NBER Reporter NATIONAL BUREAU OF ECONOMIC RESEARCH

A quarterly summary of NBER research No. 3, September 2020

The 2020 Martin S. Feldstein Lecture

Journey across a Century of Women

Claudia Goldin

My talk will take us on a Journey across a Century of Women — a 120- year odyssey of generations of college-graduate women from a time when they were only able to have either a family or a (sometimes a ), to now, when they anticipate having both a family and a career. More women than ever before are within striking distance of these goals. Fully 45 percent of young American women today will eventually have a BA degree, and more than 20 percent of them will obtain an advanced degree above an MA. More than 80 percent of 45-year-old college-graduate women have children, either biological or adopted. More women than men graduate Claudia Goldin from college, and there is greater similarity in their ambitions and achieve- ments than ever before. This should all make for a very pleasant ending to the ALSO IN THIS ISSUE journey. But that happy ending doesn’t seem to be happening. A few clarifica- Black Officeholders and tions: my evidence concerns the and the history of its college- Spending on Public Goods 8 graduate men and women. I will focus on college-graduate women because they have the greatest opportunities to achieve “career.” Career is achieved Capital Market Risks in Emerging Markets 12 over time, as the etymology of the word — meaning to run a race — would Integrating Renewable Generation imply. A career generally involves advancement and persistence and is a long- into Electricity Markets 16 lasting, sought-after , the type of work — writer, teacher, doc- Are US Treasury Bonds Still a Safe Haven? 20 tor, accountant, religious leader — which often shapes one’s identity. A career NBER News 29 needn’t begin right after the highest educational degree; it can emerge later in life. A career is different from a job. generally do not become part Conferences 26 of one’s identity or life’s purpose. They are often solely taken for generating Program Meeting 34 income and generally do not have a clear set of milestones. NBER Books 35 I recently finished most of a book on this century-long journey. But my book, like the Old Testament, was written in a BCE world — in this case,

*Claudia Goldin is the Henry Lee Professor of at Harvard and director of the NBER’s Gender in the Economy Study Group. She was director of the NBER’s Development of the American Economy Program from 1989 to 2017, and is a past president of the American Economic Association.

Reporter Online at: www.nber.org/reporter-2020-03 Before the Corona Era. Many inequities have NBER Reporter been exposed by the COVID-19 economy and NBER Reporter society, most notably those concerning social justice and our criminal justice system. The The National Bureau of Economic Research is a private, nonprofit research organization COVID economy has also magnified gender founded in 1920 and devoted to objective quantitative analysis of the American economy. Its differences at work and in the home. Women officers and board of directors are: The National Bureau of Economic Research is a private, nonprofit research orga- are essential workers, but cannot be that at Presidentnization and Chief founded Executive in 1920 Officer and —devoted to objective quantitative analysis of the American economy. Its officers andJames board M. of Poterba directors are: home and at work simultaneously. The burden Controller — Kelly Horak CorporatePresident Secretary and — Chief Alterra Executive Milone Officer — James M. Poterba of school closings on working parents that will BOARDController OF DIRECTORS — Kelly Horak continue into the coming year could erase years Corporate Secretary — Alterra Milone of career gains by young women in a way we Chair — BOARDJohn Lipsky OF DIRECTORS Vice Chair — Peter Blair Henry have rarely seen. That is where my talk will take TreasurerChair — Robert — John Mednick Lipsky us. But first, we must journey to the beginning. DIRECTORSVice Chair AT — LARGE Peter Blair Henry I’ll begin the journey 120 years ago, when col- Treasurer — Robert Mednick Susan M. Collins Robert S. Hamada Robert T. Parry lege-graduate women were faced with the stark KathleenDIRECTORS B. Cooper AT LARGEPeter Blair Henry Douglas Peterson choice of family or career (sometimes a job). CharlesSusan H. Dallara M. Collins KarenRobert N. Horn S. Hamada AliciaJames H. M. Munnell Poterba Five distinct groups of women can be dis- GeorgeKathleen C. Eads B. Cooper LisaPeter Jordan Blair Henry RobertJohn S. T. Reed Parry Jessica P.Charles Einhorn H. Dallara JohnKaren Lipsky N. Horn DouglasHal Varian Peterson cerned across the past 120 years, according to their MohamedGeorge El-Erian C. Eads LaurenceLisa Jordan H. Meyer JamesMark M.Weinberger Poterba changing aspirations and achievements. Group Diana FarrellJessica P. Einhorn KarenJohn Mills Lipsky JohnMartin S. Reed B. Zimmerman One graduated from college between 1900 and Helena MohamedFoulkes El-Erian MichaelLaurence H. Moskow H. Meyer Jacob A.Diana Frenkel Farrell AliciaKaren H. MunnellMills Mark Weinberger 1919 and achieved “Career or Family.” Group Helena Foulkes Michael H. Moskow Martin B. Zimmerman Two was a transition generation between Group DIRECTORSJacob A. BY Frenkel UNIVERSITY APPOINTMENT One, which had few children, and Group Three, Timothy Bresnahan, Stanford Samuel Kortum, Yale which had many. It achieved “Job then Family.” Pierre-AndréDIRECTORS Chiappori, BY UNIVERSITY Columbia APPOINTMENTGeorge Mailath, Pennsylvania Group Three, the subject of Betty Friedan’s MaureenTimothy Cropper, Bresnahan, Maryland Stanford JoelBenjamin Mokyr, Hermalin, Northwestern California, Berkeley , graduated from college Alan V.Pierre-André Deardorff, Michigan Chiappori, Columbia CeciliaSamuel Elena Kortum, Rouse, Yale Princeton The Feminine Mystique GrahamMaureen Elliott, California,Cropper, Maryland San Diego RichardGeorge L. Mailath, Schmalensee, Pennsylvania MIT\ between 1946 and 1965 and achieved “Family EdwardAlan Foster, V. Deardorff,Minnesota Michigan LarsJoel Stole, Mokyr, Chicago Northwestern then Job.” Group Four, my generation, graduated Bruce Hansen,Graham Wisconsin-Madison Elliott, California, San Diego IngCecilia o Wa lElena t er, New Rouse, York Princeton between 1966 and 1979 and attempted “Career BenjaminEdward Hermalin, Foster, California Minnesota, Berkeley DavidRichard B. Yoffie,L. Schmalensee, Harvard MIT Lars Stole, Chicago Ing o Wa l t er, New York then Family.” Group Five continues to today and DIRECTORSBruce Hansen, BY APPOINTMENT Wisconsin-Madison OF OTHERDavid ORGANIZATIONS B. Yoffie, Harvard desires “Career and Family.” TimothyDIRECTORS Beatty, Agricultural BY APPOINTMENT and Applied Economics OF OTHER Association ORGANIZATIONS College-graduate women in Group One Martin TimothyGruber, American Beatty, Agricultural Finance Association and Applied Economics Association aspired to “Family or Career.” Few managed both. Philip Hoffman,Martin Gruber, Economic American History Finance Association Association In fact, they split into two groups: 50 percent Arthur Kennickell, Philip Hoffman, American Economic Statistical History Association Association never bore a child, 32 percent never married. In Robert ArthurMednick, Kennickell, American American Institute ofStatistical Certified Association Public Accountants Dana M.Lynn Peterson, Reaser, The National Conference Association Board for Business Economics the portion of Group One that had a family, just Lynn Reaser,Robert National Mednick, Association American for Institute Business of EconomicsCertified Public Accountants a small fraction ever worked for pay. More Group Peter L. Rousseau, Peter L. Rousseau,American American Economic Economic Association Association Two college women aspired to , but the Gregor GregorW. Smith, W. CanadianSmith, Canadian Economics Economics Association Association WilliamWilliam Spriggs, Spriggs, American American Federation Federation of Labor of and Labor and intervened, and this transi- CongressC ofongress Industrial of Industrial Organizations Organizations tional generation got a job then family instead. Dana M. Peterson, The NBER depends on fundingThe Conference from individuals, Board corporations, and private foundations As America was swept away in a tide of early mar- to maintain its independence and its flexibility in choosing its research activities. Inquiries riages and a subsequent baby boom, Group Three concerningThe NBERcontributions depends may on befunding addressed from toindividuals, James M. corporations, Poterba, President and private & CEO, college women shifted to planning for a fam- NBER,foundations 1050 Massachusetts to maintain Avenue, its independence Cambridge, MAand its02138-5398. flexibility inAll choosing contributions its to the NBERresearch are tax-deductible. activities. Inquiries concerning contributions may be addressed to James ily then a job. Just 9 percent of the group never M. Poterba, President & CEO, NBER, 1050 Massachusetts Avenue, Cambridge, married, and 18 percent never bore a child. Even MA 02138-5398. All contributions to the NBER are tax-deductible. The Reporter is issued for informational purposes and has not been reviewed by the Board though their labor force participation rates were of Directors of the NBER. It is not copyrighted and can be freely reproduced with appro- low when they were young, they rose greatly — to priate attributionThe Reporter of issource. issued Please for informational provide the purposesNBER’s Publicand has Information not been reviewed Department with copiesby the of Board anything of Directors reproduced. of the NBER. It is not copyrighted and can be freely 73 percent — when they and their children were reproduced with appropriate attribution of source. Please provide the NBER’s older. But by the time these women entered the RequestsPublic for Informationsubscriptions, Department changes ofwith address, copies ofand anything cancellations reproduced. should be sent to Reporter, National Bureau of Economic Research, Inc., 1050 Massachusetts Avenue, workplace, it was too late for them to develop Cambridge,Requests MA for 02138-5398 subscriptions, (please changes include of address, the current and cancellations mailing label), should or be by sent email to their jobs into full-fledged careers. [email protected] Reporter,. Print National copies Bureau of the of Reporter Economic are Research, only mailed Inc., to1050 subscribers Massachusetts in the US “Career then Family” became a goal for many and Canada;Avenue, those Cambridge, in other MA nations 02138-5398 may request (please includeelectronic the currentsubscriptions mailing atlabel), https:// my.nber.org/email_preferences.or by email to [email protected]. Print copies of the Reporter are only mailed to sub- in Group Four. This group, aided by the Pill, scribers in the US and Canada; those in other nations may request electronic sub- delayed marriage and children to obtain more scriptions at www.nber.org/drsubscribe/.

2 NBER Reporter • No. 3, September 2020 ians, and administrators after the kids were Five Groups of College-Graduate Women: A Century of Work, Marriage, and Children sufficiently grown. For Group Four, the

Group, College Years, Never Married, Never Married, No Children Labor Force Labor Force Pill and its dissemination to young, single (Birth Years) by Age 30 by Age 50 (by Age 45) Rate at 25-29, Rate at 45-49, women enabled the delay of marriage and Achievement/Aspiration if Ever Married if Ever Married family and helped boost their investment Group 1: 1900-19 (1878-1897) 53% 32% 50% ~20% 30% in a career. But the biological clock ran out Career or Family for many of these women. Group Five has Group 2: 1920-45 pushed back marriage and family even fur- (1898-1923) 38% 19% 36% 28% 58% Job then Family ther, but birth rates have risen, in part due to assisted reproductive technologies that Group 3: 1946-65 (1924-1943) 16% 9% 18% 35% 73% have enabled this group to “beat the clock.” Family then Job The transition wasn’t swift and it Group 4: 1966-79 wasn’t due mainly to dissent. Instead, it (1944-1957) 21% 9% 27% 76% 85% Career then Family was often due to technological advances,

Group 5: 1980-2000* increased earnings, and greater . (1958-1978) 27% 12% 21% 83% 84% Career and Family Aspirations and achievements of col- lege women greatly changed across the

*Group 5 extends to the present but is listed here as having an upper birth year limit of 1978 to track its members to their early forties past century, with increased income, the Source: Researcher’s calculations (from voluminous data sets) mechanization of the household, and tech- nological improvements in fertility control Table 1 and assisted reproductive methods. But the education and a promising professional effect by enabling women to be financially structure of work and the persistence of trajectory. Consequently, the group had independent — they didn’t have to marry. social norms, no matter how much weaker high employment rates when young. But After Group One, as women’s potential they have become, have limited the success the delay in having children led 27 percent earnings rose and as substitutes for house- of college-graduate women in achieving to never have children. Now, for Group hold goods became cheaper, husbands’ career and family. Five the goal is career and family, and preferences, rather than necessarily chang- An important accompaniment to although they are delaying marriage and ing, became more expensive. Though fam- the transition across the Groups concerns childbirth even more than Group Four, ily came first for Group Three, college changes in customs and norms. For the just 21 percent don’t have children. women planned their home confinement past 50 years, the General Social Survey You may be thinking that, because and their eventual escape. They trained to has asked respondents whether they agree of large increases in college graduation, be teachers, nurses, social workers, librar- more or less strongly with the statement: most of the differences “Preschool children are across the groups con- likely to suffer if their cern selection into who Evolving Sentiment: Mothers’ Employment and Childrens’ Wellbeing mother works.” The goes to and graduates responses are depicted Share of males and females who agree with the statement: from college. The sur- “Pre-school children are likely to suffer if their mother works” by the respondents’ prising finding is that 100% birth years. As can selection is not that 90 be seen, agreement is important. I’ve tracked 80 always less for women college entrance classes 70 than for men, and from the 1890s to the 60 Males decreases for both men

1990s of women who 50 and women by birth have similar ability 40 Females cohort. It also increases and parental resources. 30 with age, since ear-

Their marriage ages and 20 lier birth cohorts were birth fractions track generally older when 10 those of the total col- interviewed than more 0 lege- graduate group 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 recent birth cohorts. astoundingly well. Birth year of respondent Norms became more Treatment, not selec- expensive to sustain, Data represent 5-year moving average tion, dominates. Source: Researcher’s calculations using data from the General Social Survey and they changed. At College for Group the same time that One had a treatment Figure 1 the cost of not work-

NBER Reporter • No. 3, September 2020 3 ing rose, childcare became more avail- Even though a succession of passed the baton to the next, and as able, more commonly used, and gener- women, group after group, advanced actual barriers fell and social norms ally more acceptable. on this journey, women’s careers still changed, the real underlying problem To measure the degree of success often take a back seat to those of that fuels differences in occupations, at achieving both promotions, and pay career and family has been revealed. that women in these Fraction of College-Graduate Women with Career and Family Unquestionably, groups achieved, I 35% classic discrimina- created definitions. tion, bad actors, sex- Family means having 30 Age group ual harassment, and 35–39 a child, biological 25 biased workers and or adopted, but not 50–54 exist. But necessarily a husband 20 most of the differ- or partner. (Sorry, 15 ence is due to some- dogs do not count as thing else. surrogate children). 10 To paraphrase

