THE SNOW GLOBE ECONOMY - Federal Reserve Bank of ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
INTERVIEW: Kleinberg on
algorithms, bias, and democracy
SPOTLIGHT: Analyzing labor
market dynamics
DATA DIVE: Returns on
homeownership
SPRING 2022
THE MAGAZINE OF THE OPPORTUNITY & INCLUSIVE GROWTH INSTITUTE
THE SNOW GLOBE
ECONOMY
The pandemic has changed our
relationship to work and shaken up
the job-matching process.
2 UPDATE For All
CONTENTS
Conferences provide opportunity to cultivate the The magazine of the
Opportunity & Inclusive
Institute’s community and research agenda. Growth Institute
3 SCHOLAR SPOTLIGHTS Federal Reserve Bank
of Minneapolis
Four Institute scholars discuss their research on labor incentives,
Abigail Wozniak
trade shocks, social benefits programs, and parental safety nets. Director and Senior Research Economist,
Opportunity & Inclusive Growth Institute
6 THE SNOW GLOBE ECONOMY Amy Phenix
Frictions new and old have contributed to two years of labor Senior Vice President
market tumult. How has the pandemic changed our relationship to Dominick Washington
work and shaken up the employment matchmaking process? Assistant Vice President,
Strategic Communications
Paul Wallace
14 SOCIAL POLICY IN THE AGE OF ALGORITHMS Director, Strategic Communications
Cornell computer scientist and Institute advisor Jon
Andrew Goodman-Bacon
Kleinberg discusses how modern computing can Institute Senior Research Economist
reinforce or reveal biases in decision-making. Lisa Camner McKay
Writer/Analyst
22 RESEARCH DIGESTS Brandon Hawkins
Summaries of Institute research on the impact of parental Institute Research Assistant
income following a child’s job loss, how earnings risk Kuma Okoro
Institute Research Assistant
has changed in the U.S., and the impact of telework on
mothers’ decisions about work during the pandemic. Jeff Horwich
Senior Economics Writer
Alyssa Augustine
32 DATA DIVE Media and Content Strategy Supervisor
Economic uncertainty changes the homeownership
Tu-Uyen Tran
equation for Black households. Senior Writer
Alexis Akervik
34 FINAL THOUGHT Manager, Web & Creative
From William A. (“Sandy”) Darity Jr. Nina Leo
Creative Director
Allison Bertelson
Digital Designer
Kara Witzmann
Project Manager
Kristi Anderson
Art Director & Designer
INTERVIEW: Rucker Johnson
on school finance reform,
quality pre- , and integration
SPOTLIGHT: Addressing
bias in the classroom
DATA DIVE: A blueprint for
job training programs
FALL 2021
Neel Kashkari
THE MAGAZINE OF THE OPPORTUNITY & INCLUSIVE GROWTH INSTITUTE President & CEO
CHASING
OPPORTUNITY Mark L.J. Wright
Senior Vice President &
What shapes the chances we
get in our lifetimes?
Director of Research
Connect with us
www.minneapolisfed.org/for-all
Email: Mpls.ForAllEditor@
mpls.frb.org
Phone: 612.204.5000
Mail:
Get the very latest Institute For All
Opportunity & Inclusive
news and research—follow Growth Institute
us on Twitter @OIGInstitute. 90 Hennepin Avenue
Minneapolis, MN 55401-1804
Did you miss our Fall 2021 issue?
Twitter:
Visit minneapolisfed.org/for-
@MinneapolisFed
all to catch up on past issues.
COVER ART BY L.J. DAVIDS @OIGInstitute
Sign up to subscribe to this
free magazine and never miss
another issue.
FROM THE
DIRECTOR
BY ABIGAIL WOZNIAK
LIKE MOST AMERICANS, I raised my eyebrows more than a few times over the last
year while shopping for groceries, ordering from restaurants, or restocking my family’s
medicine chest. Price inflation was last a topic of household conversation when I was a
child, but it has returned as an important issue in the pandemic recovery.
The Institute’s mission is to support the Fed in its pursuit of full employment through
our research into ways to enhance opportunity and inclusion in the U.S. economy. To
do this, we constantly seek new detail on how the full range of U.S. households are
Changing
faring. This spring, that search has led us to explore the ways inflation puts
different pressures on workers and families depending on where they live,
what they earn, and how they spend. We devoted a virtual Institute event
research for
to understanding the varied impacts of inflation and discussing how policy-
makers might respond.
a changing
The Institute is built on the idea that we need to know more and share
more—through efforts like our spring event—to expand participation in our
economy and access to its benefits. This premise has caused some to ask
economy
whether efforts like ours “distract” the Fed from fighting inflation.
I frankly think that’s the wrong question. It supposes that having a narrow-
er sense of context and trade-offs could somehow lead to better decisions.
The U.S. economy is always evolving, sometimes slowly and sometimes
very quickly. The result is that our economy now looks very different from earlier,
formative periods like the late 1970s. For example, in 1980, the ratio of household
income for those just inside the top 20 percent to those just inside the bottom 20 per-
cent was about 4 to 1. On the eve of the pandemic, this was about 5 to 1. The risk of a
lasting earnings cut at some point in one’s career has risen, and it has become harder
to move from one employer to another. Despite these challenging trends, opportunity
has in other ways expanded. The likelihood of staying with an employer for a long
time (20 years or more) has steadily increased for women. Meanwhile, Black well-be-
ing broadly measured has advanced relative to that for Whites, despite a tide of rising
income inequality more generally and persistent wealth and earnings gaps. Further-
more, efforts to address inequalities via social programs have provided opportunity
that lasts a lifetime for some.
Rather than asking whether the Fed has the bandwidth to focus on inflation and
broader opportunity, a better question is: What information do we want our deci-
sion-makers to have when they make choices that affect so many? Since its found-
ing five years ago, the Institute has become a focal point for connecting with frontier
research into how our economy does, or doesn’t, work for all Americans. As such, it
is one effort—among others at the Fed—to provide as much context as possible. As
our spring event on the different ways that American households experience inflation
We want to
shows, this approach can complement a traditional concern like inflation and can
hear from you!
even provide new motivation for combatting it.
