RESEARCH-DRIVEN INSIGHTS ON BUSINESS, POLICY, AND MARKETS SPRING 2021

Plus: What should the US do about student debt?

Lower fines could lead to higher revenues Law and order and data

Will algorithms fix what’s wrong with American justice, or make things worse? “People are going to be returning to jobs, but they’re going to be different kinds of jobs and different kinds of employers.” Page 62

2 Chicago Booth Review Spring 2021

285697_1-29.indd 2 2/11/21 2:37 PM MORE ACCURATE he US prison system is the biggest in The answer depends not only on T the world. At 655 prisoners for every coders, but principally on the key 100,000 people living in the country, decision makers in cities, states, police CRIMINAL JUSTICE the United States leads the world in both departments, and other agencies. They total number of people in prison and will both select what technologies are prison population per capita, according used, and create the framework in to World Prison Brief’s 2018 World Prison which they operate. “The biggest gains Population List. in public governance that we’ve had in The system is also incredibly expen- any country come from transparency sive—one estimate puts the total cost of and accountability, and we simply incarcerations at $1 trillion annually. And do not have that” when it comes to the racial composition of the US prison public-sector use of artificial intelli- population is skewed relative to the gence, says Chicago Booth’s Sendhil general population: Black people make Mullainathan. up 12 percent of the adult population but Elsewhere in this issue, we catalog how one-third of all prisoners. And the high researchers are analyzing other topics of financial and social cost of the current sys- great concern to policy makers, such as tem is failing to produce good results. The the effects of COVID-19 lockdowns (pages US ranks 26th of 40 countries in the safety 19 and 28), the $1.6 trillion US student-loan category of the Organisation for Economic crisis (pages 7 and 66), and bankruptcy re- Co-operation and Development’s Better form (page 12). We also highlight research Life Index, which factors in both homicide that finds measurable evidence of how rates and people’s likelihood of reporting political uncertainty in Hong Kong hurt they feel safe walking alone at night. asset prices (page 14). And in a column, There is no shortage of proposals Booth’s Jean-Pierre Dubé suggests that a aimed at improving the situation. policy at work in Switzerland, of assessing President Joe Biden’s governing plan fines on the basis of income, should be includes pushing a reform bill, as well as more widely adopted (page 49). taking “bold action to reduce our prison As always, please do let us know population, create a more just society, what you think of this issue, as well as and make our communities safer.” the articles, charts, and videos on our New technologies may help shape website. Tweet us, comment on our efforts at reform. As our cover story Facebook or LinkedIn pages, or send us explains (page 30), machine learning an email. can be used to analyze huge masses of data and make predictions relevant to Hal Weitzman criminal outcomes. It is already being Executive director, Intellectual Capital used by many police departments for Editor-in-chief, Chicago Booth Review patrolling and crime solving, and judges [email protected] are using it to guide their decisions. In theory, it could be an important piece Emily Lambert of a more efficient and equitable justice Director, Intellectual Capital system. But could there be a trade-off, Editor, Chicago Booth Review even the germ of another problem? [email protected]

Editor-in-chief Multimedia Lucy Hewett Brian Wallheimer Hal Weitzman producer Rose Jacobs Alice G. Walton Editor Josh Stunkel John Kenzie Faculty advisory Emily Lambert Video producer Maggie Li committee Deputy editor Ray Zane Michael Maiello John R. Birge Jeff Cockrell Contributors Gary Neill Lubos Pastor Data editor Peter Arkle Brett Nelson Jane L. Risen Chuck Burke Noma Bar Dan Page Social media editor Eric Butterman Michael Rapoport Blake Goble Michael Byers Bob Simison Copy chief Francesco Ciccolella Vanessa Sumo Molly Heim Áine Doris Sebastien Thibault Designer Chris Gash Meena Thiruvengadam Nicole Dudka Glen Gyssler Andrea Ucini Review.ChicagoBooth.edu

COVER ILLUSTRATION BY NOMA BAR Spring 2021 Chicago Booth Review 1

285697_1-29.indd 1 2/1/21 7:01 AM DEPARTMENTS

1 Editors’ letter 4 Feedback 68 The Equation DATAPOINTS COVERSTORY

7 Canceling all student 18 Active-fund managers 30 LAW AND ORDER AND DATA debt mostly helps didn’t shine during the high earners COVID-19 crisis Will algorithms fix what’s 9 Government policy 19 Local lockdowns may help created the student- contain COVID-19 wrong with American loan crisis 19 Don’t kill a company to justice, or make things 10 Inventors are eyeing collect a debt your home office 20 Sharing experiences, even worse? 11 Without commutes, from afar, can bring By Jeff Cockrell how are Americans people closer spending their 20 How people ‘mentally extra time? launder’ unethical income 12 Matthew Notowidigdo: 22 How online retailers can Revisit bankruptcy fulfill orders better FEATURE reform 22 Do investors view stocks 13 For better output, turn as insurance? 42 When green investments pay off down the volume 23 Why exclusivity works—in In building a portfolio, sustainability is no longer 14 Political uncertainty marketing and a luxury good. But it’s not a slam dunk either. hurt Hong Kong policy making By Emily Lambert property values 24 Who is driving 15 Why COVID-19 has stock prices? widened the gender gap 26 Central bankers misjudge 15 COVID-19 stimulus forward guidance . . . checks spurred debt 27 . . . and their economists payment and saving may be overstating QE’s 16 How to spot an effectiveness overconfident CEO 27 Is the US becoming 17 Why auditors hire a plutocracy? graduates of local 28 India’s economic recovery colleges from its COVID-19 lockdown

Page 1 7 30 42

Featured Faculty

Constantine Yannelis, assistant Yueran Ma, assistant professor of professor of finance and an FMC Faculty finance and a Liew Family Junior Faculty Scholar, conducts research in finance Fellow, conducts empirical studies at the and applied microeconomics. This issue intersection of finance and macroeconom- features some of his most recent work on ics, exploring topics such as the effects student loans, which he says became an of interest rates on financial markets, area of interest when he saw “a dearth the macroeconomic implications of debt of economic analysis grounded in both markets, and the effects of financial empiricism and theory.” (Page 7) frictions. (Page 19)

2 Chicago Booth Review Spring 2021

285697_1-29.indd 2 2/9/21 4:28 PM FOOTNOTES

49 Lower fines could lead to higher revenues By Jean-Pierre Dubé 52 Entrepreneurs, remember the power of a smile By Waverly Deutsch 57 Don’t lose sight of suffering By John Paul Rollert 60 The COVID-19 trolley problem By John H. Cochrane 62 How has the pandemic changed the labor market? The Big Question 66 What should the US do about student debt? The IGM Panel

49 68

Lubos Pastor, the Charles P. McQuaid Jean-Pierre Dubé, the Sigmund E. Professor of Finance and the Robert Edelstone Professor of Marketing and a King Steel Faculty Fellow, is a director of Charles E. Merrill Faculty Scholar, directs Booth’s Center for Research in Security Booth’s Kilts Center for Marketing, is a Prices, and a member of the Bank Board faculty research fellow at the National of the National Bank of Slovakia. He Bureau of Economic Research and is has written on a broad range of topics an academic trustee at the Marketing including sustainable investing, income Science Institute. He has published inequality, populism, and many other numerous papers on quantitative aspects of financial markets and asset marketing and empirical industrial management. (Pages 18, 27, and 42) organization. (Page 49) FACULTY ILLUSTRATIONS BY JOHN KENZIE: BY ILLUSTRATIONS FACULTY GLEN GYSSLER BY ILLUSTRATION PHOTO

Spring 2021 Chicago Booth Review 3

285697_1-29.indd 3 2/9/21 4:28 PM Find the articles to which these comments refer at Review.ChicagoBooth.edu.

JOIN THE SO MUCH FOR THE FACTS, LISTEN TO BLACK PEOPLE; DON’T CONVERSATION WHERE’S THE THEORY? JUST READ THEIR TWEETS

Subscribe to our weekly email newsletter at The downfall (and possible How should companies respond to Review.ChicagoBooth.edu, salvation) of expertise (Winter Black Lives Matter? (Winter 2020/21) and follow us on social media: 2020/21) Twitter If companies and institutions his- @chicagoboothrev [Lars Peter] Hansen points torically were more open minded to Facebook to a major problem. Everyone collecting and listening to the experi- @chicagoboothrev wants to speak on the basis of facts, ences and perspectives of minorities at least in appearance. But the facts (including underrepresented minorities), Instagram @chicagoboothreview have no language. The reality must we might not need to analyze social media be examined with the help of a channels for sentiment. Since openly LinkedIn: Chicago Booth Review theoretical framework or on the basis discussing the challenges and issues of models. minorities face is on paper accepted but YouTube: —@mrjahan49 in reality frowned upon, I guess Twitter is Chicago Booth Review the next best source. And keep in mind, To understand the world, it is Black experiences are extremely diverse, not enough to look at data on a far more so than television portrays. WE WELCOME graph or run a regression. You need a —Lawrence W. Flack LETTERS theory (model) to know what data to look at to answer a question and how Send email to to interpret regressions and other DON’T DILUTE DIVERSITY [email protected] statistical methods. or send letters addressed to —@andyneumeyer What actually helps women at work Chicago Booth Review at any (Winter 2020/21) of the following addresses: Experts have their limits, but more and broader inputs can help. I agree with Marianne Bertrand 5807 S. Woodlawn Ave. —@brandongiella that including African Americans Chicago, IL 60637 in the mainstream of our social and eco- One Bartholomew Close The mistrust comes from the far nomic fabric is a special intractable Barts Square right in every country. . . . issue. My observation is, though, that by London EC1A 7BL —Chris Lir Hawley bringing the legitimate disaffections of United Kingdom all groups together under the umbrella I think these corrupt bureaucrats of diversity, we generalw ly, as both in- 168 Victoria Road need to come up with a better word dividual Americans and as the business Mount Davis than science. Like a lawyer shopping community, become lax about fixing this Hong Kong for an expert witness, they too can pay big issue. We all must do more to focus Comments may be edited scientists to validate their agendas. on meeting this single great challenge. for clarity and/or space. —Kelly Nelson Metke —Andrew Jacknain

4 Chicago Booth Review Spring 2021

285697_1-29.indd 4 2/9/21 4:29 PM You’re Invited ON BOARD

VIRTUAL NONPROFIT BOARD SERVICE CONFERENCE

ON BOARD IS GOING VIRTUAL

APRIL 2021 Connect with a network of Chicago Booth alumni and nonprofit leaders to examine key trends and insights in board service during informative and interactive online sessions throughout the month of April.

STAY UP-TO-DATE Want to be the first to know when On Board registration opens? Email Prentiss Smith at [email protected].

285697_1-29.indd 5 2/9/21 4:29 PM Raising the bar on high-impact, custom Executive Education Offered in various formats such as online, inperson, or in a hybrid learning experience.

ACCESSIBLE WORLDWIDE TO MEET YOUR REMOTELEARNING NEEDS.

COVID-19 has permanently changed how organizations operate. To help themmove ahead, Chicago Booth designs and delivers custom executive education PARTNER WITH programs to meet an organization’s specific objectives. We work closely with our clients CHICAGO BOOTH to explore their remote-learning needs, and offer various formats such as inperson, online, or hybrid Contact us to learn more about our customized approach. Our programs have bolstered senior managementsÜ leadership, strategy, and decision-making skills—and helped drive growth and innovation for companies globally. +1 312.464.8732 We’ve helped organizations ranging from top consumer packaged goods companies to [email protected] global financial services. Working with both publicand privatesector clients, we’ve ChicagoBooth.edu/custom-program createdgeneral management programs for executives spanning the Americas, Asia, Europe, Africa,and the Middle East.

Our programs allow leaders to learn from renowned Chicago Booth faculty, and help  them elevate their skills. Partner with us and empower your organization with The Chicago Approach™.

285697_1-29.indd 6 2/1/21 7:02 AM Inventors are eyeing your home office Page 10 // For better output, turn down the volume Page 13 How to spot an overconfident CEO Page 16 // Who is driving stock prices? Page 24

uring this year’s Democratic But who benefits from such forgiveness Canceling all Dpresidential primaries, numerous would depend largely on how it’s candidates championed the idea structured. Some policy approaches of forgiving some or all of the student could chiefly benefit the highest earners, student debt debt held by the federal government. suggests research by University of Massachusetts senator Elizabeth Warren Pennsylvania’s Sylvain Catherine and mostly helps proposed canceling up to $50,000 Chicago Booth’s Constantine Yannelis. in debt for nearly all borrowers, and In the United States, about 43 million Vermont senator Bernie Sanders borrowers collectively owe nearly $1.6 high earners advocated wiping away all student debt. trillion in outstanding federal student Tying repayments to income Now that President Joe Biden, who debt, according to the Department of has endorsed some measure of debt Education’s office of Federal Student Aid. could provide more-targeted forgiveness, has taken office, the notion That balance has more than tripled since relief for US borrowers of canceling student debt has only 2007, while the number of borrowers has gained momentum. only increased by about 50 percent.

ILLUSTRATION BY CHRIS GASH Spring 2021 Chicago Booth Review 7

285697_1-29.indd 7 2/9/21 4:29 PM There are a number of ways policy average person in the bottom decile, Focusing on the balance of a loan ignores makers could go about relieving some of and almost half of all relief would go its present value—the total value today of this burden, and Catherine and Yannelis to people in the top 30 percent of the all future payments on the loan, factoring focus on three broad approaches to distribution. “Patterns are similar under in the rate of return that money would debt cancellation: universal forgiveness policies forgiving debt up to $10,000 or earn on a risk-free investment. (canceling all student debt for everyone), $50,000,” they write, “with higher-income That gets to the second reason: the capped forgiveness (canceling up to households seeing significantly more researchers find that because borrowers $10,000 or $50,000 of every borrower’s loan forgiveness.” further down the income distribution student debt), and targeted forgiveness There are two primary reasons why are less likely to repay their loan in full, (wherein borrowers’ relief is tied to their those who are better off would benefit the ratio of a debt’s present value to its income). Using data from the Federal most, Catherine and Yannelis explain. balance increases with income. For those Reserve Board of Governors’ 2019 Survey First, as prior analyses have demon- in the bottom income decile, the present of Consumer Finances, the researchers strated, loan balances are correlated value of student debt is about 40 percent examine how student debt is distributed with income. Broadly speaking, before of the balance, whereas present value throughout the income spectrum, and surgeons, lawyers, and executives and debt balance are almost equivalent how that would define the beneficiaries embark on their lucrative careers, they for those in the top decile. of various debt-relief plans. tend to amass large amounts of debt from “Once you factor that in,” Yannelis They find that under universal advanced-degree programs. But just look- says, “universal student-loan forgiveness student-debt forgiveness, the average ing at balances understates the degree policies are actually much more regres- person in the top decile of the earnings to which the benefits of student-debt sive than if we simply look at balances, distribution would receive more forgiveness would pool at the top of the because the ratio of present values to than five times as much relief as the income distribution, the researchers say. balances is much lower at the bottom of Canceling student-loan debt could exacerbate US economic inequality While the highest-income groups have about twice the student debt as the lowest-income groups, research finds that across- the-board loan forgiveness would disproportionately benefit the rich, saving them well more than twice as much money.

IF STUDENT LOANS WERE FULLY CANCELED IF UP TO $10,000 WERE CANCELED Dollars per capita, 2019 Dollars per capita, 2019 Amount in loans Actual benefit Amount in loans Actual benefit being forgiven to borrowers being forgiven to borrowers

$9k$6k $3k$0 $3k $6k $9k $3k $0 $3k Top 10% Top 10% income income group group

2.1× 5.5× 1.8× 4.1× the as much the as much amount actual amount actual of loan monetary of loan monetary dollars benefit dollars benefit forgiven forgiven

Bottom Bottom 10% 10%

Note: Per capita dollar amounts and Benefit imbalance: While the bottom group had $3,028 in loans, the researchers find that income deciles are calculated for the full population, not just those with student loans. only $1,085 was going to be repaid over time, thanks to existing loan-forgiveness programs. Catherine and Yannelis, 2020

8 Chicago Booth Review Spring 2021

285697_1-29.indd 8 2/9/21 4:29 PM the income distribution relative to GOVERNMENT 1992 led a significant number the top.” of these institutions to leave Catherine and Yannelis find POLICY CREATED the student-loan programs. that targeted forgiveness policies This tightening of credit known as income-driven repayment THE STUDENT- accounted for as much as 95 programs are much more effective at percent of the subsequent concentrating debt relief toward the LOAN CRISIS decrease in loan defaults, the bottom and middle of the income researchers estimate. distribution. IDRs, as the name IN THE PAST couple of “Between 1981 and 1992, suggests, tie monthly loan payments decades, student-loan debt in credit gets expanded and to borrowers’ income: under typical the United States ballooned to then tightened, leading terms, an IDR participant might $1.5 trillion. It is now the larg- to a high exit rate among pay 10 percent of whatever income est nonmortgage source of for-profit institutions from she has that falls above 150 percent US household debt, ahead of loan programs,” Yannelis says. of the poverty line. After a certain credit-card or auto-loan debt. “After 1992, the cycle repeats, number of years—20 or 25, depend- The average student-loan however, with credit access ing on the plan—her remaining debt is $35,000. And those loosening up again as pre- balance is forgiven. loans have the highest default 1992 reforms get rolled back.” Enrolling all borrowers in an IDR rate of any form of household By 2011, he says, borrowers at program would result in the bottom debt, according to the Federal these high-default, for-profit earnings decile receiving four times Reserve. schools accounted for almost as much forgiveness (in terms of This problem reflects half of all student-loan dollars forgiven) as those in the top decades of government defaults in the US. decile, the researchers calculate, policy, according to For-profit schools typically with 69 percent of benefits going Brookings Institution’s Adam market themselves to non- to borrowers in the third through Looney and Chicago Booth’s traditional students, offering seventh deciles. Increasing the Constantine Yannelis. two- to four-year degree threshold below which individuals The researchers tapped into programs, or diplomas in don’t pay any of their income—to three decades of data from the fields such as health adminis- 300 percent of the poverty line, up US Department of the Treasury tration, culinary arts, or beauty from 150 percent—would produce and the National Student Loan treatments. The students they similar distributional results, but Data System, assembling a attract tend to be people over with more relief going to middle- data set representing almost the age of 25, often without income borrowers. 12,000 institutions in the a high-school degree, from The analysis also suggests that US that were eligible for minority backgrounds. They’re IDR-based forgiveness would have a student-loan programs more likely to be women or smaller fiscal impact than universal between 1970 and 2014. single parents with families or or capped forgiveness. Enrolling To access most student a job to balance. all borrowers in an IDR program loans, students have to be en- These borrowers face two with a 300 percent threshold would rolled in qualified educational problems after enrollment entail slightly less forgiveness—as institutions. Federal policy that increase their risk of measured by present values—than has alternately expanded default, Yannelis says. First, a $10,000-capped policy, and far and contracted institutional they’re saddled with more less than a $50,000-capped plan. But eligibility, contributing to the debt. Secondly, they have a under that same IDR policy, the bottom current student-loan situa- harder time finding a job once 60 percent of the income distribution tion, the analysis suggests. the loans need to be repaid. would receive more forgiveness In the 1980s and 2000s, Virtually all the peaks in than with a $10,000-capped policy, education policy increased student-loan defaults are and the bottom 30 percent would eligibility for loan programs driven by changes in federal see more forgiveness than with a and raised borrowing limits policy that expand access $50,000-capped policy. for older students. This drove to the high-risk, for-profit Calls for debt relief are likely to a surge in new institutions, institutions, the researchers continue. But Yannelis says that the especially for-profit schools, argue. Addressing this means policy discussion about student-debt the researchers argue. About working through the complex forgiveness often misses the “simple 85 percent of the increase trade-off between enabling point that we’re not really forgiving in student-loan defaults access to credit—and edu- debt. We’re simply transferring debt Federal policy between 1980 and 1990 cational opportunities—and from student borrowers to taxpayers, has alternately was driven by new schools attenuating default risks for which makes it important to think expanded and entering loan programs, the borrowers and taxpayers about the distributional effects of this contracted researchers find. alike.—Áine Doris transfer.”—Jeff Cockrell institutional Then a change in federal Adam Looney and Constantine Yannelis, eligibility. policy imposing restrictions “The Consequences of Student Loan Credit Sylvain Catherine and Constantine Yannelis, “The Distributional Expansions: Evidence from Three Decades of Effects of Student Loan Forgiveness,” Working paper, November 2020. on high-default schools in Default Cycles,” Working paper, June 2020.

Spring 2021 Chicago Booth Review 9

285697_1-29.indd 9 2/1/21 7:02 AM 2 8 5 6 9 7 _ 1 10 W Inventors are eyeing your home office September 3, 2020. filed from January 7, 2010, through than 3.5 million patentapplications an algorithm toanalyze themore from anywhere office workplace conferencing, telecommuting phrases including ofrelevant termsanda dictionary review theybuilt ofnews articles, to working from home.Through a any trends ininnovation related tospotof USpatentapplications YuliaPhD candidate Zhestkova. Booth’s Steven J. Davis, andBooth to Stanford’s NicholasBloom,Chicago workingsupport remotely, according towardapplications that technologies shift innew patenta significant by the COVID-19 pandemicsetoff Thelockdownssuggest. triggered - 2 9

. The researchers parsedthetext i n d Chicago Booth ReviewChicago Booth , d

nomadic employee

what USpatentapplications that’sthe pandemic—atleast, orking from homewilloutlast 1 0 , distance work . Theythencreated telework , and , virtual , , video Spring 2021 flexible working patent applications referencing work-from-home technologysurged. patent applicationsreferencingwork-from-home Within weeksofthefirstpandemic-relatedUSlockdownsinearly2020, thenumberof economy inaremote-work seepromise filers Patent Bloom etal.,2020 0.6% Share ofUSpatentapplicationsrelatedtoadvancementsinremotework 0.4% 0.8% 0.2% 1.2% 1% 002015 2010 2010–19 average: 2010–19 0.56% 2020 2 / 1 / 2 1

7 ILLUSTRATION BY DAN PAGE : 0 2

A M If a patent application included at WITHOUT Survey data by state, industry, least one relevant term in its descrip- and earnings. tion or main-features section, the COMMUTES, HOW Thirty-seven percent of researchers marked the application as respondents said they worked a WFH (working from home) invention. ARE AMERICANS from home, almost 35 percent WFH-related patent filings began still went to business prem- rising in 2020, just a few weeks SPENDING THEIR ises, and the rest were not after Chinese authorities imposed a EXTRA TIME? working at all. The researchers’ lockdown in Wuhan to limit the spread estimates of time savings of the novel coronavirus, Bloom, apply just to those who were Davis, and Zhestkova find. As the virus DUE TO the COVID-19 pan- actually working from home. began to spread across the United demic, millions of Americans Those without children States, patent filings referencing the have swapped commuting for were far more likely to devote key terms surged. The share of patent working from home. newly freed-up time to leisure applications related to advancements The WFH phenomenon is activities such as reading, in remote work rose from 0.53 percent saving Americans more than watching TV, and being in January to 0.74 percent in February 60 million hours of commut- outdoors. The researchers to 1.07 percent in July. ing time a day, according to find that those with children “We find clear evidence that Mexico Autonomous Institute devoted 18 percent of their COVID-19 has shifted the direction of of Technology’s Jose Maria time savings to their kids. innovation toward technologies that Barrero, Stanford’s Nicholas Analyzing the survey support WFH,” the researchers write. Bloom, and Chicago Booth’s responses, Barrero, Bloom, They note another indication that Steven J. Davis. “The time sav- and Davis find that employees the pandemic sparked innovation in ings are approaching 10 billion who are able to work from this area: even before these patent fil- hours as of mid-September home tend to be in jobs that ings began climbing, COVID-19 started [2020] for American workers require more education and driving up stock prices of companies alone. The global reduction pay higher wages. People who focused on cloud computing, telecom- in time spent commuting is are saving the most time by munications, and semiconductors—key surely many times larger,” not commuting during the components of the WFH economy. they write. pandemic tend be those By spring 2020, more than half of So what are people doing who were doing well before paid work in the US was being done with the 54 minutes a day, the pandemic. from home, according to research on average, that they’re They also find no large by Jose Maria Barrero of the Mexico saving? Devoting 35 percent differences on the basis of Autonomous Institute of Technology, to more work on their primary gender or race, but they do along with Bloom and Davis. And jobs and 8 percent to more observe that workers with less WFH is showing its mettle: during the work on secondary jobs, education are likely to devote COVID-19 crisis, the market looked the researchers suggest. more extra time to work. Those favorably on companies whose Household chores, childcare, with a high-school diploma operations required less physical and other work activities or less poured 32 percent of contact and whose employees could at home eat up another 25 time savings into their primary easily work from home. These percent. The rest of their time jobs and an additional 15 companies’ stock prices outper- savings go to indoor and percent into a second job, the formed those of their counterparts outdoor leisure activities, researchers find. That’s almost whose operations made them less including exercise. “The twice as much as among resilient to social-distancing rules, picture, then, is one in which people with more education, according to research by University those [who] WFH devote according to the research. of Naples Federico II’s Marco Pagano most of their savings in The surveys turned up and Vienna University’s Christian commuting time to non- one anomaly. Those working Wagner and Josef Zechner. Plus, their leisure activities—work for from home reported an expected rate of return was lower, as pay, but also chores, home average of 32 hours a week on investors seemingly perceived them improvement, and childcare,” the job, compared with 36.4 to be less risky going forward, the the researchers write. hours before the pandemic. same research finds. They draw their conclusions The researchers suggest Says Davis, “The average person from three surveys of 10,000 respondents were considering was positively surprised by their Americans between the ages how they used their former ability to be productive at home. of 20 and 64 who earned at specific commuting time We can anticipate the technology, Chores, childcare, least $20,000 in 2019. The slot without accounting for platforms, and products that support and other work researchers took the sound- additional off-hours work. working from home will get better activities at home ings in May, July, and August —Meena Thiruvengadam over time.”—Meena Thiruvengadam eat up 25 percent 2020, and reweighted the Jose Maria Barrero, Nicholas Bloom, and Steven of the time sampling to match the Census J. Davis, “60 Million Fewer Commuting Hours per Go to Review.ChicagoBooth.edu to see citations for research Day: How Americans Use Time Saved by Working mentioned in this article. savings. Bureau’s Current Population from Home,” Working paper, September 2020.

