in
principle, we should be able to compare the rates paid by the
largest banks with those paid by smaller banks for evidence
of whether Dodd–Frank was successful in eliminating markets’ bailout expectations. But as this review will explain,
the many differences between large and small banks make it
hard to know whether we are comparing apples with apples.
We review studies that address this apples-to-apples problem
and help determine whether large banks still receive what is,
in effect, a government subsidy.
A primary stated goal of Dodd–Frank is to get rid of the
perception that the largest banks are too big to fail (TBTF).2
It aims to do so through a number of mechanisms. An
annual stress test is required for banks with assets greater
than $50 billion. The test uses hypothetical economic and
financial market scenarios of varying severity to measure
the impact on the value of banks’ capital. If the test indicates that a bank’s capital levels would fall below regulatory
requirements under the severe stress scenario, the bank
might be prohibited from making any dividend payments

or other capital distributions.3 The results of banks’ stress
tests are posted on the Federal Reserve Board of Governors
website and widely reported. Maintaining capital levels that
internally absorb economic shocks strengthens public confidence that big banks will not need to be bailed out during
an economic or financial downturn.4
Title II of Dodd–Frank gives the Federal Deposit Insurance Corporation (FDIC) authority to resolve a large,
complex financial institution that is close to failing. Among
other things, it prohibits the use of taxpayer funds and imposes losses on shareholders and creditors.5
Furthermore, in 2015 the Federal Reserve Board approved a rule requiring firms it deems global systemically
important banks (GSIBs) to maintain a larger capital cushion — more than that required of smaller banks — in order
to increase their resiliency against financial distress. This
so-called capital surcharge is based on the amount of risk a
GSIB poses to financial stability, or its “systemic footprint,”
and provides a stronger buffer against capital shortfalls that
a large bank may experience.6
Although Dodd–Frank has made significant progress
toward strengthening the financial system, some analysts and
policymakers have argued that markets still perceive the largest banks as TBTF. In particular, they have argued that the
largest banks have a funding advantage over smaller banks
because of this perception.
Lingering perceptions
that some banks remain TBTF Ryan Johnston is a banking
structure associate in the
might be a concern for a few
Research Department of
the Federal Reserve Bank of
notable reasons. First, deposiPhiladelphia. The views expressed
tors, bondholders, and other
in this review are not necessarily
creditors that perceive large
those of the Federal Reserve.

16 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

banks as TBTF may not monitor the banks’ activities as
closely as they normally would. They may also accept lower
returns from large banks. In turn, this advantage may encourage too much risk-taking by large banks. TBTF funding
advantages may also encourage banks to become too large or
promote other inefficiencies such as monopoly profits or too
little lending. Apart from these inefficiencies, policymakers
might be concerned that a funding advantage for large banks
could create unfair competition for smaller banks.
On the face of it, determining whether some banks
have a funding advantage should be easy. Banks fund themselves with a mixture of deposits, bonds, and equity. Why
not just compare the funding costs of large banks versus
smaller banks? But as former Federal Reserve Governor

Although Dodd–Frank has made
significant progress toward strengthening
the financial system, some analysts and
policymakers have argued that markets
still perceive the largest banks as TBTF.
Randall Kroszner has said, to know whether any funding
difference is due to TBTF perceptions, we need to be comparing apples with apples.7 There is a lot of evidence that
large banks have advantages from economies of scale.8 In
addition, their funding mix and business models differ from
those of small banks.
How can we solve the apples-to-apples question? What
evidence is there for the existence of a TBTF subsidy prior
to the financial crisis? What about post-Dodd–Frank? In
this article, we focus on the evidence from two rigorous approaches to the apples-to-apples issue. We are most interested in results for the post-Dodd–Frank period.
ARE BIG BANKS DIFFERENT FROM OTHER BIG FIRMS?

