NBER Reporter NATIONAL BUREAU of ECONOMIC RESEARCH

NBER Reporter NATIONAL BUREAU OF ECONOMIC RESEARCH

Reporter OnLine at: www.nber.org/reporter 2010 Number 3 The 2010 Martin Feldstein Lecture

Remarks by Roger W. Ferguson, Jr.*

I am honored to deliver the 2010 Martin Feldstein lecture. It is especially appropriate that I will be discussing the issue of retirement security, because throughout the course of his distinguished career, Martin Feldstein has improved our understanding of the U.S. Social Security system and of retirement policy more broadly. We clearly are at a pivotal moment in the national discussion on retirement security. Over the past 30 years, the responsibility for funding retirement and the associated risks have shifted from employers to individuals. Today, Americans are recovering from a deep plunge in financial markets and a recession that left people less confident about their ability to achieve financial security. Roger W. Ferguson, Jr. In discussing the future of retirement security in America, I will examine how we arrived at the current situation, outline a few core IN THIS ISSUE features that could be built into a retirement security plan or system, and consider the question that is increasingly asked by researchers: The Martin Feldstein Lecture 1 how can we design retirement plans that increase the likelihood of generating an adequate and secure lifetime income? Research Summaries … Life Chances of Disadvantaged Children 6 The Changing Retirement Landscape … Individual Saving and Investment … 9 …Time Use in Twentieth Century America 13 The contours of the U.S. retirement system have changed sub- … Change for the U.S. Auto Industry … 15 stantially over the past few decades, as the defined benefit pension systems that previous generations relied on for secure retirement NBER Profiles 18 income have become increasingly rare. According to the Employee Conferences 20 Benefits Research Institute, only 33 percent of employees working NBER News 23 for large and medium businesses had access to a defined benefit pen- Program and Working Group Meetings 24 sion plan in 2008 — down from 84 percent 30 years ago.1 In their Bureau Books 26 place, a patchwork of individual accounts has placed greater respon- sibility and risk on individual workers. Initially envisioned as a way for Americans to supplement

* Roger W. Ferguson is President and Chief Executive Officer of TIAA-CREF. This is a written and abbreviated version of the Martin Feldstein Lecture given on July 28, 2010. A video of the full lecture is at http://www.nber.org/feldstein_lecture_2010/feldsteinlecture_2010.html

NBER Reporter • 2010 Number 3 the pensions made available by their employers, 401(k) plans have instead become most work- NBER Reporter ers’ primary means of saving for retirement. As a result, 401(k)-type products have fostered a focus on asset accumulation rather than income in retirement. The decline in financial markets in 2008 and The National Bureau of Economic Research is a private, nonprofit research orga- the ensuing global recession have caused many nization founded in 1920 and devoted to objective quantitative analysis of the Americans — especially those nearing retire- American economy. Its officers and board of directors are: ment — to question whether they will be finan- President and Chief Executive Officer — James M. Poterba cially secure after they stop working. Indeed, can Controller — Kelly Horak they stop working and enjoy anything approach- BOARD OF DIRECTORS ing the standard of living to which they are Chairman — John S. Clarkeson accustomed? Vice Chairman — Kathleen B. Cooper The evidence is mixed, but for many people Treasurer — Robert Mednick the answer seems to be no. Last year, research DIRECTORS AT LARGE from McKinsey and Company found that the Peter Aldrich Jessica P. Einhorn Michael H. Moskow average American couple will face a savings gap Elizabeth E. Bailey Mohamed El-Erian Alicia H. Munnell of $250,000 at the time of retirement.2 Richard Berner Jacob A. Frenkel Robert T. Parry Why has the 401(k) framework failed to John Herron Biggs Judith M. Gueron James M. Poterba John S. Clarkeson Robert S. Hamada John S. Reed adequately prepare workers for retirement? Its Don R. Conlan Peter Blair Henry Marina v. N. Whitman shortcomings include: Kathleen B. Cooper Karen N. Horn Martin B. Zimmerman • Lack of participation among many eli- Charles H. Dallara John Lipsky George C. Eads Laurence H. Meyer gible workers; • Insufficient employer and employee DIRECTORS BY UNIVERSITY APPOINTMENT contributions; George Akerlof, California, Berkeley Mark Grinblatt, California, Los Angeles • The failure or inability of many partici- Jagdish W. Bhagwati, Columbia Marjorie B. McElroy, Duke pants to implement an appropriate asset Glen G. Cain, Wisconsin Joel Mokyr, Northwestern allocation strategy; Alan V. Deardorff, Michigan Andrew Postlewaite, Pennsylvania Ray C. Fair, Yale Uwe E. Reinhardt, Princeton • The failure to preserve assets for Franklin Fisher, MIT Craig Swan, Minnesota retirement; John P. Gould, Chicago David B. Yoffie, Harvard • And a lack of annuitization of accumu- lated assets in retirement to produce a DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS lifelong income stream. Bart van Ark, The Conference Board Fundamentally in the 401(k) context, retire- Jean Paul Chavas, Agricultural and Applied Economics Association Martin Gruber, American Finance Association ment risk burdens — funding, investment, lon- Arthur B. Kennickell, American Statistical Association gevity, and mortality — fall disproportion- Thea Lee, American Federation of Labor and ately, often entirely, upon workers who are not Congress of Industrial Organizations William W. Lewis, Committee for Economic Development equipped to manage such risks. Robert Mednick, American Institute of Certified Public Accountants On the other hand, employers have benefited Alan L. Olmstead, Economic History Association over the past three decades by jettisoning defined Harvey Rosenblum, National Association for Business Economics John J. Siegfried, American Economic Association benefit pensions. For instance, it was only by Gregor W. Smith, Canadian Economics Association reshaping their retiree health and savings plans The NBER depends on funding from individuals, corporations, and private foun- that the big three U.S. auto manufacturers could dations to maintain its independence and its flexibility in choosing its research avoid extinction. And in the public sector, where activities. Inquiries concerning contributions may be addressed to James M. defined benefit plans are still common, employ- Poterba, President & CEO, NBER 1050 Massachusetts Avenue, Cambridge, MA 02138-5398. All contributions to the NBER are tax deductible. ers are encouraging newly hired workers to select defined contribution retirement options. The Reporter is issued for informational purposes and has not been reviewed by the Board of Directors of the NBER. It is not copyrighted and can be freely repro- While the decline of defined benefit pen- duced with appropriate attribution of source. Please provide the NBER’s Public sions and the rise of defined contribution plans Information Department with copies of anything reproduced. have removed an element of security from most Requests for subscriptions, changes of address, and cancellations should be sent Americans’ retirement equation, the resulting to Reporter, National Bureau of Economic Research, Inc., 1050 Massachusetts individualized retirement system is more closely Avenue, Cambridge, MA 02138-5398. Please include the current mailing label. aligned with the way Americans work today.

2 NBER Reporter • 2010 Number 3 With more frequent job changes, of retirement plan contribution lev- well-diversified investment portfolios. including spells of independent work, els relative to other factors for ensur- More choices could be confusing and it makes less sense for Americans to ing an adequate level of retirement might actually lead people to choose have their retirement savings tied to a income.3 Workers who want to main- less-diversified investments.6 single employer. tain a standard of living close to what Workers often lack the knowledge So, three facts emerge: they enjoy at the end of their working to choose appropriate investments • First, defined benefit pen- years should be aiming to replace at and to diversify their savings. Olivia sion plans proved too expen- least 70 percent of their final salary in Mitchell of the Wharton School and sive for the vast majority of retirement. This means that individu- Stephen Utkus of Vanguard have writ- American businesses, and the als should save at least 10–15 percent ten that plan participants tend to use tide now appears to be turn- of their gross annual income, mea- “a naïve heuristic (avoid extremes, pick ing in the public sector as sured by the combined contribution the middle option) rather than main- well; of both employers and employees. tain a consistent set of well-ordered • Second, defined contribution Currently, contributions aver- risk preferences to select from the retirement plans, which shift age less than 6 percent of pay for investments offered.” 7 responsibility to individu- non-highly compensated workers Participant confusion underscores als, offer less security than and 7 percent for highly compen- the need for reliable, independent defined benefit plans and put sated workers. Sponsor contributions advice. Historically, there has been a much emphasis on asset accu- average about 3 percent of wages.4 legislated firewall between plan admin- mulation rather than retire- Half of American workers do not have istration and plan advice. Recent leg- ment income planning; access to an employer-sponsored plan. islative and regulatory changes have • And third, Americans’ work Among those who don’t have a work- lowered the firewall and, as a result, patterns have changed, so place retirement plan, fewer than 10 the defined contribution market has that portability and individ- percent have an individual account, been moving to provide individual- ual control are attractive to such as a traditional or Roth IRA. ized investment advice. The percent- workers. Asset retention is another con- age of 401(k) plans offering invest- Given these facts, the challenge for cern. Savings can only grow if they ment advisory services has increased policymakers, financial services com- remain in the plan. Because they allow from 37 percent in 2005 to about 50 panies, economists, and employers is for loans, hardship withdrawals, and percent in 2009.8 how to design retirement systems that lump-sum distributions when workers Guaranteed Income offer flexibility and individual choice, change jobs, 401(k) plans are replete yet still provide genuine security to with opportunities for savings to leak The third core element of a retire- individual savers. In a sense, these out and be used for other purposes. ment security plan is guaranteed life- retirement plans would be grounded A major expense looming for time income.9 Guaranteed income in in the realities of the present, while retirees — one that requires advanced the form of annuities — whose guar- incorporating a measure of security planning and saving — is health care. antee is subject to the claims-paying associated with the past. Without an employer-sponsored ability of the insurance company writ- health plan, a couple retiring at age ing the contract — could re-introduce Core Elements of 65 today is projected to need between the element of security that has been Retirement Security $200,000 and $800,000 to supple- missing from most private sector 401k ment Medicare and cover out-of- plans for the past three decades. Which factors are most critical to pocket health care expenses during Annuities can be made available enjoying a secure retirement? While retirement.5 That is a staggering sum within a retirement plan as an accu- there are many individual needs, three for most people — and it reinforces mulation vehicle, as a distribution elements are core: sufficient retire- the need to accumulate adequate sav- option upon retirement, or through ment plan funding by participants and ings for retirement. both the accumulation and distribu- sponsors; appropriate diversification tion phases. Individuals also have the and asset allocation; and guaranteed Appropriate Diversification and option of purchasing an annuity out- lifetime income in the form of a low- Asset Allocation side their retirement plan. cost, relatively transparent annuity. The second core element of a A good rule of thumb is to annui- Sufficient Funding retirement security plan is appropri- tize at least enough savings, so that, ate diversification and asset alloca- combined with Social Security, a Recent research has clearly dem- tion. Fifteen to 20 fund options typi- retiree will have an income stream onstrated the overriding importance cally give savers the ability to create to meet his or her basic expenses in