Spring 2021 Chicago Booth Review 11

285697_1-29.indd 11 2/9/21 4:29 PM OUR SEASONALLY Q1 ADJUSTED Q&A

Matthew Notowidigdo: Revisit bankruptcy reform Professor of Economics

What’s something that you think will or should be a priority for the Biden administration? I’m looking forward to renewing the debate about bankruptcy reform. Right now the focus is on student loans, which to my knowledge are the only debt that can’t be discharged in bankruptcy. But [Massachusetts] senator Elizabeth Warren, toward the end of her presidential campaign, produced a detailed policy brief about how she would reform the bankruptcy system, and that is a template for what I imagine the progressive wing of the Democratic party would view as a starting point. Warren, who was then a Harvard law professor, and President Biden, then a senator from Delaware, were on opposite sides of the 2005 Bankruptcy Abuse Prevention and Consumer Protection Act, which made it harder to file for bankruptcy. Filing used to be a rare occurrence, but in the 1980s and ’90s, bankruptcy rates started to take off. Around the early 2000s, about 1 percent of households every year were filing, plus there were a couple of high-profile examples of abuse. Policy makers started to feel like the system was being used by people who didn’t need it, and their concern was that the bankruptcy rate was spiraling out of control. The bill made it costlier, more

12 Chicago Booth Review Spring 2021 PHOTOGRAPH BY LUCY HEWETT

285697_1-29.indd 12 2/9/21 4:29 PM FOR BETTER classes. Dean manipulated the level of noise exposure, OUTPUT, TURN staying well below the amount that would cause DOWN THE hearing loss. Participants working VOLUME in the noisy room sewed 3 percent fewer pockets com- MANY PEOPLE working pared with those stationed from home have become in the quieter setting, he of an ordeal, and less financially perhaps acutely aware of finds. Productivity declined beneficial to file for bankruptcy. noise in and around their 5 percent for every 10-deci- Biden supported it. Warren environments—and in many bel increase in (or perceived was a critic of it, and its passage places in the world, such doubling of) noise. In inspired her to run for senate. My noise is increasing. In India, a second experiment, research with Boston University’s for example, the number involving a similar group of Tal Gross, Harvard’s Ray Kluender, of cars per capita tripled participants, Dean confirms the Consumer Financial Protection between 2001 and 2015. Not that the noise specifically Bureau’s Feng Liu, and University of only has engine noise risen affected tasks that involve Illinois’s Jialan Wang examines what accordingly but drivers are cognitive function. bankruptcy reform actually did—and purchasing custom-built car Workers don’t fully who was right about the BAPCPA. horns to cut through the din. understand how noisy In different ways, both of them The European Union environments affect them, were: Biden was right in that has taken steps to restrict the research demonstrates. consumers benefited. Some of the noise pollution. There may Participants were willing to cost savings got passed along as be a compelling economic give up only a small fraction lower credit-card interest rates and incentive for doing so: of their pay for a quieter, lower borrowing costs. But Warren increased productivity. more productive envi- was right in that the law was titled To better understand how ronment. If workers don’t abuse prevention, and it’s hard to noise affects productivity understand how much noise prevent abuse. and cognitive function, is costing them, it seems Our research suggests that Chicago Booth’s Joshua unlikely they’ll take steps to reductions in filings have happened Dean conducted two mitigate its effects. across the board, and even people randomized experiments in If people are unable or who have had some adverse event Kenya. In one, he selected unwilling to take such steps, such as an illness or injury have 128 people to participate there may be unnecessary had a much harder time filing for in a 10-day sewing course economic costs, Dean bankruptcy. It is unfortunate that outside Nairobi, recruiting warns. A growing academic the BAPCPA passed in 2005, and just day laborers who were literature argues that a few years later we had a financial waiting for work at the gates various things associated crisis. Many people affected by the of local textile factories. with poverty—such as the crisis ended up facing insolvency Sewing can be done amount of sleep people and foreclosure. But even now, many independently and doesn’t get, the temperatures to people whom we would expect to require people to commu- which they’re exposed, file for bankruptcy don’t, and they nicate with each other. Yet and the amount they worry aren’t able to get a fresh start. Dean wondered if noise about money—can impair I don’t think that’s what the could still disrupt key tasks task-management skills. proponents intended. The admin- that require strong cognitive If people prove to be as istration should start there and roll functions, such as sewing unaware of these threats to back parts of the law so that people in a straight line or moving their productivity as they are can get back on their feet after they fabric through a machine of the harm from noise, the experience adverse shocks. Now a with both hands. unnecessary costs could be lot of people suffering from illness, While participants sewed significant, according injury, or divorce aren’t able to pockets, Dean randomly to Dean. repay their debts. exposed them to noise. The “Future research,” he The idea of targeting people study participants were writes, “should provide who are abusing the system is a assigned a randomly gen- estimates of these costs good one. It’s just a hard problem erated schedule that had and how policies can be to solve. Using income as a basis Noise levels may them working sometimes designed that account for to determine who can file didn’t have serious in a quiet room and other these cognitive constraints.” work well. Maybe, 15 years after that economic costs, times near a car engine —Meena Thiruvengadam initial stab at it, we can find other particularly for that was otherwise used Joshua Dean, “Noise, Cognitive Function, and Worker ways of looking at this. poor people. for auto-mechanic training Productivity,” Working paper, February 2020.

Spring 2021 Chicago Booth Review 13

285697_1-29.indd 13 2/9/21 4:29 PM leases are guaranteed to be extended until Political uncertainty hurt 2097. They also find that properties that still have land leases granted by the British colonial government sold at a further dis- Hong Kong property values count of about 8.5 percent compared with similar leases granted by the successor nvestors widely believe that political He, Hu, Wang, and Yao tracked Hong Kong government. Moreover, the Iand social unrest, including the threat housing values from January 1998 through discounts appear to have widened since of revolution, drives down asset values. February 2020, a period of increasing 2004, as confidence in Hong Kong’s future Because so many interrelated economic political tensions, and built a model as an autonomous region has declined. and social factors are always at play, isolat- for property values on the basis of the Recent demonstrations and controver- ing the influence of political uncertainty on 2047 transition date. In Hong Kong, the sial moves by the mainland government asset prices can be difficult. colonial government owned all of the might further lower housing values, as Hong Kong, however, provides a land, leasing out the rights to use it under Hong Kong’s political and social future is laboratory for doing so. Chicago Booth’s long-term contracts, a practice that the top of mind for residents. The researchers’ Zhiguo He, Booth postdoctoral scholar special administrative region continued. data, running through February 2020, Zhenping Wang, Chinese University of The researchers used as a control group don’t include the effects of Beijing’s imposi- Hong Kong’s Maggie Hu, and Georgia State land leases that expire the day before the tion of a national security law in June 2020 University’s Vincent Yao, by comparing current political system is set to end, and that expanded government powers and property valuations before and after the they measured whether those properties limited civil rights. political authority is set to change in 2047, traded at a premium compared with real “Political uncertainty regarding the fate find that rising uncertainty in Hong Kong estate that had later expirations. of Hong Kong has been lingering throughout has significantly lowered property values. “We exploit a unique feature in Hong society” for decades, the researchers write. Uncertainty is a feature of life in Hong Kong’s housing market: Land leases expir- “But its severity and gravity have been Kong now because, as the researchers ing on June 30, 2047, have been promised evolving over time and often manifest more explain, the region, “caught in the middle a 50-year extension protection, while those significantly when public sentiment rises.” of the conflict between China and the expiring immediately after that date are left Hong Kong’s near- and long-term future western world, has become a political unprotected legally,” the researchers write. will tell a social and human-rights story battleground for the fate of the unprece- When it comes to leases that expire after that will resonate worldwide. For now, dented political experiment ‘One Country, the transition to mainland law—say in 2050 they also provide a rare glimpse into how Two Systems,’ especially since its ongoing or 2055—the government may extend those political unrest can interact with asset 2019–20 social unrest.” In 1997, the United leases under the current terms, change the valuations. And the finding that uncertain- Kingdom transferred the former colony terms, or not extend them at all. ty has a measurable cost on asset prices back to China, and both governments By tracking real-estate transactions, may apply more broadly as social unrest agreed to temporarily make Hong Kong the researchers demonstrate that housing and ideological divisions intensify in many a special administrative region that with land leases that expire immediately parts of the world.—Michael Maiello would maintain a capitalist system. That after June 30, 2047, sold at an 8 percent Zhiguo He, Maggie Hu, Zhenping Wang, and agreement expires on June 30, 2047, when discount compared with properties Vincent Yao, “Valuation of Long-Term Property Rights under Political Uncertainty,” Working paper, the Hong Kong Special Administrative belonging to the control group, whose August 2020. Region (HKSAR) will cease to exist, and Beijing will have authority. The researchers seized upon this Home buyers hedged against Beijing’s takeover predetermined transition date to calculate Buyers paid more for Hong Kong homes whose land leases will become eligible for a the effects of political uncertainty on long- 50-year extension before Beijing assumes control of the region in 2047. term asset values, residential properties in this case, which lend themselves to a Home sale prices natural experiment. In Hong Kong, land Index: 100 = price levels in 2004 Homes whose leases tenure is granted by the government for a come up for renewal fixed term. As a result, residential property the day before Beijing takes control owners will need to apply for extensions 400 when their leases end. Currently the of Hong Kong government provides a 50-year lease extension with no additional charge. Homes with leases What’s more, the leases that have 300 that expire afterward been granted by one government, say the British colonial government or the HKSAR, can be renewed by another, 200 such as the government that takes over in 2047. This exposes many lessees to substantial political risk—namely, will current leases that expire after 2047 be 100 extended at no additional cost, as has 2005 2010 2015 2020 been the case? He et al., 2020

14 Chicago Booth Review Spring 2021

285697_1-29.indd 14 2/1/21 7:02 AM COVID-19 stimulus checks spurred debt payment and saving

hen the US Congress rushed rebates, Claudia Sahm, formerly at the W through the $2 trillion Federal Reserve, along with Shapiro and Coronavirus Aid, Relief, and Slemrod, also finds that only a fifth Economic Security (CARES) Act in of survey respondents said they spent March 2020, they were trying to stave the funds. off the potentially devastating economic The July 2020 survey revealed WHY COVID-19 consequences of COVID-19 shutdowns. Americans planned to spend an average They hoped that issuing Treasury checks of 40 percent of their relief checks—lower HAS WIDENED would help people with immediate than the 50–90 percent of the 2008 tax needs, such as food, but also spur rebates. At the same time, there were THE GENDER GAP spending by businesses and individuals. stark differences among households, the But the stimulative boost was researchers report. “This has been likely less than intended, according to “Many spent their entire stimulus a huge negative University of Texas’s Olivier Coibion, payment, and just as many saved their shock to women’s University of California at Berkeley’s Yuriy entire check or used it to pay off debt,” Gorodnichenko, and Chicago Booth’s they write. Men and Hispanics were more progress in the Michael Weber, who find that the vast likely than the average respondent to labor market. There majority of Americans saved the bulk of spend. Black respondents were more likely was a brief period their checks or used the money to pay to pay off debt. That was also true down debt. Only 40 percent of the money of liquidity-constrained respondents, of time when I went toward purchases, and most in defined as people who said they would thought: Because industries that needed the least help. struggle to cover an unexpected bill equal both parents are However, the research also finds that to one month of their after-tax salary. the stimulus payments helped some people People whose earnings fell because of staying at home without jobs search harder for work. COVID-19 tended to use more of their now, maybe in The researchers base their findings checks on spending. some cases gender on an analysis of a detailed, 16-question Coibion, Gorodnichenko, and Weber national survey in July 2020 of 12,000 also investigated how the checks might norms or people’s participants in the Nielsen Homescan affect the labor market. They find that 20 gender views will panel, a long-running representative percent of people out of work reported shift a little bit sampling of consumers who track their that the check led them to search harder daily purchases. for a job. because they’ll see Their study reveals that only one Part of the argument against another how tough it is to in seven Americans—or just under 15 relief act has been that giving people have to take care of percent—reported spending or planning money deters them from looking for to spend most or all of their government work. However, “contrary to demotivating kids and so on. But checks. Half said they were using them to this group, the checks might have meant it seems like it just pay off debt, and 33 percent put the money they now had money for babysitters or went further in the into savings. transportation, the sort of things you might When people spent or planned to need in order to find a job,” Weber says. direction of women spend the funds, it was in sectors that least As for the checks’ limited success in staying home and needed additional support, such as food or spurring spending, the relief measures men going back to personal-care products. By summer 2020, were up against a big challenge in the form these industries had already experienced of lockdowns, which damped the ability work. . . . This could fillips from Americans stocking up in to, say, dine out or the desire to book a be due to existing response to lockdowns. The stimulus holiday package. But uncertainty related gender imbalances. checks did little to prompt people to to the pandemic may have also played buy bigger-ticket items such as cars or a role—as might have the mechanism of And I think what appliances or to book flights or hotels. one-off payments itself. it means is this The proportion of spenders to savers There may be a limit, the researchers gender gap that and debt payers was roughly consistent write, “on how much stimulus can be with past stimulus efforts, the researchers generated through direct transfers to we already see is find. For example, with the 2001 tax households. In the face of large crises, going to be rebates, between a fifth and a quarter government may want to consider a broad much larger.” of Americans said they mostly spent the range of policies targeting aggregate money, according to a study by University demand.”—Rose Jacobs —HEATHER SARSONS, of Chicago Booth, of Michigan’s Matthew D. Shapiro and on The Pie, a podcast from the Becker Go to Review.ChicagoBooth.edu to see citations for research mentioned Friedman Institute Joel Slemrod. In research on the 2008 tax in this article.

Spring 2021 Chicago Booth Review 15

285697_1-29.indd 15 2/9/21 4:29 PM allowing the holder to pocket the How to spot an difference between the market price and the exercise price. In theory, risk-averse CEOs looking to overconfident CEO diversify their portfolios would exercise in-the-money options well before their usiness leaders need to be confident the money—tend to have personality expiration date, while more-confident B in their decisions, but having too traits associated with overconfidence CEOs might hold their options nearly to much confidence can lead them to and lesser managerial ability, observe expiration, believing the shares will rise make poor decisions, such as ill-timed Chicago Booth’s Steve Kaplan and further. But confidence isn’t the only investments and ill-fated corporate Anastasia A. Zakolyukina and reason to hold an in-the-money option; mergers. So being able to sniff out an Dartmouth’s Morten Sørensen. tax considerations, board rules, and overconfident CEO could save investors Executive stock options generally other variables come into play. some pain. offer the right but not the obligation To confirm a connection between One common, albeit indirect, to buy shares at a defined price by overconfidence and longholding bellwether of overconfidence is the a certain date. An option is “in the behavior, the researchers analyzed tendency to hold deep “in-the-money” money” when the current stock price is personality assessments of more than stock options rather than exercise them greater than an option’s exercise price. 2,600 aspiring CEO candidates, drawn and pocket the gain, research suggests. From that point until the contract’s from four-hour interviews performed So-called longholder CEOs—who hold expiration date, the option can be by ghSMART, a management consulting

options that are at least 40 percent in exercised and converted to cash, company, between 2001 and 2012. The MICHAEL BYERS BY ILLUSTRATION

16 Chicago Booth Review Spring 2021

285697_1-29.indd 16 2/1/21 7:02 AM assessments rated each candidate on WHY AUDITORS state licensing. Big Four 30 specific personality traits and abil- companies generally want ities. Of the 67 candidates who later HIRE GRADUATES graduates to sit for state became CEOs of public companies, 58 exams soon after joining, were not longholders and nine were OF LOCAL and local graduates are longholders, allowing the researchers better prepared to do so. to compare the personalities of COLLEGES But the research also both groups. identifies a measurable Longholders tended to score lower PEDIGREE degrees are advantage: local recruits than nonlongholders in a number great, but so is hiring are associated with of areas—from analytical ability and locally. The Big Four audit higher-quality audits. Using organizational skills to acceptance firms often hire graduates 2000–16 data from Audit of feedback and calmness under from feeder schools that Analytics and Compustat pressure. Many of those characteristics are nearby, according to databases, the researchers are typically associated with overconfi- research by University of find that around 17 percent dence, according to various psycholo- Melbourne’s Gladys Lee of their observations in- gy studies, the researchers note. and Vic Naiker and Chicago volved client misstatements, Those traits also tend to move Booth’s Christopher and having a feeder school in sync and, taken together, paint Stewart. They also find within 60 miles reduced the a useful picture of a CEO’s “general that offices close to likelihood of a misstatement ability,” argue Kaplan, Sørensen, universities produce by 9 percent. and Zakolyukina. Using a regression higher-quality work. Local hires may also be analysis, the researchers find that Finding and hiring less likely to leave, which longholding is “significantly negatively talented auditors is a could reduce significant related” to overall ability—implying priority for accounting turnover costs. While the that longholders tend to be not only firms in the United States, researchers didn’t have overconfident, but also less talented where some research data to study turnover, it than nonlongholders. “Combined, has suggested that many can be hard to acclimatize then, our findings are consistent with certified public accoun- to a new city or to be far overconfidence being associated with tants leave the profession from friends and family, lower general ability,” they write. after just their first few Stewart conjectures. And The researchers also fleshed out the years. And when it comes when offices are near connection between overconfidence to recruiting, the Big Four colleges, there are more and managerial talent by looking accounting enterprises ap- opportunities—through at patterns in earnings forecasts. parently focus on colleges networking and recruiting Using a sample of 31 CEOs of public close to their offices. events and the like—for companies who provided a total of The researchers a firm to meet and assess 216 quarterly forecasts, they find that obtained a US campus graduates. rosier predictions (relative to actual map for PwC, which Granted, in the COVID-19 earnings figures) were correlated shows that the firm has era, many such events with lower ability, further suggesting recruiters stationed at 256 have been canceled, overconfident CEOs are less talented colleges, or 6 percent of all and interviews are often than those who aren’t. US universities. All these conducted online rather Finally, the researchers studied generally have recognized, than in person—meaning the link between short-term oper- well-ranked accounting local graduates may not ating performance and corporate programs. More than two- have the same leg up. investment decisions. Longholder thirds of Big Four recruits Stewart predicts that CEOs are more likely to boost capital came from these same 256 when candidate searches expenditures when cash flows are feeders, and 22 percent are done via Zoom, hiring strong and pull back harder when more came from accredited managers may revert to they’re weak, Kaplan, Sørensen, and business programs, the pedigree to distinguish Zakolyukina suggest. researchers find. between candidates. “I “We confirm that investments by Using LinkedIn, Lee, wonder if turnover will be firms with longholder CEOs are signifi- Naiker, and Stewart tracked higher,” he adds, noting cantly more sensitive to cashflows,” almost 12,000 people who that because applicants the researchers write. “Moreover, we graduated from the feeder come across differently in find investments by firms with less colleges in 2016. Almost person than they do online, talented CEOs are also significantly half who were recruited by new hires may fail to meet more sensitive to cash flows.” the Big Four went to offices managers’ expectations. —Brett Nelson Local recruits are within 60 miles. —Emily Lambert associated with There’s a practical reason Steve Kaplan, Morten Sørensen, and Anastasia A. Gladys Lee, Vic Naiker, and Christopher Zakolyukina, “What Is CEO Overconfidence? higher-quality for hiring locally: in the Stewart, “Audit Office Labor Market Evidence from Executive Assessments,” Working paper, Proximity and Audit Quality,” Working paper, August 2020. audits. US, accounting requires September 2020.