Career is achieved 5 Betty Friedan, the by exceeding a level new “problem with of income for three 0 no name” is the 1931–37 1938–44 1945–50 1951–57 1958–65 years in each five- Birth years notion of Greedy year period where Work — that there the income level is Career is defined for a woman as earning above the 25th percentile of the full-time, full-year distribution for comparable are large nonlineari- males for any three years out of five in the age bracket. Family is defined as having at least one child. given by the income Source: Researcher’s calculations using data from the US Bureau of Labor Statistics, and the US Census Bureau, CPS ties and convexities of a man of the same in pay. To have a fam- age and education at ily takes the time of the 25th percentile Figure 2 at least one parent. of the male distribution. Several lon- their husbands. The most recent group There is no way to contract out all gitudinal datasets are used, namely has expressed its disappointments and childcare, and one wouldn’t want to the National Longitudinal Survey of frustrations by focusing on issues such do that, or why have children in the Youth (1979) and the Health and as bias, pay inequity, transpar- first place? Parents have children to Study (HRS) linked to ency, and . spend time with them. For college Social Security and tax data. But as each group progressed and graduates, the gender gap in earnings Interestingly, suc- is an indication and cess at career and Ratio of Female to Male Median Annual Earnings, 1960–2017 a symptom of career family for women blockage. increased both across 0.85 Women earn and within cohorts. less than men, on The success rate for 0.80 average. The ratio women in their mid- All workers of women’s earn- 0.75 50s is around 30 per- ings to men’s, often cent — half that for 0.70 adjusted by hours of men — for the lat- College graduate workers work, is referred to est group that can be 0.65 as the gender earn- observed until that 0.60 ings gap, since it is age. This is the group often given as the born between 1958 0.55 log of the ratio. The and 1965. But the suc- ratio for all workers cess rate for that birth 0.50 narrowed consider- group of women was 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 ably from the early just 22 percent when Data is for full-time, full-year workers. Three-year centered moving averages are used for both series. 1960s until today, Source: Researcher’s calculations using data from the US Bureau of Labor Statistics, the Current they were in their late Population Survey, and the US Census Bureau but is still around 30s, or 40 percent of 0.8. That for college- the success rate for graduate women to Figure 3 comparable men. college-graduate men

4 NBER Reporter • No. 3, September 2020 followed a similar path until the late are on-call at work. The reasons for both In consequence, he earns more than 1980s, when it flattened out. Some gender inequality and couple inequity she does. That gives rise to a gender of the clues as to why the ratio is still are the same. The issues are the two sides gap in earnings. It also produces cou- substantial and why the ratio for col- of the problem mentioned above. ple inequity. If the flexible job were lege-graduate women to men became Many jobs, especially the higher- more productive, the difference would smaller than for the aggregate after the earning ones, pay far more on an hourly be smaller, and family equity would be late 1980s are: basis when the work is long, on-call, cheaper to purchase. Couples would rush, evening, weekend, and unpre- acquire it and reduce both the fam- • The gender gap in earnings dictable. And these time commitments ily and the aggregate gender gap. They exists for both annual earnings interfere with family responsibilities. would also enhance couple equity. and those on an hourly basis, The problem is illustrated here. Note that even if these were same- so it is not just due to the fact One job (the gray line) is flexible sex couples, there could still be couple that women work fewer hours. and has a constant with respect to inequity without gender inequality. And hours. The other job (blue line) is not even if couples wanted a 50-50 relation- • Women with children earn less so flexible and has a wage rate — the ship, high earnings for the position that than women without children. slope of the earnings curve — that rises had less-controllable hours could entice with hours. A couple with children them to engage in a new version of an old • Earnings gaps increase with can’t both work at the blue dot. They division of household labor. age up to a point, and they could both work at the gray dot. But if What are the solutions? First off, increase any solution must with joy- Gender Inequality and Couple Inequality involve lowering ous events the cost of the ame-

like births Earnings per week nity — temporal flex- and, often, ibility. The simplest

marriage. is to create good sub- Less flexible position stitutes. Clients could • Gaps are be handed off with no greater at loss of information. the upper Inflexibility premium Successfully deployed end of male IT could be used to earnings and pass information with education More flexible position little loss in fidelity. levels. Teams of substitutes, not teams of comple- • The more ments, could be cre- unequal ated, as they have Hours per week H* earnings are been in pediatrics,

for an occu- For illustrative purposes only anesthesiology, veter- pation, the inary medicine, per- lower are Figure 4 sonal banking, trust women’s and estate law, soft- earnings relative to men’s. they did, they would be leaving a lot of ware engineering, and primary care. money on the table: for each of them The cheaper the amenity, the more lin- • The gender earnings gap it is given by the distance between the ear total pay becomes by hours worked. is greater in occupations two dots. So one works the flexible, less But the tale I have been telling that have more demands on remunerative job and the other works has been set in a BCE world. In mid- employee time and where face the less flexible, more remunerative job. March 2020, suddenly and swiftly, we time and client relationships More often than not, the man takes the descended into a DC (During Corona) matter most. less flexible, higher-paying job. world. Most workers sheltered in place For many highly educated couples and worked from home. Fortunate The flip side to gender inequality is with children, she’s a professional who children had online schooling and at- couple inequity. Working mothers are is also on-call at home. He’s a profes- home help. Less fortunate workers were on-call at home whereas working fathers sional who is also on-call at the office. deemed essential and often worked in

NBER Reporter • No. 3, September 2020 5 unhealthful circumstances. Less fortu- temporarily furloughed care work- the AC/DC world will be the BCE nate children lost valuable schooling. ers who had worked in their homes. world on steroids. How does our understanding of the That greatly increased the child-care Here, I must go even further out BCE world of career and family help us demands on mothers. But there was on a data limb since, even though the understand the impact of the new DC also more parental sharing, since many first day of school is imminent as I am era, and what will come after? I will focus households had both parents at home writing this, districts are still debat- on college-graduate, employed adults full time. Consequently, the fraction ing what they will do. One possible and their families. These parents and of performed by women scenario is that in the AC/DC world, their children are clearly in the more for- fell, even as the absolute number of total child-care and home-schooling tunate category. I use hours will be halfway the American Time between what existed Weekly Childcare Provided by Employed, College-Graduate Mothers Use Survey (ATUS) For married mothers with employed, college-graduate husbands in the BCE and DC from 2010 to 2019 worlds. That makes 45 hours to compute child-care Before Corona Era (BCE) sense if schools and 61% 75% hours of mothers and 40 During Corona (DC) child care are avail- fathers by the age of 35 After Corona and During Corona (AC/DC) able half the week. their youngest child. But because of non- 30 70% The child-care 66% 54% linearities in work, 25 Fraction of total hours of the moth- childcare hours that are one member of the 70% 20 59% 54% provided by the mother ers in the BCE world 52% 74% couple will go back 15 are given by the dark 60% to work full time and 10 gray bars, and the 58% 60% 83% the other will work 5 fraction of the total 65% part-time from home parental child-care 0 and take care of the hours is given above Youngest age <1 Age 1<3 Age 3<6 Age 6<13 Age 13<18 kids whenever in-per- the bars. In the BCE son school is out and Daily childcare x7. BCE hours come from a sample of women in the ATUS who were currently employed, college graduates with world, college-gradu- at least one child younger than 18. DC hours are estimated by increasing BCE hours by 1.51 for mothers and 1.9 for fathers virtual school is in. (based on studies from April 2020) and then adding four additional hours per day for school-aged children. AC/DC hours for the ate employed mothers couple are the average of BCE and DC hours, but fathers are given only BCE childcare hours (assuming full-time work). Mothers If history is any are assumed to be doing the rest of the childcare. with college-graduate Source: Researcher’s calculations using data from the US Bureau of Labor Statistics guide, men will go back employed husbands to work full time and and school-aged chil- Figure 5 revert to their BCE dren were working around 60 percent hours greatly increased. For those with childcare levels. Women will take up the of total child-care hours. That fraction sick relatives, other care hours also slack and do a greater share of the total. was higher for mothers with the young- increased, and for single mothers, care The bottom line may be that there est children and lowest for those with hours must have been overwhelming. will be no net gains for working women the oldest children. We are now moving to an entirely in the AC/DC world. What they gain The descent into the DC world new AC/DC (After Corona/During from minimal school and daycare open- almost doubled total child-care hours Corona) world. Draconian pandemic ings, they lose from less parental help for working couples with children. restrictions have been partially lifted at home. Because of convex hourly pay, The dark blue bars denote the hours in some schools and daycare facilities. couple equity remains expensive for that mothers contributed to child care Daycare centers opened in most states the family unit. That expense persists and are derived from various surveys by the early summer. There will probably from the BCE world, and the careers done in the United States and else- be full-time in-place schooling in smaller of many young women will take a back where in April 2020, together with school districts, part-time schooling seat on this journey. many assumptions. (The ATUS was together with part-time virtual at-home The corrective in the BCE world not administered in March and April schooling in most of the larger, urban was to change the workplace by driv- 2020, and although it was restarted in districts, and entirely virtual at-home ing down the price of flexibility. The May, those numbers won’t be available schooling in other districts. corrective in the AC/DC world must for some time.) But full-time work has returned to change the care place by driving down In mid-March 2020, almost 90 many offices, stores, workplaces, con- the cost of child care and other fam- percent of school-aged children were struction sites, factories, and elsewhere. ily demands. But how can one do that not physically in a school, and most What can we expect to happen to the safely and equitably? child-care facilities for younger chil- child-care and home-schooling burdens When public and free elementary dren were shuttered. Many families placed on parents? For most mothers, schools spread in the United States in the

6 NBER Reporter • No. 3, September 2020 19th century, and when they expanded cially those from lower-income fami- employ those without jobs, meaning, during the high school movement of the lies. They could free parents, especially and direction and give them some- early 20th century, a coordinated equi- women, to return to work. I’ll repur- thing worthy to do: educate the next librium was provided by good govern- pose a name and call them the “Civilian generation and help women go back to ments. Good government today could College Corps” — a new CCC. work full time, either in their homes or do the same thing. We need to find safe Some of the Corps’ educational on-site. ways to have classes for children — for work could be done remotely. The In the BCE world, if the cost of their futures and for their parents’. Corps could support beleaguered par- flexibility were much lower, we would As in the Great Depression, we ents too exhausted to correct their solve the problem of Greedy Work and have unemployed labor. Today, many children’s essays and too confused achieve both gender equality and cou- of the unemployed are highly educated to help their children with algebra. ple equity. In the AC/DC world, we recent college graduates and gap-year Other Corps members could be in the must also reduce the cost to parents college students with little to do. They classroom, helping districts cope with of educating and caring for children of could be harnessed in a new Works having fewer teachers because some all ages. The original journey was from Progress Administration manner and older teachers don’t want to return to career or family to career and family. put to work educating children, espe- a school building. The Corps could The Journey continues.

NBER Reporter • No. 3, September 2020 7 Research Summaries

Black Officeholders and Spending on Public Goods

Trevon D. Logan

The United States has long had cians related to policy in a way that was racially homogeneous political leader- unexplained by the demographics of the ship. For example, 97 percent of all voting population? Is there any evidence Republican elected officials and 79 that their policies were effective? And percent of all Democratic elected offi- was that leadership related to the subse- Trevon D. Logan is a cials are White.1 Would more Black, quent violence of the Jim Crow era? research associate in the NBER’s Hispanic, and Asian officials advance Reconstruction offers a unique Development of the American different policies in office, or would their opportunity to answer these questions Economy Program. He is the policies be similar since those officials because the enfranchisement of Black Hazel C. Youngberg Trustees would represent the same communities? men was both unexpected and univer- Distinguished Professor of Theoretically, the race of a politician sal. In fact, enfranchisement came to Economics and interim dean of may or may not matter. Under a classic the formerly enslaved before the 15th social and behavioral sciences in median voter model, specific candidate Amendment of the Constitution in 1870 the College of Arts and Sciences at demographics have no effect: policymak- by way of the first Reconstruction Act in The Ohio State University. He is ers reflect the preferences of the elector- 1867. The Act required new state consti- currently director of the American ate.2 At the same time, a candidate’s race tutions in the former Confederacy that Economic Association’s Mentoring may affect the electorate by increasing or allowed for manhood suffrage irrespec- Program and member of the edi- depressing turnout, leaving the median tive of race, color, or religion, and for torial boards of the Journal of voter endogenous to the demographics the establishment of governments under Economic Literature and the of the candidates for office.3 In mod- those new constitutions. In many areas Journal of Economic Perspectives. ern citizen-candidate models, however, of the South, Black turnout for con- His current research focuses politicians differ from the median vot- stitutional ratification and subsequent on racial inequality and economic er’s preferred policy, and there would elections exceeded 90 percent.6 Along history. Logan received his BS in be a race effect to the degree that politi- with the enfranchisement of one million economics from the University of cians of the same race have similar policy Black men, the widespread election of Wisconsin-Madison in 1999, his preferences.4 Blacks in the South was a striking revolu- MA in demography in 2003 and his Racial segregation, the geographic tion in both American and global politi- PhD in economics in 2004, both concentration of Blacks in cities in cal history. from the University of California, regions outside of the South, endog- Building on the narrative work of Berkeley. His research has received enous incorporation and municipal historians of Reconstruction, I con- support from the National Science boundaries, and political districting and structed a database that matched every Foundation and the Washington redistricting make it difficult to disen- Black official to his county of service Center for Equitable Growth, tangle contemporary candidate demo- during Reconstruction to quantitatively among others. graphics from the communities they rep- measure the impact of Black officehold- Logan was born and raised in resent today. I have therefore turned ing.7 Going further, I used the narrative St. Paul, MN and currently lives to history, particularly the history of approach pioneered in macroeconomic in Columbus, OH. He enjoys Reconstruction, to answer questions history to qualitatively identify the areas tweeting (@TrevonDLogan), about the cause and effect of Black polit- of policy agreement among Black elected Marvel Comics, Legos, and col- ical representation.5 I ask three related officials that were at odds with those of lege football. questions about the role of race in polit- Whites at the time.8 I found two areas of ical representation: Were Black politi- broad agreement: land policy and pub-