NBER Reporter • 2010 Number 3 3 retirement — housing, utilities, taxes, servative funds.12 To overcome this tize is affected by the way the choice food, and health care to the extent second-stage inertia, plan sponsors is framed. According to their research, those costs are knowable. In addition, increasingly are adopting auto-esca- survey participants are more than ideally the value of these annuitized lation policies: automatically increas- three times as likely to prefer a life payments should be protected — at ing an individual’s contribution rate annuity to a savings account when the least partially — against erosion by over time. choice is framed in consumption terms inflation. One such effort is the Save More rather than in investment terms. They Incorporating these three core ele- Tomorrow program, developed by explain: “When consumers think in ments — sufficient funding, appropri- Richard H. Thaler of the University terms of consumption, they perceive ate asset allocation, and guaranteed of Chicago and UCLA’s Shlomo the life annuity as offering valuable income — into retirement plans would Benartzi. The program allows work- insurance against the risk of outliv- enhance efforts to help all Americans ers to schedule automatic increases in ing one’s resources. However, when save for a secure retirement. their savings rate for future dates. In they think in investment terms, they Behavioral Aspects of the Retirement Thaler and Benartzi’s first case study, view life annuities as increasing risk Challenge participants increased their set-aside without increasing return, because of rate from 3.5 percent to more than 13 the potential for variation in the total Even if we design retirement percent. Benartzi recently has written value of payments based on how long plans that encourage sufficient sav- that more than half of large employ- they live.”16 ings, appropriate diversification, and ers in the United States now offer the The desire to avoid what is per- opportunities to turn savings into program.13 ceived as a loss has been identified as guaranteed income, will individuals If we accept that at least par- a powerful motivator for individu- take advantage of these options? tial annuitization — or the purchase als. Recently, Columbia University’s Behavioral economists have been of guaranteed lifetime income — is Eric Johnson uncovered what he calls examining how system design can the optimal choice for people entering “hyper loss aversion” among retir- influence participant behavior. Much retirement, how can we influence their ees, who were up to five times more of the literature has focused on over- decision-making and encourage them loss averse than the average person.17 coming, or leveraging, apparently neg- to move in that direction? Employers Interestingly, this hyper sensitiv- ative tendencies — such as inertia and have been reluctant to include annui- ity to loss does not translate into a risk aversion — with new plan fea- ties as a distribution option. And, all desire to purchase guaranteed income. tures and approaches, including auto- 401(k) plans offer a lump-sum distri- Instead, Johnson has found, retirees enrolling workers in a plan and fram- bution option, but only 14 percent who exhibit hyper loss aversion are ing choices in a way that motivates offer the ability to annuitize assets.14 less likely to annuitize because they see optimal decision making. A commonly cited reason for plan giving up immediate control of their According to the Government sponsors’ reluctance to offer annuities savings as another type of loss. Accountability Office, auto-enroll- is fiduciary uncertainty. Regulatory ment can increase participation clarity could go some way toward The Need for Further Research rates to as high as 95 percent.10 The encouraging more employers to make Employee Benefits Research Institute annuities available. Despite actions taken to increase reports that auto-enrollment has If annuities are more widely avail- savings and highlight the benefits of increased the number of near-retir- able, will employees purchase them? guaranteed income, a distressingly ees who are on track to have enough Paul Yakoboski of the TIAA-CREF large fraction of people pay little heed. money to pay for basic expenses and Institute has found that retirees who Annamaria Lusardi of Dartmouth health care costs — from about 41 have annuitized their retirement sav- College and Jason Beeler of Harvard percent in 2003 to a little over half ings are more than twice as likely, University found that in the year today.11 compared with retirees who have not before the financial crisis, 30 percent However, automatically enrolling annuitized, to have saved through of Baby Boomers — the people closest individuals in a retirement plan is not an annuity in a defined contribution to retirement — had given no thought necessarily a panacea. Some research plan while working. Furthermore, rec- to retirement planning.18 has suggested that while participation ommendations of financial advisors How can we reach the nonplan- rates increase when employers auto- have a measurable impact on the deci- ners? And how can we reach more enroll employees, the default contri- sion to annuitize. 15 of the individuals who are planning, bution levels tend to be fairly low, and Jeffrey Brown of the University but who lack the knowledge to make employees often remain at these low of Illinois and his collaborators have informed decisions and may feel par- contribution levels and in very con- explored how the decision to annui- alyzed by the process? We need fur-

 NBER Reporter • 2010 Number 3 ther research to drive innovations in Security: A Shared Responsibility,” Madrian, and A. Metrick, “For Better retirement plan design, to aid poli- McKinsey & Company, October 19, or For Worse: Default Effects and cymakers in strengthening the legal 2009. 401(k) Savings Behavior,” NBER and regulatory framework that sup- 3 Brett P. Hammond and David Working Paper No. 8651, December ports retirement planning, and to cul- P. Richardson, “A New Look at 2001. tivate broader financial knowledge in Retirement Savings and Adequacy: 13 Shlomo Benartzi, “Behavioral America. Individual Investment Risk Finance and the Post-Retirement Among the questions needing fur- Management and the Asset Salary Crisis,” introduction to Behavioral ther exploration: Ratio,” prepared for the Pension Finance and the Post-Retirement • What is the appropriate mix Research Council Annual Meeting, Crisis, prepared and submitted of automatic plan features April 30, 2009. on behalf of Allianz in response with education or advice? 4 51st Annual Survey of Profit to Department of the Treasury/ • At what point in a career is it Sharing and 401(k) Plans, Profit Department of Labor Request for advisable for a participant to Sharing/401k Council of America, Information regarding lifetime income stop being an “auto-bot” and 2008. options in retirement plans, April become a planner who saves 5 Employee Benefit Research Institute 2010. and invests according to his (EBRI), June 2009. 14 Hewitt Associates, “Trends and or her plan? 6 Roderick Crane, Michael Heller, Experience in 401(k) Plans,” 2009 • And how much of their and Paul J. Yakoboski, “Defined survey. income should people save? Contribution Pension Plans in 15 Paul J. Yakoboski, “Retirees, Consensus on this figure the Public Sector: A Best Practice Annuitization, and Defined has been elusive, but if we Benchmark Analysis,” TIAA-CREF Contribution Plans,” TIAA-CREF can clarify the goal for most Institute, April 2008. Institute, April 2010. workers, we may have more 7 Olivia S. Mitchell and Stephen 16 Jeffrey R. Brown, Jeffrey R. Kling, success in helping people P. Utkus, “Lessons from Behavioral Sendhil Mullainathan, and Marian reach that benchmark. Finance for Retirement Plan Design,” V. Wrobel, “Framing and Annuities,” Economists are making an essen- in Pension Design and Structure: TIAA-CREF Institute, January 2009. tial contribution to the future of New Lessons in Behavioral Finance, 17 Shlomo Benartzi, “Hyper Loss retirement by exploring not only how O.S. Mitchell and S.P. Utkus, eds., Aversion: Retirees Show Extremely rational people should act given a cer- Oxford University Press, July 2004. High Sensitivity to Loss, But Shy tain set of facts, but also how they 8 Hewitt Associates, “Trends and Away from Guarantees that Require do act, as individuals prone to biases, Experience in 401(k) Plans,” 2009 Giving Up Control” based on inter- passions, and proclivities that are per- survey. view with Eric Johnson in Behavioral haps even more determinative of their 9 Lifetime income is a guaranteed Finance and the Post-Retirement actions than reason is. With a clear stream of income subject to the claims- Crisis, prepared and submitted view of the possibilities and limita- paying ability of the issuing insurance on behalf of Allianz in response tions of retirement plan design and a company. to Department of the Treasury/ stronger understanding of how people 10 U.S. Government Accountability Department of Labor Request for make financial decisions, we can point Office, “Retirement Savings: Information regarding lifetime income the way toward a more secure finan- Automatic Enrollment Shows Promise options in retirement plans, April cial future. for Some Workers, but Proposals to 2010. Broaden Retirement Savings for Other 18 Annamaria Lusardi and Jason 1 EBRI Databook on Employee Workers Could Face Challenges,” Beeler, “Saving Between Cohorts: Benefits, Chapter 10: Aggregate report GAO-10-31, October 23, 2009. The Role of Planning,” in Redefining Trends in Defined Benefit and 11 The EBRI Retirement Readiness Retirement: How Will Boomers Defined Contribution Retirement Rating™: Retirement Income Fare? B.C. Madrian, O.S.Mitchell, Plan Sponsorship, Participation, and Preparation and Future Prospects, and B.J.Soldo, eds. Oxford University Vesting, updated December 2009. Issue Brief No. 344, July 2010. Press, 2007. 2 “Restoring Americans’ Financial 12 J.J. Choi, D.Laibson, B. C.

NBER Reporter • 2010 Number 3 5 Research Summaries

Improving the Life Chances of Disadvantaged Children

Jens Ludwig*

Improving the schooling outcomes evident in the pre-school years, in part (HUD) sponsored the Moving to for disadvantaged children is central because of disparities in early learning Opportunity (MTO) residential mobil- to efforts to reduce overall inequal- environments. By age three, children in ity experiment. Started in 1994 in five ity and for increasing economic professional families have larger vocab- cities (Baltimore, Boston, Chicago, Los growth. Around 78 percent of white ularies than the parents of children in Angeles, and New York), MTO enrolled high school students graduate within families on welfare.4 a sample of 4600 public housing fami- four years, compared to 58 percent of My research and that of other NBER lies with children and via random lot- Hispanics and 55 percent of blacks.1 In family members suggests that segre- tery offered some families the chance the federal government’s 2007 National gation, poverty, and other aspects of to use a housing voucher to move into a Assessment of Educational Progress, the out-of-school environment, partic- less distressed neighborhood. Random only 16 percent of fourth-grade stu- ularly early in life, indeed seem to mat- assignment in MTO generated very dents who were eligible for free lunch ter for children, but apparently more so large changes in neighborhood con- scored at proficient levels in reading, for behavioral outcomes like schooling ditions among otherwise comparable compared with 44 percent of those with attainment and criminal behavior than groups of families. For example, fami- higher family incomes.2 These large dis- for achievement test scores. lies with MTO vouchers moved into parities understandably have intensified census tracts with average poverty rates concern about how to improve our sys- Social Context of just 12 percent in the year 2000, tem of public schools. much lower than the average baseline The possibility that some of the Since at least the 1920s, social sci- tract’s poverty rate of 50 percent. most effective ways to improve school entists have thought that child devel- Data collected on MTO families outcomes might not have anything to do opment may be heavily influenced by about five years after a baseline revealed with elementary or secondary schools the child’s social context, including the no detectable differences in average first was raised in a landmark 1966 interactions with peers that shape the achievement test scores across ran- study named after its lead investiga- returns to different behaviors, the infor- domly assigned MTO mobility groups. tor, the distinguished sociologist James mation that local adult role models However, my study with Jeffrey Kling S. Coleman.3 The “Coleman Report” convey about the value of schooling and Lawrence Katz shows that arrest made several remarkable claims, includ- and formal labor market involvement, rates for violent crime among youth ing: the black-white gap in school and the quality of local institutions who relocated through MTO were “inputs” was much smaller than gener- such as schools and police. These around 40 percent lower than those ally perceived; school inputs were only beliefs are consistent with the substan- for youth in the control group.5 MTO weakly correlated with student test tial cross-sectional variation observed also reduced arrest rates for other types scores; among the strongest correlates in children’s learning and other out- of crimes among young females, but of test scores were family background comes across schools and neighbor- it seems to have increased property- and the socio-economic composition hoods of differing socio-economic and crime arrests for young males. Other of the child’s school; and, disparities in racial compositions. Yet in practice, iso- studies using data from randomized test scores open up very early in life, so lating the causal effects of social con- public-school choice lotteries also have that for example the black-white test text on children’s life chances has been found that moving to a higher-qual- score gap was already 1.5 standard devi- quite difficult because of the endoge- ity or less segregated school has more ations by first grade. Subsequent stud- nous sorting of families across schools pronounced effects on behavioral out- ies have shown that these disparities are and neighborhoods. comes, like crime, than on achievement To identify and estimate the causal test scores. However, the school choice *Ludwig is a Research Associate in the NBER’s Program on Children and a pro- effects of neighborhoods on children studies do not find signs of adverse fessor at the University of Chicago. His and families, the U.S. Department of effects on property offending or other Profile appears later in this issue. Housing and Urban Development criminal behaviors of male youth.6