Spring 2021 Chicago Booth Review 17

285697_1-29.indd 17 2/9/21 4:29 PM Nevertheless, they demonstrate that each Active-fund managers extra star corresponded to a higher crisis-pe- riod annualized benchmark-adjusted return of nearly 6 percent—meaning that a five-star didn’t shine during the fund outpaced a one-star fund with the same investment style by 23 percent a year. Globes and stars were also reliable COVID-19 crisis predictors of changes in funds’ assets under management. One-globe funds shed 2.6 arket volatility is supposed to give While active funds underperformed percent of their assets as customers with- M sophisticated, active traders an on average, some fared better than drew funds, while five-globe funds suffered edge over passive investors—an others—specifically those with higher no net leakage, the researchers find. opportunity to shine and prove that active sustainability scores and performance “A popular perspective in traditional managers are worth the fees they charge. ratings, both measured by Morningstar. neoclassical economics is that sustainability But that’s not how things panned out Sustainable investing looks beyond raw issues, such as environmental quality, during the COVID-19 crisis, according financial returns by including environ- are ‘luxury goods’ that are likely to be of to Chicago Booth’s Lubos Pastor and mental, social, and corporate governance concern only to those whose more basic Booth PhD student M. Blair Vorsatz. (ESG) factors in the security-selection needs for food, housing, and survival are The researchers document crisis-period process. Morningstar assigns each fund a adequately met,” the researchers write. underperformance, which creates an sustainability score, measured in “globes,” “That investors retain their focus on additional headwind for the active-man- with more globes denoting greater sustainability during a major crisis indicates agement industry. sustainability. Funds with more globes that they view sustainability as a necessity, Many active managers have been generated higher benchmark-adjusted not a luxury.” As usual, funds with more struggling to justify their services. Low-cost returns than those with fewer globes, the stars also attracted more assets on the index investing has been gaining ground researchers find. In particular, four- and whole than those with fewer stars. for years, and in August 2019, passively five-globe funds had a 14 percent better If most pricey money managers don’t managed assets hit $4.27 trillion, edging out average annualized benchmark-adjusted add value, even in a crisis, why are so many actively managed assets for the first time. return than similarly styled funds with still in business? One reason, the research- One explanation for the continued fewer globes. (For more, see “When green ers suggest, is that investors believe active prevalence of active management is that investments pay off,” page 42.) managers “face decreasing returns to while active funds underperform passive Morningstar’s performance ratings— scale”—that is, as funds underperform, they funds on average, this average underper- scored in stars, not globes—were another shed assets, which in turn improves their formance is tolerated because active funds harbinger of returns. “It is not clear a priori future performance. Whether active funds’ do better in bear markets, when investors why Morningstar’s star ratings, which are performance improves after the COVID-19 value better performance the most. computed before the crisis from historical crisis remains to be seen.—Brett Nelson However, this explanation does not risk-adjusted returns, should have such Lubos Pastor and M. Blair Vorsatz, “Mutual hold up for the COVID-19 crisis. Pastor and strong predictive power for fund performance Fund Performance and Flows during the COVID-19 Crisis,” Review of Asset Pricing Studies, Vorsatz analyzed daily returns from 3,626 during the crisis,” the researchers observe. December 2020. equity funds between February 20 and April 30—a tumultuous 10 weeks when the Little safety as pandemic assailed the markets S&P 500 index collapsed by a third before Active-fund managers underperformed not only the S&P 500 over the early months of gaining nearly all of it back. The research- the COVID-19 crisis but also benchmark indexes with comparable holdings. ers find that, net of management fees, a large majority of actively managed Actively managed funds’ performance compared funds lagged behind their respective with benchmark indexes assigned by Morningstar benchmark indexes. Cumulative percentage difference for the average fund from February 19, 2020 Three-quarters of active-fund managers underperformed the S&P 500, with 1% average underperformance of 5.6 percent during the period, or 29 percent on an 0% annualized basis, the researchers find. The broad index isn’t a suitable gauge for every fund, but active funds also trailed −1% a variety of better-tailored benchmarks, 1.5% below though by smaller margins. −2% benchmark Depending on which valuation model by the end they chose, the researchers find that be- of April tween 60 percent and 80 percent of active −3% managers lost money on a risk-adjusted basis relative to their passive benchmarks. “Regardless of how we look at the data, we −4% see active funds underperforming during March April May the crisis,” write Pastor and Vorsatz. Pastor and Vorsatz, 2020 Trading days

18 Chicago Booth Review Spring 2021

285697_1-29.indd 18 2/9/21 4:29 PM DON’T KILL A pharmaceuticals—might not Local lockdowns have much of a market out- COMPANY TO side of that industry and can’t may help contain be easily repurposed. Even if COLLECT A DEBT there is a willing buyer, moving COVID-19 the equipment can be costly ocal lockdowns offer a promising THERE’S A SIZABLE gap and eat up value. Some assets, L solution in the effort to contain the between what a company is such as restaurant inventory, virus that causes COVID-19, suggests worth in liquidation and what may also have short shelf lives research by Nottingham University’s it’s worth while still operating, and depreciate quickly. John Gathergood and Chicago Booth PhD according to University of The researchers estimate candidate Benedict Guttman-Kenney, California at Berkeley’s Amir that, on average, the com- which finds that local lockdowns are less Kermani and Chicago Booth’s bined liquidation value of damaging to the economy than uniform Yueran Ma. Companies fixed assets, inventory, and national shutdowns. going through Chapter 11 receivables is 23 percent of the In March 2020, after the pandemic restructuring are worth about total book value of a nonfinan- struck, UK prime minister Boris Johnson twice as much when they are cial company—or 44 percent ordered a lockdown for England, which going concerns rather than if cash holdings are included. shut shops, restaurants, and pubs and liquidated, they write. Either way, it’s far less than 81 limited travel except for essential work and The finding is part of a percent, which is the average grocery runs. By contrast, local lockdowns, larger study of corporate debt, amount recovered in Chapter implemented later, after the national in which Kermani and Ma 11 for going concerns. lockdown had been lifted, aimed to restrict examine the size and compo- The gap varies across personal interactions without hampering sition of the debt loads held industries. Companies’ overall commerce—for example, by keeping by nonfinancial companies. liquidation values are higher restaurants open but having patrons dine They distinguish between in industries such as transpor- outside and only in small groups. Each asset-based debt (issued tation, wholesale/retail, and household had to keep to itself as well. against discrete assets) and hotels than they are in most “You couldn’t invite your neighbor ’round cash flow–based debt (issued other sectors. for tea,” says Guttman-Kenney. against the operating value of Liquidation values tend Unlike the initial national lockdown, local a company). In doing so, they to be much lower in other lockdowns—also orchestrated by the nation- explore companies’ cash-flow industries because not only al government—went into effect in different and asset values—essentially, may some companies’ assets places and at different times, depending how much more a company be immobile and specialized, on the rise in cases. By September, about might be worth alive rather but those companies may one in four people in the United Kingdom than dead. benefit from organizational was subject to a local lockdown, according It took the researchers capital, such as employees’ to the researchers, who find that these more than a year to ability to innovate, that would lockdowns slowed the spread of the virus amass the data needed to be lost in liquidation. while allowing the economy to chug along. answer the question. They To realize the full value of “We do not find large spending declines hand-collected information a company that continues in response to these local lockdowns,” the from 387 public, nonfinan- operating, creditors need to researchers write. “Instead, we find little cial companies that filed understand the business and (if any) decline in local spending—and any for Chapter 11 restructuring be able to oversee it effectively declines appear to be temporary.” between 2000 and 2016, plus and possibly discipline its The researchers combined COVID-19 pulled from other databases managers, the researchers case reports with spending figures provided including Compustat. argue. To some extent, this by Fable Data, an information provider that Assessing the value of can happen with financial cov- records hundreds of millions of credit-card assets took quite a bit of effort, enants that give creditors the transactions and checking-account flows. Ma says. She and Kermani right to intervene in business Those data—tracked by postal code and were able to find compre- decisions, often depending on available in real time—are highly correlated hensive appraisals that were company performance. with official aggregated statistics published disclosed in court cases. They “For most industries, there’s months later, and the faster availability also performed extensive a substantial loss to killing a of useful information gives policy makers checks using data from firm,” says Ma. That’s not to valuable time to make adjustments. other sources. say it makes sense to keep all One caveat to relying on local lockdowns Nonfinancial compa- distressed companies alive if is the need for a better-coordinated nies’ assets may fetch they have no hope of turning a national testing effort to isolate new cases little value on a stand-alone profit. But in most cases, there and mitigate any spread.—Brett Nelson “For most basis if liquidated, they is substantial value to doing industries, there’s conclude. Some specialized so.—Emily Lambert John Gathergood and Benedict Guttman-Kenney, “The English Patient: Evaluating Local Lockdowns Using a substantial loss equipment—for example, a Real-Time COVID-19 and Consumption Data,” CEPR Covid Amir Kermani and Yueran Ma, “Two Tales of Economics, January 2021. to killing a firm.” machine used to manufacture Debt,” Working paper, November 2020.

Spring 2021 Chicago Booth Review 19

285697_1-29.indd 19 2/9/21 4:29 PM SHARING EXPERIENCES, EVEN FROM AFAR, CAN BRING PEOPLE CLOSER How people

AS SOCIAL creatures, people the events happened on the ‘mentally tend to want to experience same day, the researchers good times—and get through find. The same trend was on launder’ the tough times—in the pres- display when the researchers ence of others. modified the circumstanc- But we also prefer to expe- es, making the monetary unethical rience separate events, good amounts larger or smaller and or bad, at the same time, sug- considering both wins and gest University of California losses. Participants showed income at Los Angeles’ Franklin a consistent preference for umerous experiments have Shaddy and University of experiencing similar-but-sep- N demonstrated that people are more Florida’s Yanping Tu, both arate events on the same day generous with money they’ve earned recent graduates of the as other people, but the pref- unethically. For instance, a salesperson Chicago Booth PhD Program, erence was weaker when they who earns a big commission by mislead- and Booth’s Ayelet Fishbach. didn’t share the experience. ing a client may mentally cordon off that For example, knowing that A desire to connect with money and be more willing to spend it your friend is having a root other people drives the phe- on others. canal on the same day as you nomenon, the research in- But research by Chicago Booth’s may make it feel like a shared dicates. In follow-up ex- Alex Imas, Carnegie Mellon’s George experience—and make it a lit- periments, the desire for Loewenstein, and Carey K. Morewedge tle easier to handle. integration faded when par- of Boston University finds that there’s a In one study, the research- ticipants were asked to think way to avoid this self-imposed penalty: ers arranged for participants, of a person they disliked. And psychologically “launder” the money by all UCLA students, to receive the wish for integration dimin- obfuscating its source. a personalized video mes- ished when events were more Researchers of behavioral finance have sage from a local celebrity, emotionally charged, such as spent decades documenting how mental the school’s men’s basket- when large sums of money accounting—our propensity to treat ball coach. Each participant were involved. money differently depending on things then had to choose a friend to The COVID-19 pandemic such as where it came from or how we receive a similar message, and has forced people to physi- intend to use it—affects decision-making. decide whether the message cally separate, but this work Mental accounting violates the economic should be sent on the same suggests that many of the tenet that money is fungible, or perfectly day or a different day. Almost same benefits of experiencing exchangeable: people may treat some of all participants, 88 percent, events with others in person their money more frivolously than the rest wanted the video message can manifest even when the of it, for example, if they acquired it easily sent on the same day. events happen at the same or by chance. Imas, Loewenstein, and In another experiment, time rather than in the same Morewedge’s findings hint at the compli- Shaddy, Tu, and Fishbach place. For example, in summer cated effects this informal bookkeeping asked participants to consider 2020, many organized races can have, as it can cause people to be a pair of events, such as win- switched to a virtual format generous with dirty money but also to ning $50 and $100 in an office whereby runners started at the find ways to avoid such generosity. lottery. Some participants same time on the same day In a series of experiments, the were asked about winning (as in previous years) but ran researchers explored how participants both awards themselves— alone, in different locations. treated ill-gotten gains under various would they prefer to win the A New York Times article ex- conditions. In one, they set up a game two prizes on the same day plained that these races ex- in which some participants were given or different days? But others ploded in popularity because a monetary incentive to lie to an anon- were asked about a situation runners could “still feel as if ymous and randomly assigned partner. where they win $50 and their they’re part of a larger group.” Lying, in these cases, meant maximizing friend wins $100. Again, Luckily for those socially the participants’ own earnings from would they want the wins to distancing, it turns out that the experiment but diminishing how occur on the same day or doing things at the same time much their partners would receive. The different days? as others can increase social researchers then gave all participants Participants wanted to ex- connection, just as doing People tend to be the option to donate some of their perience an event on the same things in the same place as more generous earnings to charity. The experiment day as a friend—but when it others does.—Alice G. Walton with earnings confirmed the finding established in past was a matter of experiencing that they research that people tend to be more Franklin Shaddy, Yanping Tu, and Ayelet Fishbach, two events on their own, they “Social Hedonic Editing: People Prefer to Experi- have procured generous with earnings that they have ence Events at the Same Time as Others,” Social didn’t care as much whether Psychological and Personality Science, in press. unethically. procured unethically.

20 Chicago Booth Review Spring 2021

285697_1-29.indd 20 2/1/21 7:03 AM Imas, Loewenstein, and Morewedge In further experiments, Imas, laundering and encouraging behaviors then took the experiment a step further, Loewenstein, and Morewedge find with social costs.” entering a subset of participants into evidence that mental money laundering Mental money laundering may a lottery in which they risked their also applied to situations in which ethi- also be a boon to companies whose experimental earnings but had a high cally and unethically earned money was practices or products have negative probability of receiving an identical sum pooled. When “dirty” money mixed with social effects. Such companies—which, in return. For those whose earnings had clean, participants tended to treat the prior research has shown, often have come in part through lying, processing entire pool of money as though it were to pay workers a wage premium to the money through this lottery effectively ethically earned. Moreover, people recog- overcome their qualms about the sanitized it: they donated significantly nize their tendency to treat laundered detrimental nature of their labor—may less than those who lied but didn’t play money differently than unlaundered be able to reduce that premium the lottery, and their donation behavior money, and seek out opportunities to through “greenwashing,” or putting was similar to those who told the truth sanitize it, the research suggests. some portion of their revenue toward and then participated in the lottery. The study’s findings have implications a prosocial purpose. Doing so could “Internal psychological constraints, for both individuals and businesses. allow employees to mentally launder such as negative emotions and associations “Most people . . . that are engaged in the company’s revenue, and their with moral violations, are some of the most morally questionable activities are also role in generating it, by pooling the effective deterrents to unethical behavior,” engaged in legal, ethical ventures,” the company’s harmful impact with its says Imas. “Strategies and techniques that researchers write. New technologies for positive works.—Jeff Cockrell and eliminate these constraints can potentially payments and budgeting, such as Venmo Eric Butterman make unethical behavior more likely, and Mint, “offer novel opportunities for Alex Imas, George Loewenstein, and Carey K. Morewedge, which is something that institutions and creatively pooling resources” and “may “Mental Money Laundering: A Motivated Violation of Fungibility,” Journal of the European Economic Association,

ILLUSTRATION BY SEBASTIEN THIBAULT BY ILLUSTRATION policy makers would want to prevent.” also provide a tool aiding mental money forthcoming.

Spring 2021 Chicago Booth Review 21

285697_1-29.indd 21 2/1/21 7:03 AM to determine how to split up the DO INVESTORS How online shipments, the researchers write. Amazon’s reported building spree is VIEW STOCKS AS of smaller warehouses, or FDCs. retailers can Zhao, Wang, and Xin constructed INSURANCE? a model assuming multi-item orders, no demand forecasts, and a two-layer ACCORDING to many stud- fulfill orders RDC-FDC distribution system. Their ies, people treat their invest- model also allows orders to be split ments like insurance policies. better into two shipments at most, which Those who want insurance aligns industry practice. They then against a tanking economy or mid the pandemic, online conducted a series of numerical the threat of unemployment Aretailers face a massive and experiments using the model to are willing to pay more for rapidly expanding problem: analyze solutions for different an asset if they think it’s less their costs have been spiraling situations. They find that companies likely to have bad returns in as they struggle to fill millions of should go with the least expensive the future times that coincide orders from relatively few ware- fulfillment option in the moment with these bad events. houses. Some are responding with a without considering the impact on This insurance mindset warehouse-building spree: Amazon future orders, a strategy they call a has been a longstanding reportedly plans to open 1,000 of “myopic policy.” tenet of financial models— them in cities and suburbs across Consider a hypothetical cus- yet research by University the United States. tomer in Maryland who orders two of Illinois’s Alex Chinco, But rather than build more ware- best-selling items used for remote along with Chicago Booth’s houses, online retailers should focus learning amid the pandemic: a Blue Samuel Hartzmark and on reducing costs by routing orders Yeti microphone and a Logitech Abigail Sussman, finds that more cheaply, according to Chicago C920s webcam. In a perfect world, it’s wrong. Booth PhD candidate Yanyang Zhao, the distribution center closest to the The idea of linking invest- Alibaba Group’s Xinshang Wang, customer would carry both items, ment decisions to future and Booth’s Linwei Xin. Their and the cheapest option would be economic performance dates research findings could boost for the retailer to bundle and ship to the 1970s. University of profitability for such companies. the products together. Chicago’s Robert E. Lucas US online retail sales of physical However, it’s possible that Jr. presented theories— goods in 2019 totaled $365.2 billion, the nearby distribution center for which he later won the write the researchers, and the cost carries the microphone but not the Nobel Prize for Economic of filling orders is taking a big bite. webcam. Meanwhile, a distribution Sciences—that tied investing Amazon’s shipping and warehouse center in California carries both, to fundamental economic costs surged to almost 28 percent of but bundling and shipping the items risk, the premise being that net sales in 2019 from less than 16 cross-country would be expensive. rational investors treat a percent in 2009, they note. If it’s cheaper for the retailer to split portfolio like a hedge against Some of the growing cost reflects the order so that the microphone future economic events. the often-complicated calculations ships from the center in Maryland But people haven’t typically faced by online retailers, including and the webcam travels from talked about investing in terms Amazon. Should they take more time California, the retailer should do of hedging economic moves, deciding how to fill orders most effi- that, the research suggests. and neither has the indus- ciently at the cost of slower deliveries? For many e-tail scenarios, the try, note Chinco, Hartzmark, Should they try to factor in demand model achieved consistent perfor- and Sussman. Mutual-fund forecasts? Should they just ship goods mance results comparable with prospectuses and news out as orders come in, even if they more-complex algorithms that rely reports often discuss cred- have to eat the higher costs of splitting on demand forecasts. The findings it risk, market risk, and the orders into two or more shipments may offer a key for keeping large risk that interest rates could from multiple warehouses? retailers in the black, and helping change—but not broader E-tailing giants such as Amazon smaller ones to stay open, Zhao says. economic risk. and China’s Alibaba use a two-layer That could be especially meaningful With this contradiction in distribution system that involves for retailers struggling to cope with mind, the researchers de- forward distribution centers and unreliable demand forecasts. signed a survey to compare regional distribution centers. FDCs “How many businesses will investors’ thought process- are closer to customers but tend to survive as just traditional stores?” es with financial theory. In have lower inventories and carry he asks. “The margins are thin, so it a simple experiment, they fewer items. This makes it “notori- may sadly be efficiency or close your asked a wide range of inves- ously challenging to make effective doors.”—Eric Butterman tors, including sophisticated real-time fulfillment decisions” finance professionals, how Yanyang Zhao, Xinshang Wang, and Linwei Xin, when an FDC can fill only part of an “Multi-Item Online Order Fulfillment: they would invest a hypo- A Competitive Analysis,” Working paper, order right away and the retailer has August 2020. thetical $1,000 in the market.

22 Chicago Booth Review Spring 2021

285697_1-29.indd 22 2/1/21 7:03 AM Participants typically in- wanted the T-shirt. In the third group, vested more in stocks when Why exclusivity anyone could bid. average stock returns were Average bids in the random-exclusion higher, and less when stocks works—in group were nearly double what they were were more volatile, which the in the other groups, supporting the theory researchers took as evidence marketing and that the perception of exclusivity—that that investors were consid- some people who wanted it couldn’t have ering risk and return. “Their policy making it—elevated people’s desire for the T-shirt. reactions to changes in these In a second experiment involving 95 two parameters are consistent he idea of exclusivity has long been participants, the researchers created two with textbook logic,” the re- T used to jack up prices on luxury cars, groups. In the first, everyone wrote down searchers write. “Participants liquors, and fashion, but exactly how much they’d be willing to pay for the understand their investment how much do people value exclusive T-shirt, and those who bid more than a task and respond to some pa- goods? Enough to pay a premium of 50 certain price were allowed to buy it. In rameters in our experimental percent or more, according to Chicago the other group, some participants were setup exactly as expected.” Booth’s Alex Imas and London School of randomly excluded, and the remainder However, the same text- Economics’ Kristóf Madarász. The reason wrote down how much they’d be willing to book logic suggests that par- reflects a deep-seated aspect of human pay. The median bid in the random-exclusion ticipants would notice and nature: we put greater value on things group was $5, twice the median in the first care about links between the that other people want but can’t have, just group, the researchers find. timing of their investment because they can’t have them. “These results provide direct support returns and the economy’s “It isn’t intentional meanness,” Imas for the role of mimetic dominance in health, which they did not. says. “The desire to possess something private valuations,” they write. “People When asked why they invest- that others want exclusively is an intrinsic significantly increase their willingness to ed the way they did, most par- part of human nature.” pay for a good if they know that there are ticipants said that they hadn’t Related ideas on the human quest for others whose intrinsic tastes for the good considered the correlation dominance and superiority have been may well be in excess of theirs, but . . . between stock returns and around for centuries in philosophy and they cannot obtain the good.” consumption growth. Among religion. Traditional economic models, The findings may help explain some those who did report thinking however, have yet to consider such situations that vex economists and political about this, three out of four motives as driving important aspects of scientists, the researchers suggest. For wanted more stocks when both individual behavior and markets. example, mimetic dominance can explain returns were more correlat- So Imas and Madarász developed a the decision by some companies to artifi- ed with consumption growth. theoretical framework and conducted cially restrict supply in the face of surging Thus, there was more demand two experiments involving almost 400 demand. In these cases, exclusion may be a for stocks that were worse in- participants to explore the impact and feature rather than a bug—increasing supply surance and were more likely value of exclusion and exclusivity. would eliminate the gratification people to lose value in a recession. The findings could have broad implica- get from knowing that others want what That’s not in line with the tions not only for marketing products but they have but cannot get it, which would theory of investors having an also for explaining exclusionary policies by decrease their valuation of the product. insurance mindset. businesses and governments. The motive Imas and Madarász also point to “Almost no one seems to they identify can potentially rationalize research from Princeton’s Ilyana Kuziemko, pay attention to what aca- political attitudes on redistribution, Harvard’s Ryan W. Buell and Michael I. demic finance has said is immigration, and trade. Norton, and Yale’s Taly Reich that finds arguably the most important The researchers argue that the quest for some of the fiercest opposition to raising variable,” Hartzmark says. superiority through exclusion, which they the American minimum wage comes from Even with this fundamental call “mimetic dominance-seeking,” may be people earning just over the minimum. The premise of economic models a fundamental psychological force. Mimetic mimetic-dominance motive suggests that in question, standard asset basically means “imitative” or “reflective”; this could be because higher pay for those at pricing models remain useful, they argue that a person’s desire for some- the bottom would eliminate the advantage the researchers write. A model thing is a reflection of how much others that the next-lowest group perceives having, doesn’t need to explain ex- want it. The dominance-seeking comes Imas and Madarász write. actly how the world works to from wanting what others can’t have. “A policy which sustains exclusion and still make helpful predictions. For the first experiment, 274 partici- inequality may be more popular than a Moreover, if investors aren’t pants took part in an unrelated study for redistributive policy even for those who using their portfolio as a which they were paid $15. The participants would gain materially from greater social hedge against a future down- were told they could bid as much as insurance because it may lead to a loss of turn, it might be a good idea $15 in an auction for a custom T-shirt mimetic dominance,” they write. They for them to consider. created exclusively for the study. Imas and argue that this idea also applies to policies —Emily Lambert Madarász split the participants into three on immigration, economic nationalism, groups. In one, a certain number were and barriers to trade.—Bob Simison Alex Chinco, Samuel Hartzmark, and Abigail Sussman, “Necessary Evidence for a Risk randomly excluded. In another, they were Factor’s Relevance,” Working paper, Go to Review.ChicagoBooth.edu to see citations for research April 2020. permitted to bid only if they signaled they mentioned in this article.

Spring 2021 Chicago Booth Review 23

285697_1-29.indd 23 2/1/21 7:03 AM Overall, small, active investment Who is driving stock prices? advisers have the largest influence on valuations, according to the research- hen stock prices fluctuate, The latter assume that markets are ers. Controlling for size, they find W commentators often attribute the highly elastic, Koijen explains—if prices that hedge funds tend to be the most moves to demand from certain deviate slightly from their fair values, influential. “Per dollar of capital, they groups of investors. A radio report might investors rush in to arbitrage such small are much more influential than pension attribute a daily rise in the S&P 500 to mispricings away. But the market is funds and insurance companies,” sentiment-driven retail investors—or far less elastic than thought, a growing says Koijen. maybe hedge funds, pension funds, or literature demonstrates. In this case, In addition to seeing who has the sovereign-wealth funds. differences in investor demand have a most influence in the market overall, But some investors drive valuations meaningful impact on prices. the researchers zoomed in to see who more than others do, suggests research by If asset prices reflect differences in has the most influence in specific areas Chicago Booth’s Ralph S. J. Koijen, NYU’s demand for the shares, who is driving that? of the market. When it comes to green Robert J. Richmond, and Princeton’s Koijen, Richmond, and Yogo developed stocks—so called because they are Motohiro Yogo. In traditional stock-valuation a framework to trace back differences associated with the environment—foreign methods, it doesn’t matter who owns in valuation ratios and expected returns investors have the most influence and a stock. Indeed, someone valuing a to various investors. They assembled are driving stock valuations, they find. company’s stock typically estimates the investors into eight groups, from passively Meanwhile, for companies associated company’s expected profits, and then dis- managed behemoths such as the Vanguard with good governance, smaller and active counts these profits using an appropriate Group, to smaller, actively managed asset managers take the lead. discount rate as implied by, for instance, investment advisers and hedge funds. For asset managers, understanding the capital asset pricing model (CAPM). They then modeled how valuations would how specific investors—or groups of The demand of a particular group of shift if all the assets of one group were them—influence the $38 trillion US stock investors matters only to the extent that to be redistributed to other institutional market is illuminating. Koijen and his it affects the market risk premium and investors in proportion to their assets—if all collaborators are now turning their therefore the discount rate, producing a hedge fund assets, for example, were held attention to the $40 trillion bond market, typically small effect. But some research instead by other institutions in the market. expecting that a similar effect is at work is starting to chip away at the gap between The effect of that would depend on an there.—Emily Lambert narratives about investor-driven market investor’s size and strategy compared with Go to Review.ChicagoBooth.edu to see citations for research

swings and traditional finance models. others in the market, the researchers show. mentioned in this article. CHRIS GASH BY ILLUSTRATION

24 Chicago Booth Review Spring 2021

285697_1-29.indd 24 2/9/21 4:30 PM Apple Effect on the company’s The investors with the most influence Market cap at Q3 close: market cap if the investor over companies’ market capitalization $1.966 trillion were to lose 10 percent In a study of stocks at the close Investor of its assets 2,300+ more of the third quarter in 2020, the stocks online 1. Berkshire Hathaway -2.17% researchers calculated how much a company’s market cap would Explore a database 2. BlackRock -0.97% have changed if any investor had of stocks and their 3. Vanguard -0.87% lost 10 percent of its assets. They 10 most influential ranked the results to identify investors on our 4. State Street -0.84% those with the most individual website. 5. Capital World Investors 0.29% power to alter a company’s value.