8 NBER Reporter • No. 3, September 2020 lic schools. Black policymakers sought A highly disproportionate number of used to provide a range of public goods, to use tax policy to induce the sale Black officials were people who gained such as schools and improved roads, of unimproved land. The basic idea their freedom before the end of slav- which were not available in the ante- was not to use taxes on land to seize ery. At the same time, both narrative bellum era. The results suggest that property for nonpayment of taxes, but and quantitative evidence shows that each additional Black politician could rather to alter the opportunity costs of free Blacks in 1860 correlated well explain roughly half of the increase in large landholding. Black officials were with Black officials. This correlation per capita taxes from the antebellum also strong supporters of public educa- is free from potential errors in the esti- era to Reconstruction, an increase that tion, which was not widely provided in mating equation, such as preferences was the focus of intense contemporane- the South before the Civil War. for redistribution. As examples, free ous political protests. Blacks were disenfranchised for more Tax Effects of Black Politicians than 25 years before the Civil War Black Politicians and Policy began, and they are uncorrelated with Since there was little use — or even any Reconstruction political outcomes Given that Black politicians appear scope — for munic- to have had a sizable ipal taxation at the effect on tax revenue, time, I use county Geographic Distribution of Black Politicians During Reconstruction, 1865–1877 the open question is taxes per capita as the whether those reve- measure of the pub- nue effects impacted lic finance effect of the two areas of pol- Black political lead- icy agreement among ership. Ordinary Black politicians. For Least Squares (OLS) land policy, I test regression estimates whether higher taxes imply that each addi- 0 politicians in 1870 led to more tional Black politi- 1 politician farms in the same 2-4 politicians cian was correlated 5+ politicians county, which would with a $0.09 increase be suggestive evi- in per capita county dence of the breakup tax revenue in 1870. of existing farms. Although this corre- The results, however, lation is inherently show that higher interesting and the taxes in 1870 led to Data represent all major Black officials at the national and state levels and a majority of local officeholders first evidence of tax Source: T. D. Logan, NBER Working Paper 26014 fewer farms in 1880, effects of Black pol- exactly the opposite iticians at the time, Figure 1 effect of the tax pol- the number of Black icies advocated by officials could be related to electoral such as party vote shares. The instru- Black politicians. Recalling the poten- preferences for redistribution. Black mental variable estimates are twice as tial effects of tax policy, higher taxes political leadership could have been large as the OLS estimates that do not on property could have spurred land- more likely in areas which had more correct for the endogeneity problem, owners to put more land into produc- progressive attitudes toward racial rela- suggesting that each additional Black tion, and this could have altered the tions and/or redistribution, which politician increased per capita county terms of labor negotiations for Black would overstate the relationship. On revenue by more than $0.20 per cap- farmers. I find that taxes had a positive the other hand, if White resistance and ita. This is more than a ’s hourly effect on the share of all rental farms preferences for low levels of redistri- wage at the time. that were tenant farms, with each addi- bution were an increasing function of Is this a large effect? From the tional dollar of per capita tax revenue Black political success, the relationship contemporaneous response, yes. For increasing the share of tenant farming between politicians and tax revenues example, in 1874, Whites in South by 4 percent. could be understated. Carolina organized a Taxpayers’ For education, county taxes were To overcome this endogeneity, I Convention to protest high local taxes positively related to school enrollment use an instrumental variable for Black in Congressional Reconstruction. The for both Blacks and Whites. Estimates policymakers during Reconstruction: commissioned report noted that taxes from historians suggest that one-fifth the number of free Blacks in 1860. increased by $0.38 per capita, and were of local taxes were used to fund pub-

NBER Reporter • No. 3, September 2020 9 lic education. Using the relationship lence. For example, George Barber fled his Given the extensive violence of the between Black politicians and taxes home in Fairfield County, South Carolina, time, I focus on the violence visited upon and between taxes and school out- over Ku Klux Klan death threats in 1871. Black politicians in the Reconstruction comes, I estimate that a one standard Theophilus Steward of Georgia received era to see if it was related to the public deviation increase in the number of death threats after he asserted that juries finance policies in their local communi- Black politicians resulted in an addi- should involve both Black and White ties. More specifically, what was the role tional 34 Black students enrolled in citizens. Charles Caldwell of Mississippi of taxation in the likelihood of attacks school and an additional 125 White was murdered in 1875, months after on Black politicians? I use the narrative students enrolled in school. The results he escaped an armed mob by fleeing to history of political attacks against Black also imply that the politicians to test this same change led to Black Politicians, County Taxes, and Violence During Reconstruction relationship. Overall, a decline in Black Beginning in the early 1870s, Southern Whites began a violent the likelihood of a vio- illiteracy: a one and wide-spread campaign to undo the Reconstruction process lent attack increased by standard deviation more than 25 percent All counties Counties with Black Counties with Black Counties with Black increase in Black politicians politicians who were politicians who were for each additional dol- County taxes per capita not victims of violence victims of violence politicians reduced $2.0 lar in per capita tax rev-

Black illiteracy $1.93 enue collected in 1870. at age 10 by more Even when restricting 1.5 than 30 persons $1.56 the analysis to counties and illiteracy at age $1.38 with Black representa- 15 by 15 persons. 1.0 $1.14 tion, larger tax revenues Interestingly, there $0.97 were strongly correlated $0.90 $0.89 $0.86 is no similar effect 0.5 with an increased likeli- of Black politicians hood of a violent attack on White illiteracy. against Black policy- As a further check, 0.0 makers. All of this meant I use adult liter- 1870 1880 1870 1880 1870 1880 1870 1880 an end to the tax policies acy after the end Source: T. D. Logan, NBER Working Paper 26014 of Black political leaders. of Reconstruction, The removal of Black looking at the liter- politicians at the end of Figure 2 acy of those 21 and Reconstruction more above in 1900, who would have been Jackson. Simon Coker of South Carolina than reversed the differences in taxes lev- of school age during the time of Black was killed in 1876 by White Democrats ied due to Black politicians in 1870. Tax officeholding. The results show that in the Ellenton riot; he was kneeling in revenue per capita for counties without a one standard deviation increase in prayer after being captured. a Black politician went from $0.96 in Black politicians increased adult Black While a complete accounting is 1870 to $0.98 in 1880, while for coun- male literacy by 1.6 percent. Given that impossible, congressional testimony and ties with Black politicians, revenue went the baseline literacy rate of Black men local accounts in newspapers speak to the from $1.56 in 1870 to $0.89 in 1880, above the age of 21 was 50 percent in profound regularity of racial violence in on average. Further, I find that the vio- 1900, this implies a more than 3 per- the South during this time, a significant lence against Black politicians was precise. cent increase in Black male literacy.9 portion of which was politically moti- These attacks were unrelated to other acts vated. Black voter turnout declined more of racial violence, such as lynching, which Things Fall Apart than 20 percent between the late 1860s acted as a means of voter suppression.11 and the 1880s.10 Although Blacks sought Beginning in the early 1870s, protection from this political violence Conclusions Southern Whites began a violent and and voter intimidation, and prosecution widespread campaign to undo the of the perpetrators, they found little will Analysis of the beginning and end of Reconstruction process. After the 1872 to defend their rights. By the last decade Black political leadership in the 19th cen- election cycle, the relationship between of the 19th century, funding for schools tury provides a cautionary tale about the violence and politics was revived and open to Blacks was reduced substantially, stability of democracy after expansions extended. One-third of all of the race riots taxes were significantly lowered, and the of the franchise. While Black politicians in 1873 occurred the week before a local range of public goods offered in the South were effective in advancing some of their election. More than 10 percent of Black again stood in stark contrast to the rest of policies, they were undone by violence officials at the time were victims of vio- the nation. and legislative maneuvers.12 Once Blacks

10 NBER Reporter • No. 3, September 2020 were intimidated from voting, the legis- of Public Economics 109, January 2014, “Historical Perspectives on Racial lature was captive to White supremacists pp. 101–113. Differences in Schooling in the United who promulgated racially hostile poli- Return to Text States,” Collins W, Margo R. NBER cies. Most Southern states moved to block 4 “Credibility and Policy Convergence Working Paper 9770, June 2003, Blacks from voting and subsequently in a Two-Party System with Rational and Handbook of the Economics of enacted a White supremacist agenda with Voters,” Alesina A. American Economic Education, Vol. 1, Elsevier, 2006. new state constitutions, poll taxes, grand- Review 78(4), September 1988, pp. Return to Text father clauses, and other restrictive mea- 796–805. 10 The Wars of Reconstruction: The sures. It would take more than 50 years Return to Text Brief, Violent History of America’s Most to begin dismantling these policies, and 5 “Do Black Politicians Matter?” Progressive Era, Egerton D. New York: the violence left a lingering negative effect Logan T. NBER Working Paper 24190, Bloomsbury Press, 2014. on Black political participation.13 The January 2018. Published as “Do Black Return to Text end of those racial voting restrictions also Politicians Matter? Evidence from 11 “Racial Segregation and Southern improved outcomes for Black citizens.14 Reconstruction,” Journal of Economic Lynching,” Cook L, Logan T, Parman Overall, these findings suggest there is a History 80(1), March 2020, pp. 1–37. J. NBER Working Paper 23813, measurable effect of Black political lead- “Whitelashing: Black Politicians, September 2017, and Social Science ership on the level and composition of Taxes, and Violence,” Logan T. NBER History 42(4), Winter 2018, pp. 635– public spending. Working Paper 26014, June 2019. 675. Return to Text Return to Text 6 Black Reconstruction in America, 12 “A Poll Tax by Any Other 1 “Reflective Democracy Research 1860–1880, Du Bois WEB. New Name: The Political Economy of Findings, Summary Report,” Reflective York: Harcourt, Brace and Co., 1935. Disenfranchisement,” Jones D, Democracy Campaign, October 2017. Reconstruction: America’s Unfinished Troesken W, Walsh R. NBER Working Return to Text Revolution, 1863–1877, Foner E. New Paper 18612, December 2012. 2 “An Economic Theory of Political York: Harper & Row, 1988. Return to Text Action in a Democracy,” Downs A. Return to Text 13 “Historical Lynchings and Journal of Political Economy 65(2), 7 Freedom’s Lawmakers: A Directory Contemporary Voting Behavior of 1957, pp. 135–150. of Black Officeholders During Blacks,” Williams J. Forthcoming in Return to Text Reconstruction, Foner E. Baton Rouge: American Economic Journal: Applied 3 “Electoral Acceleration: The Effect Louisiana State University Press, 1996. Economics. of Minority Population on Minority Return to Text Return to Text Voter Turnout,” Oberholzer-Gee F, 8 “Does Monetary Policy Matter? A 14 “Valuing the Vote: The Waldfogel J. NBER Working Paper New Test in the Spirit of Friedman and Redistribution of Voting Rights and 8252, April 2001. Published as Schwartz,” Romer C, Romer D. NBER State Funds Following the Voting “Strength in Numbers: Group Size and Working Paper 2966, May 1989, and Rights Act of 1965,” Cascio E, Political Mobilization,” Journal of Law NBER Macroeconomics Annual 1989, Washington E. NBER Working Paper & Economics 48(1), April 2005, pp. Vol. 4, Blanchard O, Fischer S, eds. 17776, January 2012, and Q uarterly 73–91. “How Black Candidates Affect Return to Text Journal of Economics 129(1), February Voter Turnout,” Washington E. NBER 9 “Accounting for Racial Differences 2014, pp. 379–433. “The Franchise, Working Paper 11915, January 2006, in School Attendance in the American Policing, and Race: Evidence from and Q uarterly Journal of Economics South, 1900: The Role of Separate- Arrests Data and the Voting Rights 121(3), August 2006, pp. 973–998. But-Equal,” Margo R. NBER Working Act,” Facchini G, Knight B, Testa C. “Race and the Politics of Close Paper 2242, May 1987, and Review NBER Working Paper 27463, July Elections,” Vogl T. NBER Working of Economics and Statistics 69(4), 2020. Paper 18320, August 2012, and Journal November 1987, pp. 661–666. Return to Text

NBER Reporter • No. 3, September 2020 11 Capital Market Risks in Emerging Markets

Anusha Chari

International financial integration tions from the US Treasury to decompose has increasingly exposed capital markets the quantitative impact of US emerging in emerging nations to shocks that orig- market portfolio holdings into one com- inate outside their domestic economies. ponent due to flows and another due to Theoretical open-economy macro models valuation changes. The findings shed light propose a variety of mechanisms by which on the link between US monetary policy global financial conditions are transmit- shocks, net capital flows, and emerging Anusha Chari is a research asso- ted from developed to emerging markets. market equity and bond market returns. ciate in the NBER’s International Candidate mechanisms include advanced- An important feature of our identification Finance and Macroeconomics economy monetary policy spillovers, the strategy is that the monetary shocks origi- Program. She is a professor of eco- risk-bearing capacity of international finan- nate in the United States and are exogenous nomics and finance and director of cial intermediaries, liquidity in interna- to the destination-country fundamentals. the Modern Indian Studies Initiative tional capital markets, and global exchange The method allows us to disentangle at the University of North Carolina at rate configurations. the channels through which US mone- Chapel Hill. Her research interests lie Testing the empirical validity of theo- tary policy shocks alter yields and risk pre- in international finance, with a specific retical open-economy macro models using mia in the term structure of US interest focus on emerging market economies. micro datasets is relatively new to interna- rates. Using traded derivatives contracts on Chari received her PhD in eco- tional finance. It is a focus of my research. Treasury futures, we decompose the inter- nomics from the University of Microeconomic data are important because est rate impacts of monetary policy shocks California, Los Angeles in 2000, and aggregate data can mask underlying hetero- into the effect of revisions in market par- holds a BA in philosophy, politics, geneity useful in testing theories about a ticipants’ expectations about the path of and economics from Balliol College range of international capital market phe- short-term interest rates, and the effect at Oxford, where she was the recipient nomena, such as the effectiveness of inter- of changes in required risk compensation. of a Radhakrishnan Scholarship, and national risk-sharing and the welfare gains Our results confirm that changes in US in economics from Delhi University. from international capital movements. In a yields and risk premia significantly impact She has held faculty positions at series of papers, I use microdata to explore equity prices and bond yields in emerging the University of Chicago’s Booth the mechanisms by which global finan- markets via the portfolio rebalancing and School of Business, the University cial conditions impact capital markets in signaling channels. of Michigan, and the Haas School emerging economies. In particular, there is a significant of Business at the University of correlation between US monetary policy California, Berkeley. Capital Flows and shocks, bond yields, equity prices, and Chari has served as a special International Spillovers capital flows to emerging markets. The adviser to the Indian Prime Minister’s quantitative easing (QE) and tapering Economic Advisory Council and as a The massive surge of foreign capital to phases of unconventional monetary pol- member of the Indian finance minis- emerging markets in the aftermath of the icy in the aftermath of the GFC represent ter’s advisory group of eminent persons global financial crisis (GFC) of 2008–09 salient examples of the mechanisms at on G20 issues. She is the incoming focused attention on the substantial spill- play. During the QE period, falling term associate chair and director of men- over effects of developed-market mone- premia affected the market for long-dura- toring for the American Economic tary policy for emerging market economies. tion assets domestically and internation- Association’s Committee on the Karlye Dilts-Stedman, Christian Lundblad, ally. This bears out the importance of US Status of Women in the Economics and I examine the effects of (unconven- term premia for emerging market assets, . tional) US monetary policy on emerging particularly equities. Conversely, the A native of India, she trained market capital flows and asset prices using taper tantrum episode of 2013 led to ris- in Indian classical dance and is flu- financial derivatives to identify monetary ing bond yields, falling equity prices, and ent in English, Hindi, and Tamil. She policy shocks.1 Our high-frequency iden- significant retrenchments out of emerg- lives in Chapel Hill with her husband, tification strategy allows us to extract mon- ing market assets, increasing the cost of Nathan, and daughter, Dharma. etary policy shocks and use a dataset on external finance. disaggregated global capital flows and posi- In contrast to previous work, extract-