 NBER Reporter • 2010 Number 3 These various studies of random- model programs can be taken to scale Head Start’s health services. Our esti- ized housing-voucher or school-choice effectively. mates imply that a 50–100 percent lotteries identify partial-equilibrium Head Start is the main example increase in Head Start funding reduces effects by focusing on those who move of such a scaled-up program, and has mortality rates from relevant causes by to a new social context. To learn more consistently generated debate about 33–50 percent of the control mean, about the general-equilibrium effects whether it produces lasting benefits enough to drive mortality rates from on crime from large-scale government to program participants. Head Start these causes in the treatment counties efforts to re-sort people across social was launched in 1965 by the Office of down to about the national average. contexts, David Weiner, Byron Lutz, Economic Opportunity (OEO) and There do not appear to be drops for and I study the largest and arguably provides low-income children aged 3– other causes-of-death or birth cohorts most important policy initiative in this 5 years, and their parents, with school- that should not be affected by Head area: court-ordered school desegrega- ing, health, nutrition, and social welfare Start. We also find suggestive evidence tion, which has been of increasing inter- services. The first study arguing that of a “jump” at the OEO threshold in est to economists in recent years.7 Most Head Start benefits to children fade educational attainment, but no sta- of the nation’s largest urban districts out rapidly was released in 1966, which tistically significant discontinuities in were forced to desegregate by local fed- meant there was a very short honey- achievement test scores measured dur- eral court order; differences across these moon period. The main concern with ing middle school.10 districts in the timing of the court that early study, and many subsequent orders provide our source of identify- ones, is the possibility that relatively Implications for Policy ing variation. Our analysis suggests that more disadvantaged families may select and Next Steps re-sorting children across social settings (or be selected) into program participa- is not just a zero-sum game. Court- tion, so that naïve regressions that sim- The growing body of research about ordered school desegregation seems to ply compare participants and non-par- the beneficial effects on disadvantaged generate substantial declines in homi- ticipants may understate the benefits of minority children from reducing seg- cide victimization and offending among the program. regation of schools and neighborhoods black youth and, interestingly, seems to My work on Head Start with is relevant to ongoing policy and legal generate beneficial spillovers to other Douglas Miller tries to identify its causal debates about government efforts in groups as well (such as whites and black effects on children’s life outcomes by this area. While there would be great adults), at least in the short term. taking advantage of a discontinuity in value in learning more about the gen- program funding across counties that eral-equilibrium effects of large scale re- Early Childhood Education resulted from the way the OEO initially sorting policies, the evidence we have implemented the program.9 During the to date suggests that helping poor fami- Early disparities in children’s out- spring of 1965, OEO provided techni- lies move out of high-poverty high-rise comes and the possibility that certain cal assistance to the 300 poorest coun- public housing projects may help to learning can take place only at specific ties in the United States to develop improve at least certain aspects of child times in a child’s development have Head Start funding proposals. We show well-being. generated considerable interest in early that program funding and participa- What else policy might do to reduce childhood interventions. Because get- tion rates are 50–100 percent higher in the segregation of low-income minority ting parents to behave in more devel- counties with poverty rates just above children in schools or neighborhoods is opmentally productive ways seems to OEO’s cutoff (the “treatment” group) not clear. While many public housing be quite difficult in practice, most of than in those just below (the control families appear eager to move to less- the policy attention has been devoted group). This funding difference, which distressed areas when given the chance, to center-based early childhood educa- is the key to our regression discontinu- some of my ongoing work with Brian tion (ECE) programs. Intensive, small- ity (RD) research design, appears to Jacob suggests that other low-income scale model programs from the 1960s have persisted through the late 1970s. families who are already in the private and 1970s — such as Perry Preschool The estimated discontinuity in other housing market are reluctant to move and Carolina Abecedarian — have been federal social spending is small and not out of their old neighborhoods, even shown to improve important adult eco- significant. when provided with large rental subsi- nomic and other outcomes, despite Our main finding is that this large dies. Re-sorting children across schools some “fade out” in test score gains. “jump” in Head Start funding at the without changing residential patterns is While these programs seem to gener- OEO threshold is mirrored in a large difficult given how segregated our cities ate benefits far in excess of their costs,8 “drop” in mortality rates to children 5 are, and given past U.S. Supreme Court there remains the important policy to 9 years of age over the period 1973– decisions that make it extremely dif- question of whether these small-scale 83 from causes addressed as part of ficult to re-sort children across school

NBER Reporter • 2010 Number 3 7 district boundaries. Consider, for exam- be less than ideal to have to wait 30 7 D.A. Weiner, B.F. Lutz, and ple, that in the Chicago Public School or 40 years to understand the long- J. Ludwig, “The Effects of School system, just 9 percent of students are term effects of today’s early childhood Desegregation on Crime”, NBER white, and fully 86 percent of students interventions. Working Paper No. 15380, September are eligible for free or reduced price 2009. See also, for example, J. Guryan, lunches. 1 C.B. Swanson, Cities in Crisis, “Desegregation and Black Dropout Whether local, state, or federal gov- 2009 — Closing the Graduation Gap: Rates,” NBER Working Paper No. ernments will increase investments in Educational and Economic Conditions 8345, June 2001, and American early childhood education despite their in America’s Largest Cities, Bethesda, Economic Review, 94(4) (2004), pp. current budget difficulties remains to MD: Editorial Projects in Education, 919–43. be seen. At least as important for public 2009. 8 See, for example, C.R. Belfield, policy is the question of whether Head 2 The Nation’s Report Card, M. Nores, W.S. Barnett, and L. Start is as beneficial for today’s poor Reading 2007: National Assessment Schweinhart, “The High/Scope Perry children as it was in the past. In prin- of Educational Progress at Grades 4 Preschool Program: Cost-Benefit ciple, the net effects of Head Start may and 8, National Center for Education Analysis Using Data from the Age- have changed over time, as the develop- Statistics 2007-496, Washington, DC: 40 Follow-up,” Journal of Human mental quality of the program and its U.S. Department of Education, Institute Resources, XLI(1), (2006), pp.162– alternatives have changed substantially. for Educational Sciences, 2007. 90; and W.S. Barnett and L.N. Masse, The federal government recently 3 J.S. Coleman, E.Q. Campbell, C.J. “Comparative Benefit-Cost Analysis of sponsored a randomized experimental Hobson, et al, Equality of Educational the Abecedarian Program and its Policy study of Head Start that found impacts Opportunity, Washington, DC: Office Implications”, Economics of Education on test scores measured at the end of the of Education, U.S. Department of Review, 26, (2007), pp. 113–25. program year on the order of 0.1 to 0.2 Health, Education, and Welfare, 1966. 9 J. Ludwig and D.L. Miller, “Does standard deviations. These results led 4 B. Hart and T. Risley, Meaningful Head Start Improve Children’s Life to considerable criticism of Head Start Differences in the Everyday Experience Chances? Evidence from a Regression for not doing more to eliminate the test of Young American Children, Discontinuity Design”, NBER Working score gap between minority and white Baltimore, MD: Paul Brooks, 1995. Paper No.11702, October 2005, and children or between rich and poor. But 5 See L. Sanbonmatsu, J.R. Kling, Quarterly Journal of Economics, Deborah Phillips and I note that these G.J. Duncan, and J. Brooks-Gunn, 122(1) (2007), pp. 159–208. initial impacts are about the same as “Neighborhoods and Academic 10 See also J. Currie and D. Thomas, what was found for previous cohorts of Achievement: Results from the “Does Head Start Make a Difference?”, children, for whom we observed last- Moving to Opportunity Experiment”, NBER Working Paper No.4406, ing benefits into adulthood.11 More NBER Working Paper No. 11909, July 1003, and American Economic puzzling are the latest results from January 2006, and Journal of Human Review, 85(3) (1995), pp. 341–64; the experiment’s first-grade follow-up, Resources, XLI, (2006), pp. 649–91; E. Garces, D. Thomas, and J. Currie, which showed almost complete “fade J.R. Kling, J. Ludwig, and L.F. Katz, “Longer Term Effects of Head Start,” out” of these initial gains — a more “Neighborhood Effects on Crime for NBER Working Paper No. 8054, rapid decline in Head Start effects than Female and Male Youth: Evidence December 2000, and American what was observed for previous cohorts from a Randomized Housing Voucher Economic Review, 92(4) (2002), pp. of program participants. Experiment”, NBER Working Paper 999–1012; and D. Deming, “Early The recent Head Start experiment No. 10777, September 2004, and Childhood Intervention and Life-Cycle highlights the great value for social pol- Quarterly Journal of Economics, Skill Development: Evidence from Head icy in learning more about the mapping 120(1), (2005), pp. 87–130. Start,” American Economic Journal: between short- and long-term ECE 6 See J.B. Cullen, B.A. Jacob, and S. Applied Economics, 1(3) (2009), pp. impacts. Ideally, we would be able to Levitt, “The Effect of School Choice 111–34. use short-term effects from ECE stud- on Student Outcomes: Evidence from 11 J. Ludwig and D.A. Phillips, “The ies in a manner analogous to what med- Randomized Lotteries”, NBER Working Benefits and Costs of Head Start”, ical researchers call “surrogate clinical Paper No.10113, November 2003, and NBER Working Paper No.12973, endpoints” (for example, using changes Econometrica 74(5) (2006), pp.1191– March 2007, and Society for Research in blood cholesterol levels to under- 1230; and D. Deming, “Better Schools, in Child Development Social Policy stand effects on long-term risk for car- Less Crime?” working paper, Carnegie- Report, 21(3) (2007), pp. 3–18. diovascular disease). It would certainly Mellon University, 2009.

 NBER Reporter • 2010 Number 3 The Determinants of Individual Saving and Investment Outcomes

Brigitte C. Madrian*

Over the past 30 years, employer My research with several differ- tribution rate and asset allocation for provided defined contribution (DC) ent collaborators, most notably David employees who don’t actively choose savings plan largely have displaced tra- Laibson, James Choi, Andrew Metrick, otherwise. In companies without auto- ditional defined benefit (DB) pensions and John Beshears, shows that changes matic enrollment, the modal contribu- in the private sector. In 1975, there in the nature of savings plan defaults tion rate tends to be the match thresh- were 2.4 active defined benefit plan par- have a tremendous impact on real- old (the contribution rate at which ticipants for each participant in a pri- ized outcomes. We examine savings employees receive the full employer vate sector defined contribution sav- plan participation rates for employ- match). In contrast, the modal contri- ings plan. By 2007, these proportions ees hired before and after several firms bution rate of participants hired under had almost reversed, with 3.4 active instituted automatic enrollment and automatic enrollment is the automatic defined contribution savings plan par- find that participation is substantially enrollment default chosen by the com- ticipants for each defined benefit plan higher under automatic enrollment.1 pany (initial defaults of 2 percent or 3 participant. As this shift puts more and One concern with automatic enroll- percent of pay, usually below the match more individuals in the position of hav- ment is that it may “coerce” employ- threshold, are typical). This shift in ing to self-manage the process of sav- ees into savings plan participation. If the modal contribution rate is driven ing for retirement, a natural question is so, we would expect that many par- not only by the increased participation just how well are individuals doing, and ticipants under automatic enrollment generated by automatic enrollment what factors affect their retirement sav- should eventually opt out of the sav- (which moves people from zero to a ing outcomes. My research over the past ings plan. But we observe very low positive contribution rate), but also by several years has tried to address these attrition rates under either an opt-in individuals who would have otherwise broad questions. or an opt-out participation regime. contributed at a higher rate but who High participation rates and low attri- instead remain at the automatic enroll- Institutional Features and tion rates under automatic enrollment ment default. Savings Outcomes suggest that most employees do not Similar patterns hold with respect object to saving for retirement. In the to asset allocation. A large fraction of Much of my recent research evalu- absence of automatic enrollment, how- savings plan participants stick with ates the effects of different institutional ever, many simply delay joining their the employer-chosen default asset allo- features on individual savings and invest- savings plan. cation under automatic enrollment, ing outcomes. One example of such a Interestingly, the impact of auto- even when the default is an alloca- feature is the default — that is, what matic enrollment on savings plan par- tion that very few savings plan partic- happens if an individual does nothing? ticipation is not very dependent on the ipants actively elected prior to auto- As an example, in a typical employer- existence or generosity of an employer matic enrollment. Asset allocation sponsored savings plan, individually are match.2 This finding is significant defaults also matter outside the context only enrolled if they actively elect to join because many extensions of automatic of automatic enrollment; in compa- the plan: the default is non-participa- enrollment (for example, the recently nies that direct matching contributions tion. Some companies, however, have a adopted KiwiSaver program in New to employer stock, very few employ- different default — they automatically Zealand, or the Automatic IRA pro- ees actively change their allocation ex enroll employees in their savings plan posals in the United States) do not post, even when they have the ability unless employees actively opt-out. require an employer match but none- to do so.3 theless allow individuals to opt out. Why do defaults have such a per- * Madrian is a Research Associate in the Automatic enrollment also affects sistent effect on outcomes? One expla- NBER’s Program on Aging, co-director of savings plan contribution rates and nation is that the default is perceived the Working Group on Household Finance, and Aetna Professor of Public Policy and asset allocations. In an opt-in regime, as an endorsement of a particular out- Corporate Management at the Kennedy employees must choose a contribution come. There is some evidence consis- School of Government at Harvard rate and asset allocation when they tent with this notion.4 First, savings University. Her profile appears later in enroll. Under automatic enrollment, plan participants who were themselves this issue. the company specifies a default con- not affected by automatic enrollment