Research does little good if it is not shared—with policymakers, with other research-
ers, and with business leaders, community activists, technical innovators—everyone
who participates in the economy. We hope that For All provides a way for you, our
readers, to connect with our research in an engaging way. We’d like to hear from you
Please take our about what you find useful and interesting, and what you’d like to see more of. Please
reader survey. use this QR code to access our short reader survey and tell us what you think.
SPRING 2022 / FOR ALL 1
UPDATE
Growing the conversation
Conferences are an opportunity to cultivate community and research agenda
he Institute is always at work tute Director Abigail Wozniak and featured
shaping and sharing scholarship Assistant Director Alessandra Fogli and Senior
on economic opportunity and Research Economist Illenin Kondo as pan-
inclusion. Presentations and dis- elists. The AEA meeting is the largest annual
cussions at scholarly conferences gathering of academic economists in the U.S.,
provide one venue for these efforts. and the session provided an opportunity to dis-
Last fall, the Opportunity & Inclusive cuss best practices in research to better reflect
Growth Institute held its inaugural research and account for racial economic inequality.
conference. This was not the first Institute The Institute also supports the work and
conference, but it was the first to showcase development of the next generation of research-
the range of frontier-style research that the ers. To this end, the fall research conference was
Institute engages with in pursuit of its mission preceeded by a full-day mentoring workshop to
to conduct and promote scholarship that will hone projects by seven scholars who recently
increase economic opportunity and inclusive completed their Ph.D.s. Milena Almagro, a for-
growth for all Americans. mer Institute visiting scholar who presented
This is a broad goal, one that is served by at the mentoring conference, appreciated the
embracing breadth and diversity in researchers workshop’s focus on constructive feedback. “I
and research topics. Several presentations spoke think this ‘en petit comité’ [small group] format
to the ways monetary policy impacts different really encouraged questions, suggestions, and
dimensions of inequality. Others considered interactions,” Almagro said.
how market competition, laws, and government Promoting teaching and training within
enforcement actions affect discrimination in economics is one of the goals of the Mid-
specific spheres, research that helps illuminate west Economics Association, which held its
remedies for racial discrimination. annual meeting in March. Wozniak chaired
The conference keynote panel, “Race in the session “Doing Inclusion in Economics,”
Economic Research: From One Dimension to which drew inspiration from “Reaching Our
Many,” continued the Institute’s commitment Full Potential,” an Institute report that reflects
to examine how researchers can enrich and on broad themes and concrete actions that
expand their study of questions related to race emerged from last year’s “Racism in the Econ-
in economics. It is a conversation that Institute omy: Focus on the Economics Profession”
economists took to the American Economic event. Panelists discussed engaging students
Association’s (AEA) annual meeting in Jan- from a broad range of backgrounds, improv-
uary in a session titled “Incorporating Racial ing inclusion in journals’ editorial processes,
Inequality into Macroeconomic Models.” For- and researching pressing questions of eco-
mer visiting scholar Trevon Logan convened nomic inclusion.
the session, which was moderated by Insti- —Lisa Camner McKay
2 FOR ALL / SPRING 2022
SCHOLAR SPOTLIGHTS
The research community
at the Institute includes
visiting scholars, consultants,
economists, research analysts,
and research assistants. These
scholars bring a diversity of
backgrounds, interests, and
ROB VALLETTA
expertise to research that Senior Vice President and Associate Director of Research,
deepens our understanding Federal Reserve Bank of San Francisco
of economic opportunity FOLLOWING THE DATA
and inclusion as well as ON LABOR INCENTIVES
policies that work to improve
When he joined the Federal Reserve Bank of San Francisco in
both. We talked with four 1995, Rob Valletta remembers feeling “on the margins of the
of them about their work. conversation”—an applied microeconomist at an institution focused on
macroeconomic concerns.
Times have changed. Today, questions about worker and firm behav-
ior that have long fascinated Valletta have moved center stage. “Over
the last couple decades there’s been a shift in the Fed system to think
about how the economy operates on a microeconomic basis,” said Val-
letta, now a key policy advisor to San Francisco Fed President Mary Daly.
“The topics I’ve focused on in terms of labor market dynamics—
things that cause people to change jobs, what they think about when
they’re deciding to accept a job, how long they stay unemployed—are
crucial for understanding how the labor market works and the Fed’s
maximum employment goal,” Valletta said.
Valletta serves on the Institute’s System Affiliates “It’s common to think
Board, where he helps shape the Institute’s research about workers as
and conference agendas. Each member brings a career empty vessels. ...
steeped in topics related to inclusive growth.
But people are
For Valletta, that includes critical scrutiny of the
notion that government social supports dissuade people
smart. They know to
from working. From Medicaid expansion under the Af- plan for the future.”
fordable Care Act to extended unemployment insurance —Rob Valletta
(UI) benefits amid COVID-19, Valletta’s research has
found that while such “moral hazard” might feel intuitive and powerful,
the effect is often modest or missing in real-world data.
One lesson for economists and pundits: Circumstances matter.
“Unemployment insurance might have large disincentive effects in
normal times,” Valletta said. “By contrast, when the economy is weak,
unemployment insurance is likely to have a different impact. It’s a way of
keeping people alive—bridging them from an economic shock to a labor
market recovery.”
On the heels of an unprecedented expansion of U.S. jobless ben-
efits, Valletta’s ongoing analysis suggests workers were aware that the
supports were temporary and that skills could go stale if they stayed
out of work for too long. His findings, along with work by others, suggest
that the impact of the UI expansions on the labor market was limit-
ed. This research offers a needed reminder, Valletta said, that human
beings generally want to work—not coast.
“It’s common to think about workers as empty vessels, making deci-
sions without any meaningful, forward-looking thinking,” Valletta said.
“But people are smart. They know to plan for the future.”