Microsoft Effect on Amazon Effect on Market cap: $1.591 trillion the company’s Market cap: $1.581 trillion the company’s Investor market cap Investor market cap

1. BlackRock -0.79% 1. BlackRock -0.7% 2. Vanguard -0.71% 2. T. Rowe Price -0.61% 3. State Street -0.69% 3. FMR -0.59% 4. FMR -0.51% 4. State Street -0.59% 5. Royal Bank of Canada 0.3% 5. Vanguard -0.53%

Facebook Effect on Alphabet (Class C) Effect on Market cap: $630.1 billion the company’s Market cap: $486.7 billion the company’s Investor market cap Investor market cap

1. FMR -1.52% 1. T. Rowe Price -0.93% 2. T. Rowe Price -1.13% 2. BlackRock -0.58% 3. Capital Research Global Investors -0.71% 3. State Street -0.5% 4. Capital International Investors -0.69% 4. Wellington Management 0.32% 5. State Street -0.42% 5. TCI Fund Management -0.32%

Alphabet (Class A) Effect on Tesla Effect on Market cap: $440.6 billion the company’s Market cap: $406.7 billion the company’s Investor market cap Investor market cap

1. FMR -1.01% 1. Capital World Investors -1.33% 2. BlackRock -0.44% 2. Baillie Gifford -0.9% 3. State Street -0.41% 3. BlackRock -0.65% 4. Capital World Investors 0.34% 4. Jennison Associates -0.51% 5. Royal Bank of Canada 0.3% 5. Vanguard -0.4%

Walmart Effect on Johnson & Johnson Effect on Market cap: $396.5 billion the company’s Market cap: $392 billion the company’s Investor market cap Investor market cap

1. Vanguard -0.77% 1. State Street -1.42% 2. State Street -0.7% 2. BlackRock -1.12% 3. BlackRock -0.69% 3. Vanguard -1.01% 4. Royal Bank of Canada 0.24% 4. FMR 0.6% 5. Bank of America -0.18% 5. State Farm Mutual Auto Insurance -0.55%

Koijen and Yogo, 2019; Koijen et al., 2020

Spring 2021 Chicago Booth Review 25

285697_1-29.indd 25 2/1/21 7:03 AM theoretical promise of such guidance with Central bankers misjudge the subdued empirical results. More recently, studies by D’Acunto, Hoang, Weber, and another researcher, Maritta Paloviita of the forward guidance . . . Bank of Finland, suggest that only people with a higher-than-average IQ can correctly ne of the best ways to spur an nonexistent, the researchers find. In the translate central-bank utterances into O economy is to get people spending, monthly European Commission surveys astute personal financial decision-making. and policy makers have a number of conducted in 2013 and 2014, respondents Weber warns that the public’s nonre- tools to do that. Yet growing evidence did not adjust their inflation expectations sponse to forward guidance could cast a suggests a favored approach of late— after the announcement or their assessment shadow over another policy idea currently forward guidance by central banks— of whether it was a good time to buy. And on the table: price-level targeting, which doesn’t work. Such guidance, usually the researchers did not find a delayed also relies on nonexperts understanding focusing on the outlook for interest rates, reaction that might suggest the information the mechanisms by which central-bank is meant to make clear to consumers that was trickling down via financial institutions targets affect consumer prices—and acting prices are likely to rise soon, so buying big or markets. on that understanding. items now would be smart. Both the German government and the The German VAT experience suggests While people may agree with the ECB were forced into the moves, though that a US national sales tax could be buy-now logic, they still may not for different reasons. Berlin realized it a fast-acting and effective alternative, react as economists and policy makers faced fines from the European Union during the current pandemic or any expect, according to Boston College’s unless it rapidly decreased its deficit, so other time, Weber says. Preannounced Francesco D’Acunto, Karlsruhe Institute of it sought to raise revenue as well as cut tax increases could do the work that Technology’s Daniel Hoang, and Chicago government spending. Germany’s VAT central bankers hoped forward guidance Booth’s Michael Weber. That’s because remained at 19 percent until July 2020, would do, he says. Whether politicians they don’t understand the signal, the when the government reduced it to 16 would be eager to employ such a tool is researchers find. percent for a six-month period as part of another question. “If you’re an economist too much a response to the pandemic. “You can pretty much only do it if stuck in your model world, this is very Draghi, meanwhile, was facing a prob- you’ve just been elected,” Weber says. surprising to you,” Weber says. On the lem shared by many central bankers since “You lose a lot of voter goodwill right away, other hand, he acknowledges that not the 2008–09 financial crisis. Normally, and it will take some time for the economy everyone can follow the logic chain that central banks can stimulate consumer to grow enough to earn that back.” leads from a central banker predicting spending simply by cutting interest rates. A more general lesson would be depressed interest rates, to lower But with rates hovering at or near zero for to translate opaque Fed speak into borrowing costs, to higher inflation, to many years after the crisis, this became a language that large swathes of the the urgency of buying now. “If you’re impossible, and unconventional tools population can grasp. Failing that, it not too detached from reality, it’s not such as forward guidance emerged as a may be time to stop placing much hope surprising,” Weber says. possible solution. in forward guidance as a way to spur The researchers analyzed two events Clearly, that hasn’t worked. Economists consumer spending.—Rose Jacobs in which governments or central banks at the New York Fed talked in 2012 about a Go to Review.ChicagoBooth.edu to see citations for research signaled that prices were set to rise. “forward guidance puzzle,” contrasting the mentioned in this article. One was a 2005 announcement by the German government that the country’s One way to make it obvious that prices are set to rise value-added tax (similar to the US sales Consumers clearly detected the inflation implications of a tax-hike announcement in tax) would increase from 16 percent to 2005, but not the message encoded in forward-guidance statements in 2013–14. 19 percent in 2007. The second was a 2013 statement by then European Central Share of Germans who said they expect higher inflation Bank president Mario Draghi that interest Case A: Unconventional signal Case B: Formal policy rates would stay low or decline further Germany announces VAT hike European Central Bank issues for some time. To economists, this 14 months in advance forward-guidance statements statement was a clear signal that price inflation would soon follow. 50% Analyzing monthly surveys conducted on behalf of the European Commission 40% both ahead of and after the announcement of the VAT increase, the researchers find 30% that Germans quickly recognized the price 20% implications. After the 2005 announce- ment, a higher portion of survey respon- 10% dents said they expected price inflation and that now was a good time to buy big-ticket 0% items such as furniture or electronics. Sep 2005 Mar 2007 May 2013 Aug 2014 Germans’ response to Draghi’s forward guidance, on the other hand, was D’Acunto et al., 2020

26 Chicago Booth Review Spring 2021

285697_1-29.indd 26 2/9/21 4:30 PM Central-bank papers tended to find . . . and their QE to be more effective at boosting prices, the researchers find. On infla- economists may tion, the central-bank articles found that QE had a peak effect 1 percentage be overstating point higher than all-academic articles reported, and a cumulative effect that QE’s effectiveness was around the same. And articles with central-bank student asked to grade her own researchers used more favorable IS THE US A work might give herself all As, language in the abstracts of the papers or at least higher grades than than those by academics, with more a teacher would. A similar thing is positive adjectives and fewer negative BECOMING A happening with central bankers, adjectives, according to the study. PLUTOCRACY? research suggests—and with potentially Managerial influence at central significant economic consequences. banks could be the reason. Bank When researchers at central banks management both influences the “You need a tension evaluated quantitative easing, one research process and decides between a dem- of the banks’ key responses to the who gets promoted. Central-bank ocratic system, 2008–09 financial crisis, they found a researchers who found larger effects bigger impact than did independent of QE on output were more likely which tends to economists from academia, according to get promotions, the study finds, redistribute wealth to the National Bank of Slovakia’s and there’s some evidence that the to people at the Brian Fabo, the European Central connection was strongest for senior Bank’s Martina Jančoková, and researchers—those who may be bottom, and a Chicago Booth’s Elisabeth Kempf especially reliant on the support of capitalist system, and Lubos Pastor. the bank’s top management to which tends to Conflicts of interest may explain be promoted. the rosier conclusions of the bank Fabo, Jančoková, Kempf, and disproportionately researchers. “Central bankers evalu- Pastor report one counterexam- reward people at ating their own policies is not unlike ple: researchers at Germany’s the top. If you go pharmaceutical firms evaluating their Bundesbank suggested that QE had own drugs,” the researchers write. less of an impact on output than did too much in one “Both have skin in the game.” articles by academics. However, the direction, you have Fabo, Jančoková, Kempf, and Pastor Bundesbank’s top brass has been a populist system examined 54 research articles released critical of QE, suggesting its research- between 2000 and 2018 about the ers may have been operating under a that does not effects of QE, in which central banks different line of scrutiny. reward enough on buy securities in the open market to When a central bank evaluates enterprise, merits, help bolster the economy during and its own policies, that can lead to after a financial crisis. Research studies overly optimistic research outcomes, and so forth. If you that included central-bank officials which in turn can shape the public’s go in the oppo- among their authors found QE to perception of how effective the site direction, you have a higher impact on economic central bank’s policy was, the study output and inflation than articles indicates. This is particularly relevant have a plutocracy, written solely by academicians, the as central-bank efforts play a key role which is extremely researchers find. in bolstering the economy amid the negative for a Articles with central-bank authors COVID-19 recession. found the peak impact of QE on However, the researchers stop large fraction of economic output to be about 0.7 short of concluding that central-bank the population. percentage points higher than articles studies are tainted by bias, calling it My fear is that the without central-bank authors, the “an important but messy question.” researchers find. The cumulative Central banks may have an incentive pendulum has impact of QE at the end of 2018 was to view their own policies favorably, swung too much found in central-bank research to be but they also want to preserve their in the direction of half of a percentage point higher, Fabo, credibility. The researchers write that Jančoková, Kempf, and Pastor report. they hope their study will help central plutocracy in the Those are significant differences, banks “think through the implications past 20 years in they write. All of the research with of this conflict for their research the United States.” central-bank authors reported a processes.”—Michael Rapoport statistically significant effect of QE —LUIGI ZINGALES, of Chicago Booth, Brian Fabo, Martina Jančoková, Elisabeth Kempf, on the Capitalisn’t podcast, hosted by on output; only half of the academic and Lubos Pastor, “Fifty Shades of QE: Conflicts of Interest Booth’s Stigler Center for the Study of the in Economic Research,” Working paper, papers did so. September 2020. Economy and the State

Spring 2021 Chicago Booth Review 27

285697_1-29.indd 27 2/9/21 4:30 PM India’s economic recovery from its COVID-19 lockdown

Jammu & n response to COVID-19’s rise, India Kashmir Income during ordered most of the country’s 1.3 billion Chandigarh Himachal the lockdown I residents to stop working and remain Pradesh April 2020 indoors starting in March 2020—the world’s Haryana Punjab largest lockdown. The government began National Capital Uttarakhand relaxing restrictions in June, and research Sikkim Territory of Delhi finds that while India’s economy improved Jharkhand rapidly in the following months, the Uttar Rajasthan Assam outlook for a return to prelockdown levels Pradesh Bihar Meghalaya remained unclear. In a report for Chicago Booth’s Rustandy Center for Social Sector Innovation, Gujarat Madhya Booth’s Marianne Bertrand and Rebecca Pradesh Tripurap Dizon-Ross, Centre for Monitoring Indian Economy’s Kaushik Krishnan, and Univer- OdishaOdiOd sha WestWest Bengal sity of Pennsylvania’s Heather Schofield ChhattisgarhChhatthhaattaat isgisgarararh Maharashtrashtht examined household-level survey data to establish a more comprehensive view of TelanganaTelangaana India’s initial recovery than national economic indicators could provide. These charts and maps highlight a selection of Goa AndhraAnndhra PraPradeshdedeshdes their main findings.—CBR

KarnatakaKarnataka

PPuducherryuducherry Tamilmilil NaduNadu Income during the first CHINA KerKKeralaala month of the recovery Chandigarh June 2020

PAKISTAN BHUTAN New Delhi NEPAL Jaipur Guwahati Lucknow The lockdown Patna affected income Ahmadabad Bhopal MYANMAR differently Kolkata across states Surat Raipur Per capita income in states BANGLADESH and union territories compared Mumbai with the same month in 2019 Percentage difference Hyderabad Increase 0% to −20%

−20% to −40% Bangalore Chennai

−40% to −60% Puducherry −60% to −80%

Thiruvananthapuram Marianne Bertrand, Rebecca Dizon-Ross, Kaushik No data Krishnan, and Heather Schofield, “Employment, SRI Income, and Consumption in India during and after the Lockdown: A V-shape Recovery?” Rustandy Center Disputed borders LANKA for Social Sector Innovation report, November 2020.

28 Chicago Booth Review Spring 2021

285697_1-29.indd 28 2/1/21 7:03 AM Employment remained down Income levels remained depressed as the economy opened back up as the lockdown was initially lifted

Share of entire population who were employed Monthly income per capita In 2019 rupees (5,000 = approximately US$70) Employed Employed, but worked 0 hours* 7,000 50% Lockdown relaxed Lockdown relaxed 6,000

40% 5,000

30% 4,000

Wages

20% 3,000

2,000 10% Business profit 1,000

0% Other sources Sep Jan 2020 Jun 0

*Data unavailable for portion of employees working 0 hours in September and October 2018 2019 2020

Top earners’ income didn’t drop as Spending on basic food items sharply during the lockdown months remained down after the reopening

Change in monthly household income per capita Monthly spending per capita Percentage difference from January 2020 levels In 2019 rupees (70 = approximately US$1)

Grouped by 0% 150 Lockdown relaxed monthly income Lockdown 140 before lockdown relaxed

Top −10% 130 group 12k (10,800+ 120 rupees) −20% 10k 110

100 8k Eggs, fish, meat, and milk −30% 90 Lockdown 6k Other food started 80 −40% 4k About 70 70% of popu- 60 2k lation −50% Jan Jun 50 0 2020 2020 2018 2019 2020

Spring 2021 Chicago Booth Review 29

285697_1-29.indd 29 2/1/21 7:03 AM COVERSTORY

30 Chicago Booth Review Spring 2021

285697_30-47.indd 30 2/9/21 3:45 PM LAW AND ORDER AND DATA

Will algorithms fix what’s wrong with American justice, or make things worse?

BY JEFF COCKRELL ILLUSTRATIONS BY NOMA BAR

Spring 2021 Chicago Booth Review 31

285697_30-47.indd 31 2/9/21 3:46 PM COVERSTORY

s it was for so many things, 2020 was a strange year for crime. Cities around the United States saw their overall crime rates drop considerably, according to data collected by University of Pennsylvania’s David S. Abrams, a trend perhapsA driven by the general decline of economic activity and face-to-face interaction amid the COVID-19 pandemic. At the same time, Abrams’s data also show rates of shootings and homicides in many cities that were at or even well above historical averages. But one aspect of criminal justice in the US was disturbingly familiar in 2020: the deaths of Black Americans at the hands of police. The murder of George Floyd in May and a series of other high-profile incidents fueled nationwide protests and calls for change, including from then presidential candidate Joe Biden, whose platform included expanding the powers of the Department of Justice “to address systemic misconduct in police departments and prosecutors’ offices.”

32 Chicago Booth Review Spring 2021

285697_30-47.indd 32 2/11/21 3:06 PM President Biden’s platform also included aspirations to reduce How police and criminal courts the prison population and “root out the racial, gender, and put algorithms into practice income-based disparities in the [criminal justice] system,” reflecting a sense that even apart from policing, America’s Algorithmic decision aids—tools that use data to predict justice system is bloated and unfair. the likelihood of a given outcome—are increasingly being These demands for reform intersect with advancements put to use in criminal-justice applications. in technology, namely software that uses algorithms and, in particular, machine learning to analyze huge masses of data and Select uses in policing make predictions relevant to criminal outcomes. Many police departments, municipalities, and states are adopting these tools, or increasing their use of them. Some observers hope such Recommend specific locations for tools will reduce the kind of racial inequity that has historically patrol on the basis of crime data and plagued justice in the US, but others worry they will only other information, helping police use exacerbate those problems. resources efficiently Algorithms are already being used in criminal-justice applications in many places, helping decide where police Flag high-risk officers for early departments should send officers for patrol, as well as which intervention on the basis of behavioral defendants should be released on bail and how judges should and other data, helping prevent police hand out sentences. Research is exploring the potential from harming themselves and others benefits and dangers of these tools, highlighting where they can go wrong and how they can be prevented from becoming a new source of inequality. The findings of these studies Select uses in criminal courts prompt some important questions such as: Should artificial intelligence play some role in policing and the courts? If so, Gauge defendants’ flight risk what role should it play? The answers, it appears, depend in to help determine whether they large part on small details. should be released on bail or their own recognizance while awaiting trial A persistently flawed system Even before the pandemic, crime in the US had broadly been Predict the risk of recidivism on the decline for decades. Data from the Federal Bureau of for a defendant or prisoner to Investigation show a 49 percent drop in the violent crime rate assist in bail, sentencing, and from 1993 to 2019 and a 55 percent drop in the property crime parole decisions rate. Survey data from the Bureau of Justice Statistics, which include both reported and unreported crimes, show even steeper downward trends. Public outrage over problems such as these has led to an But however encouraging some aggregate trends may be, often rancorous discourse over whether, and how, the US criminal justice in the US suffers from massive shortcomings—in should reform its approach to fighting crime. The conversations terms of both keeping people safe and treating people fairly. are wide ranging. They encompass concerns such as the Despite the improvement in its crime rates, relative to other militarization of police departments; the allocation of public developed countries, the US still faces significant crime-related funds to policing relative to other social services, including challenges, especially when it comes to its homicide rate: mental-health care; the fundamental aspects of how police according to the Organisation for Economic Co-operation approach and build relationships with the communities they and Development, the US had 5.5 reported homicides per serve; and the debate over whether incarceration should focus 100,000 people in the latest year for which data are available, on punishment for the crime or rehabilitation of the prisoner. as compared with other wealthy countries such as Canada (1.3 The use of algorithms by both police and the courts reported homicides per 100,000 people), Germany (0.5), and is becoming a more important part of these discussions. the United Kingdom (0.2). Algorithmic decision aids that use machine learning to predict Meanwhile, police departments in many US cities have been the likelihood of a given outcome are increasingly pervasive in found to engage in practices that discriminate against people of many settings. They play a role in weighty decisions such as color or other groups, and police brutality in the US, particularly who should receive a loan from a bank, or be interviewed for against Black people, is a source of persistent social upheaval. a job. But there may be few contexts in which their application Add to this the fact that among countries for which data are has greater consequence than in the criminal-justice system, available, the US maintains what is by far the world’s biggest where they can help shape a community’s exposure to policing prison system, numerically dominated by prisoners of color. and play a part in decisions about freedom and imprisonment.

Spring 2021 Chicago Booth Review 33

285697_30-47.indd 33 2/9/21 3:47 PM COVERSTORY

Such tools work by analyzing massive stores of data, locations of schools and parks, to identify discrete zones for police including data about past criminal events and outcomes, to to patrol on the basis of the likelihood of a crime occurring there. predict where crimes will occur, who’s most likely to fail to The system creates risk assessments for numerous types of crime, appear in court, and who’s most likely to be a repeat or violent from homicide to auto theft, and even suggests tactics officers offender. The promise of these tools is better outcomes using should use on their patrol. less resources while locking up fewer people. The risk is that There are various concerns about algorithms in policing, and they will make existing inequalities even more pervasive and one is bias. The American Civil Liberties Union, among others, difficult to address. worries that because ML policing tools rely in part on historical The academic literature contains support for both notions: data, they will be influenced by human bias and will exacerbate cautionary tales and warnings about unintended consequences, and further entrench historical patterns of inequality and as well as promising glimpses of new possibilities. These widely disparate treatment in the justice system. This has some backing varying outcomes hint at the heterogeneity that underlies the in research. broad banner of machine learning. “It is a common fallacy that police data is objective and University of Chicago Harris School of Public Policy’s Jens reflects actual criminal behavior, patterns, or other indicators of Ludwig illustrates this point by contrasting algorithms with concern to public safety in a given jurisdiction,” write Rutgers’s vaccines. Unlike a vaccine—which is a single, clearly defined Rashida Richardson, NYU’s Jason M. Schultz, and Microsoft thing—A.I. is not a “thing” but rather an umbrella term for a Research’s Kate Crawford. “In reality, police data reflects the collection of tools that are heterogeneous in their design and in practices, policies, biases, and political and financial accounting their implications for crime, fairness, and the harm the justice needs of a given department.” system can inflict. “In practice, A.I. algorithms actually vary Richardson, Schultz, and Crawford examined 13 jurisdictions enormously,” he says, including in their overall quality and in the US that adopted or used predictive-policing tools while in the amount of attention that gets paid to anticipating and subject to investigations, court-monitored settlements, mem- addressing concerns that may be general (such as discrimina- oranda of agreement, or consent decrees related to corrupt, tion) or specific to their use. “The key challenge for the field,” biased, or otherwise illegal police practices. They find that in he says, “is figuring out how we can get more of the good ones nine of those jurisdictions—including Chicago, New Orleans, and and fewer of the bad ones.” Maricopa County, Arizona—data that may have been shaped by such practices were available to train or otherwise affect the Predictive policing and its risks algorithms. “In these jurisdictions, this overlap presented at least The criminal-justice pipeline starts with policing, where machine some risk that these predictive systems could be influenced by or learning is being deployed in some cases to detect or solve in some cases perpetuate the illegal and biased police practices crimes. Noise sensors installed throughout many cities feed reflected in dirty data,” the researchers write. ML algorithms trained on audio data to listen for and report And if predictive-policing algorithms work in part by analyzing gunshots—a significant task, as research by Purdue’s Jillian B. data on past police actions, they seem bound to absorb whatever Carr and Texas A&M’s Jennifer L. Doleac has found that the vast bias shaped those data. As expressed by the ACLU and 16 other majority of gunshots go unreported. Some departments also organizations in a 2016 joint statement on predictive policing, “the use ML-powered facial recognition systems to help identify the data driving predictive enforcement activities—such as the location perpetrators of crimes. and timing of previously reported crimes, or patterns of communi- ML is widely used for crime prediction too. The Chicago Police ty- and officer-initiated 911 calls—is profoundly limited and biased.” Department is using products sold by the policing-technology Richardson says that algorithms used to predict criminal company ShotSpotter, including gunshot sensors but also a patrol outcomes will face inherent problems until we can trust the management software called ShotSpotter Connect. Connect uses data used to fuel them. “I think if there is to be a future with a combination of local crime and gunshot-detection data, as well automation in government decision-making and policy imple- as other data such as weather information, census data, and the mentation, there also needs to be fundamental changes around

One threat is that predictive-policing systems will create feedback loops as they ingest the data generated by their own predictions.