NBER Reporter • 2010 Number 3 9 are more likely to have an asset alloca- plans, such an approach also influ- Compared to the effects of the tion that mirrors the automatic enroll- ences outcomes relative to the typ- different approaches to savings plan ment default in effect for more recently ical norm of non-participation. For enrollment discussed above, standard hired employee cohorts if they them- example, research on a company that economic incentives have a surprisingly selves did not elect savings plan par- changed its savings plan enrollment weak impact on savings plan participa- ticipation until after automatic enroll- regime from one that required employ- tion. Having an employer match does ment was adopted. Second, savings ees to fill out a form either affirma- increase participation in a savings plan, plan participants who were subject to tively electing or affirmatively rejecting but many eligible employees still fail automatic enrollment but who take savings plan participation to a “stan- to sign up in the absence of automatic action to move away from the auto- dard enrollment” (for example opt-in) enrollment even with such a match.8 matic enrollment default have asset regime finds that savings plan partici- Choi, Laibson, and I examine a group allocation outcomes that are closer to pation three months after hire declined of workers who face particularly strong the default portfolio than do partici- from approximately 70 percent (when financial incentives for savings plan pants not affected by automatic enroll- an active decision was required) to participation: employees over the age ment — that is, their movement away approximately 40 percent (when no of 59 ½ who are vested, who have an from the default is complete. active decision was required).6 employer match, and who, by virtue of A second explanation for the per- Requiring an active decision has their age, can make unrestricted savings sistence of defaults is that opting-out an impact on asset allocation out- plan withdrawals with no tax penalty. of a default may be cognitively diffi- comes as well. In a recent paper, Choi, Even for this group, we find that a size- cult. For example, initiating savings Laibson, and I 7 study a company at able fraction (20 percent to 60 percent plan participation in the absence of which employer matching contribu- in the seven firms we study) fail to fully automatic enrollment is a complicated tions were originally made in the form exploit the employer match, either by choice that involves electing both a of employer stock, but with no restric- not participating in the savings plan contribution rate and an asset alloca- tions on subsequent diversification. or by contributing less than the match tion. Automatic enrollment simplifies At some point, the firm decided to threshold. We conclude that employer this decision by decoupling participa- require employees instead to explicitly matching is less effective at increasing tion from these other ancillary choices. choose their own asset allocation for savings plan participation than other Evidence that such complexity mat- matching contributions upon enroll- institutional approaches, such as auto- ters comes from two recent papers that ment in the plan (this allocation could matic enrollment or requiring an active evaluate a low-cost manipulation called differ from that chosen for employ- decision. “Quick Enrollment”. This intervention ees’ own contributions). Because there An employer match has its most reduces the complexity of savings plan were no constraints on trading out of significant effect on the distribution enrollment by allowing employees to employer stock before this active deci- of contribution rates rather than on elect participation at a contribution sion was required, savings plan partici- participation. Savings plan contribu- rate and asset allocation pre-selected pants could effect the same asset alloca- tion rates are heavily influenced by the by their employer.5 At one company tion for matching contributions under employer-chosen match threshold.9 studied, Quick Enrollment tripled par- either regime. In practice, however, very For example, in one firm that increased ticipation among new hires relative to few participants in the initial matching its match threshold from 5–6 percent a standard opt-in regime. When Quick regime ever actively reallocated their of pay to 7–8 percent of pay, the frac- Enrollment was made available to pre- match balances; in contrast, under the tion of new participants choosing to viously hired employees who were not active decision regime, participants save 7–8 percent increased from 8 to participating in their savings plan at tended to choose an asset allocation 33 percent of participants, whereas the two different firms, the subsequent for their matching contributions that fraction of new participants choosing enrollment rates of these non-partici- largely mirrored that chosen for their to save 5–6 percent of pay decreased pants increased by 12 to 25 percent- own contributions, and overall expo- from 43 to 19 percent. age points relative to what would have sure to employer stock fell dramatically been predicted in the absence of the as a result. In addition to highlight- Information Provision intervention. ing the difference in outcomes that and Savings Outcomes In many settings, it is hard to avoid occurs under a default versus an active- having a default outcome. One alter- decision-making regime, the results in Information provision and educa- native, however, is to require individu- this paper also suggest that individuals tion also can be useful in influencing als to make an active choice for them- engage in mental accounting and nar- individual behavior, and the savings selves — an “active decision.” In the row framing when making their asset domain is no exception. In a series of context of employer-sponsored savings allocation choices. papers with different collaborators, I

10 NBER Reporter • 2010 Number 3 examine the impact of information Investment prospectuses are does not meaningfully alter subjects’ on savings and investment outcomes. another source of information for indi- investment choices relative to the lon- These papers find that information vidual investors. In an investing exper- ger prospectus. Average portfolio fees provision alone is often not very effec- iment, Choi, Laibson, and I evalu- and past returns are similar regard- tive, and that sometimes individuals ate the impact of information salience less of the type of prospectus partici- can respond to information in perverse on investment outcomes.12 Subjects pants received. We find some weak ways. were asked to allocate a hypothetical evidence, however, that providing the In an analysis with Choi, Laibson, $10,000 across four S&P 500 index Summary Prospectus makes subjects and Andrew Metrick of an employer- funds. Subjects were randomized feel more confident about their port- sponsored financial education initia- across three information conditions: folio choices. tive, we find that compared to non- prospectuses only (control), prospec- And in a very recent paper, the attendees, employees who attend tus plus a short summary of the fees four of us and co-author Katherine financial education seminars are more charged by the mutual funds, or pro- Milkman evaluate the effect of pro- likely to sign up for their employer’s spectus plus a short statement of the viding individuals with information savings plan, to increase their contri- returns since inception attained by the on their coworkers’ behavior in an bution rate, and to make changes to mutual funds. The two treatment con- employer-sponsored savings plan. We their asset allocation.10 The magnitude ditions reduce information gathering find conflicting evidence on the impact of these effects, however, is small, both costs and increase the salience of either of receiving peer information. For one in an absolute sense, and compared to fees or returns since inception, because sub-group of workers — non-union- employees’ intentions regarding their both of these variables are reported ized non-participants — peer informa- future behavior after attending the in the prospectus. Subject payments tion increases the likelihood of subse- seminars. were tied to the actual performance quent savings plan enrollment. But for In another study, Choi, Laibson, of the chosen portfolio. Because pay- another sub-group of workers — union- and I study the impact of information ments were made by the experiment- ized non-participants — we find that provision from the news media using a ers, services like financial advice were peer information actually reduces sub- natural experiment: the media barrage effectively unbundled from portfolio sequent enrollment. The effects of so- on the risk of being over-invested in returns. And, because all of the mutual called social norms marketing are not employer stock that followed the cor- funds in the choice set had the same as predictable as some of the previous porate accounting scandals and stock objective, that is to mimic the returns literature has suggested.14 market decline of 2000–1 (and which of the S&P 500 index, the surest way has become relevant once again follow- to maximize returns was to choose the Market Experience and ing the more recent market decline).11 fund with the lowest fees. We find that Savings Outcomes Three companies received particular subjects overwhelmingly failed to min- attention over that time period: Enron, imize index fund fees. When fees were Finally, Choi, Laibson, Metrick, WorldCom, and Global Crossing. For made salient, average portfolio fees fell, and I examine the impact of previ- example, the New York Times ran 1,364 but most subjects still did not mini- ous market experience on savings out- stories on Enron during the last quarter mize fees. In contrast, when returns comes. In one paper, we study the of 2001 and the first quarter of 2002, since inception (an irrelevant statis- relationship between employee allo- of which 112 ran on the front page. We tic when comparing index funds with cations to employer stock and past show that employer stock holdings in different inception dates) were made employer stock returns. We find that other companies’ savings plans fell by salient, subjects chased these returns. high past returns induce participants only a small amount as a result of the Overall, we find small effects from the to allocate more of their contributions news media. Even in Houston — Enron’s salience manipulations in this experi- to their employer’s stock.15 In a sec- headquarters — where the Houston ment, although we find these effects ond paper, we show that past returns Chronicle ran 1,122 stories on Enron in both for information that should nor- not only impact asset allocation, but the six months surrounding the firm’s matively matter, and for information also individual savings rates.16 High collapse, employees at other companies that should not. unpredictable and idiosyncratic lagged did not diversify their employer stock In a related experiment, Beshears equity returns in an individual’s port- holdings. These results are consistent and I evaluate the effect of provid- folio predict subsequent savings rate with individual inertia (as described ing investors with a traditional invest- increases. This contradicts the relation- above), and also with a mistaken per- ment prospectus relative to the sim- ship predicted by standard economic ception on the part of individuals that pler and shorter summary prospectus theory, but can be explained by extrap- their employer’s stock is less risky other recently approved by the SEC.13 We olative beliefs. When investors expe- equity investments. find that the Summary Prospectus rience high past returns, they forecast

NBER Reporter • 2010 Number 3 11 high future returns. This will lead to Participation and Savings Behavior”, of Employer Matching on Savings increased savings if their elasticity of op. cit.; J. J. Choi, D. Laibson, B.C. Plan Participation Under Automatic intertemporal substitution is greater Madrian, and A. Metrick, “For Better Enrollment,” NBER Working Paper than one. or for Worse: Default Effects and No. 13352, August 2007, and in 401(k) Savings Behavior”, op. cit.; J. Research Findings in the Economics 1 B.C. Madrian and D. Shea, “The Beshears, J. J. Choi, D. Laibson and of Aging, D. A. Wise, ed., Chicago, Power of Suggestion: Inertia in 401(k) B.C. Madrian, “The Importance of IL: University of Chicago Press, 2010, Participation and Savings Behavior”, Default Options for Retirement Savings pp. 311–327. Also, in G. Engelhardt NBER Working Paper No. 7682, Outcomes: Evidence from the United and B.C. Madrian, “Employee Stock May 2000, and Quarterly Journal of States,” op. cit. Purchase Plans,” NBER Working Economics, 2001, 116: pp.1149–87; 5 J. J. Choi, D. Laibson, and B.C. Paper No. 10421, April 2004, and J. J. Choi, D. Laibson, B.C. Madrian, Madrian, “Reducing the Complexity National Tax Journal, 57(2), 2004, pp. and A. Metrick, “For Better or for Costs of 401(k) Participation Through 385–406, we document very high levels Worse: Default Effects and 401(k) Quick Enrollment™,” NBER Working of non-participation in employer stock Savings Behavior”, NBER Working Paper No. 11979, January 2006, and purchase plans, despite the fact that the Paper No. 8651, December 2001, and in Developments in the Economics financial benefits available from par- in Perspectives on the Economics of of Aging, D. A. Wise, ed., Chicago, ticipation in these plans are often non- Aging, D. A. Wise, ed., Chicago, IL: IL: University of Chicago Press, trivial as well. University of Chicago Press, 2004, 2009, pp. 57–82; J. Beshears, J. J. 9 J. J. Choi, D. Laibson, B.C. pp. 81–121; J. Beshears, J. J. Choi, Choi, D. Laibson, and B.C. Madrian, Madrian, and A. Metrick, “Defined D. Laibson, and B.C. Madrian, “Simplification and Saving,” NBER Contribution Pensions: Plan Rules, “The Importance of Default Options Working Paper No. 12659, October Participant Decisions, and the Path of for Retirement Savings Outcomes: 2006. Least Resistance,” op. cit. Evidence from the United States,” 6 G. Carroll, J. J. Choi, D. Laibson, 10 Ibid. NBER Working Paper No. 12009, B.C. Madrian, and A. Metrick 11 J. J. Choi, D. Laibson, and B.C. February 2006, and in Lessons from “Optimal Defaults and Active Madrian, “Are Empowerment Pension Reform in the Americas, S. Decisions: Theory and Evidence from and Education Enough? J. Kay and T. Sinha, eds., New York, 401(k) Saving,” NBER Working Underdiversification in 401(k) Plans,” NY: Oxford University Press, 2008, pp. Paper No. 11074, January 2005, and Brookings Papers on Economic Activity, 59–87. Quarterly Journal of Economics, 2 (2005), pp. 151–98. 2 J. Beshears, J. J. Choi, D. Laibson, 124(4), (2009), pp. 1639–74. 12 J. J. Choi, D. Laibson, and B.C. and B.C. Madrian, “The Impact 7 J. J. Choi, D. Laibson, and B.C. Madrian, “Why Does the Law of One of Employer Matching on Savings Madrian, “Mental Accounting in Price Fail? An Experiment on Index Plan Participation Under Automatic Portfolio Choice: Evidence from a Mutual Funds,” NBER Working Paper Enrollment,” NBER Working Paper Flypaper Effect,” op. cit. No. 12261, May 2006, and in Review No. 13352, August 2007, and in 8 For evidence on the impact of the of Financial Studies, 23(4), (2010), Research Findings in the Economics employer matching and savings plan pp.1405–32. of Aging, D. A. Wise, ed., Chicago, IL: participation in 401(k)-like savings 13 J. Beshears, J. J. Choi, D. Laibson, University of Chicago Press, 2010, pp. plans, see J. J. Choi, D. Laibson, B.C. and B.C. Madrian, “How Does 311–327. Madrian, and A. Metrick, “Defined Simplified Disclosure Affect Individuals’ 3 J. J. Choi, D. Laibson, and B.C. Contribution Pensions: Plan Rules, Mutual Fund Choices?”, NBER Madrian, “Are Empowerment Participant Decisions, and the Path Working Paper No. 14859, April 2009, and Education Enough? of Least Resistance,” NBER Working and forthcoming in Explorations in the Underdiversification in 401(k) Plans,” Paper No. 8655, December 2001, in Economics of Aging, D. A. Wise, ed., Brookings Papers on Economic Activity, Tax Policy and the Economy, Vol. 16, University of Chicago Press. 2:2005, pp. 151–198; J. J. Choi, D. J. M. Poterba, ed., Cambridge, MA: 14 J. Beshears, J. J. Choi, D. Laibson, Laibson, and B.C. Madrian, “Mental MIT Press, 2002, pp. 67–113; J. J. B.C. Madrian, and K. Milkman, “The Accounting in Portfolio Choice: Choi, D. Laibson, and B.C. Madrian, Effect of Providing Peer Information on Evidence from a Flypaper Effect,” “$100 Bills on the Sidewalk: Violations Retirement Savings Outcomes.” NBER Working Paper No. 13656, of No-Arbitrage in 401(k) Accounts,” 15 J. J. Choi, D. Laibson, B.C. November 2007, and American NBER Working Paper No. 11554, Madrian, and A. Metrick “Employees’ Economic Review, 99(5), (2009), pp. August 2005, and forthcoming in The Investment Decisions About Company 2085–95. Review of Economics and Statistics; Stock,” NBER Working Paper No. 4 B.C. Madrian and D. Shea, “The and J. Beshears, J. J. Choi, D. Laibson, 10228, January 2004, and in Pension Power of Suggestion: Inertia in 401(k) and B.C. Madrian, “The Impact Design and Structure, O. S. Mitchell