—Jeff Horwich
SPRING 2022 / FOR ALL 3
SCHOLAR SPOTLIGHTS
RICARDO REYES-HEROLES MARIANNE BITLER
Senior Economist, Federal Reserve Board of Governors Professor of Economics, University of California, Davis
HOW TRADE SHOCKS RIPPLE VISIBILITY THROUGH POLICY ANALYSIS
ACROSS GENERATIONS
As a Ph.D. student in the University of Minnesota’s mathe-
Most economists agree that, on balance, free trade matics department in the mid-1990s, Marianne Bitler found
benefits the U.S. economy. They also agree that it creates herself thinking about politics and policy in addition to math.
winners and losers. Those who gain access to new markets “I was the person who would be listening to Fresh Air while
abroad thrive. Those undersold by overseas rivals suffer. doing my complex analysis homework and thinking, `I’d rath-
But that’s what happens at the start of a trade deal, er be studying that,’” she said.
when businesses and workers are still adjusting, according So, she moved first to the Federal Reserve Board as a
to Ricardo Reyes-Heroles, a senior economist with the research assistant and then to the economics department
Federal Reserve Board of Governors. He wanted to know at MIT, where policy-focused economists were talking about
what happens next. When new workers join the labor the controversial 1996 welfare overhaul.
force, surely they would look for work in thriving industries This reform restricted welfare benefits
and avoid declining ones? and encouraged low-income parents to
Yes, they would, but inequality work. Three senior Clinton administration
among new workers means some officials resigned in protest when it was
adjust faster to trade shocks than signed, arguing that the bill would hurt
others, according to an empirical children. “We were getting data, though,”
analysis and an economic model Bitler recalled, “that would let us answer
he and two co-authors developed questions about whether there were that
based on trade between the U.S. many more kids in poverty.”
and China. Over the last 20 years, Bitler, now a professor of eco-
Because most U.S. industries nomics at the University of California, Davis, has answered
that thrive on trade with China require college degrees— more questions about how welfare reform has and has not
education and financial services, for example—young succeeded than almost any other researcher. Her work has
workers are more likely to attend college after seeing how explored how welfare reform changed marriage, divorce, and
trade affected their parents’ generation. But young work-
living arrangements and hindered women’s access to medical
ers who can’t afford tuition will likely end up in low-wage
care. She has also spotlighted the varied economic experienc-
jobs for another generation.
es of women affected by policy changes. Connecticut’s pro-
“The wealth of your parents actually matters a lot,” said
gram, for example, increased average income yet made some
Reyes-Heroles, who presented a paper describing the
women poorer. And Bitler has shown how the shift toward
project at the Institute’s fall mentoring workshop.
safety-net programs available only to people who are working
His model predicts that workers who missed out on
leaves families more vulnerable to recessions, when they may
college will save so their children can go, allowing that
not only lose a job but also many important public benefits.
generation to finally gain from trade.
Bitler’s career also mirrors the trend toward visibility and
Growing up in Mexico City as the North American Free
inclusion of LGBTQ+ research and researchers in economics.
Trade Agreement (NAFTA) went into effect, Reyes-Her-
Bitler, a bisexual woman, recalled when she was a graduate
oles said he was exposed to discussions about trade
student that “there were queer people in economics…but
between his economist father and his father’s colleagues.
research was hamstrung by the fact that you couldn’t identify
One of the inspirations for his latest paper is how NAFTA
someone’s sexual orientation or gender identity or sexual be-
was initially greeted in Mexico. People complained that
havior or sexual attraction in data.” As datasets improved and
only border regions benefited from new factories, while
regions farther south lost farm jobs. By the time his gener- laws relating to queer families changed, research on econom-
ation entered the labor force, though, few thought NAFTA ic outcomes for LGBTQ+ people, such as family dynamics,
was a bad thing, he said. income gaps, and wage/hiring discrimination, grew—and so
It’s important to consider these multigenerational tran- did the community of queer economists.
sitions, Reyes-Heroles said. Looking at just one generation The maturation of research and inclusivity in the profession
makes it seem like the benefits of trade will never reach ev- need not go hand in hand, Bitler emphasizes, but they can.
eryone. It also obscures potential policies to limit hardship Mentoring sessions organized by a new American Economic
for those suffering losses. One policy to consider, he said, Association committee and the ability—helped by the necessity
is subsidizing college tuition for those harmed by trade, of Zoom seminars—to meet regularly with LGBTQ+ econo-
which could shorten the transition period by a generation. mists, Bitler said, “have been huge for community-building.”
—Tu-Uyen Tran —Andrew Goodman-Bacon
4 FOR ALL / SPRING 2022
2021-22 Institute
Visiting Scholars
The Institute annually invites selected
scholars from many disciplines to pursue
research while in residence at the
CORINA BOAR Minneapolis Fed.
Assistant Professor of Economics, New York University
Marianne Bitler
PARENTAL SAFETY NETS, Professor of Economics, University
PLEASURABLE JOBS of California, Davis
Corina Boar
Assistant Professor of Economics,
As a student in her native Romania, Corina Boar was advised New York University
by some of her instructors that a degree in economics was a Sarah Cohodes
useful way to secure a job. Her reasons were different. Associate Professor of Economics and Education,
“During my undergraduate studies, other professors Teachers College, Columbia University
talked about economics with passion and humility,” Boar Jamein Cunningham
said. “They instilled in me a curiosity that went beyond the Assistant Professor of Policy Analysis and
Management and Economics, Cornell University
practical aspects of economics that are needed to get a job.”
Boar’s research interests as an economist were influ- Diego Daruich
Assistant Professor of Finance and Business
enced by a later conversation with her father. She men- Economics, Marshall School of Business,
tioned that earning a Ph.D. typically takes longer than five University of Southern California
years—the amount of time her pro- “People talk a lot Johannes Fleck
gram offered funding. Her final year Economist, Federal Reserve
about equality of Board of Governors (2022)
would have to be paid out-of-pocket.
“I’ll set some money aside,” her opportunity, but I have John Grigsby
father said. And that was that. an interest in equality Assistant Professor of Economics and
Public Affairs, Princeton University
Recognizing her own freedom of well-being and how
to pursue her passion prompted a Ayşe Imrohoroğlu
that intersects with Professor of Finance and Business
research question: In what ways do
economic outcomes.” Economics, Marshall School of Business,
parents affect their children’s labor University of Southern California
market outcomes? Boar’s research —Corina Boar
Ilse Lindenlaub
demonstrates that people with higher-income parents are Associate Professor of Economics, Yale University
literally able to afford jobs that people from lower-income Emi Nakamura
families cannot. Chancellor’s Professor of Economics,
High-income parents tend to beget high-income chil- University of California, Berkeley
dren. However, there are also nonmonetary benefits of some Emily Nix
occupations that children of wealthy parents can trade off Assistant Professor of Finance and Business
Economics, Marshall School of Business,
against a higher income.