The first approach aims to get around the apples-to-apples issue by examining the differences in size-related funding costs for financial and nonfinancial institutions. This
approach asks whether large banks have a greater funding
advantage over small banks than other large firms have over
small firms in their industries. The underlying idea of this
comparison is that many of the factors that give large banks

a funding advantage over smaller banks — such as broader
access to public debt markets — also give large nonfinancial
firms a funding advantage over smaller nonfinancial firms.
However, there is no reason to expect government bailouts
in most nonfinancial industries because they do not have the
extensive interconnectedness and systemic footprint that the
financial industry has. So, this comparison helps isolate any
TBTF subsidy. Since nonfinancial firms do not take deposits,
these studies focus on the costs of bond financing.
Javed Ahmed, Christopher Anderson, and Rebecca
Zarutskie compare bond funding costs for commercial banks
and investment banks with bond funding costs for 14 other
nonfinancial industries. 9 They examine three periods: before (2004 Q1–2008 Q2), during (2008 Q3–2009 Q2), and
after (2009 Q3–2013 Q2) the financial crisis. They find that
there is a size-related funding advantage in all industries,
including commercial banks and investment banks. But they
do not find a size-related bond-funding advantage for commercial and investment banks when compared with other
industries in any period.10
They also compare the size effect separately for commercial banks, investment banks, and 12 other industries.
Out of those 14 industries, commercial banks and investment banks rank only ninth and 10th in size-related bond
funding advantage — below, for example, business equipment and chemicals. Interestingly, they find that the category of “other financial” industries, which includes insurance and asset management firms, ranks high in size-related
funding advantage.
While the comparison of larger and smaller firms across
industries is designed to control for a wide range of sizerelated differences that would affect bondholders’ perceived
risk of default, the authors of this study — and all the other
studies I discuss — also seek to control for default risk more
directly. In this study, they include a measure of the default
risk on a firm’s bonds from Moody’s Analytics. So, for example, regulatory factors such as higher capital requirements
for larger banks will reduce the likelihood that bondholders
will bear losses, and this lower likelihood will be reflected in
Moody’s measure of default risk.
A different study seeks to compare apples with apples
through a variation on that same approach: Viral Acharya,
Deniz Anginer, and Joseph Warburton ask whether the
sensitivity of bond spreads to various measures of credit
risk differs for large financial firms compared with large
nonfinancial firms. Note that unlike in the study by Ahmed
and his colleagues, financial firms in this study include
insurance companies and asset management companies.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 17

Their idea is that a TBTF subsidy would make bond yields
for the largest financial firms less sensitive to measures of
credit risk compared with smaller financial firms, while this
would not be true for nonfinancial firms.11
Their main finding is that while a decrease in risk
leads to a large reduction in yields for banks below the 90th
percentile in size, banks above the 90th percentile have
much less sensitivity to credit risk. Meanwhile, there is no
such change in the risk sensitivity of yields for the largest
nonfinancial firms. They calculate a subsidy of around 20
basis points before the crisis, rising above 100 basis points in
2009, and falling to around 30 basis points in 2012. So unlike the prior study, they estimate that there is a significant
TBTF subsidy, even following the passage of Dodd–Frank.
Why do the results of these two studies differ? There
are a few possibilities. First, the sample period in the first
study ends one year later, so perceptions about TBTF could
have evolved as regulatory changes continued after Dodd–
Frank. Another reason could be that the two studies divide
the financial and nonfinancial firms differently. The first
study separates commercial banks and investment banks
from other financial institutions, while the second study includes all financial firms as one group. And it was precisely
the other financial firms in the first study that appeared to
have a size-related funding advantage.
The difference in results is illuminated by another analysis, which uses a substantially similar methodology to the
one by Acharya and his coauthors. A study by John Lester
and Aditi Kumar focuses on only the very largest commercial and investment banks, and the sample period extends
through 2013. They find a 36 basis point funding benefit for
the largest banks in 2012 — not so different from Acharya
and his coauthors — but essentially no funding benefit to
being a very large bank in 2013.
DO LARGE BANKS PAY LESS FOR UNINSURED DEPOSITS?