12 NBER Reporter • No. 3, September 2020 ing the magnitude of the monetary sur- in 2019, credit to nonfinancial corporations Asian financial crisis (AFC), corporate debt prises directly from the futures data allows in emerging markets more than tripled to vulnerability indicators, such as Altman’s us to quantify the impact of large versus $30 trillion, or 98.8 percent relative to GDP. Z-score, provide a benchmark for compari- small monetary policy shocks over and Different from the past, the composition son. Firm-level data show that post-GFC, above the qualitative statements about the of credit to emerging markets shifted from more countries are close to or in the “vulner- direction of impact that result from the use sovereign to corporate debt. Rising shares able” range of this metric, and average lever- of event dummies alone. Using monetary of debt held by troubled firms in conjunc- age for the entire emerging market sample is surprise magnitudes, we can also directly tion with the surge in leverage have led to higher in the post-GFC period than during estimate changes to US investor positions growing concerns that a wave of corporate the AFC. There is a deterioration in many in and flows to emerging markets while defaults in emerging markets could pose a indicators that measure corporate solvency controlling for a variety of global-push and grave risk to financial stability. Tightening and the ability to service debt. destination-specific pull factors. external financial conditions on the heels of Moreover, large firms appear to play an To illustrate the impact on equity easy money and a corporate leverage boom outsized role and disproportionately drive flows, Figure 1 pres- risks in emerging capi- ents the effects of a tal markets. In a second one standard deviation US Monetary Policy Shocks and Flows to Emerging Market Funds paper, we examine the monetary policy shock links between the surge Monthly impact, decomposed as flows and valuation changes, associated in the pre-GFC, QE, with a one standard deviation monetary policy shock, millions of USD in corporate leverage and taper tantrum peri- 1200 in emerging markets, ods, respectively. Using Pre-Great-Financial-Crisis Period Quantitative Easing Period Taper Tantrum Period weakened corporate the disaggregated US 800 financial fragility indi- Treasury International 600 cators, and firm charac- Capital data, the fig- 400 teristics.3 We find that ure decomposes the 200 firm size plays a critical impact in emerging 0 role in the relationship market positions into -200 between leverage, firm changes in flows and -400 fragility, and exchange -600 valuations. For exam- Valuation changes rate movements. While ple, a one standard -800 Flows the relationship between deviation equity shock -1000 firm-leverage and dis- in the taper period elic- -1200 tress scores varies over its an equity effect that, time, the relationship on average, represents Source: A. Chari, K. Dilts Stedman, C. Lundblad, NBER Working Paper 23474 between firm size and about 11 percent of corporate vulnerability monthly market capi- Figure 1 is relatively time-invari- talization changes. Given ant. All else equal, large the low average levels of liquidity in these can drive up expected default probabilities. firms in emerging markets are more finan- markets, these effects are consequential. We know relatively little about the cially vulnerable and also systemically impor- determinants of corporate debt and distress tant. Consistent with the granular origins of Debt and Distress in in emerging markets. Research in interna- aggregate fluctuations, there is a positive and Emerging Markets tional finance has primarily focused on the significant correlation between idiosyncratic real and financial impacts of sovereign credit shocks to large firms’ sales growth and GDP Periods of expansionary advanced- risk, and how external shocks that affect sov- growth in our emerging markets sample. economy monetary policy can facilitate a ereign default risk impact the macroecon- Relatedly, the negative impact of exchange rapid buildup of leverage at firms in emerg- omy. My research brings new micro evidence rate shocks has a more acute impact on the ing markets, as witnessed in the aftermath to bear on the link between global finan- sales growth of the more highly leveraged of the GFC. Emerging market corporate cial conditions and corporate default risk in large firms. leverage and debt levels surged during the emerging markets. period of QE and unconventional mone- Laura Alfaro, Gonzalo Asis, Ugo Forecasting Corporate Default tary policy in the United States, consistent Panizza, and I explore the rapid expansion of Risk in Emerging Markets with a fall in the cost of external finance credit to firms in emerging markets follow- brought about in large part by foreign capital ing the GFC.2 We document a set of styl- Using a novel multicountry data- inflows and increased cross-border financial ized facts about leverage and financial fra- set on corporate defaults from the intermediation. In the decade concluding gility for emerging market firms. During the Credit Research Initiative of the Risk

NBER Reporter • No. 3, September 2020 13 Management Institute at the National for the underperformance of standard tors in the identification of distressed University of Singapore, in two papers, bankruptcy models that have not been firms. They display the curves associated Asis, Adam Haas, and I study factors tailored to the emerging market context. with our main specification that includes that drive corporate distress in emerging Models that only include accounting and global variables, one that only has local markets. Established bankruptcy models market variables have better predictive macroeconomic variables, and one that based on developed-market experience do power during normal times. includes only accounting variables. The not capture many idiosyncrasies that affect The existing literature uses several benchmark model that includes global emerging market firm solvency, particu- measures of a model’s predictive power, variables performs substantially better in larly emerging market vulnerabilities to mostly ranking firms by their estimated identifying at-risk firms. Over 80 percent global financial conditions. We suggest probabilities of default. However, stud- of the defaulting firms are included in that the extent of global exposure influ- ies differ in the number of firms and the top 10 percent of firms when ranked ences whether deteriorating international defaults, the size of the quantiles to group by predicted default probability (blue credit market conditions reduce emerging firms, and the allocation of distressed firms curve). The specification that includes market firms’ ability to repay their debts. across quantiles making cross-model com- only accounting variables takes nearly 40 Consequently, it is not surprising that parisons difficult. We rely on the receiver percent of the sample to get to this same models developed in the US context per- operating characteristics (ROC) score, rate of identification (light gray curve). form poorly when applied to the emerging also known as the “area under the curve” Our findings suggest that increased market firms. They may be missing perti- (AUC), that uses the cumulative fraction corporate leverage can make firms more nent factors that drive corporate default of defaults as a function of the ordered exposed to adverse shocks to their cash risk. flows and asset val- Given the universe ues, driving up In-Sample Predictive Power: Receiver Operating Characteristics Curve of accounting, finan- default probabilities. cial market, domestic, Percentage of failed firms Tightening global and global macro vari- 100 financial conditions ables that could poten- 90 can exacerbate roll- tially influence emerging 80 over and currency market firms’ solvency, 70 risks. In the absence selecting the most rele- 60 of optimal hedging, vant variables for gaug- Models: exchange rate depre- 50 ing corporate resiliency Benchmark ciation can impose a and forecasting distress is 40 Only local macroeconomic variables significant strain on a challenging task. In the 30 Only accounting variables the ability of emerg- first paper, Asis, Haas, 20 ing-market firms to and I adopt a machine- 10 service foreign cur- learning approach to 0 rency-denominated directly confront the 0 10 20 30 40 50 60 70 80 90 100 debts. Therefore, issue of variable selec- Percentage of firms (sorted from most to least likely to fail, as predicted by the model) deteriorating global tion and show how a financial conditions machine-learning tech- Source: G. Asis, A. Chari, and A. Haas, NBER Working Paper 27213 can directly impact nique, LASSO, can help credit risk, elevat- corporate distress pre- Figure 2 ing corporate default diction in emerging markets.4 By limiting population of firms from most to least probabilities, particularly for firms with the analysis to the most relevant subset likely to fail as predicted by the model. The weak fundamentals. The data suggest from a much broader set of accounting, ROC curve plots the true positive rate, that global financial variables, such as US market, and global variables, machine based on the number of firms that default, interest rates, shifts in global liquidity, learning can increase a model’s predictive against the false positive rate, based on the and foreign investor risk aversion, have power while maintaining straightforward number of firms predicted to default that significant predictive power for corporate interpretation. do not, for different threshold settings. distress risk in emerging markets. In terms of forecasting corporate dis- The extant literature on distress risk We also explore the asset pricing tress, the variable selection exercise using does not account for the exposure of implications of our measure of distress machine learning suggests that models emerging market firms to global finan- risk. Do riskier firms command a higher that include global financial conditions cial shocks. This is the focus of our sec- risk premium? Prior literature using US perform best in times of crisis. The omis- ond paper.5 The ROC curves in Figure data documents an inverse correlation sion of these variables may provide a clue 2 illustrate the importance of global fac- between distress risk and future stock

14 NBER Reporter • No. 3, September 2020 returns, referred to as the “distress risk pervasive use in the financial press and tional features of global investor risk premium puzzle.” In contrast, we find among policymakers since the GFC. We appetite have important implications for strong evidence of a positive distress view RORO shocks as a reflection of the capital flows and return distributions in risk premium in emerging market stocks. variation in global investor risk aversion. emerging markets, and lead us to con- Future twelve-month stock returns are Since investors rebalance their portfo- clude that focusing only on measures of monotonically increasing in the prob- lios towards safe assets in the face of risk central tendency is incomplete. ability of corporate default, and the pat- aversion shocks, RORO variation has tern is robust to the inclusion of a vari- important implications for asset price 1 “Taper Tantrums: QE, Its ety of standard asset pricing controls. determination, particularly for so-called Aftermath and Emerging Market The impact of a global “risk-off ” envi- “risk assets.” Capital Flows,” Chari A, Dilts- ronment on default risk is higher for We build a multifaceted RORO Stedman K, Lundblad C. NBER firms whose returns are more sensitive to index to capture realized variation Working Paper 23474, June 2017, a composite global factor that measures in global investor risk appetite. Our and forthcoming in The Review of external financing conditions. index, along with constituent subindi- Financial Studies, April 2020. ces, exhibits significant skewness and Return to Text Risk-on/Risk-off fat tails. With fat tails, extreme events 2 “Lessons Unlearned? Corporate and Tail Events become both more probable and poten- Debt in Emerging Markets,” Alfaro tially more destabilizing. As examples, L, Asis G, Chari A, Panizza U. NBER My research on corporate default we observe sharp risk-off movements Working Paper 23407, May 2017. risk illustrates how changes in global during the GFC, the European debt Return to Text financing conditions 3 “Corporate can affect emerging Debt, Firm Size and Risk-on/Risk-off Index and Emerging Market Bond Flows market firms. It is clear Financial Fragility in Risk-on/risk-off index, normalized that tail events such as to 100 at onset of risk-off event Cumulative emerging market bond flows, billions US dollars Emerging Markets,” sudden stops present 3500 14000 10 Alfaro L, Chari A, 3000 12000 pressing challenges. 0 Asis G, and Panizza, 2500 10000 Further, the map- -10 U. NBER Working ping from global risk- 2000 8000 Paper 25459, January on/risk-off position- 1500 6000 -20 2019, and Journal ing to capital flows 1000 4000 -30 of International 500 2000 and returns is con- -40 Economics 118, May ditional on the mea- 0 0 2019, pp. 1–19.​ -50 sure of risk used and -500 -1000 Return to Text the severity of the epi- -1000 -3000 -60 4 “Emerging 0 8 16 24 32 40 48 56 64 0 14 28 42 56 70 84 98 112 126 sode. Identifying the Days since risk-off event Weeks since risk-off event Markets at Risk,” Asis transmission of global G, Chari A, Haas Global Financial Crisis 2015 Chinese stock market crash financial conditions to Taper Tantrum COVID-19 (Right-hand axis on left panel) A. AEA Papers and local markets requires Proceedings 110, May different sources of “Capital Flows in Risky Times: Risk-On/Risk-Off and Emerging Market Tail Risk,” A. Chari, K. Dilts 2020. risk and plausible exo- Stedman, C. Lundblad, Kansas City Fed RWP 20-08. 2020, and NBER Working Paper 27927 Return to Text geneity with respect 5 “In Search of to local fundamentals. Figure 3 Distress Risk in However, existing research has primarily crisis, the 2015 Chinese stock mar- Emerging Markets,” Asis G, Chari A, focused on the first moment of the rele- ket crash, the taper tantrum, and the Haas A. NBER Working Paper 27213, vant distributions of aggregated risk mea- COVID-19 crisis. May 2020. sures. In ongoing work, Dilts-Stedman, Figure 3 shows cumulative outflows Return to Text Lundblad, and I turn our attention to from emerging market bond funds dur- 6 “Capital Flows in Risky Times: the full distribution of emerging market ing these widely recognized, risk-off Risk-On, Risk-Off and Emerging capital flows and returns. We character- events along with corresponding cumu- Market Tail Risk,” Chari A, Dilts- ize how extreme capital flow and returns lative changes in our RORO index. The Stedman K, Lundblad C. NBER realizations are tied to global risk appe- episodes are associated with severe out- Working Paper 27927, and Federal tite (“risk-on/risk-off ” or RORO).6 flow realizations, highlighting the neces- Reserve Bank of Kansas City RWP While imprecisely defined, the sity of modeling tail risk specifically. 20-08. 2020. RORO terminology has come into Our results show that the distribu- Return to Text

NBER Reporter • No. 3, September 2020 15 Integrating Renewable Generation into Electricity Markets

Frank A. Wolak

Electricity supply industries in Reliability of Supply with many parts of the world are under- an Increasing Renewable going disruptive change because Generation Share Frank A. Wolak is the Holbrook of policymakers’ desire to reduce Working Professor of Commodity energy-sector greenhouse gas (GHG) Managing an electricity supply Price Studies in the Economics emissions. Electrifying energy ser- industry with a large share of wind and Department and the director of the vices such as transportation and space solar generation capacity involves many Program on Energy and Sustainable heating can significantly reduce these new operational challenges, as dem- Development at . emissions globally. The transporta- onstrated by the rolling blackouts of His recent research and teaching tion, electricity, and space-heating August 2020 in California. By replac- focuses on design, performance, and sectors currently account for 28, 27, ing natural gas-fired and nuclear gen- monitoring of energy and environ- and 9 percent of US GHG emissions, eration capacity with solar and wind mental markets in the United States respectively. It follows that reduc- generation capacity, California substi- and globally. ing GHG emissions from these sec- tuted on-demand generation capacity From April 1998 to April tors by significantly increasing wind with generation capacity that only pro- 2011, he was chair of the Market and solar energy production is likely duces when the underlying resource — Surveillance Committee of the to be the most economically viable wind or sunshine — is available. Figure California Independent System pathway to sizeable reductions in 1 shows the declining shares of natural Operator. In this capacity, he testified global GHG emissions. gas and nuclear generation and increas- numerous times at the Federal Energy Regulatory Commission, and at vari- ous committees of the US Senate and House of Representatives on issues California’s Electric Generation by Fuel Type, 2001–2019 relating to market monitoring and market power in electricity markets. GWh (000s) Wolak has also worked on design and 250 regulatory oversight of electricity mar-

kets in Europe, Australia, Asia, Africa, 200 Latin America, and Africa. Wind Wolak was a member of Solar the Emissions Market Advisory 150 Committee (EMAC) for California’s Hydro Market for Greenhouse Gas Emissions 100 allowances from January 2012 to December 2014. This committee Natural gas advised the California Air Resources 50 Nuclear Board on the design and monitoring Biomass and of the state’s cap-and-trade market for geothermal 0 Coal and greenhouse gas emissions allowances. petroleum coke In 2017, Wolak and other EMAC 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019

members provided an assessment of Source: California Energy Commission the extension of the California’s cap and trade market to 2030. Figure 1