University of Southern California
“When you choose your career, you’re going to try to bal-
Eric Ohrn
ance two things: how much money I make versus how much Associate Professor of Economics, Grinnell College
I like my job,” said Boar. With co-author Danial Lashkari, Boar
Jane Olmstead-Rumsey
finds children of rich parents are more likely to pursue jobs Assistant Professor of Economics,
with high “intrinsic quality.” These have higher levels of au- London School of Economics (2022)
tonomy, respect, and control, and require less physical effort. Monika Piazzesi
Occupations high on intrinsic value include post-secondary Joan Kenney Professor of Economics,
teacher, architect, writer, artist, entertainer, and athlete. Stanford University
With family wealth as a backstop, children have the free- Martin Schneider
dom to make less money but be happier in their work (or to Professor of Economics, Stanford University
gamble on, say, a long-shot acting career). Children from less Benjamin Schoefer
privileged backgrounds face a bigger financial risk to follow a Assistant Professor of Economics,
University of California, Berkeley
passion heedless of the paycheck.
Boar’s work takes on a challenge in traditional economics Chelda Smith
Associate Professor of Elementary Education,
to recognize that utility means more than dollars and cents College of Education, Georgia Southern University
and to create models in which people respond to nonmone-
Jón Steinsson
tary incentives. Chancellor’s Professor of Economics,
“People talk a lot about equality of opportunity, but I have University of California, Berkeley
an interest in equality of well-being and how that intersects Mallika Thomas
with economic outcomes,” said Boar. “Your chances of be- David M. Rubenstein Fellow, Economic
coming what you want to become shouldn’t depend on how Studies, Brookings Institution
rich your family is.” Christopher Tonetti
—Alyssa Augustine Associate Professor of Economics, Graduate
School of Business, Stanford University
Robert M. Townsend
Elizabeth & James Killian Professor of Economics,
Massachusetts Institute of Technology
SPRING 2022 / FOR ALL 5
THE
SNOW
GLOBE
ECONOMY Two years of COVID-19 have upended
our world of work. When—and
where—will we come down?
BY LISA CAMNER MCKAY
AND JEFF HORWICH
Big economic moments deserve a proper name.
Some have labeled this one “the great reallocation.”
Another way to put it: COVID shook up our
economy like a snow globe. Workers and businesses
are not gliding gently back to their old positions:
Work environments, job responsibilities, child care,
life priorities—all have been set swirling by the
pandemic experience.
Even in the before-times, the process for employ-
ees and employers to find the right fit was an ordeal
for both sides. Researching, applying, and negotiat-
ing take effort and time, oftentimes unfruitful. Unlike
ILLUSTRATIONS
most markets, in which just one side of a transaction is BY L.J. DAVIDS
6 FOR ALL / SPRING 2022
SPRING 2022 / FOR ALL 7
making a choice (consumers care which The extreme signals from the data tell
restaurant they eat at; restaurants do us the labor market is in a high-friction
not generally care which consumers eat moment. Last year saw a record 47 mil-
there), the labor market is complicated lion people quit their jobs, the highest
by “two-sided differentiation,” said Uni- since the Bureau of Labor Statistics start-
versity of Minnesota labor economist ed collecting the data in 2000. Mean-
and former Institute visiting scholar Aar- while, the number of people who have
on Sojourner: Employers and workers taken themselves out of the labor force
both size up each other, and either can remains elevated, particularly for people
scuttle the deal if the match isn’t right. over age 55.
The employment process is naturally At the same time, there have been an
full of frictions, which Sojourner defines extraordinary number of job openings,
as anything “that gets between you and nearly 11 million at the end of 2021—
the best job out there for you.” The term more than the number of people looking
“frictional unemployment” refers to the for work. The Beveridge Curve shifted
inevitable share of people who are navi- outward into unknown territory and
gating this matchmaking process. took on an unfamiliar vertical shape (see
A wild ride for workers along the COVID-19 Beveridge Curve
The Beveridge Curve (named in honor
Oct 21 Jul 21
of British labor economist William 7% Dec 21
Beveridge) shows the relationship Jun 21
Nov 21 Sep 21 Aug 21
between the unemployment rate
and the job vacancy rate, and how 6%
May 21
Apr 21
that relationship changes over time.
Each dot represents one month. The Mar 21
cluster of blue dots to the left show Feb 21
each month from January 2001 to 5%
Oct 20 Jul 20
March 2020, displaying an intuitive, Jan 21 Sep 20
Job vacancy rate
negative relationship: When job Nov 20/
Aug 20
Jun 20
Dec 20
openings are high, unemployment 4% May 20
is low, and vice-versa. The red line Apr 20
traces the curve since the start of the Mar 20
pandemic, starting with almost 15%
3%
unemployment in April 2020. The
COVID-era curve has moved into
territory unseen in modern times.
And in 2021, it turned nearly vertical 2%
as job openings soared to the highest
level in decades, but workers declined
to fill them—or added to vacancies
1%
by quitting in record numbers. 3% 4% 5% 6% 7% 8% 9% 10% 11% 12% 13% 14% 15%
Unemployment rate
Jan. 2001–Feb. 2020 COVID era (so far)
Source: U.S. Bureau of Labor Statistics
8 FOR ALL / SPRING 2022People appear to gain
valuable new information
about employment
options when colleagues
quit to look for a new job.
As a result, quitting seems
to beget more quitting.
spending patterns, and altered firms’ use
of technology, all of which have impact-
ed which businesses are more and
less productive—AMC Theatres versus
Netflix, say, or a restaurant with a large
dining room versus a takeout counter.