The second approach analyzes deposit rates to compare
the differences in funding advantages between large and
small banks. If large banks have a funding advantage because of TBTF perceptions, it should show up as a smaller
differential between rates on uninsured deposits compared
with insured deposits. Unfortunately, only one study uses
this approach to measure the subsidy in the postcrisis
period, although a second study is helpful for putting the
results in perspective.
William Bassett compares the interest rate differential paid by large and small banks on small time deposits

— which are fully insured — and interest-bearing transactions and saving accounts — which are not fully insured.12
The main comparison is between the largest banks and
large regional banks. Bassett argues that this comparison is
more relevant than comparing large and small banks if we
are interested in TBTF versus other reasons why we might
observe a size-related funding differential.
Bassett compares the funding differential for banks
with assets of more than $125 billion and banks with assets
of $20 billion to $125 billion. First, he demonstrates that
the interest rates on small time deposits are not sensitive
to measures of bank risk for either large or smaller banks,
evidence that rates on insured deposits do not include a
premium for default risk. He then compares the rates on
interest-bearing savings and time deposits. Consistent with
the view that these deposits are not viewed by depositors

If large banks have a funding advantage
because of TBTF perceptions, it should
show up as a smaller differential between
rates on uninsured deposits compared
with insured deposits.
as fully insured, he shows that rates on these deposits are
sensitive to risk.
Bassett compares the difference in the rates on uninsured and insured deposits for large and smaller banks in the
precrisis and postcrisis periods. He finds a statistically insignificant funding advantage of 10 basis points in the precrisis
period and no advantage in the postcrisis period. While
Bassett’s analysis provides no evidence of a TBTF subsidy
— particularly in the postcrisis period — he notes that any
such subsidy may be difficult to find in the environment of
low interest rates and stable conditions that has prevailed
since the Great Recession.
Stefan Jacewitz and Jonathan Pogach provide no evidence of a TBTF premium for the post-Dodd–Frank period,
but their research helps to put bounds on the size of any preDodd–Frank TBTF subsidy.13 Like Bassett, they compare
the differential between rates paid on insured and uninsured
funding sources by large and small banks. They focus on a
narrower type of funding, money market deposit accounts
(MMDAs), and consider the different interest rates paid on

18 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

insured versus uninsured MMDAs. Prior to 2009, MMDAs
in excess of $100,000 were uninsured. Their main test
compares the differential for banks with assets exceeding
$200 billion and all other banks. This is a relatively clean
comparison, because regulatory restrictions impose uniformity on both large and small MMDAs. It is also economically important because MMDAs account for 35.3 percent
of banks’ liabilities.14
Jacewitz and Pogach’s main finding is that prior to the
crisis, banks with assets greater than $200 billion had a 40
basis point funding advantage, but the spread declined to
nearly zero when all MMDAs began to be insured during
the financial crisis. This decline to zero once the larger accounts were insured is evidence that the measured differential reflects a TBTF subsidy. But the fraction of the differential that can reasonably be ascribed to TBTF is probably too
large, as Jacewitz and Pogach themselves suggest.
They also try out a range of specifications to better
understand the underlying source of the precrisis funding
advantage for large banks. In particular, they find a significant premium of 21 basis points for banks with assets above
$10 billion compared with all other banks. Then again, few
would argue that a $10 billion bank would ever be considered important enough to the stability of the financial

system to be bailed out. This reality suggests that up to 21
basis points of the measured funding advantage can’t be
explained by TBTF and leaves us with an estimate of the
TBTF subsidy prior to the crisis ranging from a modest 20
basis points to a more significant 40 basis points.
CONCLUSION

There is evidence supporting and disputing the continued existence of TBTF subsidies. There are also many
methods that can be used to find evidence of a TBTF subsidy that go beyond the studies reviewed here. The weight of
the evidence is that while there may have been significant
TBTF subsidies prior to and during the financial crisis, following the crisis any subsidies are small. In addition, there
is evidence that funding costs now more accurately measure actual bank risk.15 This apparent absence of meaningful postcrisis subsidies could be partly due to the rules and
regulations resulting from Dodd–Frank. Investors may now
believe that they would have to take a hit to their wallets
if a large bank were to fail. However, the low interest rate
environment and relatively stable conditions in banking
markets make it difficult to disentangle any subsidy by examining funding costs.