16 NBER Reporter • No. 3, September 2020 ing shares of wind and solar generation on and off more frequently. Turning on a the electricity market designs that existed in California since 2013, the first compli- natural gas or coal generation unit incurs in many regions of the US and currently ance year of the state’s 33 percent by 2020 a significant up-front cost because fuel exist throughout Europe) are increas- renewables portfolio standard goal. is burned without injecting electricity to ingly costly to operate, particularly in In an empirical analysis of the behav- the grid. These dispatchable generation regions with a growing share of intermit- ior of the hourly output of more than units also have minimum safe operating tent renewables.2 A general conclusion 50 wind and solar generation unit loca- levels, maximum safe operating levels, and from this work is that the market designer tions in California, I found a high degree minimum up-time and minimum down- must make the market model that is used of contemporaneous correlation in the time constraints. In addition, these units to set prices and dispatch generation units hourly output of individual solar and wind have ramping constraints that restrict match as closely as possible the model that facilities.1 This implies a “feast or famine” how fast they move from one output level system operators use to operate the system distribution of aggregate wind and solar to another. Finally, transmission network in real time.3 energy production, which can make it capacity constraints can restrict the ability The increased energy supply risk from extremely challenging for system opera- of all generation units to supply energy to a larger share of wind and solar resources tors to meet the difference between the where it is needed. also increases the expected benefits from hourly system demand risk-management ser- and hourly renewables vices. Akshaya Jha and production, typically California’s Electricity Demand on 8/15/2020 and the Effect of a Blackout I find that the actions called the net demand. MW (000s) of purely financial par- Net demand must 50 ticipants who do not be met by natural gas, own generation capac- 45 Hour-ahead coal, or other genera- forecast demand ity or serve demand tion capacity that has can reduce the cost of Actual demand controllable instanta- 40 serving demand, par- neous output. These Net demand ticularly during high- kinds of generation 35 load conditions when units are called “dis- all of these operating 30 The discrepancy between patchable,” to distin- forecasted demand and actual constraints are likely to demand from 6:25pm to 6:47pm is guish them from inter- a rough estimate of the demand be most relevant. The mittent renewable 25 curtailed by the rolling blackouts actions of these purely units. As the amount financial participants of wind and solar 20 make the generation generation capacity Midnight 4am 8am Noon 4pm 5 6 7 8pm Midnight schedules that emerge in a region increases, from the day-ahead during the majority Source: The California ISO market closer to how of hours of the year these generation units many dispatchable Figure 2 actually operate in real units are no longer needed to meet the An increasing the amount of inter- time, thereby reducing the need for costly net demand. However, because of the mittent generation in region makes increases or decreases in their output in feast or famine nature of wind and solar the non-convexities and indivisibilities real time.4 energy production, there are still likely in production described above increas- to be hours during the year when each ingly relevant. In addition, the require- Strategies for Active of these units is required to meet the ment to deliver all electricity through a Participation of Final Demand net demand. Figure 2 presents a graph transmission network with finite trans- in the Wholesale Market of system demand, the hour-ahead fore- fer capacity between locations in the grid cast of system demand, and net demand becomes increasingly important because As the share of energy from wind for August 15, 2020, when 470 MW of the substantial increase in volatility and solar resources increases, system of load shedding occurred 6:25 pm to in net demand versus system demand. operators have fewer supply options to 6:47 pm. The rapid disappearance of Christoph Graf, Federico Quaglia, and deploy to maintain real-time system bal- solar energy production in the evening I demonstrate that electricity market ance at all locations in the transmis- implies a rapid increase in net demand. designs that employ simplified models of sion network. Consumers can no longer The intermittency of wind and solar the transmission network operation that be passive participants in the wholesale energy production also implies that these ignore many transmission and genera- market. By shifting the demand for grid- dispatchable units will have to be switched tion unit operating constraints (such as supplied electricity from hours when

NBER Reporter • No. 3, September 2020 17 the wind and solar resources are not how individual appliances are used customers in the control group during producing to the hours when they are throughout the day and which uses are the dynamic pricing experiment, what allows system operators to maintain sys- flexible can provide important input into we call the “Campaign effect.” The pro- tem balance with fewer dispatchable determining the efficient deployment of file labelled “Full Event-day effect” is the resources, thereby reducing the cost of these technologies. Jiyong Eom and I difference in the mean daily load pro- serving demand. Laura Andersen, Lars used a field experiment involving com- file between customers in our treatment Gårn Hansen, Carsten Jensen, and I per- mercial customers in South Korea to group and customers in the control group formed a field experiment involving resi- measure the typical pattern of appliance- during a dynamic pricing event day. It dential consumers in to inves- level electricity use and the appliance- measures the combined impact on appli- tigate the willingness of consumers to level responsiveness of these customers ance use of participating in the experi- shift consumption away from or into cer- to dynamic prices.6 We find an impor- ment and a dynamic pricing event day. tain hours of the day if prodded at short tant difference between the how commer- notice using cellphone text messages.5 Rooftop versus Grid-Scale This experiment provided Danish Representative Hourly Load Solar Generation Investment residential consumers with dynamic Profile for an Appliance price and environmental signals aimed at Rooftop solar systems are an alter- causing them to shift their consumption Full event-day effect native source of renewable energy that either “into” or “away” from certain time kWh many customers find attractive because periods. We found that the same mar- of how the sunk cost of the transmis- ginal price signal caused substantially sion and distribution networks have his- larger consumption shifts “into” target Campaign effect torically been recovered. A per-kilowatt- hours compared to consumption shifts hour (kWh) charge is typically assessed “away” from target hours. We also found on all energy withdrawn by the customer that consumption is reduced in the hours to recover these sunk costs. This raises before and after the “into” target hours. the customer’s opportunity cost of con- There is weaker evidence of increased suming grid-supplied electricity, which consumption in the hours surrounding Full event-day effect makes an investment in a rooftop solar the “away” target hours. The same “into” system more attractive, despite the fact versus “away” results hold for the envi- that utility-scale solar units produce elec- ronmental signals, although the abso- 1h 12h 24h tricity at a significantly lower average cost lute size of the effects is smaller. For both than a rooftop solar system. Currently the price and environmental treatments, in Northern California, the average cost the same qualitative results are obtained, Hourly treatment relative to control group profile of electricity to residential consumers is but with uniformly smaller quantitative close to 20 cents per kWh, despite the magnitudes. We use these estimates to fact that the average marginal cost of grid- perform counterfactual experiments in Figure 3 supplied electricity in 2019 was less than which all of an electricity retailer’s resi- cial versus residential customers respond 5 cents per kWh. This creates an incen- dential customers are assumed to face to dynamic prices. Rather than reducing tive for households to install a rooftop these dynamic price signals. We find sub- their consumption in response to indi- solar system that produces electricity at an stantial wholesale energy cost savings for vidual dynamic pricing events, commer- average lifetime cost of 15 cents per kWh the retailer from declaring “into” events cial customers appear to reconfigure their in order to avoid consuming more expen- designed to shift consumption from mode of operation in response to fac- sive, grid-supplied electricity. Although higher to lower demand hours within ing dynamic prices. Consistent with our this decision is privately profitable for the day, which suggests that such a pric- reconfiguration hypothesis, small busi- the household, it increases the total cost ing strategy could significantly reduce nesses primarily curtailed their electric- of supplying electricity to all customers, the cost of increasing the share of wind ity usage during peak periods of the day including those that install rooftop solar and solar electricity generation. during all days of the experiment period. systems, because the sunk costs of the The declining cost of electronic Appliances not critical to a positive cus- transmission and distribution network devices enables automated customer-level tomer experience were the major sources must now be recovered over a smaller demand response actions. A number of of the energy savings from these recon- quantity of grid-supplied electricity. companies have designed machine learn- figuration actions. Figure 3 shows a rep- Because installing a rooftop solar ing-based technologies that use Wi-Fi- resentative difference between the mean system requires substantial up-front enabled plugs to control electricity use on daily load profile for an appliance for costs and a house for the customer to the customer’s premises. Understanding customers in the treatment group and install it on, these systems tend to be

18 NBER Reporter • No. 3, September 2020 clustered in wealthier neighborhoods. the performance of three matched car- 1 “Level versus Variability Trade- During many hours of the day, rooftop bon-tax/cap-and-trade pairs with equiv- offs in Wind and Solar Generation solar systems produce more electric- alent emissions targets, mean emissions, Investments: The Case of California,” ity than the customer consumes. This and mean carbon prices respectively.8 Wolak FA. NBER Working Paper introduces reverse energy flows, which Across these matched pairs, the 22494, August 2016, and The Energy can require expensive distribution net- cap-and-trade mechanism produced Journal 37 (Special Issue 2), 2016, pp. work upgrades to accommodate. Using much higher wholesale electricity 185–220. data from the three major California prices (38.5 to 52.6 percent higher) Return to Text distribution utilities, I find that distri- and lower total electricity production 2 “Simplified Electricity Market Models bution network prices for each of these (2.5 to 4.0 percent lower) than the with Significant Intermittent Renewable utilities more than doubled between “equivalent’’ carbon tax, without any Capacity: Evidence from Italy,” Graf C, 2003 and 2019. lower carbon emissions. Market par- Qualia F, Wolak FA. NBER Working Using time series data on utility- ticipants that forecast a lower price of Paper 27262, May 2020. level quarterly rooftop solar capacity, carbon in the cap-and-trade games ran Return to Text I find that virtually all of the increase their generation units more frequently 3 “Wholesale Market Design,” Wolak in these distribution network prices than those that forecast a higher price FA. Forthcoming in Handbook on the can be explained by distribution net- of carbon, which caused emissions from Economics of Electricity, Glachant JM, work upgrades to accommodate the the dirtiest generating units — coal and Joskow P, Pollitt M. Northhampton: distribution network flows associated natural gas-fired units with high heat Edward Elgar Publishing. with rooftop solar investments. The rates — to be significantly higher (15.2 Return to Text mechanical effect of fewer withdrawals to 33.0 percent higher) than in the 4 “Can Financial Participants Improve of grid-supplied electricity to recover carbon tax games. This highlights an Price Discovery and Efficiency in the same sunk costs of the distribu- important advantage of the carbon tax Multi-Settlement Markets with Trading tion network explains only a small frac- as a policy. With a carbon tax, the car- Costs?” Jha A, Wolak FA. NBER tion of the distribution network price bon is a known input to the suppli- Working Paper 25851, May 2019. increase.7 This paper concludes with er’s production process and there is Return to Text a description of a distribution net- no disagreement among market par- 5 “Can Incentives to Increase Electricity work pricing scheme that eliminates ticipants about the price of this input. Use Reduce the Cost of Integrating the incentive for economically ineffi- Under a cap-and-trade mechanism, Renewable Resources?” Andersen LM, cient bypass of grid-supplied electricity. market participants can hold different Hansen LG, Jensen CL, Wolak FA. beliefs about the price of carbon, and NBER Working Paper 25615, February Carbon Pricing and these differences will typically result in 2019. Electricity Supply higher wholesale electricity prices. Return to Text 6 “Breaking Routine for Energy Carbon pricing is an important Directions for Future Research Savings: An Appliance-Level Analysis of part of any climate policy for reducing Small Business Behavior under Dynamic GHG emissions from fossil fuel genera- There are many difficult remaining Prices,” Eom J, Wolak FA. NBER tion units. Although a carbon tax and a economic and engineering challenges Working Paper 27263, May 2020. cap-and-trade market can be shown to associated with reducing GHG emissions Return to Text be equivalent under certainty, they can from the electricity sector. These increase 7 “The Evidence from California on lead to different outcomes under uncer- rapidly as the share of wind and solar gen- the Economic Impact of Inefficient tainty. Over the past 10 years, Trevor eration rises above 50 percent. Addressing Distribution Network Pricing,” Wola k Davis, Mark Thurber, and I have devel- them will require more active involve- FA. NBER Working Paper 25087, oped a web-based Energy Market Game ment of consumers in the wholesale mar- September 2018. (EMG) in which students own a portfo- ket, an increasing range of financial tools Return to Text lio of thermal and intermittent renew- to manage supply risk, investments in 8 “An Experimental Comparison of able generation units and compete to both short-term and long-term storage Carbon Pricing under Uncertainty in sell electricity in an offer-based whole- technologies, spatial and temporal pric- Electricity Markets,” Davis T, Thurber sale market with uncertain demand with ing of access to distribution networks, and M, Wolak FA. NBER Working Paper either a cap-and-trade market or a car- new protocols for operating the transmis- 27260, May 2020. bon tax. Using the EMG, we compared sion and distribution network. Return to Text

NBER Reporter • No. 3, September 2020 19 Are US Treasury Bonds Still a Safe Haven?

Zhiguo He and Arvind Krishnamurthy

Safe assets are integral to the func- bonds. US Treasuries have been the this relation that the average conve- tioning of banks, financial markets, premier safe asset around the world nience yield on Treasury debt over their and the international financial system. for the past several decades, although sample is 75 basis points.1 Financial market participants use safe events in the recent COVID-19 cri- There is ample evidence that some assets to meet liquidity and transac- sis raise the specter that this reign may private safe assets carry convenience tion needs, as high-quality collateral end, as we review later in this article. yields, the most significant example for loans and derivative contracts, and The left panel of Figure 1 plots being short-term debt issued by finan- as default-free stores of value. These the spread in yields between long-term cial institutions, including banks. Gary services imply a nonpecuniary return AAA corporate bonds and Treasury Gorton argues that this shapes the on safe assets, a convenience yield, bonds against the total stock of pri- structure and operation of the banking that drives up the prices of safe assets vately held US Treasury bonds, using system.2 Figure 1’s right-side panel plots and lowers their expected return in annual data from 1919 to 2008. The the quantity of outstanding short-term equilibrium. figure traces out a convenience-demand financial sector debt against the sup- The clearest example of the exis- function for safe assets, akin to a money- ply of government safe assets, including tence of a safe-asset convenience yield demand function. Krishnamurthy and gold certificates in the early part of the comes from examining US Treasury Annette Vissing-Jorgensen infer from sample, from 1874 to 2014. The figure

Zhiguo He is the Fuji Bank and Heller Professor of Finance, Jeuck Faculty Fellow, and faculty codirector of the Fama-Miller Center for Research in Finance at the University of Chicago’s Booth School of Business. He is also director of the Becker Friedman Institute for Economics in China. Previously the Dean’s Distinguished Visiting Scholar at Stanford University’s of Business, He is serving as Special-term Alibaba Foundation Professor at Tsinghua University’s School of Economics and Management, and as a member of Academic Committee of Luohan Academy. His research focuses on agency frictions and debt maturities in financial markets and mac- roeconomics, with a special focus on contract theory and banking. He is also conducting active academic research on Chinese financial markets and on FinTech, especially the potential business applications of blockchain technology. He holds bachelor’s and master’s degrees from Tsinghua University in Beijing, and received a PhD degree from the Kellogg School of Management at Northwestern University in 2008. He was named a 2014 Alfred P. Sloan Research Fellow, and has won numerous scho- lastic awards, including the Lehman Brothers Fellowship for Research Excellence in Finance in 2007, the Swiss Finance Institute Outstanding Paper Award in 2012, the Smith-Breeden First Prize in 2012 and the Brattle Group First Prize in 2014.