As a result, “there’s another firm out
with labor demand and labor supply there now that is more productive and
adapting to the COVID economic shake- can make better offers” to prospective
up, one with no recent precedent, it can employees, Sojourner said.
be tricky to answer that eternal, nagging But how people learn about these new
figure): Unemployment stopped falling question: Are we there yet? opportunities depends upon “informa-
while openings continued to rise. This tion frictions”—the fact that information
imbalance is driving up wages, particu- about job openings, employer quality, and
larly for lower-wage jobs. That there are so You can’t fire me—I quit! even wages is not easily or equally avail-
many job openings and so many job seek- Quitting has been having a moment. able. Research by Institute visiting schol-
ers implies there are matches to be made. Quits tend to naturally rise with tight ar and University of California, Berkeley
That they aren’t happening, or aren’t hap- job markets. But “they’re not just high,” economist Benjamin Schoefer shows that
pening quickly, suggests that frictions, old said economist Steven Davis of the Uni- people don’t seem to have a good sense
and new, are getting in the way. versity of Chicago Booth School of Busi- about how their compensation com-
What do these unusual trends say ness. “They’re higher than any period in pares to workers in similar jobs at other
about how the labor market itself has the history of the data.” employers—whether you are paid well or
changed? What do they say about us, The reasons for this churn are likely poorly compared to your peers, you likely
as workers? What is the endgame of both structural—where will the swirling think you’re close to the middle.
the great reallocation? Understanding snowflakes fall?—and informational— “If you get stuck in a low-wage job,
is critical for the Federal Reserve as it how long will it take them to land? Struc- you might think all jobs are low-wage
sets policy to support the conditions for turally, the pandemic introduced major jobs and therefore you never switch,”
reaching maximum employment. Yet health concerns, changed consumer Schoefer said. But if these workers are
SPRING 2022 / FOR ALL 9pushed to search for new jobs—as many
who were laid off or left their jobs during
the COVID recession did—they learn,
which may lead them to seek out and
obtain higher wages in the future.
In addition, Schoefer’s research sug-
gests, people appear to gain valuable new
information about employment options
when colleagues quit to look for a new
job. As a result, in the near term at least,
quitting seems to beget more quitting—a
kind of multiplier effect. “One of the top
things in the news—and in everyday
conversation—is people talking about,
‘Oh, wow, employers are really bidding
up wages and a lot of people are quitting
their jobs,’” said Nick Bunker, director of
economic research for the Indeed Hiring
Lab. “People might sort of say, ‘Wait a
minute—let me think about this.’”
Power to the people?
In the quits-rate and other aspects of
the tight, post-pandemic labor market, it
is tempting to see a shift in power from
employers to employees. “I do think
what we’re seeing right now is a tilting of
the bargaining table more toward work- and the higher wages that resulted—will
ers,” said Bunker. “There’s more power tip the scales in favor of automation,
for job-seekers because of the kind of predicts economist Andra Ghent of the
outside options they have. If you’re an University of Utah’s David Eccles School economist Arie Kapteyn of the Center
employed person…you can go to your of Business. for Economic and Social Research at the
current employer and say, ‘Hey, look, “This technology was available prior University of Southern California, which
there’s all this demand out there and all to the pandemic, but for firms it wasn’t runs the ongoing Understanding Coro-
these people are quitting their jobs. It’d cost-effective to invest in it. And now it is,” navirus in America survey. Health risk
be a shame if I left!’” Ghent said. “Long term, this is not good is mediated not only by interaction with
Household savings surged during the news” for many lower-wage workers. others but also by employer decisions:
pandemic, thanks to government stim- Have they put a mask mandate or a vac-
ulus and lower household spending. cine mandate in place? These concerns
The feeling of extra money in the bank Rethinking our relationship to work add to the criteria that job seekers and
could provide a temporary wealth effect, Wages are not the only piece of our work prospective employers must match on,
empowering and emboldening workers. lives that the pandemic put in relief. increasing job search frictions.
The increased bargaining power of Most obviously, the pandemic caused But health concerns are not the only
workers may be short-lived, however, a newfound awareness of health risks driver of new expectations, Kapteyn
especially where automation (essential- on the job. “It turns out to be hard to noted. “Another story is that people are
ly, substituting capital for people) is an fill a number of jobs that require daily reevaluating their lives: Is this really
option. Ironically, the pandemic itself— or intensive contact with others,” said what you want to do?”
10 FOR ALL / SPRING 2022Health concerns are not the only driver of
new expectations, Kapteyn noted. “Another
story is that people are reevaluating their
lives: Is this really what you want to do?”
of it might give you a new perspective on nia’s Bay Area, for example, in hopes of
alternatives that life has to offer—pos- a remote-working life in Boise. A more
sibly with one alternative being idle or substantial shift was the movement away
unemployed.” from city centers into the suburbs. While
Loewenstein worries about a dark- renters are more mobile than home-
er effect of pandemic unemployment, owners, physical moves are not quickly
with long-term consequences for the job undone, leaving a sticky situation where
market: “I suspect that there is a massive jobs and workers are not in the same
mental health crisis that we’re not fully places as office work returns.
aware of.” Even before the pandemic, “What might have looked like an
Loewenstein said, going back to work attractive job when I only had a 20-minute
after an extended absence “was a very commute, now doesn’t look so attractive
daunting prospect for a lot of people—a if it’s a 75-minute commute,” noted Davis
lot of insecurity about whether they had of the University of Chicago. He says the
the right skills. The pandemic has led to “spatial mismatch” works both ways,
this re-entry issue on a mass scale.” with employees in the suburbs who don’t
For some, living through the pandem- want to go into the city and city-dwellers
ic has shifted the place of work and earn- finding that service jobs have followed
ings in our priorities. By introducing more white-collar workers into the suburbs.