REFERENCES
Acharya, Viral V., Deniz Anginer, and A. Joseph Warburton. “The End
of Market Discipline? Investor Expectations of Implicit Government
Guarantees,” working paper (May 2016).

Jacewitz, Stefan, and Jonathan Pogach. “Deposit Rate Advantages at the
Largest Banks.” FDIC Center for Financial Research Working Paper 2014–02
(February 2014).

Ahmed, Javed I., Christopher Anderson, and Rebecca E. Zarutskie. “Are the
Borrowing Costs of Large Financial Firms Unusual?” Federal Reserve Board of
Governors Finance and Economics Discussion Series 2015–024 (2015).

Kroszner, Randall S. “A Review of Bank Funding Cost Differences,” Journal of
Financial Services Research, 49:2 (June 2013), pp. 151–174.

Bassett, William F. “Using Uninsured Deposits to Refine Estimates of the
Large Bank Funding Advantage,” Journal of Law, Finance, and Accounting,
1:1 (2016), pp. 44–91.
Gandhi, Priyank, Hanno Lustig, and Alberto Plazzi. “Equity Is Cheap for Large
Financial Institutions: The International Evidence,” Swiss Finance Institute
Research Paper 16–22 (June 2016).
Hughes, Joseph P., and Loretta J. Mester. “Who Said Large Banks Don’t
Experience Scale Economies? Evidence From a Risk-Return-Driven Cost
Function,” Federal Reserve Bank of Philadelphia Working Paper 11–27
(July 2011).

Lester, John, and Aditi Kumar. “Do Bond Spreads Show Evidence of Too Big
to Fail Effects?” Oliver Wyman working paper (April 2014).
O’Hara, Maureen, and Wayne Shaw. “Deposit Insurance and Wealth Effects:
The Value of Being ‘Too Big to Fail,’ Journal of Finance, 45:5 (December
1990), pp. 1,587–1,600.
Santos, Joao, A.C. “Evidence from the Bond Market on Banks’ ‘Too-Big-toFail’ Subsidy,” Federal Reserve Bank of New York Economic Policy Review,
20:2 (December 2013), pp.29–39.
Strahan, Philip E. “Too Big to Fail: Causes, Consequences, and Policy
Responses,” Annual Review of Financial Economics, 5 (2013), pp. 43–61.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 19

NOTES
The term bailout refers to a government intervention in which the bank is
kept from failing and uninsured claimants are made whole.

1

7

See Randall Kroszner’s survey of the evidence.

See the article by Joseph Hughes and Loretta Mester for evidence of
significant scale economies.
8

While size is one feature that might make a bank TBTF, other factors such
as organizational complexity, dependence on funds that might disappear in
a crisis, and interconnectedness with other financial institutions can affect
banks’ systemic risk. The notion of TBTF incorporates all of these factors.
2

3
Regulators incorporate a bank’s stress test results into their quantitative
assessment in an annual Comprehensive Capital Analysis and Review
(CCAR), which evaluates the bank’s “capital adequacy, capital planning
process, and planned capital distributions, such as any dividend payments
and common stock repurchases. As part of CCAR, the Federal Reserve
evaluates whether BHCs [bank holding companies] have sufficient capital
to continue operations throughout times of economic and financial market
stress and whether they have robust, forward-looking capital-planning
processes that account for their unique risks. The Federal Reserve may object
to a BHC’s capital plan on quantitative or qualitative grounds. If the Federal
Reserve objects to a BHC’s capital plan, the BHC may not make any capital
distribution unless the Federal Reserve indicates in writing that it does not
object to the distribution.” See http://www.federalreserve.gov/newsevents/
press/bcreg/bcreg20160623a1.pdf.