34 Chicago Booth Review Spring 2021

285697_30-47.indd 34 2/1/21 7:30 AM Spring 2021 Chicago Booth Review 35

285697_30-47.indd 35 2/1/21 7:30 AM COVERSTORY

ML COULD HELP IDENTIFY secondary employment. It factored in This is an area of focus where ML details about the neighborhoods in proponents and critics appear to agree. AT-RISK OFFICERS which the officers worked, as well as “Police could use predictive tools to incidents that may have been particu- anticipate which officers might engage in Early-intervention systems, designed to larly stressful, such as those involving misconduct, but most departments have identify officers likely to use excessive young children or gang violence. not done so,” write the ACLU and 16 other violence or engage in other behaviors Using historical data to analyze the organizations in a 2016 joint statement. harmful to others or themselves, have results of both the existing approach been popular with police departments and the ML model, the researchers Early experiences from Chicago and across the country for decades. Both find the department’s existing system elsewhere show that police misconduct academic researchers and private-sector was only slightly better than chance at follows consistent patterns, and that businesses have begun exploring distinguishing high-risk officers from offering further training and support machine learning’s potential to perform low-risk ones, but that the ML system to officers who are at risk can help to this task. could improve the true-positive rate (the avert problems. Police should be at In 2017, a research team led by the rate at which it correctly identifies high- least as eager to pilot new, data-driven Center for Data Science and Public risk officers) by 75 percent while cutting approaches in the search for miscon- Policy—formerly housed at the University down on false positives (the rate at which duct as they are in the search for crime, of Chicago, now at Carnegie Mellon it flags low-risk officers) by 22 percent. particularly given that interventions University—published the results of The University of Chicago has licensed designed to reduce the chances of work it had conducted in North Carolina the technology developed in the study misconduct do not themselves pose in collaboration with the Charlotte- of the Charlotte-Mecklenburg Police risk to life and limb. Mecklenburg Police Department, in Department to Benchmark Analytics, which it had used an ML system to a data-science company of which the According to Jay Stanley, a senior identify at-risk officers. The department’s university is part owner, and which policy analyst at the ACLU, the fact that existing early-intervention system was specializes in helping police forces collect some departments are experimenting triggered when “behavioral thresholds” and analyze data about their officers. with using algorithms to flag potential were met; if officers were involved in Benchmark’s early-intervention system, trouble spots is an encouraging develop- complaints or use-of-force events a First Sign, is being used or tested in cities ment. He says that because of manage- certain number of times within a given including Albuquerque, New Mexico; ment practices, union contracts, and time period, they would be flagged for Dallas, Texas; Nashville, Tennessee; and other factors, there haven’t been robust possible intervention. The thresholds San Jose, California. systems in place to identify problem were chosen on the basis of expert In September 2020, the City of Chicago officers. He cautions, however, that if intuition, and the decision to intervene launched its own early-intervention algorithms are involved, “all the same was subject to supervisor approval. system, this one created in partnership fairness issues” that the organization The researchers’ ML-based approach with the police department and the has raised about algorithms in predictive factored in events related to officer University of Chicago Crime Lab. Like the policing still apply, for the protection of behaviors—things such as the dis- Mecklenburg system, the Officer Support the police officers. “Decisions should not charge of a firearm, vehicle accidents, System also leverages ML to help flag be made [entirely] algorithmically but or citizen complaints—as well, but also high-risk officers—with the goal of then subject to human review.” drew upon the department’s millions of connecting them with resources and Jennifer Helsby, Samuel Carton, Kenneth Joseph, Ayesha Mahmud, records related to its officers’ training, support to prevent future problems. The Youngsoo Park, Andrea Navarrete, Klaus Ackermann, Joe Walsh, Lauren Haynes, Crystal Cody, Major Estella Patterson, and Rayid the traffic stops and arrests they made, system is being implemented gradually Ghani, “Early Intervention Systems: Predicting Adverse Interactions between Police and the Public,” Criminal Justice Policy Review, the citations they issued, and their across the city. March 2017.

data collection and use and sharing within government,” she crimes—so if an algorithm sends a patrol to a particular neighbor- says, adding that those changes will be especially challenging to hood, it becomes more likely that new data will be generated for effect given that they’ll involve organizations at the local, state, that location. The data will then be used to update the algorithm, and federal levels. which becomes more likely to send a future patrol to the same There’s also the threat of predictive-policing systems creating area, which can lead to overpolicing. feedback loops as they ingest the data generated by their own Danielle Ensign, formerly of the University of Utah, Sorelle predictions. Crime data come in part from police observations of A. Friedler of Haverford College, University of Michigan PhD

36 Chicago Booth Review Spring 2021

285697_30-47.indd 36 2/9/21 3:49 PM Greater predictive power could help focus pretrial detention on individuals who pose the greatest risk, and thereby minimize the number of people who have to endure this disruptive experience.

student Scott Neville, University of Arizona’s Carlos Scheidegger, result in any further action. Goel, Rao, and Shroff developed an and University of Utah’s Suresh Venkatasubramanian analyzed algorithm that, had it been used, would have enabled the police a model based on PredPol, a widely used policing platform, and to recover 90 percent of the weapons they had confiscated using find that it is vulnerable to such feedback loops. only 58 percent of the stops. The algorithm would also have They also find that by filtering the data used to update the improved the equity of the racial balance of stops. algorithm over time, it’s possible to mitigate the problem. Under In Chicago, ML is used in Strategic Decision Support Centers, their proposed solution, the probability that a crime is fed back which were introduced in 2017 following a civil-rights investiga- into the algorithm goes down as the probability of the area being tion into the police department by the Obama administration’s patrolled goes up, thereby counteracting the tendency toward Department of Justice. The centers were meant to provide a way overpolicing. However, the findings underscore the threat posed to combine better integration and use of technology with new by feedback, Venkatasubramanian says, “because we have to management and deployment policies and procedures. District 7, throttle the feedback to prevent the system from drifting away.” in the Englewood neighborhood on the city’s south side, was one of the first to receive an SDSC; there, police leaders and analysts Encouraging evidence on bias trained by the University of Chicago Crime Lab work side by side, To investigate whether predictive policing in Los Angeles using ShotSpotter products in real time to create more targeted resulted in more arrests of Black and Latinx people, University of deployment strategies for the district. Though gun violence in California at Los Angeles’ P. Jeffrey Brantingham, Louisiana State the city fell generally in 2017 after an exceptionally brutal 2016, it University’s Matthew Valasik, and George O. Mohler of Indiana fell almost twice as steeply in District 7. The CPD has so far rolled University–Purdue University Indianapolis reviewed the data from out SDSCs to 21 of its 22 police districts. a randomized controlled experiment conducted earlier by Mohler, The deployment recommendations made in each SDSC don’t Brantingham, and five coauthors. (Brantingham and Mohler are divert resources from other districts—they are used to determine cofounders of PredPol.) In that 2014 experiment, officers in select how to allocate patrols within districts, not across them. divisions of the Los Angeles Police Department were given a list Therefore, even if the people who live in a particular district are of 20 target areas to patrol and told that crime was expected to be overwhelmingly Black—many neighborhoods in Chicago, as in highest in those locations. Some lists were generated by human other cities, are highly segregated by race—there’s reason to think crime analysts using “all of the technological and intelligence the algorithm isn’t moving the needle much on the racial balance assets at their disposal,” and some were generated by an algorith- of police exposure. And to the extent the SDSC helps reduce mic forecasting tool. Whether officers received a human-generated crime within that district, ML is helping to reduce disparities in list or an algorithm-generated list varied randomly by day. public safety within Chicago. The algorithm outperformed the human analysts in terms of Jay Stanley, a senior policy analyst at the ACLU, agrees that the its impact on crime: for patrols of average duration, the algorith- narrower that predictive policing is used, the less concern there mically generated ones resulted in a 7.4 percent drop in crime, is. “It’s reasonable that changing the area reduces the concerns,” as compared with a 3.5 percent drop for human-directed patrols. he says, “but I don’t think it eliminates them.” Furthermore, in their follow-up analysis, Brantingham, Valasik, and Mohler find that “there is no significant difference in the Algorithms and the prison system arrest proportions of minority individuals” between the human Algorithms’ reach into criminal justice goes beyond policing, and algorithmic patrols. extending into the US’s large and expensive prison system, and ML could even make policing less biased, other research into decisions about who goes into it. suggests. Stanford’s Sharad Goel, Booking.com’s Justin M. Some judges use algorithmic tools to help them decide Rao, and NYU’s Ravi Shroff used ML to derive more effective who should await trial at home. This decision—often come to guidelines for New York City’s controversial stop-and-frisk policy. on the basis of the defendant’s flight risk, risk of recidivism, Data from 2008 to 2012 show that the overwhelming majority or both—has enormous consequences, as Chicago Harris’s of stop-and-frisk incidents involved people of color and didn’t Ludwig explained in a 2018 presentation as part of the Talks at

Spring 2021 Chicago Booth Review 37

285697_30-47.indd 37 2/1/21 7:30 AM COVERSTORY

using only two variables—age and and where algorithms have been called WHO IS MOST number of past convictions—performed into use. In Pennsylvania, the state LIKELY TO COMMIT slightly better than COMPAS. parole board began using ML-generated Stanford PhD student Zhiyuan “Jerry” forecasts of future criminal behavior in ANOTHER CRIME? Lin, data scientist Jongbin Jung, and 2013 to inform decisions about which Stanford’s Sharad Goel, with University of prisoners to release on parole. Each In sentencing defendants, judges often California at Berkeley’s Jennifer Skeem, of these forecasts, which came from a factor in how likely a person is to com- replicated the experiment by Dressel and custom-built assessment tool developed mit another crime in the future. And as Farid and find similar results. by University of Pennsylvania’s Richard technology advances, this assessment However, further experimentation Berk, was accompanied by an assessment sometimes involves algorithms. suggests algorithms may still have of the prediction’s reliability, which varied Criminal-justice nonprofit Recidiviz’s an advantage over human predictors, from case to case. Research by Berk finds Julia Dressel and University of or at least nonexperts. In Dressel and that recidivism among parolees declined California at Berkeley’s Hany Farid Farid’s experiment, human predictors significantly after the board began studied COMPAS, an algorithmic were given feedback after each predic- considering the ML forecasts, particularly tool that predicts defendants’ risk of tion—whether they were right about the when it came to violent crimes. In late recidivism. Using a database of defen- defendant they’d just evaluated, as well 2019, the board ceased using the tool after dants from Broward County, Florida, as an update on their overall accuracy; analysis by the Pennsylvania Department the researchers find that COMPAS when Lin, Jung, Goel, and Skeem didn’t of Corrections raised concerns about achieved roughly 65 percent accuracy offer predictors this instantaneous possible racial bias in the risk forecasts. in its predictions of who would commit feedback, depriving them of the chance Berk says those concerns reflect a another crime. However, a set of human to learn from their errors and more misinterpretation of the results, and predictors with no criminal-justice closely mimicking the conditions faced that the data in fact show that “the risk expertise was almost as accurate: by real-life judges, human performance instrument was implemented properly participants the researchers recruited fell well below that of both COMPAS and such that Black and white parolees were through Amazon Mechanical Turk the researchers’ own statistical model. rearrested at comparable rates.” averaged 62 percent accuracy. Dressel Parole decisions are another area Go to Review.ChicagoBooth.edu to see citations for research and Farid further find that an algorithm where the risk of recidivism is considered, mentioned in this article.

Google lecture series. “If the judge jails you, on average Kleinberg and his coresearchers examined arrest and you’ll spend two to four months in a place like the Cook bail data in New York between 2008 and 2013, and their County Jail,” he said. “You can imagine what that does findings indicate that pretrial judges were out of sync with to your job prospects. You can imagine what that does to the predictions of the researchers’ new algorithm. The judges your family.” released nearly half of the defendants the new algorithm Greater predictive power could help focus pretrial picked out as most risky—more than 56 percent of whom then detention on individuals who pose the greatest risk, and failed to appear in court. With the new algorithm’s help, the thereby minimize the number of people who have to researchers find, the judges could have maintained the same endure this disruptive experience. failure-to-appear rate while jailing 40 percent fewer people, or In 2017, Cornell’s Jon Kleinberg, Harvard’s Himabindu lowered the failure-to-appear rate by 25 percent without jailing Lakkaraju, and Stanford’s Jure Leskovec, with Ludwig a greater number of people. And they could have done all this and Chicago Booth’s Sendhil Mullainathan, constructed while reducing racial disparities. an algorithm to explore whether they could improve on That research served as a proof of concept that a new the results of the pretrial-release system then being used algorithm could potentially improve on pretrial judgments in New York City, in which judges could reference risk in practice. Following that study, New York City engaged the assessments made by an older predictive tool (rolled out University of Chicago Crime Lab and a private company called in 2003) as part of their decision-making. They find that, Luminosity to develop a new algorithmic pretrial assessment had their algorithm been used during the date range they tool, which it began using in 2019. The algorithm uses eight studied, it could have offered substantial benefits over the factors to generate a 26-point risk score judges can consider existing system. in their bail decisions—higher scores correspond to a greater

38 Chicago Booth Review Spring 2021

285697_30-47.indd 38 2/9/21 3:50 PM predicted likelihood that defendants will appear in court. How algorithms could help or hurt Data from November 2019 to March 2020 (at which time New the criminal-justice system York suspended pretrial court appearances due to COVID-19) indicate that these risk scores predicted defendants’ behavior Researchers find that algorithms can perform tasks in ways with a high degree of accuracy: appearance rates tracked risk that help address systemic flaws in policing and the courts, scores closely, with nearly 98 percent of those who received but their use is hampered by human bias—both in the data the highest possible score—a group that included roughly that feed into algorithmic systems and in the way police four in 10 defendants represented in the data—subsequently officers and judges interpret and act on their output. appearing in court. The algorithm recommended defendants in 85 percent of Potential benefits Potential dangers cases be “released on their own recognizance,” or without paying bail, in contrast to the city’s previous pretrial-release tool, which recommended such release in just 34 percent Greater predictive Biased predictions of cases. What’s more, the rate at which the new algorithm accuracy based on based on historical suggested ROR varied little across races: 83.5 percent of data rather than data tainted by white defendants, 83.9 percent of Black defendants, and 85.8 human intuition discrimination percent of Hispanic defendants were recommended for ROR— in contrast with recommendations made by the old tool, which Greater racial Veneer of objectivity recommended ROR 30 percent more frequently for white equity in policing covering systems defendants than Black defendants. The early evidence suggests and judicial susceptible to error judges’ decisions generally aligned with the new algorithm’s decisions and bias recommendations: nearly 90 percent of defendants recom- mended for ROR were ultimately released without bail. Just as pretrial judges are often asked to predict defen- More efficient Feedback loops in dants’ future behavior, trial judges may be asked during policing and fewer which systems ingest sentencing to forecast defendants’ risk of recidivism and unnecessary data generated by future harm to the community. This opens another window incarcerations their own predictions for algorithmic involvement. (See “Who is most likely to commit another crime?” on the facing page.) Greater ability to Lack of transpar- In 2013, a Wisconsin judge relied on COMPAS—an algorith- regulate transpar- ency and technical mic tool, also sometimes used in pretrial-release decisions, ency in decision understanding of that predicts the risk that a criminal defendant will commit processes how algorithms work another crime in the future—in his sentencing of Eric Loomis for attempting to flee from the police after being found driving a car that had been used in a shooting. Loomis argued that the sentencing violated his due process. The Wisconsin A 2016 ProPublica analysis of COMPAS risk scores Supreme Court disagreed, and the US Supreme Court highlights that concern: the results indicate that the tool declined to hear the case. was more likely to misidentify Black defendants as high risk The risks of allowing algorithms to weigh in on judicial than white defendants, and more likely to mislabel white decision-making—whether in pretrial decisions, sentencing, or defendants as low risk. The company that owns COMPAS has parole decisions, where they’ve also been used—are obvious. disputed those findings, and other analyses have called the Just as with predictive policing, biased data pose a threat to broader conclusion that COMPAS produces racially biased the equitability of outcomes. results into question.

“A lot of what we’re seeing is in the form of commercial products that are proprietary and opaque by design, and also probably way oversold.”

— JAY STANLEY

Spring 2021 Chicago Booth Review 39

285697_30-47.indd 39 2/9/21 3:51 PM COVERSTORY

But Loomis’s case highlighted a concern beyond bias: Regulation is the factor transparency. He argued that it violated his constitutional right to This, then, is the crux of algorithms and ML in criminal-justice due process for a trial court to rely on the results of a proprietary applications: they’re tools that, like any other, can be well instrument. He couldn’t challenge the accuracy or science of made or poorly made, used for good or used for harm. COMPAS because he couldn’t see it, much less analyze its results. Concerns about bias, transparency, and more boil down to the And what he did know about it concerned him: Loomis alleged contents of individual algorithms, some of which are better that it unjustly took gender and race into account. than others—and to how those algorithms are implemented. Transparency is another issue that the ACLU and other In theory, those contents and that implementation could be watchdogs have raised in terms of predictive policing—trans- strongly guided by a robust set of rules. parency about the algorithmic systems themselves, and “Part of the problem is that this technology is still so new, about the procurement processes by which those systems are we haven’t yet developed the right regulations for guiding its selected. “A lack of transparency about predictive policing use in public-sector policy applications,” says Ludwig. “We systems prevents a meaningful, well-informed public debate,” need to get those regulations right.” they write in a 2016 letter. Kleinberg, Ludwig, Mullainathan, and Harvard’s Cass R. Sunstein argue in 2019 research that algorithms have the Whenever automated predictions are considered for policing, advantage of being explicit in a way human decision-making all stakeholders must understand what data is being used, can never be—as long as regulation is in place that encourages what the system aims to predict, the design of the algorithm transparency. The researchers suggest that such regulation that creates the predictions, how predictions will be used in would require the producers of algorithms to store the data practice, and what relevant factors are not being measured or used to train their algorithms; to make the algorithms available analyzed. The natural tendency to rush to adopt new technolo- for regulators to test in order to see how changing certain gies should be resisted until a true understanding is reached as factors, such as a job applicant’s or criminal defendant’s race to their short and long term effects. or gender, would affect the algorithms’ predictions; and to make clear the algorithms’ “objective function,” or the specific They further argue that products and vendors need to be outcome they’re asked to predict, to scrutinize whether that subject to independent, ongoing scrutiny—and currently are outcome is fair and reasonable. not getting that, and in fact too often claim trade secrets. Access to those things, the researchers argue, would bring “A lot of what we’re seeing is in the form of commercial much-needed clarity to some of the questions that are often products that are proprietary and opaque by design, and also unanswerable in cases of suspected discrimination: What probably way oversold,” says the ACLU’s Stanley. “I think some factors were considered in making a particular decision, and of the commercial products have relatively simple nuts and why were those factors chosen? Algorithmic bias, under these bolts but use secrecy to evoke magical results that aren’t really conditions, becomes easier to detect than human bias. reflecting what’s going on under the hood.” Kleinberg, Ludwig, Mullainathan, and Sunstein note that And he warns that algorithms could aggravate discriminato- when it comes to the elements that regulators need to scru- ry patterns by giving biased police and judges the appearance tinize an algorithm, “at a minimum, these records and data of digitally sanitized objectivity. “There’s a problem of people should be stored for purposes of discovery.” In other words, reifying the data and algorithms, and both obscuring and they should be available to resolve legal questions, even if reifying decisions that are in fact highly questionable, but they’re not made public. making them seem as though they’re objective,” he says. But for algorithms used in the public sector, transparency “That is the fundamental problem: a bunch of people running can go well beyond storing information for private inspection. around playing with data and algorithms in all kinds of ways The New York Criminal Justice Agency, which administers the that have the potential for enormous destructivity. People are pretrial-release algorithm developed in part by the University fast, loose, and out of control here.” of Chicago Crime Lab, maintains a website where the general

“We should be wary about the government procuring algorithms the same way we procure phones for the police department.”

— JENS LUDWIG

40 Chicago Booth Review Spring 2021

285697_30-47.indd 40 2/9/21 3:52 PM If algorithms are to help improve American justice, the people adopting and using the tools must be fully aware of the potential dangers in order to avoid them.

public can see data about its performance, read about how it’s Algorithms aren’t everything used in practice, and even use the tool themselves. The site As machine learning and other algorithms become more also describes the agency’s plan for assessing and updating the pervasive, their presence in and influence on criminal justice will algorithm over time. likely continue to grow. Of course, it’s only part of the increas- Given that public-sector use of algorithms is an issue ingly complicated picture of law and order in the US. Embracing of not only technical and regulatory competence but also algorithms, or abolishing them, will not take the place of broad and popular acceptance, this kind of visibility into the function, thoughtful reconsideration of how the police should function within performance, and maintenance of algorithms could play a key a community, what sort of equipment and tactics they should use, role in making them palatable to a skeptical public. Another and how they should be held accountable for their actions. key could be greater public oversight of how algorithmic The question, then, is whether ML and other algorithms can products are selected for use. Given these tools’ potentially be part of the future. If algorithms are to help improve American high impact, susceptibility to negative unintended conse- justice, the people adopting and using the tools must be fully quences, and variation in quality, a thorough and transparent aware of the potential dangers in order to avoid them. process for deciding which algorithm to use, and how to Stanley concludes that while predictive policing and other use it, may be appropriate. “We should be wary about the examples of ML in criminal justice hold promise, it could take government procuring algorithms the same way we procure decades to work through the problems. Regulation is necessary, he phones for the police department,” Ludwig says. “Having a says, but it is also a blunt tool, and legislators are rarely tech savvy. private company say, ‘We can’t tell you how the algorithm He compares implementing ML to building the US transcon- works—that’s our [intellectual property]’ is not an acceptable tinental railroad system in the 19th century, which took many answer for algorithms.” years and involved many train wrecks. “There’s no question The 2016 joint statement issued by the ACLU and its 16 there are ways that this could be socially useful and helpful, but cosigners echoes this sentiment: it’s something that needs to be approached with great caution, great humility, and better transparency,” he says, adding that Vendors must provide transparency, and the police and “a lot of the institutional patterns and incentives and cultures other users of these systems must fully and publicly inform in law enforcement don’t lend themselves especially well to the public officials, civil society, community stakeholders, and kind of transparency that’s necessary. . . . It’s not foreordained the broader public on each of these points. Vendors must be that data and algorithms are going to bring some big social ben- subject to in-depth, independent, and ongoing scrutiny of their efit compared to the nuts and bolts that need to be addressed to techniques, goals, and performance. Today, instead, many fix American policing.” departments are rolling out these tools with little if any public Kleinberg, Ludwig, Mullainathan, and Sunstein acknowledge input, and often, little if any disclosure. that algorithms are fallible because the humans who build them are fallible. “The Achilles’ heel of all algorithms is the Ludwig says one solution may be to rely less on the private humans who build them and the choices they make about sector and more on in-house or nonprofit development of outcomes, candidate predictors for the algorithm to consider, algorithms. Mullainathan agrees that there’s still too little and the training sample,” they write. “A critical element of oversight of how A.I.-driven tools, from facial recognition regulating algorithms is regulating humans.” systems to pretrial-decision algorithms, are selected by Getting this regulation right could be the key to realizing public decision makers, and that there should be far greater the often striking performance benefits of algorithmic systems transparency about the performance of algorithms purchased without aggravating existing inequalities—and perhaps even by police departments and other public agencies. “The biggest while reducing them. But it remains to be seen whether gains in public governance that we’ve had in any country regulatory structures will develop that can meet this goal. Such come from transparency and accountability, and we simply structures are, after all, maintained by humans.—CBR do not have that” when it comes to public-sector use of A.I.,

Mullainathan says. Go to Review.ChicagoBooth.edu to see citations for research mentioned in this article.

Spring 2021 Chicago Booth Review 41

285697_30-47.indd 41 2/9/21 3:52 PM When green investments pay off

In building a portfolio, sustainability is no longer a luxury good. But it’s not a slam dunk either.

BY EMILY LAMBERT ILLUSTRATIONS BY GARY NEILL

42 Chicago Booth Review Spring 2021

285697_30-47.indd 42 2/1/21 7:30 AM Spring 2021 Chicago Booth Review 43

285697_30-47.indd 43 2/1/21 7:30 AM hen the COVID-19 pandemic prompted global stock marketsW to plunge in March 2020, Chicago Booth’s Lubos Pastor and Booth PhD student M. Blair Vorsatz set out to see if active managers outperformed stock indexes in choppy markets. But they stumbled on an unexpected insight.

Most active funds underperformed passive bench- of assets), and those with the highest sustainability marks during the crisis, they find, but what surprised rating received net inflows of $460 million (0.1 percent them is funds that had high sustainability ratings had of assets). Similarly, funds that apply exclusion criteria higher benchmark-adjusted returns. Morningstar, on the basis of ESG factors in their investment process, the investment research company, rates a fund’s by screening out certain companies or industries, sustainability according to environmental, social, and such as oil producers, also generated net inflows, of $1 governance (ESG) factors, and it assigns each fund a billion (1.2 percent of assets). globe rating—with funds that earn five globes being the “Our finding that investors remain focused on most sustainable. Five-globe funds outperformed four- sustainability during this major crisis suggests they globe ones, which in turn outperformed three-globe view sustainability as a necessity rather than a luxury ones, and so on, Pastor and Vorsatz observe. “This good,” write Pastor and Vorsatz. result is driven largely by environmental sustainabili- Sustainability is a new twist on an old theme in ty,” the researchers write, emphasizing the E in ESG. investing. Socially responsible investing has been And fund flows matched performance. Between around in some form for decades, if not millennia, February 20 (the day after the stock market peaked) explains Jon Lukomnik, senior fellow at High and April 30 (after the market had largely recovered), Meadows Institute, and author of a forthcoming book active funds recorded fund outflows of 1.3 percent of examining modern portfolio theory and systemic risk. assets under management as nervous investors pulled Muslims have long created investments to comply out. But the net flows for five-globe funds were around with sharia, and some churches have similarly pre- zero. Those with the lowest Morningstar sustainability ferred investment vehicles that screen out companies score had net outflows of almost $7 billion (2.7 percent in alcohol, tobacco, and gambling. The concept of sustainability has gone mainstream more recently, often associated with a subset of it, which integrates ESG into investing decisions. Funds across the globe have begun including ESG factors in the security-selection process. Several companies, including Morningstar and MSCI, score stocks and assets according to ESG, and BlackRock, the world’s largest asset manager, is integrating ESG into its risk analyses and is working to make more sustainable options available to clients. As the E in ESG, green investing is not without controversy, in part due to a lack of clarity surrounding what companies do or should report. But like climate change, green investing may be at a tipping point, and research suggests their trajectories are closely related.