Many already-tough jobs were made family time and life without a commute, For jobs with a remote-work poten-
more unpleasant by the pandemic, the pandemic could have altered the val- tial, the dance between employers and
said RAND economist Kathryn Anne ue people place on leisure, said Ghent, workers is far from over. The pandemic
Edwards. Hospitality workers have had to who studies work-from-home trends. We was a historic inflection point—similar,
enforce mask mandates. Retail workers are more willing to step onto a different said the University of Utah’s Ghent, to
do more cleaning. Restaurant staff spend path, “willing to say, long-term, maybe the telephone or email reaching criti-
time bagging take-out orders (for lower this isn’t going to increase my wages as cal mass. “These technologies have this
tips) and contend with surly customers. much, but I won’t put myself on the same characteristic of a network externality,
Nurses, bus drivers, substitute teachers trajectory to have the big increase in pro- where the benefit of them depends upon
… the altered nature of many jobs may ductivity later on [in my career].” how many other people are using them,”
lead workers to hold out for alternatives. Ghent said. “Zoom was not invented in
In addition, a “status quo” effect may 2020. But it wasn’t appropriate for an
be at work among the historically large Commuting and cubicles? accountant to say to a client, ‘Hey, you
number of people who have been out No, thank you. want to just meet over Zoom?’”
of work for an extended time, explained People did more than change their Ghent believes the productivity gains
behavioral economist George Loewen- minds during the pandemic. They from crossing this threshold—includ-
stein of Carnegie Mellon. “Continuing to changed ZIP codes. ing eliminating hours of nonproductive
work in an occupation is very different Some workers moved to an entirely commuting time—will be a “win-win”
from re-entering it,” he said. “Being out different metro area—leaving Califor- for workers and employers overall. The
SPRING 2022 / FOR ALL 11While renters are more
mobile than homeowners,
physical moves are not
quickly undone, leaving a
sticky situation where jobs and
workers are not in the same
places as office work returns.
University of Minnesota’s Aaron Sojourn-
er points out that it could also reduce job
search frictions, allowing both workers manage the organization, how you cul-
and companies to search nationally, not tivate cultural values, how you transmit
just locally, which should increase the knowledge from more-experienced to
quality of employee-employer matches. less-experienced workers.” We are still in
But for this to happen, employees the thick of experimenting, bargaining,
and employers must strike a truce on the and self-sorting our way to a new equi-
right amount of work-from-home. Davis’ librium around remote work.
ongoing Survey of Working Arrange-
ments and Attitudes finds employee and Post-COVID inflation adds another
employer expectations are converging, A hard(er) bargain wrinkle—but also a possible solution to
but workers still expect a nearly full day With workers dreading a commute and the stickiness of wage decreases. “Smart
more at home each week, on average, eager to preserve newfound work-life employers can probably get away for the
than employers. Nearly 40 percent of balance, negotiating a pay and benefits next couple of years with offering lower
recent, college-educated job-quitters, package just got more complicated. Wage wage increases than they might otherwise
surveyed in the fall, said they did so to bargaining is already prone to frictions offer,” said Davis. “It’ll be a lot easier to get
obtain greater time working from home. caused by asymmetric information; by away with a wage hike below the inflation
One economic diagnosis for the mis- the fact that negotiation typically happens rate, if you at the same time allow workers
match in expectations: Some diffuse and only annually; and by the practical reality to work from home part of the week.”
long-term benefits to office-time for the that wages are hard to adjust downward. Remote work is not an option for all
organization—innovation, collaboration, Now the pandemic has introduced jobs, of course. In particular, many low-
mentoring—do not factor into workers’ a new element into the compensation er-wage jobs in retail, hospitality, and
day-to-day calculations of costs and picture. Employees and job seekers want manufacturing must be done on site.
benefits. Loewenstein, the behavioral flexibility; but at what cost? Davis’ ongo- These are jobs at which workers tradi-
economist, suggests workers might also be ing Survey of Working Arrangements and tionally have little negotiating power
ignoring long-term benefits to themselves, Attitudes finds employees now work- over wages; they have also experienced
which he labels an “internality” problem. ing from home part of the week would anemic wage growth over the past 20
“These kinds of changes occur slow- require a raise of more than 8 percent years. Now, however, wages in a number
ly within organizations,” Davis said. “It to compensate them for returning to the of low-wage sectors are the fastest-grow-
requires a profound shift in how you office full time. ing in the economy. These jobs cannot
12 FOR ALL / SPRING 2022viders were also more likely to transition
out of employment entirely. Research by
Edwards shows that the more kids there
are in a household, the larger the decline
in mothers’ labor force participation.
Unfortunately, child care frictions
stoked by the pandemic remain elevated.
The child care sector is roughly 10 per-
cent smaller than it was before, Cascio
reports—and as many parents can attest,
availability was a challenge even in 2019.
Lower supply, compounded with post-
COVID approaches to cleaning and
crowding, will tend to make child care
even more expensive than the $15,000-
$20,000 that quality centers typically
cost before the pandemic. Schools have
reopened, but children are still subject to
unpredictable quarantines and closures.
mothers of young children to work in And while most working parents feel a
the formal labor market,” said Elizabeth child care crunch, the burden doesn’t fall
Cascio, an economics professor at Dart- evenly. Ultimately, parents with greater
mouth College. resources are better able to solve their
When the pandemic sent not just child care needs, stay in the labor force,
young children but all children home, be more productive, earn more money,
be compensated with greater flexibility, “in terms of the time parents had to and invest more in their children, U.S.
and in many cases the job responsibil- devote to these child care needs, it was Census Bureau economist Misty Heg-
ities carry both new risks and burdens. almost like having another newborn,” geness explained, making affordable,
By not accepting the old, low rates of observed Gema Zamarro, professor of accessible child care an equality issue.
pay, workers are exerting influence to economics and education reform at the (See “The Great Balancing Act,” page 26.)
increase wages. University of Arkansas. Mothers were With some 10 million open jobs, it’s
far more likely than fathers to provide also a matter of economic growth. Speak-
this care, even when both parents were ing to CBS’s “Face the Nation” about the
The child care challenge employed, according to Zamarro’s anal- challenge of finding child care, Minneap-
Flexible hours and hybrid work help ysis of survey data: In the spring of 2020, olis Fed President Neel Kashkari said, “It
keep parents in the labor force. But try- one in three working moms reported does have an effect on women’s participa-
ing to juggle work and children during being the sole provider of child care for tion in the labor force and how high our
the pandemic has proven they are no their children, compared to one in 10 labor force participation is as a whole.”
substitute for reliable child care. working dads. And this imbalance had Like the tug-of-war over remote work,
Child care has long been a friction grown even larger by fall 2020. workers’ willingness to quit, and our
for parents of young children, especially But the days didn’t get any longer. shifting relationship to our jobs, “These
mothers, affecting both whether and how “Being the only one providing child care [child care] challenges have been exac-
much they work, RAND’s Edwards said. in the spring of 2020 is associated with a erbated in the pandemic,” Kashkari said.