Banks must also conduct their own stress tests under the same scenarios
as well as tests under bank-developed scenarios. For more information on
CCAR, Dodd–Frank Act stress tests, resolution plans, and other capital
requirements, see the Federal Reserve Board’s banking and regulation web
pages at http://www.federalreserve.gov/bankinforeg/default.htm.
4

There are critics who do not believe that Dodd–Frank will prevent bank
bailouts. This article does not focus on whether Dodd–Frank will actually
prevent bailouts. Instead, it concentrates on the market’s perception that a
bank will be bailed out.
5

The Fed bases its GSIB designations on criteria developed by the Bank
for International Settlements’ Basel Committee on Banking Supervision,
which include the bank’s “size, interconnectedness, lack of readily available
substitutes or financial institution infrastructure, global (cross-jurisdictional)
activity and complexity.” See http://www.bis.org/publ/bcbs207.htm.
6

9
In addition, they examine credit default swap (CDS) spreads. A CDS is a
type of insurance contract in which the seller of the CDS promises to pay the
buyer of the contract in the event of default on the firm’s insured bonds. So,
a smaller spread means there is a lower perceived risk of default on the firm’s
bonds. I focus on their results for bond spreads to facilitate the comparison
with other studies.

10
Their evidence for CDS spreads is largely similar. However, they find
evidence that CDS spreads were lower for larger commercial and investment
banks during the crisis, potential evidence of a TBTF funding advantage at
the time.

11
To bolster their case that their results do not depend on the use of a
particular measure of default risk, Acharya and his coauthors use a number
of measures of default risk and get similar results. As in the study by Ahmed
et al., this study includes measures of default risk in regressions to control for
firms’ risk of default for reasons other than size.

12
Small time deposits are defined as deposits of less than $100,000. Before
October 3, 2008, deposits smaller than $100,000 were fully insured by
the Federal Deposit Insurance Corporation (FDIC). After October 3, 2008,
deposits of $250,000 or less became fully insured.

13
Evidence of a TBTF subsidy would not be expected after the rise in the
insurance limit for MMDAs in 2008.

14
In addition, Jacewitz and Pogach examine pricing at the branch level to
help control for differences in funding costs due to scale economies.

15

See Philip Strahan’s article.

20 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

RESEARCH UPDATE
These working papers present preliminary findings of research conducted by Philadelphia Fed economists, analysts, and
visiting scholars. Visit our website for more abstracts and papers.

Valuing “Free” Media in GDP: An Experimental Approach

“Free” consumer entertainment and information from
the Internet, largely supported by advertising revenues, has
had a major impact on consumer behavior. Some economists
believe that measured gross domestic product (GDP) growth
is badly underestimated because GDP excludes online entertainment (Brynjolfsson and Oh 2012; Ito 2013; Aeppel
2015). This paper introduces an experimental GDP methodology that includes advertising-supported media in both
final output and business inputs. For example, Google Maps
would be counted as final output when it is used by a consumer to plan vacation driving routes. On the other hand,
the same website would be counted as a business input when
it is used by a pizza restaurant to plan delivery routes.
Contrary to critics of the U.S. Bureau of Economic
Analysis (BEA), the process of including “free” media in
the input-output accounts has little impact on either GDP
or total factor productivity (TFP). Between 1998 and 2012,
measured nominal GDP growth falls 0.005% per year, real
GDP growth rises 0.009% per year and TFP growth rises
0.016% per year. Between 1929 and 1998, measured nominal
GDP growth rises 0.002% per year, real GDP growth falls
0.002% per year, and TFP growth rises 0.004% per year.
These changes are not nearly enough to reverse the recent
slowdown in growth.
The authors’ method for accounting for free media is
production oriented in the sense that it is a measure of the
resource input into the entertainment (or other content) of
the medium rather than a measure of the consumer surplus
arising from the content. The BEA uses a similar production-oriented approach when measuring GDP. In contrast, other researchers use broader approaches to measure
value. Brynjolfsson and Oh (2012) attempt to capture some
consumer surplus by measuring the time expended on the
Internet. Varian (2009) argues that much of the value of the
Internet is in time saving, an additional metric for capturing
consumer surplus. The McKinsey Institute (Bughin et al.
2011) attempts to measure the productivity gain from search