Investors prefer sustainability In 2016, Morningstar launched sustainability ratings for 20,000 funds in its database—an action that took sus- tainability from something that was opaque for investors “to being clearly displayed and touted by one of the leading financial research websites,” observe Chicago

44 Chicago Booth Review Spring 2021

285697_30-47.indd 44 2/9/21 3:54 PM Booth’s Samuel Hartzmark and Abigail Sussman, who in underperform. Perhaps the best-known example is a a 2019 paper analyzed the ratings’ effect. 2009 paper by Columbia’s Harrison Hong and Imperial Simply calling attention to sustainability prompted College London’s Marcin Kacperczyk, who looked changes in how investors allocated their assets. at “sin stocks,” those of companies in the alcohol, Money flowed out of mutual funds with the lowest tobacco, and gaming industries. Sin stocks create rating and into those with the highest rating. The negative externalities, costs that are borne by society, researchers estimate that in the 11 months after the and some investors avoid these companies on moral ratings were issued, between $12 billion and $15 grounds or as a result of the higher legal risk. Hong and billion flowed out of funds that received just one Kacperczyk find that in doing so, investors depress the globe out of five, while between $24 billion and $32 prices of sin stocks, which less-constrained investors billion flowed into funds awarded five globes. then pick up on the cheap. Thus sin stocks outperform In a follow-up experiment, Hartzmark and comparable stocks by about 0.3 percentage points per Sussman teased apart investors’ reasoning and find month, or around 3.6 percent per year. that the investment decisions were motivated both But other research arrives at a different conclu- by performance expectations and what appeared to sion, arguing that in at least some time periods, be altruistic and environmental motives. Investors investments that are more socially responsible tend to believe that more-sustainable companies will outperform. For example, Illinois State University’s outperform the market—and they place a high value on Abhishek Varma and University of Alaska’s John R. social responsibility when investing. Nofsinger studied domestic equity mutual funds be- Yet the evidence is mixed on whether sustainable tween 2000 and 2011 and find that socially responsi- investments pay off. Several papers have made ble mutual funds tend to outperform during market the case that investments made with ESG in mind crises. They also find that such funds underperform

Sustainable funds were a decent place to be when COVID-19 hit the markets After stocks began to plunge in February 2020, funds with higher sustainability ratings outperformed those with lower ratings.

Actively managed funds’ performance compared Funds grouped with benchmark indexes assigned by Morningstar by Morningstar’s Cumulative percentage difference from February 19, 2020 sustainability ratings

2% Five-globe funds Companies with the best sustainability practices 0%

-2%

-4%

-6%

One-globe funds -8% Companies with the worst sustainability practices

-10% March April May Pastor and Vorsatz, 2020

Spring 2021 Chicago Booth Review 45

285697_30-47.indd 45 2/1/21 7:30 AM during noncrisis periods and that the performance desired portfolio, they earn lower financial returns of socially responsible funds and conventional funds is than non-ESG investors. But they are willing to sacrifice roughly the same over time. more than they actually do, and thus earn what the Pastor and Vorsatz analyzed a shorter time frame researchers term an “investor surplus.” In addition, the but more funds—over 3,600 of them, representing nonfinancial benefits they earn from green investing nearly $5 trillion in net assets—in arriving at their more than compensate for the lower financial returns. conclusion that funds with higher sustainability ratings Even investors focused only on profits have some posted higher returns in the COVID-19 crisis period. reason to embrace green assets, the model predicts. The researchers consider climate risk and news When green investments can outperform associated with it. When a giant iceberg splits from Theory may help explain these seemingly contrasting an Antarctic ice shelf, it is unlikely to have a direct results. Seeing little in the academic finance literature or immediate impact on society or the stock market. on this front, particularly when it comes to investments However, as a dramatic example of the effects of that are specifically environmentally sustainable, Pastor climate change, it can attract attention and make the and University of Pennsylvania’s Lucian A. Taylor news. That can in turn compel consumers to buy and Robert F. Stambaugh have developed a model to green products, push governments to pass the kind of explain what going green could mean for investors and environmental legislation that the researchers expect society. The model formalizes Hong and Kacperczyk’s to benefit green companies and hurt brown ones observation that sustainable assets have lower expected (those that generate negative externalities), and prompt returns. In the model, if a group of investors wants regulators to take action. Politicians, notes Pastor, to hold green assets, they bid up the prices for those “always juggle multiple agendas. For the environment stocks. The risk-adjusted returns will be lower for to rise to the top, something needs to happen.” those more-expensive, green stocks and higher for the So whenever there’s bad climate news, green less-expensive, nongreen ones. stocks benefit, and such stocks effectively become However, the researchers also find that investors who a hedge against headlines about climate change. prefer green assets are prepared to earn lower financial Even if investors think global warming is a hoax, it returns. The model explains that, in order to get their represents a risk, and they will want to hold stocks that hedge market-moving bad news. In some cases, the model suggests, green stocks will Focused on sustainability outperform brown stocks, which could explain Pastor and Vorsatz’s observations during the COVID-19 crisis. Amid the COVID-19 crisis, investors were more likely to stick with funds If consumers decide, unexpectedly, that there is value with higher sustainability ratings and drop those with lower ratings. in going green—whether that means buying electric cars or seeking out plastic-free products—green stocks Investors’ buying and selling of actively managed funds can outperform in the short run, even though they Cumulative percentage difference in flows from February 19, 2020 generally have lower risk-adjusted returns. Thus, green Funds grouped assets can outperform brown ones in periods when 2% by Morningstar’s investors’ tastes are becoming greener. This of course sustainability ratings assumes that some investors will continue to prefer 1% polluters and other brown stocks. If all investors were to move green, the entire market would adjust, and green would become the new benchmark. 0% Five-globe funds Companies with the Good for society too best sustainability −1% A move to ESG investments can also be good for society, practices write Pastor, Taylor, and Stambaugh. This may seem obvious, until you look at their underlying reasoning. −2% Milton Friedman, the late University of Chicago professor and 1976 Nobel laureate, famously argued in −3% One-globe funds a 1970 New York Times article that companies should fo- Companies with the cus on profits and let shareholders direct their portion worst sustainability to social programs if they choose to. Managers have a −4% practices responsibility to shareholders—and only shareholders, March April May Friedman argued. “That responsibility is to conduct Pastor and Vorsatz, 2020 the business in accordance with their desires, which

46 Chicago Booth Review Spring 2021

285697_30-47.indd 46 2/9/21 3:55 PM will generally be to make as much money as possible while conforming to the basic rules of the society, both those embodied in law and those embodied in ethical custom,” he wrote. Harvard’s Oliver Hart and Chicago Booth’s Luigi Zingales have since argued that company managers should instead maximize shareholder welfare, which they see as broader than shareholder value. Shareholders care about more than money, Hart and Zingales write—they also have ethical and social concerns. While it might be possible in some cases for a company to separate its moneymaking from its do-gooding, in other cases it’s more complicated. How do you untangle an issue such as Walmart selling high-capacity ammunition magazines if some across an investor’s portfolio. And it gets complicated: of its shareholders favor gun control, for example? a company could be a great employer for working (For more, read the essay “It’s time to rethink mothers but also a large emitter of greenhouse gases. A Milton Friedman’s ‘shareholder value’ argument,” corporation might be favored by environmentalists but in our December 2017 issue and online at Review. attacked for poor labor practices. ChicagoBooth.edu.) Although ESG ratings make more information salient Pastor, Taylor, and Stambaugh stick with the for investors, ESG investing risks becoming diluted, Friedman way of looking at the world. In their model, says Kimberly Gladman, senior associate at ValueEdge managers couldn’t care less about ESG or being green Advisers and a lecturer at Boston University. “The num- and simply focus on maximizing market value. But ESG ber of people who claim they’re doing it now is vastly investing has a positive social impact nevertheless. If expanded,” she says. “But the definition is so weakened outside investors care about being socially responsible, that it is very, very far from the original intent.” managers may follow suit purely to increase the This concern is one reason many researchers and company’s market value. And when investors have a investors, among others, have pushed to standardize taste for green stocks, the companies they buy develop key metrics of sustainability for different industries a lower cost of capital. Managers end up gravitating to and develop a reporting framework. Chicago Booth’s ESG, enticed by higher stock prices and that lower cost Hans B. Christensen and Christian Leuz and University of capital. of Pennsylvania’s Luzi Hail have collected various The net effects are good for society, the researchers arguments and compiled them in a report and an conclude. When there are more investors who care accompanying working paper. (For more, read “Should about ESG, this boosts the price of greener stocks, sustainability disclosures be standardized?” Winter which in turn induces ESG companies to invest more 2019/20 and online.) in their operations. The result is greater positive social Despite the unanswered questions, investors seem to impact, whether in the form of cleaner rivers, happier be piling on board—particularly investors from outside employees, or healthier markets. the United States, research suggests. Booth’s Ralph S. J. Koijen, NYU’s Robert J. Richmond, and Princeton’s Who decides what’s green? Motohiro Yogo find that demand for green companies This theory on green investing is simple and breaks in the US stock market is coming at least in part from new ground. That’s because while green investing may foreign investors, including sovereign wealth funds. be established and close to mainstream, academic “Larger, passive, and foreign investors have a stronger research on the subject, and on ESG more generally, is demand for greener firms,” they write, noting that State only just getting going. Street, PRIMECAP, and Vanguard are the most influential Perhaps because of that lag, ESG is a woolly term. among US investors, and the Norwegian sovereign In the investment market, it’s not at all clear what it wealth fund is most so among foreign investors. actually means. For some investors, ESG is evolving Under the Trump administration, the US relaxed into just another investing style, a way to find an edge environmental regulations, but that trend may reverse by including material factors not yet priced by the under President Joe Biden. And at least in theory, stricter market, such as employee satisfaction, that can deliver environmental rules could benefit green stocks—and outperformance. Others focus on ESG factors that their investors.—With Brett Nelson highlight systemic, societal risks, such as environmen- tal degradation, which have an impact on companies Go to Review.ChicagoBooth.edu to see citations for research mentioned in this article.

Spring 2021 Chicago Booth Review 47

285697_30-47.indd 47 2/9/21 3:56 PM Is capitalism the engine of destruction or the engine of prosperity?

Hosts Luigi Zingales, a world-renowned economics professor, and Bethany McLean, a Vanity Fair contributing editor, explain how capitalism can go wrong, and what we can do to fix it. Listen and subscribe wherever you get podcasts, or on Capitalisnt.com.

285697_48-68.indd 48 2/1/21 7:35 AM Entrepreneurs, remember the power of a smile Page 52 // Don’t lose sight of suffering Page 57 The COVID-19 trolley problem Page 60 // What should the US do about student debt? Page 66

n January 2010, Swiss officials handed including Finland also base fines on I down what was briefly the most income. By contrast, the way municipali- expensive speeding ticket on record: ties in the United States issue most fines is JEAN-PIERRE 299,000 Swiss francs (US$290,000), the neither progressive nor effective. It hurts DUBÉ penalty for the driver of a red Ferrari poor people and fails to bring in revenue. Testarossa caught going up to 137 kph We can do better. (85 mph) through the town of St. Gallen, Consider the current situation: more Lower fines where the limit was 80 kph. The fine than a decade of economic volatility has was based not on the driver’s speed but led to budget shortfalls for governments on his ability to pay: he was reportedly at every level. Politicians are reluctant could lead worth more than 20 million francs, and to hike taxes, and there are few other had four other luxury cars besides the options for raising revenue. Many local to higher Ferrari. Later that year, officials ticketed governments have turned to fines to fill another driver, this time in a Mercedes, their coffers. Fines are convenient and about 1 million Swiss francs. politically expedient, as elected officials revenues One takeaway here is to slow down on can say that the budget will be balanced Personalized fines would be picturesque roads, at least while driving not by taxpayers as a whole but on the in the Swiss Alps. But there is also a backs of people who break the law. If you a win-win for municipalities lesson for US policy makers in the form don’t want to pay more, all you have to do and their residents of personalized fines. Other countries is follow the rules.

ILLUSTRATION BY ANDREA UCINI Spring 2021 Chicago Booth Review 49

285697_48-68.indd 49 2/9/21 3:11 PM Chicago mayor Lori Lightfoot is using Personalized fines delinquent traffic fines that had been this playbook. Until recently, anyone caught assigned to a collection agency. Ninety on camera driving 10 mph over the speed can be low enough percent of these fines were concentrated limit was fined $35 and up. But the city will in 20 zip codes populated predominantly soon also ticket drivers who are 6–9 mph for people to pay by Black Americans and Hispanics. over the limit. This plan is reminiscent of them but high And there’s one more problem with one from 2012, when then mayor Rahm uniform fines: they may not deter many Emanuel attempted to raise $16 million enough to become people from the behavior at issue. A by hiking the fine for not having a vehicle deterrents even $175 ticket is unlikely to have stopped sticker from $120 to $200. Switzerland’s joyrider. For wealthy people, This phenomenon goes beyond Chicago, for more affluent a six-figure fine might be required to deter as other cities across the US, facing shrink- behavior, whereas $175 readily deters ing tax bases, are also relying on fines to citizens. someone in poverty. balance their budgets. A 2019 investigation by Governing magazine finds that while There’s a better way the vast majority of local US governments There is another option, however, a earned less than 5 percent of their win-win strategy that would help cities general fund revenue from fines, nearly collect more while keeping the poorest 600 local jurisdictions collected more than from suffering the financial consequences 10 percent that way, and 284 governments of the current system. The answer is collected more than 20 percent. (I believe personalized fines, setting the prices for these are conservative estimates since some individuals on the basis of their ability municipalities put money in places difficult to pay. Personalized fines can be low to track, and the report doesn’t include enough for people to pay them but high locales that collect less than $100,000 per enough to become deterrents even for year in fines.) more affluent citizens. This is how some But if Chicago’s experience is any fines are already set in Switzerland, as indication, plans to raise money this way well as in Finland, which in 2002 issued won’t garner as much as expected—and will its own then-record-breaking 116,000 bury some of the poorest people in debt. euro (US$103,000) speeding ticket. These As of 2018, Chicago drivers have racked up employers. In the year after a traffic stop, countries are more likely to collect on more than $275 million in unpaid fines for he finds, poor drivers experienced a one their tickets because they realize that an city-sticker violations since 2012, far more percentage point drop in the probability of individual’s propensity to pay a fine, just than the initial revenue goal, and it’s caus- having any payroll earnings, which suggests like the demand for goods and services, ing many of them financial ruin. According an increased likelihood of unemployment is elastic. to ProPublica, 1,000 Chicago residents who or a job change. Did some of those people Consider how a company comes up claimed chapter 13 bankruptcy protection have their licenses suspended because with a one-size-fits-all price for a product. in 2007 included unpaid city tickets in their they were unable to pay the traffic ticket? When a price is low, more consumers can debt. By 2017, however, 10,000 people More than 7 million Americans may have afford the product, but profit margins per filing chapter 13 listed city fines among lost their licenses due to unpaid court or paying consumer are thin. When a price their unpayable debt. The amount they administrative debt, according to a 2018 is higher, margins per paying consumer owed jumped from $1,000 to $3,900. Washington Post estimate. rise, but the number of potential buyers Charging everyone the same fine is Without the ability to drive, many falls. The company wants to find the regressive. Fines, plus related fees and people can’t work. In New Jersey, according sweet spot, where it optimally balances costs, can devastate someone who is to COSCA, 42 percent of people whose the number of paying customers at the financially struggling. Dartmouth’s Steven licenses were suspended lost their jobs margins per paying consumer. Mello finds that for people in Florida’s because of the suspensions, and 45 percent This uniform-pricing strategy knowingly poorest quartile, a $175 traffic ticket causes of those regained employment only when excludes some potential consumers who a degree of financial distress similar to that they regained driving privileges. might like to purchase the company’s caused by a significant earnings decline. Needless to say, when a person can’t product, but not at the current price. The A ticket often comes with processing fees drive or work, it becomes less likely that same is true with one-size-fits-all fines. and administrative costs, plus failure to pay a city will be able to collect a fine. The There are many people who want to pay on time can lead to a bigger fine. Someone unintended consequence of dispropor- the fine, but they simply can’t afford it. unable to pay $175 on time could find tionate delinquencies among the poor A company that instead sets prices on herself thousands of dollars in debt within exacerbates preexisting symptoms of the basis of what a customer is willing just a few months. The Conference of State poverty—and fails to generate revenues for to pay can make far more money. Movie Court Administrators (COSCA) reports that the city. Without measures that account theaters have for decades been charging 10 million Americans owe more than $50 for an individual’s ability to pay, those lower prices to children and seniors. billion to the criminal justice system. with the lowest incomes will be the most Similarly, most B2B companies that use Mello linked data on traffic tickets prone to continue accumulating debt a sales force authorize their sales reps issued in Florida between 2011 and 2015 and penalties and paying higher fines. to negotiate discounts off the regular to recipients’ monthly credit reports I recently reviewed 2018–19 data from prices. Even many public mass transit and payroll records from a large set of the City of Chicago on individuals with services, such as city buses, adopted such

50 Chicago Booth Review Spring 2021

285697_48-68.indd 50 2/1/21 7:35 AM segmented pricing long ago by offering fine or paying rent, the water bill, and the SERVUS, that uses artificial intelligence to student and senior fares. A similar tactic grocery tab. It has become abundantly create payment options for individuals. Its is used in places such as Chicago, where clear that people in this position choose goal is to help collect as much as possible people demonstrating financial distress to house, clothe, and feed themselves and for a city, health-care facility, or utility can meet with a city representative their families. while keeping payments manageable. It to obtain payment plans and waivers A flat fine is even more regressive than a may be illegal in some states to charge a on portions of late fees. San Francisco flat income-tax rate. In Illinois, all residents personalized fine on each ticket, but a city recently implemented waivers of up to are expected to pay the state 4.95 percent can forgive a portion due, which would be 80 percent of the original fine amount. of earnings no matter their income. A $200 equivalent to personalizing for people with However, these systems do not scale and ticket represents 0.13 percent of the income financial difficulties. are prone to human error. of a person making $150,000 per year, but SERVUS currently uses ability to pay The advent of data and artificial intel- it’s more than 10 times the financial burden as a basis for determining a person’s ligence enables us to implement scalable for an individual at the federal poverty obligation, but the system could benefit solutions that optimize the manner in level. Nearly one in five Chicagoans live at from more data. Together with SERVUS, if which we target differential prices. I proved or below the poverty level. we were to get the go-ahead from a city to this concept with my Booth colleague The major difference between products conduct research, we would randomize Sanjog Misra. We ran experiments with and fine payments in terms of the elasticity fine amounts for people unable to pay. ZipRecruiter, a company that charges a of demand is the ramifications. If a new We would test how much a person who monthly fee to simplify and accelerate the blender costs more than you’re willing demonstrates financial distress would be process of finding and screening qualified to spend, don’t buy one. An unpaid fine, able to pay. This information would be used job candidates. ZipRecruiter had been however, leads to more costs. to inform an algorithm that could then offer doing well charging a uniform price to all If a fine were based on the ability to pay data-driven relief to individuals who get its clients, but we thought it might be able it, both the person fined and the city would fines that are above their means to pay. to serve more clients if it offered them benefit. If an offender who cannot afford This represents an incredible opportu- different prices. $200 can come up with $50, the amount nity. After implementing fine forgiveness, For one month, when new clients remains a deterrent, the person ticketed San Francisco saw an almost immediate signed up for ZipRecruiter, we charged isn’t saddled with insurmountable debt, increase in fine revenues. ZipRecruiter random amounts between $19 and $399. and the city collects money. found that it could nearly double its profits We collected data on the companies that I’m hoping to test the personalized-fine if it were to adopt such a personalized accepted and rejected the offers. Then we theory soon by working with a company, model for its business. (It didn’t end up used these data to build an algorithm that doing so for a variety of reasons, including would determine the price a company a shift in its product offerings and a change would be willing to pay. In a second in its competitive landscape.) experiment that tested the optimized SERVUS has a relationship with a personalized prices, revenues increased 84 company that Chicago uses to collect percent from the baseline. (For more, read fines. Considering that the City of Chicago the Spring 2018 feature “Are you ready for collected $272 million from parking and personalized pricing?” online at Review. automated camera tickets in 2018, a sizable ChicagoBooth.edu.) bump in fine collection would go a long Fine payments, too, are elastic in the way toward closing its budget gap. Even if sense that individuals have some discretion It may be illegal Chicago decided not to adopt personalized over their choice of whether or not to in some states fines or fine forgiveness for whatever pay. If a fine is affordable, it’s likely that a reason, the proposed research could help person will pay it, either out of a sense of to charge a determine an optimal fine rate to maximize duty or to avoid penalties. Since few people the amount collected—or perhaps two budget for traffic tickets, that money has to personalized fine optimal rates, which would allow the city come from disposable income. Wealthier on each ticket, but to use a different rate for citizens who can scofflaws won’t have much trouble coming demonstrate financial need. up with a few hundred dollars to pay off a city can forgive a Above all, this plan would take better a ticket, but less affluent offenders may portion due, which care of the most financially vulnerable of already be struggling to pay for basic the city’s residents. Beyond Chicago, there necessities, let alone a fine. The person would be equivalent are many places that could benefit from receiving a city-sticker fine might not have a plan that brings in additional revenue had the money for the renewal sticker to personalizing but that doesn’t lead its most financially initially, much less for the fine. And nearly for people with vulnerable citizens toward bankruptcy. We 40 percent of Americans have said they need one city to sign off on this idea, and couldn’t come up with $400 in case of an financial difficulties. the research could begin.—CBR emergency, according to a 2018 Federal Reserve report. Jean-Pierre Dubé is the Sigmund E. Edelstone More ideas for policy makers The choice for many at this point is Professor of Marketing and a Charles E. to borrow the money needed at a high Visit our website to find research Merrill Faculty Scholar. interest rate or to take it from other budget on pressing issues. Go to Review.ChicagoBooth.edu to see citations for research items. It’s a choice between paying the mentioned in this essay.

Spring 2021 Chicago Booth Review 51

285697_48-68.indd 51 2/1/21 7:35 AM ILLUSTRATION BY FRANCESCO CICCOLELLA FRANCESCO BY ILLUSTRATION

52 Chicago Booth Review Spring 2021

285697_48-68.indd 52 2/1/21 7:35 AM WAVERLY DEUTSCH Entrepreneurs, remember the power of a smile What investors really look for in startup pitches

ntrepreneurs trying to build big • Momentum: some traction metrics in companies turn to equity investors terms of users, customers, recurring E to raise capital. If a company is revenue, etc. an established ongoing concern, • Management team: an entrepreneur’s investors can look at cash flows, growth education, industry credentials, and rates, customer retention, capital assets, entrepreneurial experience leverage, and other hard metrics to deter- • Market size: enough demand that there is mine whether to invest and how to value the potential to build a really big company the company. An early-stage company has • Money: how much an entrepreneur few if any hard metrics, however. Startups will need to build the company, and the are often just emerging from the idea percentage of ownership that the investor stage into market trials, and the minimum will be able to maintain in the process of viable products and services they offer are getting to an exit being tested by beta customers. This presents a challenge for the Early-stage investors following this guide, entrepreneur. What will convince angel and or one like it, may believe they are objec- early-stage venture investors that what she tively evaluating startups on factors such is building is not only a feasible company as management team credentials, sustain- but one that has the potential to generate ability of the business model, evidence of a large return on investment, and maybe traction in the market, and large-company even become one of those rare unicorn potential. However, new research indicates companies that will make its investors rich? that regardless of what investors think If she were to Google “what investors they’re doing, the reality is subtly different. look for before funding a startup,” she Investors, these findings suggest, are more would find myriad guides, many written influenced by an entrepreneur’s likability, by angel investors and venture capital- positivity, and happiness than they are ists. Most of these guides include some aware of. One implication of this for entre- variation of the four M’s outlined by preneurs is that your smile matters very entrepreneur and investor Mark Suster much, which is a revelation that I find to be in his blog Both Sides of the Table. surprising—and somewhat disappointing.