Arvind Krishnamurthy is the John S. Osterweis Professor of Finance at the Stanford Graduate School of Business. He is affiliated with the NBER research programs in Asset Pricing, Economic Fluctuations and Growth, International Finance and Macro, and Monetary Economics. Krishnamurthy formerly taught at the Kellogg School of Management at Northwestern University. He studies finance, macroeconomics, and monetary policy. He has studied the causes and consequences of liquidity crises in emerging markets and developed economies, and the role of gov- ernment policy in stabilizing crises. Recently he has been examining the importance of US Treasury bonds and the dollar in the international monetary system. Krishnamurthy received his undergraduate degree from the University of Pennsylvania and his PhD from MIT in 1998.

20 NBER Reporter • No. 3, September 2020 illustrates that bank debt is a substi- ernment bond, swapped into dollars via LIBOR basis, constructed analogously tute for government debt: less govern- a foreign exchange forward. The black but for LIBOR (bank deposit) rates. ment debt increases the convenience line is negative, indicating the lower Because safe dollar debt carries a yield on safe assets (left panel) and yield on US Treasury bonds; hence, the convenience yield, high-grade firms induces the banking and banks domiciled sector to take advan- both in the US and tage of the higher US Debt, the Corporate Bond Premium, and Bank Debt around the world convenience yield by finance themselves AAA corporate bonds to Treasuries spread, 1919–2008 Net short-term debt to GDP, detrended increasing their issu- 2.0% 0.4 in dollar-denomi- 1975 ance of bank debt 2008 nated debt. That is,

(right panel). 2001 the dominance of 2007 1.5 0.2 2006 dollar debt, which is 2008 International 1988 a well-documented 1930 Dimensions 1998 1933 feature of the inter- 1.0 0 1981 1934 1934 national financial 1935 Although we 1993 1962 system, is a direct 1979 have reviewed evi- 1960 consequence of the dence from the 0.5 -0.2 1939 dollar safe asset phe- 1966 1948 1946 United States, it is 1950 1948 nomenon. Moreover, 1943 1955 1946 likely that govern- 1945 high levels of dollar 0 -0.4 ment and bank debt 0 0.2 0.4 0.6 0.8 1.0 1.2 -0.2 0 0.2 0.4 0.6 0.8 debt explain why US in other advanced Debt-to-GDP Government supply of safe assets to GDP, detrended monetary policy has countries also bear large international Source: “The Aggregate Demand for Treasury Debt,” Krishnamurthy A, Vissing-Jorgensen A. Journal of Political a convenience yield, Economy 120(2), April 2012, pp. 233-267 and “The Impact of Treasury Supply on Financial Sector Lending and spillover effects and and that the forces we Stability,” Krishnamurthy A, Vissing-Jorgensen A. Journal of Financial Economics 118(3), December 2015, pp. 571-600. why changes in the have described above dollar exchange rate carry over to these Figure 1 drive a global finan- countries. Marco Del Negro, Domenico figure is evidence of a higher conve- cial cycle. Finally, as Pierre-Olivier Giannone, Marc P. Giannoni, and nience yield on dollar bonds. Wenxin Gourinchas and Hélène Rey observe, Andrea Tambalotti argue that this fac- Du, Joanne Im, and Jesse Schreger the existence of a dollar convenience tor helps explain the low neutral real are the first to construct and study yield induces the US as a whole to rate of interest across advanced econo- this basis spread.4 The blue line is the run a carry trade of issuing safe dollar mies.3 Nevertheless, debt and investing US dollar safe assets The US Treasury Bond and LIBOR Premium in higher-return for- are noteworthy rela- eign assets.5 Jiang, tive to those of other Basis points Krishnamurthy, countries in carrying 20 and Hanno Lustig a higher convenience develop a model 0 yield and shaping the that connects these 6 international finan- -20 observations. cial system. Figure 2 illus- -40 Why Are US Safe trates the premium Assets Special? -60 on dollar bonds LIBOR basis Yield on US bank deposit minus compared to the average on non-US G10 bank deposits Working with -80 bonds of other coun- Konstantin Milbradt, Treasury basis tries. It plots in black -100 we have developed a Yield on a one-year Treasury bill minus the yield on the Treasury basis, the average non-US G10 one-year government bond model that empha- which is defined as -120 sizes financial coor- the yield on a one- 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 dination incentives year Treasury bill in the determination minus the yield on Source: Z. Jiang, A. Krishnamurthy, H. Lustig, NBER Working Paper 27682 of safe asset status.7 the average non-US The key idea is that G10 one-year gov- Figure 2 the safety of assets is

NBER Reporter • No. 3, September 2020 21 endogenous to investors’ actions: inves- The β of Long-Term Treasuries during the global financial crisis of 2007– tors act in a manner to make the asset 09. The right panel of Figure 3 plots safe. By having a larger Treasury bond Treasury bonds typically appreci- the daily market movement of the S&P market, liquidity and market depth in ate in times of turmoil — that is, they 500 and yields of Treasury securities in Treasury bonds are enhanced, drawing have a negative β. However, events in September 2008, covering the Lehman in investors across the globe. This in March 2020, during the COVID-19 bankruptcy on September 15. Treasury turn reduces rollover yields fall (prices rise) risk for the Treasury, when the stock mar- which enhances the US Treasury Yields and Stock Prices ket falls. safety of the bonds, It is worth high- reinforcing inves- 2020 2008 lighting that the con- tors’ desire to pur- Treasury bond yields S&P 500 Index Treasury bond yields S&P 500 Index trast in price move- chase them. The 1.5 3250 5% 1400 ments between 2020 model endogenously and 2008 is only generates the out- 1.2 3000 present in long-term 4 1300 come that the value 10 year Treasury securities. of the safe asset rises 0.9 2750 Figure 3 shows that 10 year in bad states of the 3 1200 the yield of one-year world, when investors 0.6 2500 5 year Treasury bills has endogenously fly to remained largely flat safety. 5 year 2 1100 in 2020. However, 0.3 2250 1 year The model also yields at the short 1 year offers a nuanced per- end are affected by spective on the out- 0.0 2000 1 1000 the Fed’s monetary 3/2/08 3/31/08 9/2/08 9/29/08 standing amount of policy easing. We government debt. Source: Researchers’ calculations using data from Bloomberg and the Center for Research in Security Prices next examine quan- Having too little debt tities where a clearer reduces liquidity and Figure 3 contrast between investor coordination long- and short-term incentives. Having too much debt rela- crisis, did not follow this established Treasuries emerges. tive to fiscal capacity, however, renders pattern. As in many previous periods of debt unsafe and weakens coordination financial market turmoil, stock prices Sales of Treasuries incentives. The size of the US economy fell dramatically, implied stock index enables the country to sustain a large return volatility spiked, credit spreads We plot the changes in Treasury absolute amount of liquid debt, which widened, and the dollar appreciated. holdings of a number of the key actors, is why investors coordinate around Yet, in contrast to previous episodes, comparing March 2020 and September Treasury debt as the world’s safe asset. prices of long-term Treasury securities 2008. Figure 4 shows the flows for The model is cast in real terms, and fell sharply. From March 9 to March three major institutional players in the one shortcoming of the analysis is that 18, when the US stock market fell Treasury market: foreign investors, it does not explain why global investors 19.3 percent, the 10-year Treasury yield mutual funds, and the Fed. We focus coordinate on US Treasury debt issued increased by about 60 basis points (a on foreign investors and mutual funds in nominal dollar units.8 return of -4.9 percent), resulting in an because they are liquidity sensitive hold- unusual positive correlation between ers. We separate the long-term treasuries The US Treasury Market stock and bond returns (see the left (notes and bonds) and short-term secu- in 2020: Canary in panel of Figure 3). The pattern of rising rities (T-bills), and graph the flows for the Coal Mine? yields also held for five-year Treasury each player as a percentage of the corre- notes. He, Stefan Nagel, and Zhaogang sponding total outstanding Treasuries in The large and growing size of US Song find that this behavior was not each bucket. In Figure 4, we report the deficits has raised concern that US due to rising inflation expectations and dollar volume (in billions) correspond- Treasuries and the US dollar may end inflation uncertainty, which fell during ing to each bar. We only plot the flows in their reign at the center of the inter- this episode.9 September 2008 for an analogous com- national financial system. We review The positive β of long-term Trea­sur­ parison to March 2020, although the events in the 2020 COVID-19 crisis in ies during March 2020 is particularly strik- flows were spread out over the second this context. ing when contrasted with what occurred half of 2008 for all entities.

22 NBER Reporter • No. 3, September 2020 For long-term Treasuries in 2020, US government, the market prices of bonds in March 2020 reflects long- we observe that foreign investors, includ- long-term Treasuries are endogenous. standing flaws in clearing/settlement ing foreign central banks and investors They are subject to coordination incen- market design for Treasury bonds, in tax havens, sold about 2.5 percent, or tives of market participants and expec- amplified by constraints on dealer bal- US $311 billion (the negative blue bar in tation of future fundamentals.11 The ance sheets.12 Dealers who in previous the top-left of Figure 4). In comparison, positive β of long-term Treasuries thus episodes may have absorbed these flows they bought about 0.7 did not do so because percent, or US $20 of regulatory balance billion in September Change in Treasury Holdings of Foreign Investors, Mutual Funds, and The Fed sheet constraints.13 2008. Mutual funds As a result, long- acted similarly, selling term Treasury prices in 2020 but buying in Foreign Investors Mutual Funds Federal Reserve fell in March 2020. 2008. In the COVID- Mar. Sept. Mar. Sept. Mar. Sept. Mar. Sept. Mar. Sept. Mar. Sept. The Fed’s announce- 2020 2008 2020 2008 2020 2008 2020 2008 2020 2008 2020 2008 19 crisis, the Fed Percentage Percentage Percentage Percentage Percentage Percentage ments of the purchase 4 10 0.2 20 5 2 stepped in to buy 3.4 Change in 93.6 203.5 45.5 of long-term Treasury billions of percent of outstanding dollars 4.8 3.75 419.0 1.5 bonds and subsequent 2 5 266.2 long-term Treasuries, 0.1 10 purchases were criti- 20.1 2.5 1 or US $419 billion, 0 0 cal in restoring mar- starting on March 15. -8.0 1.25 0.5 ket function. Indeed, 0 0 In sharp contrast, the -2 -5 0 as of September 0 0 Fed supplied about -310.8 2020, the long-term -3.3 $200 billion worth of -4 -10 -0.1 -11.9 -10 -1.25 -0.5 Treasury market has Long-term Short-term Long-term Short-term Long-term Short-term long-term Treasuries bonds and notes bills bonds and notes bills bonds and notes bills normalized. Yields during the period have fallen and the from March 2008 to negative β pattern for Source: Researchers’ calculations using data from the Treasury International Capital June 2009 by allow- System, the Federal Reserve Bank of St. Louis, and the Investment Company Institute long-term Treasury ing dealers to obtain bonds has been Treasuries against Figure 4 restored. Whether non-Treasury collateral the events of March in the Term Securities Lending Facility.10 raises the prospect that investors ques- 2020 were a technical aberration or the Next, consider the quantity move- tioned the safe-haven status of these proverbial canary in the coal mine of ment of short-term T-bills in March bonds. international investors shifting toward 2020. Mutual funds purchased a mas- Furthermore, although the dollar a nondollar equilibrium remains an sive amount of short-term Treasuries, did appreciate in March 2020, which open and consequential question. totaling 10.4 percent of the outstanding indicates a flight to safety and is con- stock, or $266 billion. Foreign investors sistent with past episodes of market were net sellers, although the amount turbulence, the degree of appreciation 1 “The Aggregate Demand for Treasury was negligible. Note that foreign central in the dollar against other currencies is Debt,” Krishnamurthy A, Vissing- banks were likely to have acquired dol- much smaller than that in 2008. From Jorgensen A. Journal of Political Economy lars via the Fed’s swap lines, which were the perspective of safe-haven assets, this 120(2), April 2012, pp. 233–267; expanded temporarily on March 19 to observation is at odds with the fact that “The Impact of Treasury Supply on include some central banks in emerg- the 2020 global macroeconomic shock Financial Sector Lending and Stability,” ing markets. We conclude that unlike due to the ongoing COVID-19 pan- Krishnamurthy A, Vissing-Jorgensen A. the case of long-term Treasuries where demic appears as severe as the financial Journal of Financial Economics 118(3), liquidity-sensitive investors sold in 2020, crisis shock in 2008. Moreover, the dol- December 2015, pp. 571–600. investors sought the safe haven of short- lar has depreciated since March 2020, Return to Text term Treasuries in both 2020 and 2008. alongside weakening US economic fun- 2 “The History and Economics of Safe damentals and a rising fiscal deficit. Assets,” Gorton G. Annual Review of Are Treasuries and the Dollar There are alternative perspec- Economics 9, 2017, pp. 547–586. Losing Safe-Haven Status? tives on the Treasury market behavior. Return to Text Darrell Duffie, and in another report 3 “Safety, Liquidity, and the Natural Relative to short-term T-bills, Andreas Schrimpf, Hyun Song Shin, Rate of Interest,” Del Negro M, whose values are largely determined by and Vladyslav Sushko, argue that the Giannone D, Giannoni M, Tambalotti the near-term promise to repay by the price behavior of long-term Treasury A. Brookings Papers on Economic Activity,