directly. In particular, this production-oriented accounting
has no method to account for instances in which the good
or service precedes the revenue that it eventually generates.
Over the past two decades, many Silicon Valley firms have
followed the disruptive business model described as URL:
ubiquity now, revenue later. Some firms have been creating
proprietary software or research, which is already captured
in the national accounts as investment. Other firms have
been creating intangible investments in open source software, customer networks and other organizational capital.
Despite their long-run value, none of these intangible assets
are currently captured in the national accounts as investment. If we treat these asset categories as capital, then the
productivity boom from 1995 to 2000 becomes even stronger and the weak productivity growth of the past decade
may be ameliorated somewhat.
Working Paper 16–24. Leonard Nakamura, Federal Reserve Bank of Philadelphia Research Department; Jon Samuels,
Bureau of Economic Analysis; Rachel Soloveichik, Bureau of
Economic Analysis.
Localized Knowledge Spillovers: Evidence from the
Agglomeration of American R&D Labs and Patent Data

The authors employ a unique data set to examine the
spatial clustering of private R&D labs. Instead of using
fixed spatial boundaries, they develop a new procedure for
identifying the location and size of specific R&D clusters.
Thus, they are better able to identify the spatial locations
of clusters at various scales, such as a half mile, 1 mile, 5
miles, and more. Assigning patents and citations to these
clusters, they capture the geographic extent of knowledge
spillovers within them. Their tests show that the localization of knowledge spillovers, as measured via patent citations, is strongest at small spatial scales and diminishes
rapidly with distance.
Working Paper 16–25. Kristy Buzard, Syracuse University; Gerald A. Carlino, Federal Reserve Bank of Philadelphia
Research Department; Robert M. Hunt, Federal Reserve Bank

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 21

of Philadelphia Payment Cards Center; Jake K. Carr, Ohio
State University; Tony E. Smith, University of Pennsylvania.
Supersedes Working Paper 15–03.
Borrower Credit Access and Credit Performance After
Loan Modifications

While the preventive effect of loan modifications on
mortgage default has been well-documented, evidence on
the broad consequences of modifications has been fairly limited. Based on two unique loan-level data sets with borrower
credit profiles, this study reports novel empirical evidence
on how homeowners manage their credit before and after receiving modifications. The paper has several main findings.
First, loan modifications improve borrowers’ overall credit
standing and access to credit. Modifications that provide
principal reduction, rate reduction, or greater payment relief,
as well as those received by borrowers not in financial catastrophe, lead to a larger improvement in borrowers’ credit rating than others. Second, loan modifications lead to a slight
increase in borrowers’ debts, primarily on home equity line
of credit (HELOC) accounts and auto loans. Third, borrowers’ performance on nonmortgage accounts, however, has
not been negatively impacted by modifications. This study
demonstrates that interventions designed to improve household balance sheets could have a direct and sizable impact
on borrower financial outcomes.
Working Paper 16–26. Lei Ding, Federal Reserve Bank of
Philadelphia Community Development Studies & Education.
Identity Theft as a Teachable Moment

This paper examines how a negative shock to the security of personal finances due to severe identity theft changes
consumer credit behavior. Using a unique data set of linked
consumer credit data and alerts indicating identity theft, the
authors show that the immediate effects of fraud on consumers are typically negative, small, and transitory. After
those immediate effects fade, identity theft victims experience persistent, positive changes in credit characteristics, including improved risk scores (indicating lower default risk).
The authors argue that these changes are consistent with
increased salience of credit file information to the consumer
at the time of severe identity theft.
Working Paper 16–27. Nathan Blascak, Federal Reserve Bank of Philadelphia Payment Cards Center; Julia
Cheney, Federal Reserve Bank of Philadelphia Payment Cards
Center; Robert M. Hunt, Federal Reserve Bank of Philadelphia
Payment Cards Center; Vyacheslav Mikhed, Federal Reserve

Bank of Philadelphia Payment Cards Center; Dubravka Ritter,
Federal Reserve Bank of Philadelphia Payment Cards Center;
Michael Vogan, Moody’s Analytics.
Supersedes Working Paper 14–28.
Information Spillovers, Gains from Trade, and
Interventions in Frozen Markets