12 NBER Reporter • 2010 Number 3 and S. P. Utkus, eds., New York, NY: 16 J. J. Choi, D. Laibson, B.C. Behavior,” Journal of Finance, 64(6), Oxford University Press, 2004, pp. Madrian, and A. Metrick, (2009), pp. 2515–34. 121–36. “Reinforcement Learning and Savings

Trends in Time Use in Twentieth Century America

Valerie A. Ramey*

In his 1930 essay “Economic and forecasts of long-term labor supply some of the early studies reported very Possibilities for Our Grandchildren,” behavior. detailed tabulations by characteristics, John Maynard Keynes looked beyond U.S. labor productivity rose eight- which I was able to use in cell-based the pessimism surrounding the Great fold during the twentieth century. regressions. I then used these estimates Depression and predicted that rapid Did leisure time rise significantly in to make the averages more nationally productivity growth would result in response? To answer this question, I representative and linked them to the abundant leisure and freedom from gather detailed data on the main uses available micro data from 1965 on. most economic needs within a hun- of time by major segments of the pop- I find that time spent in home pro- dred years.1 He speculated that the lit- ulation during the twentieth century. duction by housewives fell by only a few tle work left to do would be shared as Although there have been numerous hours between 1900 and 1965, con- widely as possible, so that each person studies of time use and hours of work firming earlier results by sociologists.3 could spend about fifteen hours per conducted during the early twentieth For all prime-age women, time spent in week doing a few meaningful tasks. century, most of them were focused on home production fell by only six hours Keynes was not alone in his belief a particular segment of the population. per week from 1900 to 1965, but by an that a new era of rising leisure was begin- Thus, the main challenge of my research additional twelve hours between 1965 ning. As of the 1930s, the standard fac- was to understand the particular context and 2005, with most of that decrease tory workweek had declined signifi- of each of the earlier studies and then to occurring between 1965 and 1975. cantly over the previous hundred years, combine the pieces into a mosaic that These results are surprising because the appliances were reducing the drudgery would reveal patterns in time use for the main diffusion of appliances occurred of housework, and the high unemploy- general population. before 1965, not after. Moreover, much ment rates of the Great Depression had In “Time Spent in Home Production of the decrease in time spent by women led to “forced leisure.” Numerous schol- in the Twentieth Century United States: from 1900 to 2005 was countered by an arly articles during the 1930s examined New Estimates from Old Data,” I com- increase in time spent by men. various aspects of leisure, from teaching pile information from virtually every Including all age groups, I find that children how to use leisure time wisely time-use study conducted from 1912 to average time spent in home production to a variety of time diary studies that the present in order to estimate trends actually rose slightly over the century. recorded how individuals used their in time spent on “home production” — The absence of a decline in the popu- leisure. that is, unpaid household tasks, such lation overall was in part due to the The extent to which societies as cooking, cleaning, laundry, and tak- decrease in the share of children (who respond to productivity growth by ing care of children.2 Almost all of the do little home production), the increase increasing their leisure time is fundamen- studies use detailed time diaries. While in the share of the retired elderly (who tal to numerous economic questions. most sample only a few hundred peo- do more home production than the For example, the size of the response ple, together they cover thousands of employed), and the loss of economies of affects the foundations of growth mod- individuals across the United States. scale as households got smaller. els, assessments of standards of living, The most detailed data are for farm- Interestingly, time spent in home *Ramey is a Research Associate in wives and housewives, but some of the production by prime-age individuals the NBER’s Program on Economic studies also surveyed employed women, did not decrease after the mid-1970s, Fluctuations and Growth and a Professor men, and children. Others compared although the composition of tasks at the University of California, San Diego. time use across racial groups. Although changed significantly. In particular, as Her Profile appears later in this issue. the individual-level data no longer exist, Mark Aguiar and Erik Hurst demon-

NBER Reporter • 2010 Number 3 13 strate, time spent on such activities as work for various segments of the popu- significant convergence in tasks across cooking and laundry decreased steadily, lation. To measure aggregate time spent genders. while time spent on childcare increased.4 in market work from 1900 to 1958, The groups with the greatest declines Garey Ramey and I document and assess we use the data compiled by Kendrick, in market work over the twentieth cen- possible explanations for the dramatic which covers all sectors and adjusts for tury were those under age 25 and those rise in childcare time in the United actual hours of work (rather than sched- over age 65. To understand how the States beginning in the 1990s.5 We find uled hours of work), and incorporates time was reallocated for the young, it is that the largest increases in childcare detailed estimates of the work of unpaid important to estimate time spent in for- time were among college-educated par- family members on farms.7 We then mal schooling. We estimate time spent ents, although less educated parents also use decennial census data on employ- in school using information on enroll- showed an increase. We test numerous ment and school enrollment by age and ment, the average days attended per possible explanations, such as safety con- gender to allocate the aggregate hours enrollee, and hours spent in class and cerns, income effects, and sample selec- to each demographic group. For the on homework per day attended. The lat- tion, but find that all are inconsistent later period, we use Current Population ter estimates come from numerous time- with the data. We then offer a new the- Survey data. use studies of school children beginning ory linking at least part of the increase Our results paint a very different in the 1930s. Using these estimates, we in childcare time to “cohort crowding” picture of trends in hours of market determine that the 18-hour decline in and the increase in competition to get work from the ones typically discussed. market work among children ages 14 into college. We argue that a significant For example, it is well documented that to 17 was channeled directly into time portion of the increase in childcare time the normal workweek in manufactur- spent in formal education. The story is parents trying to improve their chil- ing fell from about 60 hours per week in was qualitatively similar for those ages dren’s chances of getting into a “good” 1900 to about 40 hours in 2005. When 18 to 24. college by tutoring them and building we look at average weekly hours per cap- After subtracting the various activi- up their after-school resumes. As one ita for all prime age males, we estimate ties from the total amount of time avail- test of the theory, we turn to Canada, that they fell from 50 hours per week able, we can study the residual, which we where there is no steep hierarchy of uni- in 1900 to 41 hours in 1940 and 37 call “leisure.” We find that average weekly versities and where college admissions hours in 2005. Most previous references leisure time for those ages 25 to 54 was are less competitive. Using individual- to dramatic decreases in the workweek about the same in 2005 as it was in 1900. level data from Canadian time- use stud- either referred to the era before 1900 or All other age groups enjoyed rises in lei- ies, we show that time spent on childcare focused on manufacturing. But in 1900, sure during the twentieth century, about did not increase among college-educated manufacturing accounted for only 20 five hours per week for those between parents in Canada over the past twenty percent of all workers — agriculture and the ages of 14 and 25 and fourteen hours years. the government sector had lower average per week for those over age 65. Neville Francis and I estimate time workweeks, in part because of seasonal- We also estimate lifetime leisure for spent in a variety of activities in order ity, and accounted for 40 percent of all various cohorts. The fraction of one’s to produce estimates of work and lei- employment. Thus, the average hours lifetime spent in leisure rose only mod- sure by age and gender since 1900.6 per capita for prime age males fell less estly, from 24 percent for those born Specifically, we study home production, than the standard workweek in manu- in 1890 to 25 percent for those born in market work, formal schooling, com- facturing did. 1950. Cumulative lifetime leisure rose by muting time, personal care time, and lei- For all individuals between the almost 50 percent, though, because of the sure time. ages of 25 and 54, hours spent in mar- increase in life expectancy. Mechanically, Our home production estimates are ket work changed by only a few hours the extra years of life add to the lifetime based on the time diary estimates from between 1900 and 2005. This average endowment of time, so almost all uses my earlier paper, augmented to corre- masks the significant trends within gen- of time rise significantly when viewed spond to the age groups in this paper. der, though: women in this age group cumulatively over a lifetime. We find decreases in time spent in home increased their market hours by 18 hours On balance, my research indicates production for those under the age of per week whereas men decreased their that the response of leisure time to pro- 25, little change for those ages 25 to 64, market hours. Combined with the ear- ductivity growth in the twentieth century and some increase for those ages 65 and lier estimates on home production, these United States was very weak. In a stan- over. time-use estimates indicate a decline in dard model, matching the mild increase We had to use alternative data sources specialization by gender over time: in in leisure time to the eight-fold increase for “hours of market work” because there 1900, most men specialized in market in productivity requires that income and was not sufficient information from the work and most women specialized in substitution effects approximately cancel time diary studies on hours of market home production; by 2005, there was each other out. However, standard mod-

14 NBER Reporter • 2010 Number 3 els do not capture other possible trends Press Ltd., 1931, p. 358–74. and Quarterly Journal of Economics that may have been occurring. For exam- 2 V. A. Ramey, “Time Spent in Home 122, 3 (August 2007): pp. 969–1006. ple, society may have used its increasing Production in the 20th Century United 5 “The Rug Rat Race,” with G. Ramey, wealth to make work more pleasant, so States: New Estimates from Old Data,” NBER Working Paper No. 15284, that the demand for leisure did not rise NBER Working Paper No. 13985, August 2009, and Brookings Papers on as much as would be predicted by a sim- May 2008, and Journal of Economic Economic Activity, forthcoming. ple model. Moreover, the invention of History, Cambridge University Press, 6 A Century of Work and Leisure,” new products, and in particular medical vol. 69 (March 2009), pp. 1–47. with N. Francis, NBER Working Paper technology that could extend life expec- 3 J. Vanek, Keeping Busy: Time Spent No. 12264, May 2006, and American tancies, may have increased the weight in Housework, United States, 1920– Economic Journal: Macroeconomics, that individuals place on market goods 1970, PhD. Dissertation, University of American Economic Association, vol. and services. Michigan, 1973. 1(July 2009), pp. 189–224. 4 M. Aguiar and E. Hurst, “Measuring 7 J. W. Kendrick, Productivity Trends 1 J.M. Keynes, “Economic Possibilities Trends in Leisure: The Allocation in the United States, Princeton: NBER for our Grandchildren,” in Essays in of Time Over Five Decades,” NBER and Princeton University Press, 1961. Persuasion, London: The MacMillan Working Paper No. 12082, March 2006,