NBER Reporter • No. 3, September 2020 23 Spring 2017, pp. 235–316. 8 “Banking, Trade, and the Making of 11 “A Model of Safe Asset Return to Text a Dominant Currency,” Gopinath G, Determination,” He Z, Krishnamurthy 4 “The US Treasury Premium,” Du W, Stein J. NBER Working Paper 24485, A, Milibradt K, American Economic Im J, Schreger J. NBER Working Paper April 2018, and “The International Review, vol 109(4), pages 1230– 23759, August 2017, and Journal of Medium, of Exchange,” Chahrour R, 1262; “Beauty Contests and Iterated 112, May 2018, Valchev R. March 2020, offer mod- Expectations in Asset Markets,” Allen pp. 167–181. els that pin down this denomination F, Morris S, Shin HS. The Review of Return to Text based on firms’ pricing decisions. Financial Studies 19(3), Fall 2006, pp. 5 “International Financial Adjustment,” There is a complementarity between 719–752. Gourinchas P, Rey H. Journal of Political firms’ decisions to denominate their Return to Text Economy 115(4), August 2007, pp. 665– exports in a given unit of account and 12 “Still the World’s Safe Haven? 703. bond issuers’ decisions to denominate Redesigning the US Treasury Market Return to Text in that same unit of account. After the COVID-19 Crisis,” Duffie 6 “Dollar Safety and the Global Return to Text D. Hutchins Center Working Paper Financial Cycle,” Jiang Z, Krishnamurthy 9 “Treasury Inconvenience Yields 62, , May 2020; A, Lustig H. NBER Working Paper during the COVID-19 Crisis,” He “Leverage and Margin Spirals in Fixed 27682, August 2020. Z, Nagel S, Song Z. NBER Working Income Markets During the Covid-19 Return to Text Paper 27416, June 2020. Crisis,” Schrimpf A, Shin HS, Sushko 7 “A Model of Safe Asset Return to Text V. BIS Bulletin 2, April 2020. Determination,” He Z, Krishnamurthy 10 “Repo Market Effects of the Term Return to Text A, Milbradt K. NBER Working Paper Securities Lending Facility,” Fleming 13 “Treasury Inconvenience Yields 22271, May 2016, and American M, Hrung W, Keane F. Federal Reserve during the COVID-19 Crisis,” He Economic Review 109(4), April 2019, pp. Bank of New York Staff Reports, No. Z, Nagel S, Song Z. NBER Working 1230–1262. 426,January 2010. Paper 27416, June 2020. Return to Text Return to Text Return to Text

24 NBER Reporter • No. 3, September 2020 NBER News

John Lipsky Elected Chair of NBER Board of Directors; Peter Blair Henry Elected Vice Chair

John Lipsky was elected chair of the The board elected Peter Blair Henry, the NBER’s Board of Directors at the board’s dean emeritus and William R. Berkley Professor September 14 meeting. He succeeds Karen N. of Economics and Finance at New York Horn, a partner in the Brock Capital Group, University’s Stern School of Business, as vice former CEO of Bank One, and former presi- chair. Henry’s research in international mac- dent of the Federal Reserve Bank of Cleveland, roeconomics overturned conventional wisdom who had served since 2017. Lipsky, the Peter on debt relief, international capital flows, and G. Peterson Distinguished Scholar at Johns the role of institutions in economic growth. He Hopkins University’s Paul H. Nitze School served as head of the external economics advi- of Advanced International Studies (SAIS) sory group for then-Senator ’s and a senior fellow of SAIS’s Foreign Policy 2008 presidential campaign, led the presidential John Lipsky Institute, joined the NBER board in 1998. He transition team’s review of lending agencies, and served as first deputy managing director of the was appointed to the President’s Commission International Monetary Fund between 2006 on White House Fellowships in May 2009. and 2011. Prior to his IMF service, he was Henry is a member of the boards of vice chair of the JPMorgan Investment Bank and Nike, and is the principal investigator of the and served as chief and director of PhD Excellence Initiative, a Sloan Foundation- research at Chase Manhattan Bank. Lipsky is funded fellowship program for minority schol- cochair of The Aspen Institute’s Program on the ars seeking admission to economics doctoral World Economy and vice chair of the Center programs. He received his PhD in econom- for Global Development. He is a life member of ics from MIT, was a Rhodes Scholar at Oxford the Council on Foreign Relations. He received University, and holds a BA in economics from his BA in economics from Wesleyan University the University of North Carolina at Chapel Hill, and his PhD from Stanford University. where he was a Morehead Scholar. Peter Blair Henry

Four New Directors Elected to NBER Governing Board

The NBER Board of Directors elected four new members at its September meeting. Dana M. Peterson is the new representative of The Conference Board, succeeding Bart van Ark. She serves as chief economist at The Conference Board. Prior to joining that organization, she was the North American and global economist at Citigroup, where she analyzed global economic themes with financial market implications. She also was a researcher in the fiscal analysis section of the Federal Reserve Board. Peterson is the rising first vice chair of the New York Association for Business Economics. She received an undergraduate degree in economics with honors from Wesleyan University and a master’s in economics from the University of Wisconsin-Madison.

Dana M. Peterson

NBER Reporter • No. 3, September 2020 25 Lynn Reaser is the new representative of the National Association for Business Economics (NABE), replacing Jack Kleinhenz. She is a faculty member and the chief economist for the Fermanian Business & Economic Institute at Point Loma Nazarene University, and recently completed a four-year term as chief economist of the California Treasurer’s Council of Economic Advisors. Reaser was previously the chief economist for the Bank of America Investment Strategies Group and for First Interstate Bank. She is a past president of the NABE and has been honored as an NABE Fellow. She received her BA, MA, and PhD degrees in economics from the University of California, Los Angeles.

Lynn Reaser

Lars Stole is the new representative of the University of Chicago, taking up a position pre- viously held by John Gould. Stole is the David W. Johnson Professor of Economics at the Booth School of Business. His research interests include applied contract theory, industrial economics, and game theory. He is a past editor of the RAND Journal of Economics, and the founder of the Applied Theory Initiative at the Booth School, which he codirected from 2009–19. He has been an Alfred P. Sloan Research Fellow, a National Science Foundation Presidential Faculty Fellow, and an Olin Fellow in Law and Economics. Stole received his undergraduate degree from the University of Illinois, an MSc in Economics from the London School of Economics, and a PhD from MIT.

Lars Stole

Hal Varian is a new at-large director. He is the chief economist at Google and an emeritus professor in the School of Information, the , and the Department of Economics at the University of California, Berkeley. He has made important research contri- butions in industrial organization, consumer search, the theory of public goods, and demand theory. Varian is the author of two successful microeconomic theory textbooks, one graduate and one undergraduate, as well as an early guide to the information economy, “Information Rules: A Strategic Guide to the Network Economy,” with Carl Shapiro. He received his under- graduate degree from MIT and his PhD from the University of California, Berkeley.

Hal Varian

In addition to these new appointments, the NBER board elected Elizabeth Bailey of the University of Pennsylvania, formerly an at-large board member and board chair from 2005-08, to emeritus status.

A complete listing of members of the NBER Board of Directors https://www.nber.org/board.html

26 NBER Reporter • No. 3, September 2020 New Research Associates and Faculty Research Fellows Named, Fall 2020

The NBER Board of Directors colleges or ; their appoint- fellows, also on the advice of program appointed 42 research associates at its ments are recommended to the board directors and steering committees. September 2020 meeting. All but one by directors of the NBER’s 20 research As of October 1, 2020, there were of these appointees were previously programs, typically after consultation 1,295 research associates and 301 fac- faculty research fellows; most received with a steering committee of leading ulty research fellows. tenure recently at their home insti- scholars. The new research associates The names and university affili- tutions. One appointee is a former are affiliated with 31 different colleges ations of newly promoted and newly research associate who has returned and universities; they received gradu- appointed NBER affiliates are listed from government service. ate at 22 different institutions. below. Italics indicate the research Research associates must be tenured In addition, the NBER president associate returning from government faculty members at North American appointed two new faculty research service.

Research Associates

Rodney Andrews...... University of Texas, Dallas...... Labor Studies Anirban Basu...... University of Washington...... Health Care Christiane Baumeister...... University of Notre Dame...... Environment and Energy Economics Renee Bowen...... University of California, San Diego...... Political Economy Marshall Burke...... Stanford University...... Environment and Energy Economics Marika Cabral...... University of Texas, Austin...... Health Care Katherine Casey...... Stanford University...... Development Economics Arun Chandrasekhar...... Stanford University...... Development Economics James Cloyne...... University of California, Davis...... Monetary Economics Laura Dague...... Texas A&M University...... Health Economics Mariacristina De Nardi...... University of Minnesota...... Public Economics Tatyana Deryugina...... University of Illinois...... Environment and Energy Economics Thomas Fujiwara...... ...... Political Economy Michael Geruso...... University of Texas, Austin...... Health Care Jesse Gregory...... University of Wisconsin...... Labor Studies Catherine Hausman...... University of Michigan...... Environment and Energy Economics Joshua Hausman...... University of Michigan...... Monetary Economics Theresa Kuchler...... ...... Corporate Finance Lee Lockwood...... University of Virginia...... Economics of Aging Matteo Maggiori...... Stanford University...... International Finance and Macroeconomics Arnaud Maurel...... Duke University...... Labor Studies Isaac Mbiti...... University of Virginia...... Development Economics Pascal Michaillat...... Brown University...... Public Economics

NBER Reporter • No. 3, September 2020 27 Eduardo Morales...... Princeton University...... International Trade and Investment Andreas Mueller...... University of Texas, Austin...... Labor Studies Aldo Musacchio...... Brandeis University...... Development of the American Economy Gregory Niemesh...... Miami University...... Development of the American Economy Fernando Parro...... Pennsylvania State University...... International Trade and Investment Ricardo Perez-Truglia...... University of California, Berkeley...... Political Economy Tomas Philipson...... University of Chicago...... Health Care Giorgia Piacentino...... Columbia University...... Corporate Finance Laura Salisbury...... York University...... Development of the American Economy Yu-Chu Shen...... Naval Postgraduate School...... Health Care Adi Sunderam...... ...... Corporate Finance Eric Swanson...... University of California, Irvine...... Monetary Economics Lesley Turner...... Vanderbilt University...... Economics of Education Arthur van Benthem...... University of Pennsylvania...... Environment and Energy Economics Joseph Vavra...... University of Chicago...... Monetary Economics Andrea Vedolin...... Boston University...... Asset Pricing Danny Yagan...... University of California, Berkeley...... Public Economics Wesley Yin...... University of California, Los Angeles...... Health Care Gabriel Zucman...... University of California, Berkeley...... Public Economics

Faculty Research Fellows

Monica Deza...... Hunter College, CUNY...... Health Economics Sebastian Tello-Trillo...... University of Virginia...... Health Economics

28 NBER Reporter • No. 3, September 2020 Conferences

Gender in the Economy

An NBER conference on Gender in the Economy took place online July 24–25. Research Associates Jessica Goldberg of the University of Maryland, Claudia Goldin of Harvard University, Seema Jayachandran of Northwestern University, Claudia Olivetti of Dartmouth College, and Tom Vogl of the University of California, San Diego organized the meeting, which was sponsored by the Bill & Melinda Gates Foundation. These researchers’ papers were presented and discussed:

Papers on Women and Household Finance Issues

• Francesco D’Acunto, Boston College; Ulrike Malmendier, University of California, Berkeley and NBER; and Michael Weber, University of Chicago and NBER, “Gender Roles and the Gender Expectations Gap” (NBER Working Paper 26837)

• Simone G. Schaner, University of Southern California and NBER; Erica M. Field, Duke University and NBER; Rohini Pande, Yale University and NBER; Natalia Rigol, Harvard University and NBER; and Charity M. Troyer Moore, Yale University, “On Her Own Account: How Strengthening Women’s Financial Control Impacts Labor Supply and Gender Norms” (NBER Working Paper 26294)

Papers on Victimization, Vulnerability, and Violence against Women

• Eleonora Guarnieri, Ifo Institute Munich, and Ana Tur-Prats, University of California, Merced, “Cultural Distance and Conflict-Related Sexual Violence”

• Girija Borker, The , “Safety First: Perceived Risk of Street Harassment and Educational Choices of Women”

• Roee Levy and Martin Mattsson, Yale University, “The Effects of Social Movements: Evidence from #MeToo”

Papers on Women’s Well Being and Children’s Health

• Daniel Halim, Hillary C. Johnson, and Elizaveta Perova, The World Bank, “Preschool Availability and Female Labor Force Participation: Evidence from Indonesia”

• Sarah Miller, University of Michigan and NBER; Laura R. Wherry, University of California, Los Angeles and NBER; and Diana G. Foster, University of California, San Francisco, “The Economic Consequences of Being Denied an Abortion” (NBER Working Paper 26662)

• Wolfgang Keller, University of Colorado, Boulder and NBER, and Hâle Utar, Grinnell College, “Globalization, Gender, and the Family” (NBER Working Paper 25247)

Papers on Women and Education across the World

• Claudia Senik, University Paris IV Sorbonne, and Naomi Friedman-Sokuler, Bar-Ilan University, “From Pink-Collar to Lab Coat: Cultural Persistence and Diffusion of Socialist Gender Norms”

• Itzik Fadlon, University of California, San Diego and NBER, and Frederik P. Lyngse and Torben Heien Nielsen, University of Copenhagen, “Early Career, Life-Cycle Choices, and Gender”

NBER Reporter • No. 3, September 2020 29 • Josefa Aguirre, Pontifícia Universidad Católica; Juan Matta, New York University; and Ana María Montoya, Universidad de Chile, “Joining the Men’s Club: The Returns to Pursuing High-Earnings Male-Dominated Fields for Women”

Papers on Victimization, Gender, and COVID-19 that received study group support:

• Heidi Stöckl, London School of Hygiene and Tropical Medicine, and Gerry Mshana, National Institute for Medical Research, Tanzania, “The Effect of COVID-19 on Women, Livelihood, and Violence in Mwanza, Tanzania”

• Amalia R. Miller, University of Virginia, and Carmit Segal, University of Zurich, “Effects of the COVID-19 Pandemic on Domestic Violence in US Cities”

• Sonia R. Bhalotra, University of Essex; Emilia Brito Rebolledo, Brown University; Damian Clarke, University of Chile; Pilar Larroulet, University of Maryland; and Francisco Pino, University of Chile, “COVID-19 and Domestic Violence — Evidence from Rolling Quarantines in Chile”

• Keith Finlay, US Census Bureau; Michael G. Mueller-Smith, University of Michigan; and Brittany Street, University of Missouri, “The Determinants and Aftermath of Victimization in US Households and the Implications of COVID-19”

• Erica M. Field, Duke University and NBER, and Ursula T. Aldana, Institute for Peruvian Studies, “The Impact of COVID-19 on Intimate Partner Violence in Urban Peru”

• Rebecca Thornton, University of Illinois at Urbana-Champaign and NBER; Scott Cunningham, Baylor University; and Gregory DeAngelo, Anuar Assamidanov, and Yunie Le, Claremont Graduate University, “COVID-19, Shelter-in- Place, and Domestic Violence”