With limited access to paid family leave, 20 percentage-point increase in respon- “Long term, this is an important eco-
subsidized child care, or part-time work dents saying, ‘I had to reduce working nomic growth issue and competitiveness
schedules, “We haven’t made it easy for hours,’” Zamarro reported, and sole pro- issue for the country.”
SPRING 2022 / FOR ALL 13INTERVIEW 14 FOR ALL / SPRING 2022
SOCIAL
POLICY
IN THE AGE OF
ALGORITHMS
Institute advisor Jon Kleinberg on engaging
in dialogue with our technology to confront
hidden bias and make smarter choices
BY JEFF HORWICH
PHOTOS BY HEATHER AINSWORTH
SPRING 2022 / FOR ALL 15YOU MIGHT THINK
JON KLEINBERG, as one of the world’s
leading authorities on computer algorithms, would only surf
the web in “incognito” mode.
“I have colleagues who turn all that stuff off, and that’s a
very reasonable decision,” said Kleinberg. “But I personal-
ly found it was simply hard to navigate the world online if I
had everything switched off.” And of course, each click is an
opportunity for a prolific researcher. “I tend to try to figure out
why I’m seeing what I’m seeing. What am I going to see in the
future as a result of taking a given step?”
Use the technology; learn from the experience; adapt and
repeat. Kleinberg brings the same philosophy to his wide-rang-
ing professional work—and a certain amount of faith: With all
he knows about the risks and promise of advanced comput-
ing, he believes we can ultimately employ it as a force for good.
Kleinberg is the Tisch university professor of computer sci-
ence at Cornell University. Some of his earliest research in the
late ’90s laid the conceptual groundwork for the now-domi-
nant Google search engine. He helped establish the modern
study of networks—the science of interconnectedness and
spread, whether of ideas, illnesses, or financial panics. Along
the way, his career has traced the rise of the internet, social
media, and the unseen strings of ones and zeros that now per-
meate many aspects of life. We’ve had algorithms for much longer than we’ve had
Kleinberg is in familiar territory as an advisor to the Oppor- computers. I think that’s important because these terms have
tunity & Inclusive Growth Institute, having partnered often a way of isolating the concept, making it seem somehow weird
with economists—as well as sociologists, doctors, and legal and distinct from the rest of our lives, but it’s really blended
scholars. He dove deep with For All into a potent research through our lives. Any time a person, an organization, or a
focus: the potential for modern computing to reinforce our machine carries out a structure or procedure to solve a prob-
biases, but also to reveal them. This can help us make smarter lem, they’re running an algorithm.
economic and social policy—if we are willing to truly listen to
what our computers are telling us. You’ve done a lot of thinking about how algorithms—which
have no soul or opinions of their own, as far as I’m aware—
can be biased. What’s an example of how that happens?
ALGORITHMS: A GATEWAY TO OUR HIDDEN BIASES The simplest way would be that you have a procedure that was
Thanks in large part to Facebook and the last two U.S. made up by people who had a bias that they were trying to act
presidential elections, the word “algorithm” has become a on. And now, all the algorithm is doing is formalizing that bias
household term. What do you like to give people as a gener- procedure.
al, working definition? But I think the more subtle way this happens is with a large,
I think of an algorithm as any procedure that’s structured and emerging category of algorithms that have become quite pow-
that can be followed to solve a problem. Your GPS, when it erful over the past 20 to 30 years, called “machine learning
wants to find the shortest drive to your destination, uses an algorithms.” The idea with machine learning algorithms is that
algorithm to do that. Addition is an algorithm. Long division is there are a lot of problems that we want to solve that we don’t
an algorithm. There are a lot of analogies between algorithms actually know how to write down the rules for. We, as humans,
and recipes that we use in cooking. can solve them. But we don’t really know how we solve them.
16 FOR ALL / SPRING 2022WE’VE HAD ALGORITHMS FOR MUCH LONGER THAN WE’VE HAD COMPUTERS.
ANY TIME A PERSON, AN ORGANIZATION, OR A MACHINE CARRIES OUT A STRUCTURE OR
PROCEDURE TO SOLVE A PROBLEM, THEY’RE RUNNING AN ALGORITHM.
The problem comes in when the rule that it’s learning may So, we think that we are removing the human element,
have our own biases encoded into it. For example, people perhaps, by using a machine learning algorithm. But that
who read résumés make decisions about which ones look like algorithm is learning things that we didn’t even know about
strong résumés and which ones don’t. We wouldn’t know how ourselves—and formalizing them.
to write down a step-by-step procedure for that, but we can Right. In addition to thinking of the algorithm as producing a
feed the results to a machine learning-styled algorithm. The tool, it is also producing a diagnostic.
algorithm will now try to learn a rule that distinguishes the The algorithm almost becomes like an experiment, which
résumés that look strong from the other ones. I can probe. I can create synthetic job applicants. I can run
This is where bias sneaks in. We have several decades of them through the algorithm. I can say, “Okay, what if I change
research from behavioral sciences that when people look at a it slightly this way? What happens?”
résumé, a huge amount of their own implicit bias comes into With a human being, if I ask them, “Would you still have
the process. The algorithm—which knows nothing about the hired this person if they had gone to school X instead of school
world—now just knows, “These are the strong résumés and Y?” the person might make their best effort to give you an
these aren’t.” It’s just trying to tell you a rule that faithfully answer to that. But they can’t really know what they would’ve
describes your behavior, but your behavior was biased. It finds done in that situation. With an algorithm, we can change the
exactly the ways in which you’re biased, and it reproduces them. input from Y to X—what school the applicant went to, for
SPRING 2022 / FOR ALL 17THERE IS A SENSE IN WHICH WE HAVE A MUCH BETTER CHANCE OF UNDERSTANDING
THE PIPELINE OF DECISION-MAKING WHEN IT’S PASSED TO AN ALGORITHM
THAN WHEN IT’S A HUMAN BEING.
instance—and we can just feed it back through. And we’ll are we trying to achieve through the allocation of a stimulus?
learn something. We could have some aggregate measure of economic activi-
There is a sense in which we have a much better chance ty that we’re trying to promote. We could try to maximize the
of understanding the pipeline of decision-making when it’s number of people that we bring above some threshold that
passed to an algorithm than when it’s a human being. we’ve defined.