A Decade of Change for the U.S. Auto Industry: The Internet, Promotions, and Rising Gasoline Prices

Florian Zettelmeyer*

During the last decade the U.S. financing on all GM vehicles for up important complement to the car-buy- automotive industry has been affected to five years. While manufacturers had ing process. As early as the year 2000, 54 by a series of major changes. First, auto- used financing or price incentives before, percent of all new vehicle buyers used motive retailing, which had been firmly “Keep America Rolling” is thought to the Internet in conjunction with buying controlled by franchised automotive have started a substantial escalation of a car. My work with co-authors Fiona dealers, started to feel the effect of the average incentive amounts.1 Scott Morton and Jorge Silva Risso Internet in the late 1990s. Although Third, the dramatic increase in gaso- looks at whether and how the wide- state franchise laws require all new cars line prices from below $1 in early 1999 spread use of the Internet by consumers to be sold by dealers, the Internet has to $4 at their peak in 2008 made it much has affected auto retailing. become a major source of information more expensive for consumers to operate We first investigate the effect of about car characteristics and pricing. an automobile. This has affected manu- Internet car referral services (Autobytel. Second, the 9/11 terrorist attacks facturers differentially, depending on the com, Autoweb.com, Carpoint.com, and changed the way that automotive firms fuel efficiency of the cars they sell. In a the like) on dealer pricing of automo- compete in the United States. Eight days series of research papers, my co-authors biles in the United States in 1999.2 after 9/11, GM started an incentive pro- and I have investigated the consequences Combining transaction data with data motion with the name “Keep America for the industry of these changes. from a leading online auto referral ser- Rolling” which offered zero percent vice, we compare online transaction The Effect of the Internet on prices to regular “street” prices. We find *Zettelmeyer is a Research Associate the Auto Retailing Industry that Internet prices, controlling for the in the NBER’s Program on Industrial car purchased, on average were 1–2 per- Organization and a Professor at Even though consumers remain cent lower than those paid by conven- Northwestern University. His Profile interested in physically inspecting a tional consumers. In addition, we find appears later in this issue. car, the Internet has become a very that dealer average gross margin on an

NBER Reporter • 2010 Number 3 15 online vehicle sale was lower than an sible that our results are due to selec- tion of prices paid by consumers in the equivalent offline sale. However, these tion. This suggests an additional aspect auto industry. This result is remarkable findings do not imply that the Internet of the “digital divide”: not only are dis- because dealer franchise laws prevent is shifting rents from car retailers to advantaged minorities less likely to use direct competition from either man- consumers. If online car buyers would a computer, but they are also the group ufacturers or independent companies also have negotiated low prices in the that would most benefit from it. using the Internet to sell cars directly to offline world, then the Internet merely While these papers are informa- consumers. Nonetheless, the ease with provides an alternative channel for a tive about the overall effect of Internet which the Internet allows consum- consumer-dealer interaction. usage on new car prices, they leave ers to access information, the partial To determine whether the Internet some unanswered questions about the obfuscation of individual character- has a causal effect on car prices, we use mechanism by which the Internet low- istics when interactions are mediated instrumental variables to control for ers prices for consumers. To answer through the Internet, and the referral selection. We find that traditional buy- this question we added much more mechanism, are enough to affect the ers pay 2.2 percent more than Internet detailed data on the way that consum- distribution of surplus between con- buyers.3 This is consistent with con- ers searched offline and online, and on sumers and firms. sumers choosing to use the Internet their personal characteristics. This led because they know that they would to a paper in which we use direct mea- The Effect of Pricing pay more in the traditional channel, sures of search behavior and consumer Transparency in the U.S. perhaps because they strongly dislike characteristics to investigate how Automotive Market collecting information and bargaining the Internet affects negotiated prices in the traditional way. in car retailing.5 We match transac- After 9/11, incentive promotions This finding raises the question of tion data on 1,500 car purchases in played an increasingly important role the Internet’s effect on groups of con- California with the buyers’ responses in the U.S. automotive market. These sumers who have traditionally been to a survey that asks detailed ques- promotions also provide an opportu- considered disadvantaged in the car tions about their Internet usage, their nity for us to investigate how pricing buying process. In a follow-on paper, attitudes towards information search transparency and information asym- we analyze whether the Internet’s dual and bargaining, and their demograph- metry affects the auto industry and its role of reducing a dealer’s ability to ics. We show that the Internet lowers consumers. accurately assess a consumer’s willing- prices for two distinct reasons: first, Meghan Busse, Silva Risso, and I ness to pay and increasing consumers’ the Internet informs consumers — the exploit a natural experiment to test the ease in finding information reduces most important piece of information effect of private information on the discrimination in car buying by race that consumers glean on the Internet division of surplus between consumers and gender.4 For offline car purchases, is the invoice price of dealers; sec- and automotive dealers.6 Automobile we find a minority race premium of 2.0 ond, the referral process of online buy- manufacturers frequently use two percent to 2.3 percent when we do not ing services, a novel institution made types of promotions that give cash- control for other demographics; 1.1 possible by the Internet, also helps back payments: rebates to customers, percent to 1.5 percent when we con- consumers obtain lower prices. Our which are widely publicized to poten- trol for neighborhood characteristics; results show that the combined infor- tial customers, and discounts to deal- and 0.6 percent to 0.8 percent when mation and referral price effects are ers, which are not publicized. While we control for search costs. This dem- -1.5 percent. This corresponds to 22 the payments nominally go entirely onstrates that pricing of new cars to percent of dealers’ average gross profit to one party or the other, the real offline consumers strongly depends on margin per vehicle. We also find that division of the manufacturer-supplied individual car buyers’ characteristics. the benefits of gathering information surplus between dealer and customer Our main finding is that the Internet differ by consumer type. Buyers who depends on what price the two parties eliminates most of the variation in really dislike bargaining but who have negotiate. These two types of promo- new car prices that results from indi- collected information on the specific tions thus form a natural experiment vidual characteristics associated with car they eventually purchase will pay of the effect of information asymme- race and ethnicity: online buyers who 1.5 percent less than they otherwise try on bargaining outcomes, with the use the Internet referral service that would. In contrast, buyers who like the parties symmetrically informed in the we study pay the same prices as whites, bargaining process do not benefit from customer rebate case and the dealer even after controlling for their income, such information. having an informational advantage in education, and other neighborhood In summary, this research stream the dealer discount case. We show characteristics. Because of the way race shows that the Internet has had a sub- that customers receive approximately is measured in our data, it is implau- stantial effect on the level and distribu- 80 percent of the customer rebate and

16 NBER Reporter • 2010 Number 3 approximately 35 percent of the dealer prices decreased. We hypothesize that relative price increase for fuel-efficient discount. This is consistent with the the complex nature of auto prices, the cars is $363 for a $1 increase in gas theoretical prediction that when cus- fact that prices are negotiated rather prices; for used cars it is $2839. tomers are at an information disad- than posted, and the fact that buy- There are three reasons why these vantage, they are also disadvantaged in ers do not participate frequently in results are interesting. First, the gaso- negotiations. In this setting, the infor- the market made it possible for auto line usage characteristics of the new mation disadvantage is substantial: for manufacturers to manipulate custom- cars added to the U.S. fleet every year a promotion of average size, consum- ers’ beliefs about current versus future affect the level of gasoline consump- ers receive $500 less of the surplus if prices, even without changing prices tion (and greenhouse gas emissions) they do not know that the promotion themselves. over subsequent years. Knowing how is available. gasoline prices (and by extension gaso- The preceding papers raise a more The Effect of Gasoline Prices line taxes or carbon taxes) might affect fundamental question: how well on New and Used Car Markets what cars are sold is thus important informed are automobile consumers for policy decisions. Specifically, our about whether the price they negoti- The dramatic increase in gasoline results suggest that consumer choices ate with a dealer is a “good price,” what prices from below $1 in early 1999 are quite sensitive to gasoline price we refer to as their “price knowledge”? to $4 at their peak in 2008 made it changes. Second, our used car results The evidence suggests that consumer much more expensive for consumers reveal something about how consum- price knowledge may not be high, to operate an automobile. As concern ers trade upfront capital costs against given that information provided by about climate change has grown, econ- ongoing operating costs when they the Internet and through the format of omists have become increasingly inter- choose among cars of different fuel price promotions affect pricing. Most ested in the question of how people efficiencies. This can inform how poli- marketing studies on this topic have respond to the cost of gasoline. Fully cies intended to encourage energy con- found that consumers have poor price addressing this question is not easy, in servation more generally should be knowledge, although the marketing part because there are many margins crafted. The $2839 increase in the dif- studies generally have analyzed only over which individuals—and firms— ference between the most and the least low-priced goods (often in the context can respond, including the usage, pro- fuel-efficient quartiles of cars reflects of supermarkets).7 In contrast, buy- duction choice, customer choice, and fuel expenditure savings associated ing a car is the second largest purchase technology of vehicles. with driving the average car in the most of typical consumers and they spend Busse, Christopher Knittel, and fuel-efficient quartile, rather than the many hours engaging in price search.8 I address one aspect of this question: average car in the least fuel-efficient To determine how much consumer how gasoline prices affect the transac- quartile, for ten years assuming a 3 per- price knowledge exists in the U.S. auto tion shares and prices of new and used cent discount rate. This means that we industry, Busse, Duncan Simester, and cars of different fuel efficiencies.10 We find very little evidence that consum- I analyze an unusual event.9 During combine data on local gasoline prices ers are “myopic” in trading off upfront the summer of 2005, the Big Three and data on model-specific fuel effi- capital costs versus ongoing operating U.S. automobile manufacturers offered ciency with transaction data from a 20 costs. Third, we find that the adjust- a customer promotion: customers percent sample of U.S. new car dealers ment of equilibrium transaction shares could buy new cars at the discounted from 1999 to 2008. These dealers sell and prices in response to changes in price formerly offered only to employ- both new and used vehicles. gasoline prices differs greatly between ees. The initial months of the promo- We find that a $1 increase in gas- new and used markets. In the new car tion produced record sales for each of oline price changes the transaction market, the adjustment is primarily in the Big Three firms, suggesting that shares of the most and least fuel-effi- market shares, while in the used car customers believed that the promo- cient quartiles of new cars by +20 market, the adjustment is primarily in tional prices offered were particularly percent and -24 percent, respectively. prices. We show how this difference attractive. We show that in reality, In contrast, the same gasoline price can be explained easily by differences the rebates that had been available increase changes the transaction shares in the supply of new and used cars. before the employee discount promo- of the most and least fuel-efficient In summary, the last decade has tion were so large that many customers quartiles of used cars by only +3 per- brought significant changes to the paid higher prices following the intro- cent and -7 percent, respectively. We U.S. auto industry, culminating in the duction of the promotions than they find that changes in gasoline prices restructuring of much of that indus- would have in the weeks just before. also change the relative prices of cars try in the wake of the financial cri- Nevertheless, unit sales increased for in the most fuel-efficient and least sis. These changes have enabled us as these cars, as well as for cars whose fuel-efficient quartiles: fornew cars the researchers to learn about the effect