• Sarah J. Baird, George Washington University, and Manisha Shah, University of California, Los Angeles and NBER, “The Shadow Pandemic: COVID-19 and Violence against Adolescent Girls in LMICs”

• Bilge Erten and Silvia Prina, Northeastern University, and Pinar Keskin, Wellesley College, “COVID-19 Movement Restrictions and Domestic Violence: Evidence from the US”

Summaries of these papers, as well as a several additional papers that the organizers identified as important related studies, may be found at www.nber.org/conferences/2020/SI2020/GE/summary.html

International Trade Policy and Institutions

An NBER conference on International Trade Policy and Institutions took place online September 11–12. Research Associates Stephen J. Redding of Princeton University and Robert W. Staiger of Dartmouth College organized the meeting, which was spon- sored by the Smith Richardson Foundation. These researchers’ papers were presented and discussed:

• Samuel S. Kortum, Yale University and NBER; David Weisbach, University of Chicago; and Michael Wang, Northwestern Medical School, “Optimal Unilateral Carbon Policy”

• George A. Alessandria, University of Rochester and NBER; Shafaat Y. Khan, The World Bank; andArmen Khederlarian, University of Rochester, “Taking Stock of Trade Policy Uncertainty: Evidence from China’s Pre-WTO Accession”

30 NBER Reporter • No 3, September 2020 • Jiwon Choi, Princeton University; Ilyana Kuziemko, Princeton University and NBER; Ebonya L. Washington, Yale University and NBER; and Gavin Wright, Stanford University, “Local Employment and Political Effects of Trade Deals: Evidence from NAFTA”

• Beata Javorcik, Katherine A. Stapleton, and Ben Kett, University of Oxford; and Layla O’Kane, Burning Glass Technologies, “Unravelling Deep Integration: Local Labour Market Effects of the Brexit Vote”

• Brian McCaig, Wilfrid Laurier University; Nina Pavcnik, Dartmouth College and NBER; and Woan Foong Wong, University of Oregon, “Export Markets and Long-Run Industry Adjustment: State, Private, and Foreign Firms in Vietnam”

• Alberto Cavallo and Gita Gopinath, Harvard University and NBER; Brent Neiman, University of Chicago and NBER; and Jenny Tang, Federal Reserve Bank of Boston, “Tariff Passthrough at the Border and at the Store: Evidence from US Trade Policy”

• Pablo Fajgelbaum, Princeton University and NBER; Pinelopi K. Goldberg, Yale University and NBER; Patrick Kennedy, University of California, Berkeley; Amit Khandelwal, Columbia University and NBER; and Daria Taglioni, The World Bank, “Global Reallocations in the 2018–2019 Trade War”

• Kyle Handley, University of Michigan and NBER; Nuno Limão, University of Maryland and NBER; Rodney Ludema, ; and Zhi Yu, Renmin University of China, “Firm Input Choice underTrade Policy Uncertainty”

• Ohyun Kwon, Constantinos Syropoulos, and Yoto V. Yotov, Drexel University, “Pain and Gain: The Short- and Long- Run Effects of Economic Sanctions on Growth”

• Ralph Ossa, University of Zurich; Robert W. Staiger, Dartmouth College and NBER; and Alan O. Sykes, Stanford University, “Disputes in International Investment and Trade” (NBER Working Paper 27012)

• Emily J. Blanchard, Dartmouth College; Chad P. Bown, Peterson Institute for International Economics; and Davin Chor, Dartmouth College and NBER, “Did Trump’s Trade War Impact the 2018 Election?” (NBER Working Paper 26434)

Summaries of these papers are at www.nber.org/conferences/2020/ITPIs20/summary.html

Employer Challenges in Financing and Managing Plans

An NBER conference on Employer Challenges in Financing and Managing Pension Plans took place online September 17–18. Research Associates Robert L. Clark of North Carolina State University and James M. Poterba of MIT organized the meeting, which was sponsored by the Alfred P. Sloan Foundation. These researchers’ papers were presented and discussed:

• Olivia S. Mitchell, University of Pennsylvania and NBER, “Building Better Retirement Systems in the Wake of the Global Pandemic”

• Robert L. Clark, “Recent Developments in Public Sector Pension Plans”

• Deborah J. Lucas, MIT and NBER, and Daniel Smith, MIT, “How Much Can Collective Defined Contribution Plans Improve Risk-Sharing?”

NBER Reporter • No. 3, September 2020 31 • Dhiren Patki, Federal Reserve Bank of Boston, “Breaking the Implicit Contract: Using Pension Freezes to Study Lifetime Labor Supply”

• Sean Myers, Stanford University, “Public Employee and Municipal Insolvency”

• Maria D. Fitzpatrick, Cornell University and NBER, and Gopi Shah Goda, Stanford University and NBER, “The Prevalence of COLA Adjustments in Public Sector Retirement Plans”

• Laura Quinby and Gal Wettstein, Boston College, “Do Deferred Benefit Cuts for Current Employees Increase Separation?”

• Chuck Boyer, University of Chicago, “Public Pensions and State Government Borrowing Costs”

Summaries of these papers are at www.nber.org/conferences/2020/PPf20/summary.html

Tax Policy and the Economy

An NBER conference on Tax Policy and the Economy took place online September 24. Research Associate Robert A. Moffitt of Johns Hopkins University organized the meeting, which was sponsored by the Lynde and Harry Bradley Foundation. These researchers’ papers were presented and discussed:

• Jeffrey Clemens, University of California, San Diego and NBER; Joshua D. Gottlieb, University of Chicago and NBER; and Jeffrey Hicks, University of British Columbia, “How Would Medicare for All Affect Health System Capacity? Evidence from Medicare for Some”

• Zhao Chen and Zhikuo Liu, Fudan University; Yuxuan He, Duke University; Juan Carlos Suárez Serrato and Daniel Xu, Duke University and NBER, “The Structure of Business Taxation in China”

• Youssef Benzarti, University of California, Santa Barbara and NBER, “Estimating the Costs of Filing Tax Returns and the Potential Savings from Policies Aimed at Reducing These Costs”

• Mark Duggan and Gopi Shah Goda, Stanford University and NBER, and Gina Li, Stanford University, “The Effects of the Affordable Care Act on the Near Elderly: Evidence on Coverage and Labor Market Outcomes”

• Benjamin Lockwood, University of Pennsylvania and NBER; Afras Sial, University of Pennsylvania; and Matthew C. Weinzierl, Harvard University and NBER, “Designing, not Checking, for Policy Robustness: An Example with Optimal Taxation”

• Diane Whitmore Schanzenbach, Northwestern University and NBER, and Michael R. Strain, American Enterprise Institute, “Employment Effects of the Earned Credit: Taking the Long View”

Summaries of these papers are at www.nber.org/conferences/2020/TPE20/summary.html

32 NBER Reporter • No 3, September 2020 Economics of Artificial Intelligence

An NBER conference on the Economics of Artificial Intelligence took place online on September 24–25. Research Associates Ajay K. Agrawal, Joshua S. Gans, and Avi Goldfarb of the University of Toronto and Catherine Tucker of MIT organized the meet- ing, which was sponsored by the Alfred P. Sloan Foundation and the Creative Destruction Lab at the University of Toronto. These researchers’ papers were presented and discussed:

• Martin Beraja, MIT and NBER; David Y. Yang, Harvard University and NBER; and Noam Yuchtman, London School of Economics and NBER, “Data-Intensive Innovation and the State: Evidence from AI Firms in China” (NBER Working Paper 27723)

• Stephanie Assad and Robert Clark, Queen’s University; Daniel Ershov, Toulouse School of Economics; and Lei Xu, Bank of Canada, “Algorithmic Pricing and Competition: Empirical Evidence from the German Retail Gasoline Market”

• Wei Jiang, Columbia University and NBER, and Sean Cao, Baozhong Yang, and Alan L. Zhang, Georgia State University, “How to Talk When a Machine Is Listening: Corporate Disclosure in the Age of AI”

• Kate Bundorf and Maria Polyakova, Stanford University and NBER, and Ming Tai-Seale, University of California, San Diego, “How Do Humans Interact with Algorithms? Experimental Evidence from Health Insurance” (NBER Working Paper 25976)

• Laura Blattner, Stanford University, and Scott T. Nelson, University of Chicago, “How Costly Is Noise? Data and Disparities in the US Mortgage Market”

• Anton Korinek, University of Virginia and NBER, and Joseph E. Stiglitz, Columbia University and NBER, “Steering Technological Progress”

• Stephan T. Zheng, Alexander Trott, Sunil Srinivasa, Melvin Gruesbeck, and Richard Socher, Salesforce Research; Nikhil Naik, MIT; and David Parkes, Harvard University, “The AI Economist: Improving Equality and Productivity with AI-Driven Tax Policies”

• Simona Abis, Columbia University, and Laura Veldkamp, Columbia University and NBER, “The Changing Economics of Knowledge Production”

• Emma J. Pierson and Jure Leskovec, Stanford University; David M. Cutler, Harvard University and NBER; Sendhil Mullainathan, University of Chicago and NBER; and Ziad Obermeyer, University of California, Berkeley, “An Algorithmic Approach to Explaining Why the Underserved Feel More Pain”

• Dirk Bergemann and Tan Gan, Yale University, and Alessandro Bonatti, MIT, “The Economics of Social Data”

• Danielle Li, MIT and NBER; Lindsey R. Raymond, MIT; and Peter Bergman, Columbia University and NBER, “Hiring as Exploration” (NBER Working Paper 27736)

• Debraj Ray, New York University and NBER, and Dilip Mookherjee, Boston University and NBER, “Growth, Automation and the Long-Run Share of Labor”

• Katherine A. Stapleton, University of Oxford, and Michael Webb, Stanford University, “Automation, Trade and Multinational Activity: Micro Evidence from Spain”

NBER Reporter • No. 3, September 2020 33 • Ashesh Rambachan, Harvard University; Jon Kleinberg, Cornell University; Jens Ludwig, University of Chicago and NBER; and Sendhil Mullainathan, “An Economic Approach to Regulating Algorithms”

• Daron Acemoglu and David Autor, MIT and NBER; Pascual Restrepo, Boston University and NBER; and Jonathon Hazell, MIT, “AI and Jobs: Evidence from Online Vacancies”

Summaries of these papers are at www.nber.org/conferences/2020/AIf20/summary.html

Program Meeting Economic Fluctuations and Growth

Members of the NBER’s Economic Fluctuations and Growth Program met July 11 online. Research Associates Benjamin Moll of Princeton University and Valerie A. Ramey of the University of California, San Diego organized the meeting. These researchers’ papers were presented and discussed:

• Antonio Coppola, Harvard University; Matteo Maggiori, Stanford University and NBER; Brent Neiman, University of Chicago and NBER; and Jesse Schreger, Columbia University and NBER, “Redrawing the Map of Global Capital Flows: The Role of Cross-Border Financing and Tax Havens”

• Kevin Donovan, Yale University; Jianyu Lu, Central Bank of Chile; and Todd Schoellman, Federal Reserve Bank of Minneapolis, “Labor Market Dynamics and Development”

• Martin S. Eichenbaum and Sergio Rebelo, Northwestern University and NBER, and Mathias Trabandt, Freie Universität Berlin, “The Macroeconomics of Epidemics” (NBER Working Paper 26882)

• Vasco M. Carvalho, University of Cambridge; Stephen Hansen, Imperial College London; José Rodriguez Mora, University of Edinburgh; and Juan R. García, Álvaro Ortiz, Tomasa Rodrigo, and José Ruiz, BBVA Research, “Tracking the COVID-19 Crisis with High-Resolution Transaction Data”

• Alisdair McKay, Federal Reserve Bank of Minneapolis, and Johannes Wieland, University of California, San Diego and NBER, “Lumpy Durable Consumption Demand and the Limited Ammunition of Monetary Policy”

Summaries of these papers are at https://www.nber.org/conferences/2020/SI2020/EFGs20/summary.html

34 NBER Reporter • No 3, September 2020 NBER Books

NBER Macroeconomics Annual 2019, Volume 34 Edited by Martin Eichenbaum, Erik Hurst, and Jonathan A. Parker https://www.journals.uchicago.edu/toc/ma/2020/34/+ The thirty-fourth edition of theNBER nomic performance was affected by a bind- Macroeconomics Annual features theoretical ing zero lower bound constraint on the and empirical studies of issues in contempo- interest rate. Michael McLeay and Silvana rary macroeconomics and a keynote address Tenreyro explain why it is difficult to iden- by James Stock, a member of President tify empirically the — a key Obama’s Council of Economic Advisers element of the policy framework used by from 2013 to 2014. Matias Covarrubias, central banks — using aggregate data. The Germán Gutiérrez, and authors suggest using regional variation study the evolution of profits, investment and in and inflation to esti- market shares in the US industries over the mate the relationship between these vari- past 40 years and find evidence of inefficient ables. Nir Jaimovich, Sergio Rebelo, Arlene concentration and barriers to entry since Wong, and Miao Ben Zhang investigate 2000. Margherita Borella, Mariacristina De the role that increases in the quality of the Nardi, and Fang Yang examine the effects goods consumed (“trading up”) played in of shorter life expectancies, higher medi- the rise of the skill premium that occurred cal expenses, and lower for white, over the last four decades. Chong-en Bai, non-college-educated Americans born in Chang-Tai Hsieh, and Zheng Song exam- the 1960s on labor supply and retirement ine the “special deals” provided by Chinese savings. Davide Debortoli, Jordi Galí, and local governments to favored private firms Luca Gambetti assess whether recent eco- and their effects on economic growth.

Social Security Programs and Retirement around the World: Reforms and Retirement Incentives Edited by Axel Börsch-Supan and Courtney C. Coile https://press.uchicago.edu/ucp/books/book/chicago/S/bo45618333.html

This ninth phase of the International tude of the employment increase and its Social Security project, which studies the large variation across countries. experiences of 12 developed countries, The studies in this volume explore examines the effects of public pension how financial incentives to work at older reform on employment at older ages. In ages have evolved as a result of public pen- the last two decades, men’s labor force sion reforms since 1980 and how these participation at older ages has increased, changes have affected retirement behavior. reversing a long-term pattern of decline; Utilizing a common template to analyze the participation rates for older women have developments across countries, the findings increased dramatically as well. While bet- suggest that social security reforms have ter health, more education, and changes strengthened the financial returns to work- in labor-supply behavior of married cou- ing at older ages, and that these enhanced ples may have affected this trend, these financial incentives have contributed to the factors alone cannot explain the magni- rise in later-life employment.

NBER Reporter • No. 3, September 2020 35 Nonprofit Org. NBER U.S. Postage Reporter PAID NATIONAL BUREAU OF ECONOMIC RESEARCH National Bureau of Economic Research 1050 Massachusetts Avenue Cambridge, Massachusetts 02138-5398 (617) 868-3900

Change Service Requested