Those are human decisions that, in a democratic society,
the policy process has to actually arrive at a conclusion on.
COMPLEXITY, TRANSPARENCY, AND DEMOCRACY I almost think of the role of the algorithm, or of mathemati-
You recently authored a paper with one of your comput- cal models, as a counterparty in a dialogue about how to set
er science grad students looking at how the government objectives, how to set thresholds.
could use algorithms to more fairly and efficiently allocate You go to your model and you run a counterfactual simula-
stimulus checks. That’s a very timely question, and I will tion. You say, “What if we tried this, what would happen?” And
freely admit that—not being in the field—I could not begin then you see what happens, at least within your model. There’s
to understand the findings! this back-and-forth dialogue, where, in a sense, the compu-
The complexity of it made me wonder about how we tational model is giving you some clarity on the downstream
bridge the gap between making smarter, more efficient consequences of choices that you might make.
policy decisions, and still having the voting public under- The algorithm is telling us things that are very, very hard
stand and have faith in what’s going on. How do you see to figure out. Like, when I allocate [financial] assistance to
that balance being achieved? a particular part of the system, everything is connected in
Those are great questions and big challenges. The question of some kind of network of transactions. It’s like I poke this
making policy decisions that are informed by complex mod- spiderweb of transactions and the whole thing ripples and
els and large amounts of data—that’s a problem that began it spreads out in all sorts of different directions. Algorithms
before the widespread use of computing and algorithms, with are very good at helping you figure out what all those ripples
the introduction of large-scale mathematical and statistical will look like. But then it’s up to you to decide what it is you’re
models into policymaking. actually trying to accomplish.
But the introduction of machine learning algorithms in
computing takes us one step further because it allows us to And going through that process of querying the model,
deal with models that are, in some sense, inscrutable even to calibrating it—that itself is potentially a way to build public,
their developers. We can actually be looking at the answer, transparent faith that you are meeting whatever goals
right in front of us—we have this computer code that is doing society sets out.
this thing that we don’t know how to do, and we can’t say That’s how we hope the process works. The algorithm’s com-
how it’s doing it. It’s a profound challenge, and it’s still a very putational models are one participant in that process—and it’s
new area—the area of interpretability and explainability of a process with many participants.
machine learning algorithms.
In research with economist Sendhil Mullainathan, you make
Sticking with the example of stimulus checks: A simple the point that when it comes to algorithms, simplicity and
solution everybody can understand is to give everyone the fairness can be fundamentally inconsistent with one anoth-
same amount of money—maybe subject to some basic rules er. That sounds like a very frustrating finding. What are we
and cutoffs. I believe your paper’s point is that complex supposed to do with that knowledge?
algorithms could help target the assistance better, which We know that, as humans, if we’re operating under conditions
would be a more efficient use of taxpayer money. But that of low information or rapid decision-making, that is when
runs into problems in a democracy. How can we make people are prone to fall back on stereotypes—and often perni-
things more efficient, without just saying, “We just have to cious stereotypes that work to the detriment of people who are
trust the robots to get it right”? already at a disadvantage.
It’s a great example to work through because the first question If we take a complex model—let’s say there are thousands
you come to is one you can’t derive using an algorithm: What of pieces of information we might have about a person, and
18 FOR ALL / SPRING 2022LINDSAY FRANCE / CORNELL UNIVERSITY
we could simplify it by using only a few pieces of information. We’re now in a new kind of situation that becomes slightly
What we found in this work was that when you start removing precarious. Let’s say all the different firms in an area are all doing
the information available to an algorithm, it begins to do things a first-pass screening of résumés using the same algorithm. First,
that resemble the human process of falling back on stereotypes. if the algorithm just doesn’t like your résumé for some reason,
What this tells us is that we should be alert to opportu- you no longer have a chance for recourse or a second opinion. If
nities to strategically “un-simplify” our models in certain one doesn’t like you, then they’re all not going to like you.
targeted ways. There are many reasons to prefer simple mod- Second, if conditions change, then we could all suddenly
els: [More complex algorithms] are inscrutable; they are not start making the same set of mistakes. For example, maybe
really amenable to collaborative decision-making or refining. this is an algorithm that’s evaluating loan applications, and
But the question sometimes is: Are there ways in which, in a the underlying economic conditions change. Maybe this mod-
limited, targeted way, we can expand the models in ways that el was trained pre-pandemic, when the meanings of certain
deliberately address the dimensions where it seems to be fall- things in your financial history just look different. Then, all of
ing back on stereotype-like heuristics? a sudden, all of these algorithms are now making mistakes in
the same way because they’re operating in an environment
that they weren’t trained on.
THE ARGUMENT FOR DIVERSITY…OF ALGORITHMS These are things that become much more acute risks
Here’s another term for us—it’s kind of a mouthful: “algo- now that we have the ability to really replicate our decisions
rithmic monoculture.” What is that, and what is the danger through computing.
it can pose?
The term monoculture comes from agriculture, where if you
plant the same plant species across all of your fields, it’s at risk to COMPUTER SCIENCE MEETS ECONOMICS
being eliminated by a single pathogen that can sweep through You have research examining the power of algorithms to
the whole thing, or by a single change in weather conditions. improve the way that we distribute welfare payments, or to
Suppose that we begin introducing algorithms for some improve intergenerational mobility. You get into one of the
problem that is very complicated and that humans struggle staples of behavioral economics, looking at sunk cost bias. As
with: medical diagnosis, evaluating loan applications, eval- a computer scientist, what are you and your field bringing to
uating résumés. Maybe we could even demonstrate that we these economic and social questions—and to the Institute?
have made the system more accurate, or we have reduced the I’ve gotten a huge amount of benefit, over my whole career,
amount of bias or disparity in the system. from working with economists and social scientists. What
SPRING 2022 / FOR ALL 19You can also read