Spring 2021 Chicago Booth Review 53

285697_48-68.indd 53 2/1/21 7:35 AM Team is No. 1 What VCs look for in a startup Suster’s four-M framework is largely Institutional venture capitalists who focused on seed and early-stage startups said the echoed in extensive research by Chicago target company’s leadership team was the most important factor in their decision to invest. Booth’s Steve Kaplan, Harvard’s Paul Gompers, University of British Columbia’s Survey responses from seed and early-stage VCs Will Gornall, and Stanford’s Ilya Strebulaev. 241 respondents, 2015–16 survey In a study published in the Journal of Financial Economics, this team sought Share who included the factor among their Most important to answer the question: How do venture top three criteria for investing in a startup consideration in an capitalists make decisions? They surveyed investment decision 885 institutional VCs at 681 companies Management team worldwide from November 2015 through 96% March 2016, asking them to rate their top three investment criteria. Nearly all the Business model respondents said the target company’s leadership team was one of the top factors. 84% 53% Management The researchers narrowed in on the 241 team venture capitalists who focused on seed Product and early-stage investment, and 96 percent 81% in this group included leadership team in their top three criteria. The other popular Market criteria were the company’s business 13% Fit with portfolio model (84 percent), product offering (81 74% percent), potential market opportunity (74 12% Product percent), and fit within the VC’s existing Fit with portfolio 7% Business model portfolio (48 percent). However, when 48% asked to identify the single most important 7% Market investment criterion, more than half the 8% Other factors Deutsch, 2021; Gompers et al., 2017 investors chose the founding team, demon- strating even more clearly their reliance on founder credentials in assessing early-stage Because AngelList often included companies. Asked to define the qualities information from just one or two of they looked for in a startup team, the the categories rather than all three, investors cited ability, followed by industry the researchers were able to send out experience, passion, entrepreneurial experimental emails that randomized experience, and teamwork. the categories included, and they could A research experiment conducted see which ones made an investor more in summer 2013 by Harvard’s Shai likely to click through to learn more Bernstein, University of Southern about a company. They sent 16,981 California’s Arthur Korteweg, and emails about 21 startups to 4,494 active AngelList’s Kevin Laws further validates investors. Recipients opened 48 percent this dependence by early-stage investors of the emails, and in 16.5 percent of these on the characteristics and experiences of opened emails they clicked on the View the founding team, which serve as a quali- button, an indication that the investor ty signal. The researchers used AngelList, wanted more information about the an online platform created to match featured company. The researchers also startups with accredited investors, and When asked to collected data on how often the investor randomly selected a cohort of companies identify their then requested an introduction to the seeking funding. At the time, when inves- company, a further sign of interest. tors joined AngelList, they specified what single most Not surprisingly, as it supports the kinds of startups they were interested in, findings of Kaplan and his coresearch- and then AngelList periodically sent them important ers, emails that included information an email of “featured” companies, with investment about the team saw 13 percent the intention of attracting them back to higher-than-average click-through on the website to learn more. criterion, the View button, with a click rate of 18.7 AngelList’s emails included percent. This bump was even bigger information covering three areas: team more than half when the investor’s area of interest and credentials, current investors, and of investors expertise coincided with the industry market traction. Bernstein, Korteweg, of the startup company—then there was and Laws decided to mimic these emails surveyed chose a 20.4 percent click-through rate. The to see which of the data categories most the founding presence or absence of information in influenced an investor’s decision to the other categories did not significantly learn more about a particular company. team. alter the click rate.

54 Chicago Booth Review Spring 2021

285697_48-68.indd 54 2/1/21 7:35 AM But investors can be fooled Entrepreneurs Not at all surprisingly, investors pre- Investors are smart in prioritizing team ferred teams that looked and sounded over other criteria in their invest- who use the kind of positive. In fact, a 1-standard-deviation ment decisions, research suggests. increase in overall positivity improved a Numerous studies over the decades technical, content- team’s chance of being selected into the have found a correlation between rich language the accelerator by 35 percent, up from 8.5 human capital—education, experience, percent to 11.5 percent. knowledge, and skills—and entre- algorithm picked up But Hu and Ma also found that preneurial success, and more recent on are less likely to investors prioritized this positivity research by University of La Verne’s over all other factors in the pitch, to Byungku Lee suggests that founders’ appear friendly and their detriment. A higher score on the hard work is a significant predictor verbal-ability dimension was actually of new venture success. There is little happy, and therefore negatively correlated with selection. doubt that a great team is critical to to appeal to funders. When the researchers controlled for startup success. But how exactly are both the teams’ education and work investors evaluating teams? experience—characteristics that in Research conducted in 2020 by practice correlate with entrepreneurial Yale PhD candidate Allen Hu and success—they found that this did not Yale’s Song Ma reveals that investors change the results. Hu and Ma then may be sucked in by a smile. Hu and tracked down the companies to see Ma created an experiment to use what happened to them. Looking at a new methodological framework survival rates, jobs created, and the to identify what features in human results of later funding rounds as a way interactions matter in economic to determine the success of the compa- decision-making and to utilize nies in their sample, they discovered cutting-edge technology to empirically that more-positive entrepreneurs, represent, measure, and analyze those indicated by happier facial expressions features across vocal clues, facial and vocal emotions, built companies expressions, and language choices. that raised less follow-on funding They selected a random set of short and hired fewer employees than the startup video pitches created by 1,139 average of the cohort, while those who entrepreneurial teams as part of the exhibited higher verbal ability were process of applying to top accelerator likely to employ more people at their programs including Y Combinator, ventures—even though they were less MassChallenge, and Techstars, likely to have obtained funding. This among others. Eight and a half suggests that entrepreneurs who use percent of the teams were accepted the kind of technical, content-rich into the accelerator program for language the algorithm picked up on which they applied. are less likely to appear friendly and Hu and Ma used voice and facial rec- happy, and therefore to appeal to ognition software to deconstruct the funders, even though their language videos—capturing facial expressions at correlates with more competent teams 0.1-second intervals and tone of voice that build better businesses. and inflection on a sentence-by-sen- The study also suggests that first tence basis, as well as word-by-word impressions disproportionately affect content. This created an enormous investor decisions. When Hu and Ma data set which they then analyzed via limited their analysis to the first five machine learning. Using well-estab- seconds of the pitch, an amount of lished frameworks from psychology, time during which little of substance finance, and linguistics, they were about the business could be commu- able to create 10 distinct metrics—six nicated, the results were similar—the quantifying emotion in voice and more positive the beginning of the facial expression and four analyzing pitch, the more likely the company the content. (See “How to turn voices was to be selected. and facial expressions into data” on the following page.) They examined Implications for entrepreneurs how much time during each video a and investors team displayed positivity or negativity Many years ago, I was coaching an according to the visual, vocal, and entrepreneur through Chicago Booth’s Perfect your pitch content metrics. Weighting each of own accelerator program, the New these measurements, they created an Find past columns and more Venture Challenge. The young man overall pitch score that represented the advice on our website. was an exceptionally talented and “positivity” of the team’s presentation. experienced engineer and presented

Spring 2021 Chicago Booth Review 55

285697_48-68.indd 55 2/9/21 3:14 PM a relevant technology concept in a How to turn voices and facial expressions into data patient, thorough, and thoughtful Researchers created a set of metrics to measure a startup pitch’s persuasiveness. manner. It could not have been more boring. After his presentation, I asked EMOTION MEASURES CONTENT MEASURES him, “Ed, do you like your business?” He frowned, folded his arms, thought for a moment, and replied seriously, “I think it has a chance.” The room laughed nervously, and I said to him, Visual positive Visual negative “Ed, if that is how you feel about your own company, how are the investors Facial expressions supposed to get excited about it?” Needless to say, his startup did not Verbal positive Verbal negative receive funding. I have worked with Use of language hundreds of entrepreneurs, always en- couraging them to show their passion for their companies and to smile and engage with their potential investors. Now there is scientific evidence to back me up. Vocal positive Vocal negative Here’s some advice for early-stage Tone and energy entrepreneurs: research finds that the team you put together is the most important aspect of your business in Verbal warmth Verbal ability investor decision-making and that being Socially perceptive Industry happy and positive about your startup is positivity and execution a huge subconscious signal to investors (words such as (words such as that your team includes people they “together” “intelligence” want to invest in. In your presentation, or “help”) or “design”) immediately smile and express your Vocal valance Vocal arousal enthusiasm for this opportunity, and Range of Excitement level work the backgrounds of your team emotional quality and intensity members into the story early. There are mixed messages for female Deutsch, 2021; Hu and Ma, 2020 entrepreneurs, however. When Hu and Ma looked at the gender composition of the teams in their study, they found that their processes to gather information investor sensitivity to positivity affected Being happy and before meeting entrepreneurs. If female-only teams more intensely positive about your investors are so quick to judge on a than male-only teams. When a female smile, and if this leads them to invest entrepreneur failed to appear positive startup is a huge in underperforming companies, they and warm in her pitch, her chances of subconscious signal should avoid face-to-face meetings until acceptance into the program decreased after they have done a little research. far more dramatically than her male to investors that Request a one-page overview that counterpart’s. In mixed-gender teams, focuses heavily on team background investors were affected by the positive your team includes and company traction, and consciously visual signals from the male team people they want to overweight these criteria. To counter members, and the facial expressions of implicit bias, perhaps suggest that the women didn’t seem to change the invest in. titles be used for the founders, rather investment chances, implying that their than names, to avoid seeing names smiles or frowns were being ignored. that give away the gender or ethnicity However, when the researchers of the entrepreneurs. Investors, only analyzed the vocal signals, if a woman after you assess team competency was pitching with a male colleague, the and determine whether a company impact of her passion, enthusiasm, and fits with your portfolio should you energy switched and became negatively allow yourself to fall for the founders’ correlated with investment chances. dazzling smiles.—CBR Apparently, and sadly still, in 2021, in- vestors prefer that women not upstage Waverly Deutsch is clinical professor at or dominate their male cofounders. Chicago Booth and the Polsky Director of the There’s advice here for early-stage UChicago Global Entrepreneurs Network. investors, as well. Angel investors Go to Review.ChicagoBooth.edu to see citations for research and venture capitalists should change mentioned in this essay.

56 Chicago Booth Review Spring 2021

285697_48-68.indd 56 2/9/21 3:17 PM of higher ed, is the antithesis of scrub- IN-HOUSE bing toilets. In many respects, passing ETHICIST between these workplaces was jarring—so JOHN PAUL ROLLERT jarring as to seem slightly grotesque. Work is a curse, or at least that’s what I learned in Bible study. When Adam and Eve were driven out of Eden, they finally Don’t lose had to work for a living. We all do—well, nearly all of us—but not all jobs are created equal. That’s certainly the lesson I’ve sight of learned, though at times it seems more like a dirty secret. In fact, some jobs can be so rewarding (monetarily, intellectually, suffering even morally) that we lose sight of the burdens others shoulder. Their work Awareness of hardship is fine; becomes invisible to us—even when they understanding it is better are toiling on our behalf.

earned $4.25 an hour at my first job. Decadence and drudgery It was at Sagebrush, a now-defunct “What you don’t necessarily realize when I clothing chain that sold cheap jeans to you start selling your time by the hour is working-class Midwesterners. In the that what you’re actually selling is your 1990s, when I worked there, $4.25 was the life,” Barbara Ehrenreich notes in Nickel minimum wage, a callous phrase that aptly and Dimed, her best-selling account of captures employers’ intention to do the trying to make ends meet on low-wage very least they can by you. I took home work. Ehrenreich was writing her about 70 bucks a week after taxes—until I blue-collar correspondence right about the was held up while working the register. If time I was pulling double shifts scrubbing you can’t put a price on life, you certainly steel showers. What united our efforts—in can on labor: after reassuring me that my addition, of course, to scrawny pay stubs safety was all that really mattered, the and occupational hazards—was their regional manager reassessed my value to transitory nature. Over three months, the company and gave me a 10 cent raise. Ehrenreich tried various jobs that all paid Now I made $4.35 an hour, or nearly $72 less than $10 an hour, knowing full well for an 18-hour weekend. she was a tourist in the penal colony of My second job was cleaning dorms the working poor, whereas my hard labor as an undergraduate at Harvard. The might be regarded as either a stepping pay was considerably better, $9.65 an stone or even a rite of passage and, thus, a hour, but it was beastly work. If retail is way station in the work world rather than a never-ending war against the tedium a terminal. of monotonous tasks and the temerity In either case, it wasn’t where we of capricious customers, cleaning is “belonged,” an invidious description that an all-out assault on corporeality. You often sits comfortably on the tongues of are constantly straining or squatting those who tend to sentimentalize manual or sprinting up stairs. The work was labor as conducive to “building character.” especially gruesome in June after school Certainly, in terms of a moral education, let out; a month of 80-hour weeks left there is much to commend an extended my body hurting in ways that make me engagement with hard, repetitive, and cringe now. deeply unglamorous work, but only if the But once those four weeks were over, ultimate lesson more closely approximates I often had two grand in my pocket, “There but for the grace of God go I” than enough to fund an unpaid internship, a the bootstrapping bunkum of “Anyone summer ritual for academic overachiev- can do it!” The enthusiasm of the second ers that, in the status-conscious precincts lesson treats such work as a kind of

Spring 2021 Chicago Booth Review 57

285697_48-68.indd 57 2/1/21 7:35 AM crucible of capitalist advancement, a To forget that says. “The reward rests not in the task but challenging experience, to be sure, but in the pay.” one that individuals pass through on their someone is doing In contrast to those who so labor, way to bigger and better things—unless, Galbraith described members of a “New of course, they suffer from some kind of some terrible task Class,” a charmed group of individuals personal failing. on our behalf whose work included “exemption from The truth is that most tasks of a menial manual toil; escape from boredom and variety do not constitute “starter jobs.” relieves us of the confining and severe routine; the chance They are just jobs, the work that people trouble of taking to spend one’s life in clean and physically do to make ends meet. This is not to say comfortable surroundings; and some that the individuals who work them don’t any responsibility opportunity for applying one’s thoughts to want more, simply that the “more” in the day’s work.” This class of individuals question tends to be “more pay.” They for it. was “new” for Galbraith insofar as the don’t view the jobs they have as merely ghosts of the Great Depression still haunt- another rung on some endless career ed Eisenhower’s America, and it should be ladder. That idea is essential to the emphasized that what characterized these striver’s ethic and, perhaps, necessary people as a class was not only that work to progress in a competitive enterprise was its own reward for them, but that it system, but it can create the impression was rewarded, too, at least well enough that the waitress, the janitor, or the field to afford the upper-middle-class comforts hand—or, for that matter, the Uber driver, and bourgeois peace of mind we associate the floorwalker at Walmart, or a member with affluence. of the invisible army packing boxes at Galbraith was born in 1908 and there- Amazon—are undeserving of any larger fore was old enough to remember not only consideration than the invisible hand both World Wars and a Great Depression already grants them. but also a time before child labor laws, The danger of an ever-widening central air, and the polio vaccine. He, circle of wealth, the late John Kenneth like everyone else of his generation, was Galbraith wrote in 1958 in his most intimately familiar with a world that could famous book, The Affluent Society, is be cruel and unforgiving regardless of that “we will settle into a comfortable one’s position in society. Cash and connec- disregard for those excluded from its tions could no doubt insulate one from a benefits and culture.” What’s more, he considerable degree of grief and suffering, continued, “there is the likelihood that, but in a world where one out of every 10 as so often in the past, we will develop a children died as infants, as they did in the doctrine to justify the neglect.” US at the dawn of the 20th century, and Willful neglect of the working poor no amount of money could replace a bum may be odious, but it does have the hip, suffering was far more democratic. No benefit of its being self-aware. If often one escaped the gallows of despair. giving little evidence of being especially But the world of 1950s America not informed, the opinions sanctioning only afforded the opportunities for a such neglect, whether or not they appreciation of, or even exposure to, a far more humane existence, there also congeal into a proper “doctrine,” are workday consisting of drudgery. seemed to be a growing community of still susceptible to principled engage- What is drudgery? It’s not merely people, Galbraith contended, who had ment and possible change. The greater tedium or time-consuming tasks. escaped old Adam’s curse altogether. They danger, far greater, is a mere oblivion. Many first-year analysts at investment could take for granted that work “will Ehrenreich warns of it in the conclu- banks spend enough time staring at be enjoyable,” and if not, that it was a sion to her book. “Some odd optical spreadsheets to long for the consolations “legitimate source of dissatisfaction, even property of our highly polarized and of a jackhammer. No, if as a society we frustration.” It wasn’t so much that work unequal society makes the poor almost tend to valorize overwork and deprecate need not be drudgery; it should not be invisible to their economic superiors,” leisure in a manner that would have drudgery. It should be an activity, however she writes. “The poor can see the stunned our forebears after a summer demanding, that is filled with personal affluent easily enough—on television, day bent double in the fields, those satisfaction and even pleasure, a condition for example, or on the covers of who bumptiously embrace work-life that involved the very nature of work as magazines. But the affluent rarely see imbalance aren’t at the head of the queue well as its reward. the poor or, if they do catch sight of for sympathetic consideration, even if them in some public space, rarely know one might think they should seriously get The lives of others what they’re seeing.” their heads checked. Three generations now stand between What accounts for such blindness? It Drudgery, according to Galbraith, us and the debut of Galbraith’s book, may be due, in part, to the cradle-to-grave is work for which the only thing that and affluence in the US has only grown. comforts that, especially in the United commends it is a paycheck. “It is fatiguing The “New Class” is not new anymore, States, so many people enjoy today, a con- or monotonous or, at a minimum, a and it’s far larger. Many people are dition that distances us from an empathic source of no particular pleasure,” he several generations removed from familial

58 Chicago Booth Review Spring 2021

285697_48-68.indd 58 2/9/21 3:18 PM touchstones of toil and economic turmoil, community regards as acceptable,” a spirit of Milton Friedman, we are both free and for those fortunate enough to grow up belief that led him to conclude that “one of to choose. in such blessed conditions, they live and the central economic goals” of a civilized But choice, of either a moral or work at a kind of experiential remove that society is “to eliminate toil as a required professional variety, is only meaningful if it makes a life of drudgery, and the mishaps economic institution.” is informed. That requires, at least among of poverty that so often attend it, at best a On the other hand, they might conclude members of Galbraith’s “New Class,” a kind of academic phenomenon, something that rage is only a proper response to kind of curiosity about the lives of those one can identify and even opine on but manifest injustice, and insofar as the price less fortunate, a patient inquisitiveness that hardly understand. of one’s labor is determined like the price imparts a sense of moral perspicacity. I see this in my own classes at Chicago of any good or service in a free market, by Such an engagement recalls an obser- Booth, disproportionately attended by the intersection of supply and demand, vation by George Orwell in The Road to young people who are familiar with if the market determines one’s labor is Wigan Pier. “I had read the unemployment upper-middle-class comforts. A statistic only worth $15,080 (the annual salary for figures,” he wrote, “but I had no notion of that seems to catch my students especially someone working 40 hours a week for what they implied.” Orwell meant that such off guard is the median household income a full 52 weeks at a minimum-wage job), figures were meaningless to him in a moral in the US, which, according to the Census well, there’s nothing unjust about that. sense if he knew nothing about the lives Bureau, was just over $63,000 in 2018. You are no more required to take such a they intimated. He set about remedying his Given that many of these students have job than I am required to pay you a penny ignorance by undertaking an experiential already enjoyed annual salaries double more than I must for your services. In the regimen that, if more involved, approxi- this amount and have done so without mates the advice I give to my own students: also discharging any of the obligations seek to understand the lives of those who of parenthood or even partnership, the suffer on your behalf. number is a shock to them. (“People live As he chronicled in The Road to Wigan on this?”) And immediately on the heels Pier, Orwell went to observe the lives of of this shock is the bracing realization coal miners in England, at least as they that half of American households get by appeared in the mid-1930s. The second on less, often far less. Indeed, in 2018, for chapter of his book is extraordinary, a those under 65, the federal poverty line literary Grand Guignol of the grimmest was set at individuals making $13,064 or working conditions. Orwell concluded it less—or, for a family of four, a maximum by reflecting on what had changed for the income of $25,900. That year, 38 million miners in recent years, and what remained people made the cut, a number slightly the same for his readers, in addition to the smaller than the combined population of connection between them. “It is not long the 23 smallest states. since conditions in the mines were worse “Most of the people I write about in than they are now,” he wrote. this book do not have the luxury of rage,” David Shipler contends in the opening of There are still living a few very old his best-selling book, The Working Poor: women who in their youth have worked Invisible in America. “They are caught underground, with the harness round in exhausting struggles. Their wages do their waists, and a chain that passed not lift them far enough from poverty between their legs, crawling on all fours to improve their lives, and their lives, in and dragging tubs of coal. They used to turn, hold them back.” If such individuals go on doing this even when they were do not have “the luxury of rage,” that is pregnant. And even now, if coal could because luxuries of any sort, material or not be produced without pregnant emotional, are well out of their reach. women dragging it to and fro, I fancy we Whether they are entitled to feelings of should let them do it rather than deprive rage, however—or, more to the point, ourselves of coal. But most of the time, whether those of us a bit more blessed of course, we should prefer to forget that should vicariously indulge them on their they were doing it. behalf—is a moral question, one with sub- stantial social and political implications. To forget that someone is doing some As an ethics professor, I have strong terrible task on our behalf relieves us of the feelings on such matters, while also trouble of taking any responsibility for it. strongly believing that the lectern is not Choice, of The work is absorbed into some wonder- a pulpit. Students are welcome to come ful, mysterious world whose benefits, like to any conclusions they like. They might, either a moral manna from heaven, seem to appear from like Galbraith, resolve that, when people or professional nowhere and drop right into our laps. “cannot have what the larger community Ignorance is undoubtedly bliss, but it is regards as the minimum necessary for variety, is only always morally unbecoming.—CBR decency,” they are therefore “degraded[,] meaningful if it for, in the literal sense, they live outside John Paul Rollert is adjunct assistant professor the grades or categories which the is informed. of behavioral science at Chicago Booth.