NBER Reporter • 2010 Number 3 17 of new Internet institutions, infor- Working Paper No. 8668, December Marketing, 2008, Elgar Publishing mation, price transparency, and usage 2001, and Quantitative Marketing Ltd. cost on the U.S. auto market. And Economics, Vol. 1 (1), 2003, pp. 8 See, for example, B. Ratchford, M. 65–92. Lee, and D. Talukdar, “The Impact of 1 See, for example the “Automotive 5 F. Zettelmeyer, F. Scott Morton, the Internet on Information Search for Leasing Guide” https://www.alg.com/ and J. Silva Risso, “How the Internet Automobiles,” Journal of Marketing pdf/ND09_RVR_US.pdf Lowers Prices: Evidence from Matched Research, 40 (May 2003), pp. 193– 2 F. Scott Morton, F. Zettelmeyer, Survey and Auto Transaction 209. and J. Silva Risso “Internet Car Data”, NBER Working Paper No. 9 M. Busse, D. Simester, and F. Retailing”, NBER Working Paper No. 11515, August 2005, and Journal Zettelmeyer, “‘The Best Price You’ll 7961, October 2000, and Journal of of Marketing Research, Vol. 43 (2), Ever Get’: The 2005 Employee Industrial Economics, Vol. 49 (4), 2006, pp. 168–81. Discount Pricing Promotions in the 2001, pp.501–19. 6 M. Busse, J. Silva Risso, and F. U.S. Automobile Industry”, NBER 3 F. Zettelmeyer, F. Scott Morton, and Zettelmeyer, “1000 Cash Back: The Working Paper No. 13140, May 2007, J. Silva Risso, “Cowboys or Cowards: Pass-Through of Auto Manufacturer and Marketing Science, Vol. 29 (2), Why are Internet Car Prices Lower?” Promotions”, NBER Working Paper 2010, pp. 268–90. NBER Working Paper No. 8667, No. 10887, November 2004, and 10 M. Busse, C. Knittel, and F. December 2001. American Economic Review, Vol 96 Zettelmeyer, “Pain at the Pump: The 4 F. Scott Morton, F. Zettelmeyer, and (4), 2006, pp. 1253–70. Differential Effect of Gasoline Prices on J. Silva Risso, “Consumer Information 7 For a review of the literature, see E. New and Used Automobile Markets”, and Discrimination: Does the Internet T. Anderson and D. Simester, “Price NBER Working Paper No. 15590, Affect the Pricing of New Cars to Cues and Customer Price Knowledge,” December 2009. Women and Minorities?” NBER in Handbook of Pricing Research in

NBER Profile: Jens Ludwig

Research Associate Jens Ludwig co- Ludwig is a co-editor of the Journal directs the NBER’s Working Group on of Human Resources and a member of the the Economics of Crime and is a mem- Institute of Medicine/National Academy ber of the Program on Children and the of Sciences Board on Children, Youth and Program on Health Economics. He is also Families. In 2006 Ludwig received the the McCormick Foundation Professor of David N. Kershaw prize for “distinguished Social Service Administration, Law, and contributions to public policy analysis and Public Policy at the University of Chicago. management by age 40.” He lives in the After receiving his doctorate in eco- Hyde Park neighborhood of Chicago with nomics from Duke University, Ludwig was his wife Liz, daughter Annika, and rescue on the public policy faculty at Georgetown mutt Trixi. University. His research focuses on urban problems related to crime, housing, health, and education.

18 NBER Reporter • 2010 Number 3 NBER Profile: Brigitte Madrian

Brigitte Madrian has been an NBER Her current research focuses on Research Associate since 2001 in the household saving and investment behav- Programs on Health Care, Aging, Labor ior. Her work in this area has influenced Studies, and Public Economics. She is the design of employer-sponsored savings the Aetna Professor of Public Policy and plans in the United States and pension Corporate Management at Harvard’s reform legislation both here and abroad. Kennedy School of Government and She also has examined the impact of Director of the Social Science Program health insurance on the job choice and at the Radcliffe Institute for Advanced retirement decisions of employees and Study. on the hiring decisions of firms. Madrian received her Ph.D. in eco- Madrian is co-editor of the Journal nomics from MIT and a BA/MA in eco- of Human Resources and a recipient of nomics from Brigham Young University. the TIAA-CREF Paul A. Samuelson Before joining the Kennedy School Award for Scholarly Research on Lifelong faculty in 2006, Madrian was at the Financial Security in 2002. University of Pennsylvania’s Wharton Brigitte lives in Wellesley, MA with School from 2003–6, the University of her husband David, her two daughters Chicago’s Graduate School of Business Erika (14) and Liesel (10), and their two from 1995–2003, and in Harvard adorable Coton de Tulear puppies, Mr. University’s Economics Department Darcy and Mr. Bingley. from 1993–5.

NBER Profile: Valerie Ramey Valerie Ramey is a Research Associate spending, the source of business cycle fluc- in the NBER’s Program on Economic tuations, wage inequality, and time use. Fluctuations and Growth and the Program She is past macroeconomics co-editor of on Monetary Economics. She is also a pro- the American Economic Review, and is a fessor in the Economics Department at member of the Federal Economic Statistics the University of California, San Diego Advisory Committee. (UCSD) and is Chair of the Institute for Ramey lives in San Diego with her Applied Economics there. husband Garey Ramey (also an economics Ramey received a B.A. in Economics professor at UCSD), and their two chil- and Spanish from the University of dren: Sean, who is studying engineering Arizona in 1981 and a Ph.D. from Stanford at Stanford, and Michelle, who is a high University in 1987. She joined the UCSD school junior. When she is not studying faculty in that year. other people’s use of leisure time, she enjoys Her research has addressed such top- cooking ethnic cuisines, boogie boarding at ics as the behavior of inventories, volatil- La Jolla Shores, dancing, and attending the ity and growth, the effects of government theatre.

NBER Reporter • 2010 Number 3 19 NBER Profile: Florian Zettelmeyer

Florian Zettelmeyer is a Research he was an Associate Professor of Marketing Associate in the NBER’s Program in and chair of the marketing group at the Industrial Organization. He also holds Haas School of Business, University of the J.L. Kellogg Chair in Marketing at California at Berkeley. Northwestern University’s Kellogg School His research deals with how informa- of Management. tion asymmetries between firms and con- Zettelmeyer received a Vordiplom in sumers affect how well consumers do in business engineering from the University markets. Recently he has focused on the of Karlsruhe (Germany), a M.Sc. in eco- automobile industry, investigating both nomics from the University of Warwick informational and environmental issues (UK), and a Ph.D. in marketing from associated with cars and car purchasing MIT. Prior to his appointment at Kellogg, behavior.

Conferences

Twenty-first Annual EASE Conference

The NBER, the Australian National University, the China Center for Economic Research, the Chung-Hua Institution for Economic Research, the Hong Kong University of Science and Technology, the Korea Development Institute, the National University of Singapore, and the Tokyo Center for Economic Research jointly sponsored the NBER’s 21st Annual East Asian Seminar on Economics. It took place on June 25 and 26, 2010 at the Reserve Bank of Australia. Takatoshi Ito, University of Tokyo and NBER, and Andrew K. Rose, University of California, Berkeley and NBER, organized the conference, which focused on “A Pacific Rim Perspective on the Financial Crisis.” These papers were discussed:

• Michael B. Devereux, University of British Columbia and NBER; and James Yetman, Bank for International Settlements, “Financial Contagion and Vulnerability of Asian Financial Markets”

• Warwick J. McKibbin and Andrew Stoeckel, Australian National University, “Modelling the Impact of the Global Financial Crisis on World Trade”

• Jonathan Eaton, Pennsylvania State University and NBER; Samuel S. Kortum, Brent Neiman, and John Romalis, University of Chicago and NBER, “Trade and the Global Recession”

• Bih Jane Liu, Chung-Hua Institution for Economic Research, “Why World Exports Are so Susceptible to the Economic Crisis — The Prevailing ‘Export Overshooting’ Phenomenon Especially in Taiwan”

• Jiandong Ju, Tsinghua University, and Shang-Jin Wei, NBER and Columbia University, “When Are Trade Liberalizations and Capital Flows Substitutes or Complements?”

• Ippei Fujiwara, Nao Sudou, and Yuki Teranishi, Bank of Japan; and Tomoyuki Nakajima, Kyoto University, “Global Liquidity Trap”

20 NBER Reporter • 2010 Number 3 • Yiping Huang, Nian Lin, Tao Kunyu, Wang Bijun, and Wang Xun, CCER, “China’s Monetary Systems and Economic Performance during the Global Crises: From Great Depression to Great Crash”

• Joshua Aizenman, University of California, Santa Cruz and NBER; Yothin Jinjarak, Nanyang Technological University; and Donghyun Park, Asian Development Bank, “International Reserves and Swap Lines: Substitutes or Complements”

• Paul Bloxham, Christopher Kent, and Michael Robson, Reserve Bank of Australia, “Asset Prices, Credit Growth and Monetary Policy: An Australian Case Study”

• Hyun Song Shin, Princeton University, and Kwanho Shin, Korea University, “Macroprudential Policy and Monetary Aggregates”

• Shin-ichi Fukuda, University of Tokyo, “Money Market Integration during the Global Financial Crisis: Evidence from the Interbank Markets in Tokyo and London”

• Yongheng Deng, Jing Wu, and Bernard Yeung, National University of Singapore; and Randall Morck, University of Alberta and NBER, “Monetary and Fiscal Stimuli, Ownership Structure and China’s Housing Market”

Summaries of these papers may be found at: http://www.nber.org/confer/2010/EASE10/summary.html

NBER Conference in Beijing

The twelfth annual NBER-CCER Conference on China and the World Economy took place at the China Center for Economic Research (CCER) in Beijing on June 26–28, 2010. The conference program was jointly arranged by the National Bureau of Economic Research, the CCER at Beijing University, and Tsinghua University. After opening remarks by U.S. orga- nizer Shang-Jin Wei of NBER and Columbia University, Yang Yao of CCER, and Chong-En Bai of Tsinghua University, the following topics were discussed: Macroeconomics in China and the United States • Feng Lu, CCER, “Macroeconomic issues in China”

• Simon Johnson, NBER and MIT, “The Financial Oligarchy in the United States”

Exchange Rates and Economic imbalances

• Kenneth D. West, NBER and University of Wisconsin, Madison, “Exchange rate Economics”

• Yang Yao, “Manufacturing-Finance Comparative Advantage and China’s Trade Surplus”

• Binzheng Wu, Tsinghua University, “Income Inequality, Status Seeking, and Consumption”

• Chong-En Bai, “Declining Share of Household Income in China”

• Shang-Jin Wei, “Global Imbalances and ‘Undervalued’ Currency”

• Binkai Chen, CUFE, “The Cursed Virtue — Infrastructural Investment and Household Consumption in China”

NBER Reporter • 2010 Number 3 21 Entrepreneurship and Capital Allocation

• Erik Hurst, NBER and University of Chicago, “Entrepreneurship”

• Ping He, Tsinghua University, “Capital Allocation and Operation Efficiency in China”

Banking and Consumer Finance

• Peter Tufano, NBER and Harvard University, “Consumer Finance”

• Yan Shen, CCER, “Financial Sector Efficiency and Lending Behavior in China”

Consumer Credit and Regional Governments

• Jonathan D. Levin, NBER and Stanford University, “Recent Research on Consumer Credit Markets”

• Li-An Zhou, GSM, “Incentives of Chinese Local Officials”

Health Economics

• Daniel Kessler, NBER and Stanford University, “Health Economics”

• Ling Li, CCER, “Healthcare in China”

• Amanda Kowalski, NBER and Yale University, “Assessing the Impact of Health Care Mandates” (Presentation based on “The Impact of an Individual Health Insurance Mandate on Hospital and Preventive Care: Evidence from Massachusetts,” Jonathan T. Kolstad and Amanda E. Kowalski, NBER Working Paper No. 16012.)

Labor and Productivity

• Richard B. Freeman, NBER and Harvard University, “Topics in Labor Economics”

• Miaojie Yu, CCER, “Processing Trade, Productivity, and Firm Scope”

Summaries of these papers may be found at: http://www.nber.org/confer/2010/China10/summary.html

State and Local Pensions

An NBER Conference on State and Local Pensions, organized by Research Associate Jeffrey Brown of the University of Illinois and Robert Clark of North Carolina State University, took place on August 19 and 20, 2010. The following papers were discussed:

• Henning Bohn, University of California, Santa Barbara, “Should Public Retirement Plans be Fully Funded?”