The authors study government interventions in markets
suffering from adverse selection. Importantly, asymmetric information prevents both the realization of gains from
trade and the production of information that is valuable to
other market participants. They find a fundamental tension
in maximizing welfare: While some intervention is required
to restore trading, too much intervention depletes trade
of its informational content. The authors characterize the
optimal policy that balances these two considerations and
explore how it depends on features of the environment.
Their model can be used to study a program introduced in
2009 to restore information production in the market for
legacy assets.
Working Paper 16–28. Braz Camargo, Sao Paulo School
of Economics–FGV; Kyungmin Kim, University of Iowa;
Benjamin Lester, Federal Reserve Bank of Philadelphia Research
Department.
Supersedes Working Paper 13–20.
Declining Trends in the Real Interest Rate and
Inflation: The Role of Aging

The authors explore a causal link between aging of the
labor force and declining trends in the real interest rate and
inflation in Japan. They develop a New Keynesian search/
matching model that features heterogeneities in age and
firm-specific skills. Using the model, they examine the longrun implications of the sharp drop in labor force entry in
the 1970s. They show that the changes in the demographic
structure induce significant low-frequency movements in
per-capita consumption growth and the real interest rate.
These changes also lead to similar movements in the inflation rate when the monetary policy follows the standard
Taylor rule, failing to recognize the time-varying nature of
the natural rate of interest. The model suggests that aging of
the labor force accounts for roughly 40% of the declines in
the real interest rate observed between the 1980s and 2000s
in Japan.
Working Paper 16–29. Shigeru Fujita, Federal Reserve
Bank of Philadelphia Research Department; Ippei Fujiwara,
Keio University, Australian National University.

22 | Federal R eserve Bank of Philadelphia R esearch Department | Fourth Quarter 2016

FedinPrint—a comprehensive index to Federal Reserve research
Visit Fed in Print
www.fedinprint.org/

The Fed in Print database is the central catalog of Federal Reserve System publications. It provides access to
research published by the Board of Governors and all 12 Federal Reserve Banks.
Browse or search materials, or keep up with the latest publications with RSS feeds.
Coverage includes:
•
•
•
•
•
•
•

The System’s core economic reviews.
Specialized regional and international publications.
Working papers.
Annual report essays.
Selected speeches.
Community development publications.
Conference papers.

Librarians throughout the Federal Reserve System contribute to the content. Older publications are added as they
are made available in full text.
Our aim is to make the database a vital tool for researchers, students, and all those interested in economics, finance,
banking, and community development.
Fed in Print has its roots in the 1960s, when the librarians at the Federal Reserve Bank of Philadelphia began
compiling and publishing an index to the Federal Reserve Bulletin and each Reserve Bank’s economic review. In
1996, the Federal Reserve Bank of San Francisco introduced Fed in Print as a web-based, searchable index. The
print version ceased publication in 2000. It has been managed and hosted by the Federal Reserve Bank of St. Louis
Economic Research Division since 2013.

Fourth Quarter 2016 | Federal R eserve Bank of Philadelphia R esearch Department | 23

1918

1942

Sign up to receive e-mail
notifications of every issue
of Economic Insights at
www.philadelphiafed.org/
notifications.

Back issues of the
Business Review remain
available on our website
at www.philadelphiafed.
org/research-and-data/
publications/businessreview.

1963

1989

2015

2016

PRESORTED STANDARD
U.S. POSTAGE
PAID

Federal R eserve Bank of Philadelphia

PHILADELPHIA, PA
PERMIT #583

Ten Independence Mall
Philadelphia, Pennsylvania 19106-1574

ADDRESS SERVICE REQUESTED

www.philadelphiafed.org

Read Economic Insights and other publications online at
https://www.philadelphiafed.org/research-and-data/publications/.
Past Business Review articles can be found at
https://www.philadelphiafed.org/research-and-data/publications/business-review.
Also on our website: data and analysis on the regional and
national economy, resources for teachers, community development
initiatives and research, and more.

EXPLORE

AND

LEARN