Spring 2021 Chicago Booth Review 59

285697_48-68.indd 59 2/1/21 7:35 AM follows, then, that now that we have a THE GRUMPY vaccine for COVID-19, there should be one ECONOMIST rule guiding its distribution: give it fast JOHN H. to the people most likely to get the virus COCHRANE and spread it to others. (There is some uncertainty how much the vaccines help to stop transmission. I assume here that they do, at least to a significant degree.) It may The seem intuitive that vaccines should go first to those most vulnerable to the disease, COVID-19 but by targeting instead those most likely to spread the virus, we can end the pandemic and, in so doing, provide even greater trolley protection to those vulnerable groups. Nothing matters but the reproduction problem rate. That is not, however, the tack US public-health officials have taken US public-health policy has throughout the pandemic, including fallen behind virus growth in their approach to who should be inoculated first. again and again The recurring failure of our government response to this pandemic has been n late December, a new, more trans- to fall behind the pace of the virus’s missible strain of the virus that causes spread—to permit it to reach and sustain ICOVID-19—B.1.1.7, often called simply the exponential growth. The US Food and Drug “UK variant”—appeared in the United Administration wasted months when the States. As this strain becomes more and vaccines were known to be safe, and wasted more dominant in the US, it will mean the weeks over Thanksgiving delaying approval virus’s reproduction rate, popularly known in a public-relations gambit. We have snafu as R, will be higher. If or where R was a bit after snafu in vaccine distribution. And we below 1 (meaning that each infected person have vaccine rationing that is designed to do spread the virus to fewer than one other just about nothing to stop the spread. person, on average) and the virus was In fact, the reproduction rate has been contracting, the virus will begin moving about the last criterion in the allocation toward exponential growth again. And the scheme. The first allotments of the vaccines virus has already been spreading quickly in are going to protect old folks in nursing our second wave. homes. And this approach isn’t unique As Zeynep Tufekci explained in the to the US: the UK gave its first dose to a Atlantic, “a more transmissible variant is 90-year-old. Germany to a 101-year-old. in some ways much more dangerous than It appears kindhearted to protect a more severe variant. That’s because those who are most likely to die if higher transmissibility subjects us to a they get the virus. It is a purely private more contagious virus spreading with benefit, so one might wonder why the exponential growth, whereas the risk recipients aren’t asked to pay anything from increased severity would have for it. A policy that prioritizes the elderly increased in a linear manner, affecting is really an income transfer to old people only those infected.” in the form of a vaccine. If you want to stop a pandemic, there If those same vaccines were given to is one rule for public-health policy: lower frontline health-care workers, or to young R, the reproduction rate. With R below partiers who just can’t seem to help them- 1, the virus fades away. With R above 1, it selves from giving it to 25 other people, grows exponentially until “herd immunity” including three grandparents, we’d curb stops it, meaning in this case that most the disease, we’d address the externality

Americans have caught the disease. It that one sick person can create many new MAGGIE LI BY ILLUSTRATION

60 Chicago Booth Review Spring 2021

285697_48-68.indd 60 2/9/21 3:29 PM sick people, and we’d protect old people, upon thousands may die as a result of the What good has been achieved by banning all much more effectively. delay. That’s a lesson that should have been a private market for vaccines on top of This is a hard choice, but it is one that painfully obvious from the January–March government allocation? The government competent public-health (and military) delays in addressing travel, not allowing is really not even thinking about the prime bureaucracies are supposed to know how tests to be used, not ramping up tracing market failure—the externality that if I get the to handle. It’s a classic trolley problem with while it could do any good, and so on. virus I might give it to you. In ignoring that, it one person on the left track and thousands In a year, our bureaucracies have not is mostly just achieving an income transfer to on the right. Do you first give a vaccine wrapped their heads around a simple fact: favored groups. to old people who are likely to die if they the point is to stop the exponential spread Suppose there were a free market get sick with COVID-19—freeing them to go of a disease, not (just) to protect individ- in vaccines. The only difference is, the out a bit and freeing nursing homes from uals. Tests should be evaluated by their government would have to buy at market having to implement stringent protection usefulness in stopping spread—and there, prices. How would this world look different requirements—but, in so doing, let the imperfect is far better than nothing—not from ours? First, I think we would see much disease run rampant and the economy tank only by their usefulness in diagnosing a quicker allocation to health workers. Even if for another six months? Or do you give it in a given patient for treatment. Vaccines, used the government were not willing to buy and way that stops the pandemic, to people who right, can stop the exponential spread of give it to them, hospitals would pay whatever individually are less likely to die but, by being a disease, not just protect individuals—or, it takes to give it to staff in their COVID-19 vaccinated, will not spread the disease? The more accurately, enable them greater units immediately. They are forbidden from overall death rate is lower in the latter case, social interaction. doing that now. but only because of reduced secondary One way to distribute vaccines that Second, people and businesses that know infections, which makes it harder for the CDC would target the reproduction rate R would they are hosts to spreading events could buy to claim to have saved those lives. be to create a free market for them on top the vaccine. There is a lot of private incentive Public health knows how to do of government distribution—a suggestion I to combat an externality! this—or used to. When we were eradicating have made previously, and been lambasted Third, the government could do exactly smallpox, the response to an outbreak was for. Only the government can artfully offset what it is doing now. It just might have to to ring-fence the outbreak and vaccinate in externalities and information problems, pay more. If you are a huge fan of the CDC’s order to contain the spread of the disease. the critics say. Every single dose must be allocation scheme and its brilliant targeting They did not give the vaccine randomly to requisitioned to the government’s majestic of externalities, public goods, information the whole population of a country, taking planning effort; not one may be sold in problems, and every other fable from Econ six months to a year to get everyone, order to allow willingness to pay guide the 101, along with its new social justice and protecting people by age, and meanwhile usefulness of the vaccine. Not even a hospital equity goals, fine. It can keep doing that. The letting the disease spread exponentially. emergency room treating COVID-19 patients Treasury just might have to pay a bit more. When combating exponential growth may spring money and butt the line. So the ban on private sales comes down and heterogeneous spreaders, targeting to one thing only: money. The government matters, and speed matters even more. is keeping down the price it has to pay by George Mason University’s Alex Tabarrok forbidding you and me and businesses that noted on the blog Marginal Revolution: could hire back lots of employees from bidding for the vaccine. This is the same The FDA should have approved the Pfizer government that has borrowed $5 trillion vaccine, on a revocable basis, as soon as already, and, as of the time of writing, is the data on the safety and efficacy of its discussing $1,400 checks, far more than the vaccine were made available, around Nov. free-market cost of a vaccine. 20. But the FDA scheduled its meeting of Nothing matters but the reproduction experts for weeks later and didn’t approve The ban on private rate. R < 1, and the virus goes away. R > 1, and until Dec. 11, even as thousands of people it explodes. And given a reproduction rate, were dying daily. We could have been sales comes down nothing matters but how quickly you get on weeks ahead of where we are today. Now to one thing top of exponential growth. Time is exponen- the epidemiologists are telling us that tially valuable.—CBR weeks are critical. only: money. The John H. Cochrane is a senior fellow of the Hoover The issue is not just that thousands were government is Institution at Stanford University and distin- dying daily, tragic as those deaths are. It’s keeping down the guished senior fellow at Chicago Booth. This that the disease was growing exponentially essay is adapted from a post on his blog, The in this crucial 20-day period, and thousands price it has to pay. Grumpy Economist.

Spring 2021 Chicago Booth Review 61

285697_48-68.indd 61 2/10/21 7:55 PM THE BIG QUESTION

to these workers is a big, outstanding HOW HAS THE issue. The other factor is that this recession has been more of a “sheces- sion” than a “mancession.” And that raises issues about women’s continued PANDEMIC CHANGED engagement with the labor force.

Notowidigdo: In terms of the gendered THE LABOR MARKET? aspect of the COVID recession, some of this could reflect preexisting trends. The United States has been somewhat Chicago Booth’s Marianne Bertrand, Steven J. Davis, of an outlier among developed and Matthew Notowidigdo discuss the impact of countries. At the start of the 2000s, women worked about as much in the COVID-19, and policies created to counter it, on the US as in Germany, France, and the US workforce United Kingdom, and in the past couple of decades, those labor-force partic- ipation rates have diverged. Some of what’s going on in the COVID recession could just be a continuation of these trends, although, of course, there’s childcare and other issues that might be disproportionately affecting women during the pandemic.

Davis: I agree on how the gender impact of COVID has played out so far, Marianne Bertrand least temporarily. People are eager to but there are potentially important Chris P. Dialynas Distinguished Service get out of really densely populated areas differences over the longer run and pos- Professor of Economics and Willard for fear of infection risk. Thirdly, we had sibly big benefits for women, especially Graham Faculty Scholar some slow trends toward interacting more-educated women. My survey remotely and working from home before work with [Stanford’s] Nick Bloom Steven J. Davis the pandemic, but they were greatly and [Mexico Autonomous Institute of William H. Abbott Distinguished Service accelerated by the pandemic. Technology’s] Jose Maria Barrero shows Professor of International Business a tremendous decline in the stigma and Economics Bertrand: At the beginning of the associated with working from home. recession, people were talking about There’s also been a lot of investment Matthew Notowidigdo COVID being an economic equalizer, in making work from home better, and Professor of Economics but it’s been exactly the opposite. It has improving remote interactivity through increased inequality across income, technological advances. When we ask across socioeconomic groups, and be- people whether they would like to work Has the pandemic exacerbated trends tween genders. There’s no doubt that the from home, it’s quite uniform across that were already in the labor market lower-wage sector of the labor market is demographic groups—a lot of people or introduced a new range of issues? bearing, so far, the brunt. It’s somewhat would like to work from home two days different from what we’ve seen histori- a week. Working from home is more Davis: It’s exacerbated some trends cally in terms of industrial composition: attractive for those who can. That’s and reversed others. First, it’s been a most recent recessions were more mainly better-educated, professional sharp negative blow to less-educated focused on manufacturing and con- people. That could be favorable for workers. That’s both a continuation struction. This time, it’s service-sector more-educated women. of a decades-long trend, but also workers, gig workers, and self-employed something of a reversal of the past 2–3 workers who have been heavily impacted Bertrand: I very much agree. The other years, in which they had been seeing a and were not covered by a safety net silver lining is that among the higher better labor market. Second, we’ve had prior to the CARES [Coronavirus Aid, educated who have been working agglomeration in cities for more than a Relief, and Economic Security] Act. What from home, there are more fathers century, and that has been reversed, at the new administration does with regard engaging in childcare who might not

62 Chicago Booth Review Spring 2021

285697_48-68.indd 62 2/9/21 3:31 PM do so otherwise. I’m worried, though, Bertrand: My sense of the research on about what’s happening to the current nonpharmaceutical interventions is generation of women who may be that they do not explain much of what really missing out on a valuable time happened with small businesses. in their career and how they are going to recover from that. Also, flexibility Davis: It’s still an open question. can take very different forms. Working Either way, it doesn’t address the on the weekend and late at night is counterfactual. We spent trillions in more of a norm, and that, again, is combating the economic effects of the going to be harder for women, no pandemic. For a few billion or even matter what. tens of billions, we could have estab- lished large-scale randomized testing, How effective was the US federal both for antibodies and infections, government’s initial support for in the population. You can imagine small businesses? randomly sampling census tracts, and paying people $500 to take the test. Bertrand: It’s too early to tell whether What would we have learned if we had the Paycheck Protection Program done that? We’d have learned about has been effective. It has helped, but the prevalence of infection at a point whether it was money well spent is in time, and related it to the tests that still up in the air. Especially for very were done for health-care diagnostic small businesses—retail, food, restau- reasons. We could have informed rants—the situation is dire. There was people about their actual risk. We a quick recovery up until the middle of could have learned about the infec- June, but the data suggest that nothing tion mortality rate more quickly. We’d much has improved since then. A lot have learned about how far we were of small businesses might be sitting on from herd immunity. If we’d tested overdue rental payments. for several months, we’d have learned about the duration of immunity. We’d Davis: If your metric was “spend a lot have learned more directly about the of money quickly,” the PPP succeeded. effectiveness of these nonpharmaceu- How much better did it make things tical interventions. I’m disappointed than they would have been without that as part of the CARES Act, we policy of that particular form? You didn’t set aside resources to do that. could have had more generous funds There’s also the public infrastructure for low-income families or a broader to implement contact tracing. That’s coverage of unemployed workers. a bigger investment, but it would Would small businesses have been benefit us tremendously as we go better off if the federal government “We spent trillions forward with this pandemic and with had spent more money on trying to in combating the future pandemics. get a handle on the dimensions of the coronavirus itself by establishing economic effects What have we learned about the US mass testing and contact regimes? Just of the pandemic. unemployment safety net? knowing how big the risks are, and where they are, and what types of For a few billion or Davis: One thing that the whole nonpharmaceutical interventions are COVID pandemic has revealed most effective could be helpful to all even tens of billions, quite starkly is how inadequate our businesses. Small businesses suffered we could have unemployment-insurance system is so much because we made most of as a claims-processing entity. First, in them shut down. If we had figured out established large- many states, the system fell way behind some way to deal with the virus while in processing claims or in verifying that not hammering small businesses so scale randomized claims were legitimate. People who hard, that would have been a better testing.” lost their jobs had to wait weeks to get outcome than just giving them funds anything. That’s extreme hardship, to tide them over. — STEVEN J. DAVIS and it’s an insult. Secondly, there were

Spring 2021 Chicago Booth Review 63

285697_48-68.indd 63 2/1/21 7:35 AM THE BIG QUESTION

suggestions—about which I’m dubi- unemployment benefits became more ous—that there’s no way to provide generous during recessions, exactly additional unemployment benefits, when the unemployment rate crossed except by a fixed amount. That’s such a specific threshold, and that benefit an easy problem to fix. What does it take extension happened automatically. I for a state to send an unemployment-in- certainly think this is something the surance benefit payment to somebody? US could do. We should also think They’ve either got to know the bank about future stimulus as income account, the name, and the social-se- maintenance or income replacement. curity number—they already know The unemployment rate is continuing that—[for an electronic transfer] or to drop, the labor market is recov- they have to have a physical address, ering, and it may not be clear that a name, and a social-security number we need more stimulus. But many to send a check. Even if the state can’t Americans are still struggling, and do a finely tailored expansion of this relates to the disparate impact of unemployment-insurance benefits, the COVID recession. That’s a good it can send that information to the argument for targeted income relief. federal government, which already sends out checks to tens of millions What are the labor-market of people. This seems like a fixable challenges facing the Biden billion-dollar problem, to revamp administration? the IT infrastructure of all the UI [unemployment-insurance] systems. Davis: There are parts of the labor Thirdly, there are other administra- market that are doing quite well. At tive shortcomings. Like Marianne, the same time, there are parts of the I’m sympathetic to providing some labor market defined by skill groups, UI benefits to gig workers, but that demographics, and geographic areas part of the CARES Act is especially where things are really quite bad. prone to fraud, so we need some We talked a little bit about service investment in systems to deter fraud jobs in business districts—people and abuse, both because they’re working at restaurants, baristas costly and a misuse of the system, in coffee shops, people providing but also because fraud undermines personal services to office workers. political support for the system. That part of the economy is pretty There’s a lot of missed opportunities devastated. So there’s both a skill in improving the back end of our aspect and a spatial aspect. Part of unemployment benefit system. the job for Congress, the president, and local officials is, for those Bertrand: Many have been calling people who aren’t going back to the for building in more automatic jobs they used to have, how can we stabilizers within our UI or safe- create new job opportunities for ty-net system. Seeing the gridlock them quickly? That strikes me as the that we have in Washington, with central challenge from a labor-market people having to wait and fearing the perspective. A lot of this is going uncertainty of what will happen to to happen on the ground level. If them, it’s high time we had more of you look at cities such as Detroit, or these automatic stabilizers built into Gary, Indiana, they basically failed to the systems. The triggers could be “Many Americans adjust to a massive demand collapse. based on the unemployment rates in And of course, the massive demand a particular community, which could are still struggling, collapse in steel, in the case of Gary, automatically change the level of and this relates or the auto industry, in the case of unemployment benefits—something Detroit, was the triggering event. But that would not rely on an act of to the disparate then the local officials in a city have Congress. Think about how much the to recognize, “Look, we’ve got to well-being of all the people who have impact of the COVID repurpose our commercial space, our been affected would have increased if recession. That’s a local economies, and we’ve got to do they had not had to wait for Congress it expeditiously.” That involves busi- to get its act together. good argument for ness licensing, zoning, new ways of targeted income organizing residences and businesses. Notowidigdo: I agree. I was involved Some cities do that well; some don’t. in a project in which we surveyed relief.” I worry a lot about the left-behind countries during the Great Recession. people, the left-behind places, and the We found that in many countries, — MATTHEW J. NOTOWIDIGDO left-behind people in the left-behind

64 Chicago Booth Review Spring 2021

285697_48-68.indd 64 2/1/21 7:35 AM places. That is partly a consequence of “People are going to because higher income people were policy, not just at the national level but more likely to work from home. Even also at the local level. be returning to jobs, in November, it was still at 35–40 percent. When we ask individuals, Notowidigdo: I’m optimistic we’ll but they’re going to “What does your employer plan return to an era of incremental be different kinds for you to do after the pandemic’s reforms. What would that look like over?” they say they’ll be able to do in terms of tackling inequality? I’m of jobs and different about 23 percent of their work from guessing that it would be a modest kinds of employers.” home. In another survey, where increase in progressive taxes, and we’ve asked employers directly, they I doubt we’ll go anywhere close to — MARIANNE BERTRAND give a number closer to 20 percent. what Biden had promised during Before the pandemic, it was more the campaign, or any of the major like 5 percent of all paid workdays. changes that were suggested during So that’s a huge, persistent shift in the Democratic primary, in terms of the structure of working arrange- things like large wealth taxes. But ments that I expect will endure. I think we’ll make small steps to returning to what tax rates looked Bertrand: I agree about working like under Obama and under Clinton. from home. The crisis has also Is that going to make a huge dent in dramatically accelerated the inequality? Maybe a little bit. continued reshaping of our retail sector, a trend we’ve seen for a long Davis: There’s opportunity for time. That’s more a product-market bipartisan agreement. For example, development, but it’s going to have in our working-from-home research, implications for the labor market as when we ask people about how well. Independent restaurants have productive they are when working been badly damaged by the pandem- from home, how efficient they are, ic, while chains appear to have held there’s a lot of heterogeneity. There’s up better. So people are going to be two main reasons for that. For those returning to jobs, but they’re going who are living in a confined space with to be different kinds of jobs and kids and feel they’re less productive different kinds of employers. That’s working from home, that’s pretty hard going to be fascinating to study. for policy to address. But some people feel they’re less productive because Notowidigdo: We’re not just going they’ve got poor-quality internet to be working from home, but connections. And not surprisingly, we’re also going to be consuming those are concentrated among lower- at home more—watching movies or middle-income, less-educated folks. at home instead of going to movie We hear all this desire for spending on theaters, for example. That’s going public infrastructure, and this is a big to be another enduring impact. If gap in the quality of public infrastruc- you think about that in the context ture. Here’s an opportunity both to of health care, a lot of states almost improve the distribution of outcomes immediately started experimenting and to do it in a way that is actually with telemedicine at the start of the productivity enhancing. There aren’t COVID recession—they were reim- many of those, so hopefully we’ll bursing providers to offer telemed- take advantage of that. I can imagine icine visits. This is a sort of forced Republicans and Democrats coalescing experimentation that we’re going around an infrastructure investment to keep with us for at least several program of that sort. years after the pandemic ends. We’re going to figure out what kinds of What is the most enduring effect health care you can consume at the pandemic will have on the US home without having to go to the labor market? doctor’s office or a hospital. There could be some real positives to come Davis: The working-from-home out of that. It could increase access phenomenon is huge. At the peak for people who, prior to the COVID of pandemic-induced lockdowns recession, had difficulty making it in April and May, more than half of to a doctor’s office or going to the all paid hours worked were done at hospital. Telemedicine could provide home—more than 60 percent by our a way for them to get the health care estimates, if you weight by earnings, that they need.—CBR

Spring 2021 Chicago Booth Review 65

285697_48-68.indd 65 2/1/21 7:35 AM THE IGM PANEL WHAT SHOULD THE US DO ABOUT STUDENT DEBT?

The balance of outstanding student debt in the United States is large— nearly $1.6 trillion—and growing. Some policy makers have suggested the US government should forgive all federally held student debt, while others have advocated forgiving a capped amount per borrower. In January, an aide to then president- elect Joe Biden revealed that Biden planned to ask Congress to forgive up to $10,000 per borrower and extend the pause on loan payments that the federal government had offered borrowers throughout most of 2020. Are debt-forgiveness policies regressive (with more benefits accruing to high earners) or progressive (with more benefits accruing to low earners)? And is extending the payment pause an efficient way to encourage pandemic recovery? To investigate these questions, Chicago Booth’s Initiative on Global Markets polled its US Economic Experts Panel. (For more on student-debt relief, see “Canceling all student debt mostly helps high earners,” page 7.)

About the IGM Economic Experts Panels See more online To assess the extent to which economists agree or disagree on major public-policy issues, All responses to these Booth’s Initiative on Global Markets has assembled and regularly polls two diverse panels of polls can be seen at economists, all senior faculty at the most elite research universities in the United States and igmchicago.org. Europe. The panels include Nobel laureates and John Bates Clark medalists, among others. Polls are emailed individually to the panel members, and panelists may consult whatever resources they like before answering. Members of the public are free to suggest questions.

66 Chicago Booth Review Spring 2021 PHOTO ILLUSTRATION BY GLEN GYSSLER

285697_48-68.indd 66 2/9/21 3:33 PM Statement A: Statement B: Statement C: Having the government Having the government Extension of the issue additional debt issue enough additional suspension of payments to pay off all current debt to pay off student on student loans after outstanding student loans up to a threshold, the end of the year loans would be net for borrowers whose [2020] would support regressive. income is below a the recovery more certain level, could be effectively than devoting David Autor, MIT progressive. equivalent resources to “Alongside my kids’ general income-based student loans, I’d like Daron Acemoglu, MIT the government to pay transfer payments. “Yes, this would be much better. off my mortgage. If the Student debt is a problem Steve Kaplan, Chicago Booth latter idea shocks you, for low- and middle-income the first one should too.” “I suspect you could target aid households, not those earning Response: Strongly agree to those more in need than the more than $100k or $150k.” typical student-loan recipient.” Eric Maskin, Harvard Response: Agree Response: Disagree “Blanket loan forgiveness Judith Chevalier, Yale Larry Samuelson, Yale would help those who went “There is evidence that extant to college at the expense “Both would support recovery, income-driven repayment of those who didn’t.” but would be most effective if plans are underutilized. Some Response: Agree targeted to those in most need, form of ex post IDR strategy is and it’s not clear which would do Emmanuel Saez, University likely sensible.” this better.” of California at Berkeley Response: Agree Response: Uncertain “Depends on how the Jonathan Levin, Stanford government debt is going Richard H. Thaler, Chicago Booth “Perhaps, but it seems to be repaid: progressive unlikely if you’re providing “Seems like a weak stimulus. versus regressive taxation, funds only to people who have College grads are already saving or implicit inflation tax.” attended college.” a lot. It’s the bottom quartile Response: Uncertain that needs help the most and Response: Uncertain will spend it.” Response: Disagree

Strongly 25% 11% 0% agree

Agree 48% 80% 5%

Uncertain 27% 9% 37%

Disagree 0% 0% 55%

Strongly 0% 0% 3% disagree

Percentages are weighted by confidence ratings panelists assigned to their own responses.

Spring 2021 Chicago Booth Review 67

285697_48-68.indd 67 2/9/21 3:35 PM How to put a price on people’s appetite for exclusivity hilosophers have long mused about people’s craving for dominance and tendency to engage in activities to assert their superi- ority over others. In consumer choice, this manifests itself as a person’s tendency to want what others desire and the feeling of superiority that comes from possessing what others cannot have, according to Chicago Booth’s Alex Imas and London School of Economics’ Kristóf Madarász. This pursuit of “mimetic dominance” over others can raise people’s valuation of a product above their own intrinsic taste for it. In the researchers’ model, a consumer’s willingness to pay grows as the product’s appeal to others increases, but only if others are excluded from purchasing the product. Setting the price of an item above what many consumers can afford—as luxury brands typically do—is one way to exclude potential customers, as is restricting supply, which results in long queues, for example, to get into restaurants or events. Either way, making a product exclusive allows businesses to charge higher prices and extract greater revenues from consumers. To learn more about this research, turn to page 23.

ILLUSTRATION 68 Chicago Booth Review Spring 2021 BY PETER ARKLE

285697_48-68.indd 68 2/1/21 7:35 AM March to May 2021 See you online While COVID-19 has changed the location of many events, conferences, and programs, Chicago Booth and the University of Chicago continue to sponsor many opportunities for inquiry and for participants to gain insights. The events below will all be held virtually, and more information can be found at the sites listed.

MARCH 29–APRIL 9 APRIL 30 MERGERS AND ACQUISITIONS MANAGEMENT CONFERENCE ChicagoBooth.edu/ma ChicagoBooth.edu/managementconference In this interdisciplinary program, learn the analytical Join fellow alumni and business leaders to share best framework and tools necessary to execute mergers, practices, gain new insights, and discover solutions for acquisitions, and corporate restructurings successfully. today’s management issues.

APRIL 5–14 MAY 3–13 STRATEGIC THINKING FOR TURBULENT TIMES EXECUTIVE PROGRAM IN CORPORATE STRATEGY ChicagoBooth.edu/sttt ChicagoBooth.edu/epcs Learn the foundational skills in strategic thinking Strengthen yourself as a leader by adding strategic that are needed to effectively navigate organizations value to your organization. Learn techniques to through turbulent times. Disruptive forces were understand the competitive structure of an industry underway before the COVID-19 pandemic and have and a company’s competitive advantage. accelerated amid significant market uncertainty that will persist for some time. MAY 11–20

APRIL 12–21 RESILIENT LEADERSHIP FOR HIGH-PERFORMING ORGANIZATIONS HIGH-STAKES STRATEGIES: EXECUTIVE-LEVEL ChicagoBooth.edu/rlhpo STRATEGIES TO MANAGE SYSTEMIC RISKS FOR Tap into your inner gumption, and gain the tools to COMPETITIVE ADVANTAGE lead with courage to create an agile, high-performance ChicagoBooth.edu/hss environment. Move beyond traditional risk-management approaches to uncover fragile interconnections, anticipate potential shock waves, and develop a proactive risk-and-reward ONGOING management strategy for your organization. Gain D&I DIALOGUES high-performance leadership and communication ChicagoBooth.edu/alumni/ frameworks to protect and manage your diversity-and-inclusion-dialogues organization’s reputation. Explore diversity and inclusion-related topics in this event series. APRIL 29–MAY 1 RECONNECT 2021 EXECUTIVE MBA ADMISSIONS EVENTS ChicagoBooth.edu/reconnect ChicagoBooth.edu/exec-events Celebrate your time at Chicago Booth at our Meet students and alumni, and hear from Booth’s reunion event. Admissions team, at regionally focused events.

Since 1898, the University of Chicago Booth School of Business has produced ideas and leaders that shape the world of business. Our rigorous, discipline-based approach to business education transforms our students into confident, effective, respected business leaders prepared to face the toughest challenges. Visit ChicagoBooth.edu/programs for more information about our Full-Time MBA, Evening MBA, Weekend MBA, and Executive MBA Programs, our PhD Program, and our Executive Education courses. Chicago Booth has campuses in Chicago, London, and Hong Kong.

285697_48-68.indd 69 2/9/21 3:37 PM The University of Chicago Booth School of Business 5807 S. Woodlawn Ave. Chicago, IL 60637

WHO IS DRIVING STOCK PRICES? Page 24