• Jean-Pierre Aubry, Alicia Munnell, and Laura Quinby, Boston College, “Public Pension Funding Standards in Practice”

• Sylvester Schieber, Towers Watson, “Political Economy of Public Sector Retirement Plans”

22 NBER Reporter • 2010 Number 3 • Robert Novy-Marx, University of Chicago and NBER, and Joshua D. Rauh, Northwestern University and NBER, “Policy Options for State Pensions Systems and Their Impact on Plan Liabilities”

• George Pennacchi and Mahdi Rastad, University of Illinois, “Portfolio Allocation for Public Pension Funds”

• Eduard Ponds, Netspar, and Clara Severinson and Juan Yermo, OECD, “Funding in Public Sector Pension Plans — International Evidence”

• Leora Friedberg, University of Virginia, “Labor Market Aspects of State and Local Retirement Plans: A Review of Evidence and a Blueprint for Future Research”

• Brigitte C. Madrian, Harvard University and NBER, “Defined Contribution Plans in the Public Sector: Lessons from Behavioral Economics”

• Robert L. Clark and Melinda S. Morrill, North Carolina State University, “Retiree Health Plans In the Public Sector”

Summaries of these papers may be found at: http://www.nber.org/~confer/2010/SLPf10/summary.html

NBER News

John Kendrick Dead at Age 92 John Kendrick, a long-time mem- He also served as Chief Economist of that today is known as total factor pro- ber of the senior research staff at the the U.S. Department of Commence in ductivity. His 1961 NBER volume on NBER, passed away on November 17, the mid-1970s. Productivity Trends in the United States, 2009, at the age of 92. In addition Kendrick made pioneering con- with Maude Pech, remains a classical to his NBER role, he was a member tributions to productivity analysis, contribution to the long-term analysis of the economics faculty at George national income accounting, and the of productivity trends. Washington University — where he estimation of capital stocks. He was had earned his Ph.D. — for many years. one of the first to employ the concept

F. Thomas Juster, 1926–2010

F. Thomas Juster, who served as University in 1956 and was a member Institute for Social Research and was Vice President of the NBER from 1969 of the NBER’s research staff from 1959 the founding director of the Health & to 1972, died on July 21, 2010 at age 83. until 1973. In 1973 Juster joined the Retirement Study. Juster was an active Born in Hollis, Long Island, New York, faculty of the University of Michigan, and effective advocate for the impor- he received his Ph.D. from Columbia where he served as director of the tance of social science research.

Arnold Zellner

Arnold Zellner, who was elected the University of California and taught in 1966. Zellner founded the NBER- to the NBER’s Board of Directors in at the University of Washington and NSF Seminar on Bayesian Inference 1980 and became a Director Emeritus at the University of Wisconsin before in Econometrics and Statistics, and in 2004, passed away on August 11, joining the faculty of the University of directed this seminar for more than 2010. Zellner received his Ph.D. from Chicago’s Graduate School of Business twenty years.

NBER Reporter • 2010 Number 3 23 Thirty-first NBER Summer Institute Held in 2010

In the summer of 2010, the NBER search programs and many of the work- Vice-Chairman of the Federal Reserve held its thirty-first annual Summer ing groups, and more than 400 separate Board of Governors. A complete agenda Institute. Over 2000 economists from research presentations. The Summer and many of the papers presented at the more than 300 different colleges, uni- Institute included a set of Econometric various sessions are available on the versities, and other economic research Methodology Lectures on financial NBER’s web site at http://www.nber. organizations throughout the world econometrics, as well as the second org/confer/2010/SI2010/SI2010. attended. annual Martin Feldstein Lecture, deliv- html. There were 47 distinct meetings, ered by Dr. Roger Ferguson, President representing all of the 19 NBER re- and CEO of TIAA-CREF and former

Program and Working Group Meetings

Japan Project Meets

The NBER together with the Center on the Japanese Economy and Business, The Center for Advanced Research in Finance, and the Australia-Japan Research Centre held a project meeting on the Japanese economy in Tokyo on June 25 and 26, 2010. The organizers were: Jennifer Corbett, Australia-Japan Research Centre; Charles Horioka, NBER and Osaka University; Anil K Kashyap, NBER and the Graduate School of Business, University of Chicago; and David Weinstein, NBER and Columbia University. The following papers were discussed:

• Mary Amiti, Federal Reserve Bank of New York, and David Weinstein, “Exports and Financial Shocks”

• Kazuo Ogawa, Osaka University; Elmer Sterken, University of Groningen; and Ichiro Tokutsu, Kobe University, “Financial Distress and Industry Structure: An Inter-Industry Approach to the ‘Lost Decade’ in Japan”

• Jennifer Corbett and Kazuki Onji, Australian National University, and David Vera, Kent State University, “Capital Injection, Restructuring Targets, and Personnel Management: The Case of Japanese Regional Banks”

• Yasushi Hamao and Pedro Matos, University of Southern California, and Kenji Kutsuna, Kobe University, “Foreign Investor Activism in Japan: The First Ten Years”

• Chad Steinberg, International Monetary Fund, “Shakedown: Economic Geography Meets the Kobe Earthquake”

• Takayuki Mizuno, Makoto Nirei, and Tsutomu Watanabe, Hitotsubashi University, “Closely Competing Firms and Price Adjustment: Evidence from an Online Marketplace”

• Yuki Hashimoto, University of Tokyo, and Ayako Kondo, Osaka University, “Long-Term Effects of Labor Market Conditions on Family Formation for Japanese Youths”

• Ayako Suzuki, Waseda University, “An Empirical Analysis of Entrant and Incumbent Bidding in Electric Power Procurement Auctions”

Summaries of these papers may be found at: http://www.nber.org/confer/2010/JPMs10/summary.html

24 NBER Reporter • 2010 Number 3 Economic Fluctuations and Growth Research Meeting

The NBER’s Program on Economic Fluctuations and Growth met in Cambridge on July 17. NBER Research Associates Erik Hurst, University of Chicago, and Matthew D. Shapiro, University of Michigan, organized the meeting. These papers were discussed:

• Christopher J. Nekarda, Federal Reserve Board, and Valerie A. Ramey, University of California, San Diego and NBER, “The Cyclical Behavior of the Price-Cost Markup”

• Olivier Coibion, College of William and Mary; and Yuriy Gorodnichenko and Johannes F. Wieland, University of California, Berkeley and NBER; “The Optimal Inflation Rate in New Keynesian Models” (NBER Working Paper No. 16093)

• Jeremy C. Stein, Harvard University and NBER, “Monetary Policy as Financial-Stability Regulation”

• Raj Chetty, Harvard University and NBER, “Bounds on Elasticities with Optimization Frictions: A Synthesis of Micro and Macro Evidence on Labor Supply”

• Jonathan Eaton, Pennsylvania State University and NBER, and Samuel S. Kortum, Brent Neiman, and John Romalis, University of Chicago and NBER, “Trade and the Global Recession”

• Charles I. Jones and Peter J. Klenow, Stanford University and NBER, “Beyond GDP? Welfare across Countries and Time”

Summaries of these papers may be found at: http://www.nber.org/confer/2010/EFGs10/summary.html

The Economics of Household Saving

NBER Research Associate Erik Hurst of the University of Chicago and NBER President James Poterba of MIT, who co- direct an NBER project on “The Economics of Household Saving”, organized a meeting of that project on July 24, 2010. The following papers were discussed:

• Dean Karlan, Yale University and NBER; Margaret McConnell, Harvard School of Public Health; Sendhil Mullainathan, Harvard University and NBER; and Jonathan Zinman, Dartmouth College, “Getting to the Top of Mind: How Reminders Increase Saving”

• Mark A. Aguiar and Mark Bils, University of Rochester and NBER, “Has Consumption-Inequality Mirrored Income Inequality?”

• Garry Barrett, University of Sydney; Thomas Crossley, Cambridge University, and Kevin Milligan, University of British Columbia and NBER, “Micro and Macro Based Saving Rates: What Explains the Differences?”

• John Sabelhaus, University of Maryland, and Jae Song, Social Security Administration, “The Great Moderation in Micro Labor Earnings”

• Robert Townsend, MIT and NBER, “Saving and Risk in Developing Countries: Theory and Evidence”

The authors of each of these papers have prepared short research summaries that describe their findings and the broader impli- cations of their work. These summaries may be found at: http://www.nber.org/confer_papers/summarize?conf_id=SI0SAV

NBER Reporter • 2010 Number 3 25 Bureau Books

The following three volumes may be ordered directly from the University of Chicago Press Distribution Center, at Telephone: 1-800-621-2736 Email: [email protected]

For more information on ordering and electronic distribution, see http://www.press.uchicago.edu/Misc/Chicago/infopage.html

The Economics of Crime: Lessons For and From Latin America

The Economics of Crime: Lessons addresses a variety of topics, including Economy, and the Joseph C. Wilson for and from Latin America, edited by the impact of mandatory arrest laws, Professor of Business Administration Rafael Di Tella, Sebastian Edwards, education in prisons, and the relation- at Harvard Business School. Edwards and Ernesto Schargrodsky, is available ship between poverty and crime. It also is a Research Associate in the NBER’s from the University of Chicago Press presents research from outside Latin Program on International Trade and for $110.00. America, illustrating the broad range Investment and the Henry Ford II Crime rates in Latin America are of approaches that have been fruitful Professor of International Economics among the highest in the world, and in in studying crime in developed nations. at the Anderson Graduate School of several countries they have steadily risen The Economics of Crime should interest Management, University of California, over the past two decades. Despite this researchers, policymakers, and students Los Angeles. Schargrodsky is a profes- situation, there has been little system- of both crime and Latin American eco- sor and dean of the business school at atic study of crime in the region or of nomic policy. Universidad Torcuato Di Tella. the effectiveness of policies designed Di Tella is a Research Associate to tackle it. The Economics of Crime in the NBER’s Program on Political

Targeting Investments in Children: Fighting Poverty When Resources Are Limited

Targeting Investments in Children: effects, resulting in poor educational, interventions targeted at children of Fighting Poverty When Resources Are labor market, and physical and mental different ages, and they study a range Limited, edited by Phillip B. Levine health outcomes for adults. Numerous of programs, including child care, after- and David J. Zimmerman, is available programs are designed to alleviate or school care, and drug prevention. from the University of Chicago Press even eliminate poverty; as they com- Levine is an NBER Research for $99.00. pete for scarce resources, it is important Associate in the Programs on Labor About 17 percent of American chil- to analyze their impact. The papers in Studies and Children and the Class dren under the age of eighteen live at a this NBER Conference Report evaluate of 1919 Professor of Economics at level that meets the government’s defi- these programs using a common metric: Wellesley College. Zimmerman is nition of poverty, and the proportion their impact on earnings in adulthood. a Research Associate in the NBER’s is even greater within minority groups. The volume’s contributors explore a Program on Education and a Professor Childhood poverty can have lifelong variety of issues, such as the effect of of Economics at Williams College.

26 NBER Reporter • 2010 Number 3 The Economic Consequences of Demographic Change in East Asia

The Economic Consequences of years, leading to aging societies and and specific policy concerns in individ- Demographic Change in East Asia, economies that suffer from a declin- ual countries. The volume provides an Volume 19 in the NBER’s series on the ing working population along with fis- overview of economic growth in East East Asia Seminar on Economics, is cal deficits linked to increased govern- Asia as well as more specific studies on available from the University of Chicago ment spending. East Asia exemplifies Japan, Korea, China, and Hong Kong. Press for $105.00. The volume’s editors, these trends, and this volume offers an Offering important insights into the who also co-chaired the conference, are in-depth look at how long-term demo- causes and consequences of this tran- NBER Research Associates Takatoshi graphic transitions have taken shape sition, the book will benefit students, Ito, University of Tokyo, and Andrew there and how they have affected the researchers, and policymakers focused K. Rose, University of California, economy in the region. on East Asia, as well as anyone con- Berkeley. This seminar assembled a group of cerned with similar trends elsewhere in Almost all industrialized countries experts to explore such topics as com- the world. have experienced dramatic decreases in parative demographic change, popula- fertility and mortality rates in recent tion aging, the rising cost of healthcare,

The following volume may be ordered directly from the University of Chicago Press Journals Division. To order by telephone, call Monday through Friday, 8 am to 5 pm Central Time, (773) 753-3347; or toll-free in the U.S. and Canada, (877) 705-1878. To order by mail, the address is: University of Chicago Distribution Center, 11030 South Langley Avenue, Chicago, IL 60628, (773)702-7000

Tax Policy and the Economy, Volume 24

Tax Policy and the Economy, Volume Volume 24 includes studies of the of recent reforms in the Earned Income 24, edited by Jeffrey R. Brown, is now relative efficacy of tax cuts versus spend- Tax Credit. available from the University of Chicago ing increases as a form of economic Brown is a Research Associate in the Press Journals Division for $60.00 stimulus; a targeted analysis of the NBER’s Programs on Public Economics (clothbound) or $20.00 in paperback. Low Income Housing Tax Credit; two and Aging and a Professor in the Finance This annual series of volumes presents papers that examine different aspects of Department at the University of Illinois current academic research findings on policies designed to provide fiscal stim- at Urbana-Champaign. taxation and government spending. ulus; and an examination of the effects

NBER Reporter • 2010 Number 3 27 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