NBER Reporter NATIONAL BUREAU OF ECONOMIC RESEARCH

Reporter OnLine at: www.nber.org/reporter spring 2006

Program Report

IN THIS ISSUE

Program Report Health Care  Health Care

Research Summaries Consumption and Saving 8 Child Health and Human Capital  Alan M. Garber* Estate Taxation 4 Rules for Monetary Policy 7

NBER Profiles 22 Conferences 24 Bureau News 24 The NBER’s Program on Health Care holds two program meetings Bureau Books 36 annually, as well as NBER Summer Institute sessions, an annual “Frontiers Current Working Papers 37 in Health Policy Research Conference,” and occasional theme meetings. Program members conduct research on a diverse range of issues in the economics of health, the delivery and financing of health care, and the interactions of health and health care with other areas of economic activ- ity. Because Health Care is a large and active program, as reflected in the Working Papers issued by program members and in their presentations and publications, this report describes only a fraction of its work. Private Health Insurance

Most Americans pay for health care with private health insurance obtained through employers. The loss of employment-based insurance fre- NBER Service Brings quently leads to either enrollment in government programs like Medicaid You New Data for Free or the loss of health insurance altogether. Michael Chernew, David M. Cutler, and Patricia S. Keenan examine why the share of Americans with- A new, free NBER email ser- out health insurance rose over the 1990s, despite the relative prosperity vice gives you daily email links to of the decade. In one paper, they relate changes in health insurance cost all U.S. government data releases, growth to changes in insurance coverage rates across metropolitan areas, including unemployment, trade, accounting for a broad set of additional factors that may affect changes in interest rates, GDP, etc. We keep 1 track of your preferences and email coverage. They find that rising premiums accounted for over half of the you the requested links when they decline in health insurance coverage during the 1990s. A $1,000 increase are released. To sign up for any or in premiums is associated with a 2.6 percentage point decline in insurance all of the government releases, visit coverage rates. They also project that rising health insurance costs will cause www.nber.org/releases and register your choices. IT’S — FREE!! * Garber directs the NBER’s Program on Health Care and is a professor at Stanford University. He is also a practicing physician, holding both an M.D. and a Ph.D. in Economics. NBER Reporter Spring 2006  the number of uninsured to increase by 2 to 6 million people by 2010. In another paper, they NBER Reporter report that the availability of uncompensated care leads to greater losses of insurance coverage when premiums rise.2 Many of the uninsured say that they lack health insurance because it is unaffordable. M. The National Bureau of Economic Research is a private, nonprofit research orga- Kate Bundorf and Mark V. Pauly develop “nor- nization founded in 1920 and devoted to objective quantitative analysis of the mative” and “behavioral” definitions of afford- American economy. Its officers and board of directors are: ability, examining whether health insurance is President and Chief Executive Officer — Martin Feldstein affordable to the currently uninsured. Analyzing Vice President for Administration and Budget — Susan Colligan data from the Medical Expenditure Panel Survey, Controller — Kelly Horak they report that when a normative definition of BOARD OF DIRECTORS affordability is used for family incomes above Chairman — Elizabeth E. Bailey the poverty level, health insurance was afford- Vice Chairman — John S. Clarkeson able to 82 percent of the uninsured. Increasing Treasurer — Robert Mednick the threshold to 2 and 3 times the poverty level, DIRECTORS AT LARGE the proportions of the uninsured classified as Peter Aldrich Jessica P. Einhorn John Lipsky able to afford coverage were 55 percent and 34 Elizabeth E. Bailey Martin Feldstein Laurence H. Meyer percent, respectively. These researchers also find John Herron Biggs Jacob A. Frenkel Michael H. Moskow Andrew Brimmer Judith M. Gueron Alicia Munnell that, with a “behavioral” definition of afford- John S. Clarkeson Robert S. Hamada Rudolph A. Oswald ability, which is defined by the health insurance Don R. Conlan George Hatsopoulos Robert T. Parry purchase behavior of individuals with similar Kathleen B. Cooper Karen N. Horn Richard N. Rosett George Eads Judy Lewent Marina v. N. Whitman economic circumstances, about half of the unin- Martin B. Zimmerman sured could purchase health insurance. Thus, these economists believe that affordability is not DIRECTORS BY UNIVERSITY APPOINTMENT the sole barrier to health insurance coverage.3 George Akerlof, California, Berkeley Joel Mokyr, Northwestern Bundorf and Jay Bhattacharya also have Jagdish W. Bhagwati, Columbia Andrew Postlewaite, Pennsylvania Michael J. Brennan, California, Los Angeles Craig Swan, Minnesota examined whether there are offsetting wage Glen G. Cain, Wisconsin Uwe Reinhardt, Princeton decreases for workers with large expected health Ray C. Fair, Yale Nathan Rosenberg, Stanford care costs by studying the wage patterns for Franklin Fisher, MIT David B. Yoffie, Harvard 4 Saul H. Hymans, Michigan Arnold Zellner, Chicago obese workers. Annual medical expenditures Marjorie B. McElroy, Duke are $732 higher on average for obese individuals DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS than for normal weight individuals. In a paper on this subject, they report that obese workers Richard B. Berner, National Association for Business Economics Gail Fosler, The Conference Board with employer-sponsored health insurance pay Richard C. Green, American Finance Association for their higher expected medical expenditures Dr. Arthur Kennickell, American Statistical Association through lower cash wages. This conclusion is Thea Lee, American Federation of Labor and Congress of Industrial Organizations strengthened by their finding that obese workers Robert Mednick, American Institute of Certified Public Accountants with insurance coverage through an alternative Angelo Melino, Canadian Economics Association employer (for example, a spouse) do not expe- Jeffrey M. Perloff, American Agricultural Economics Association John J. Siegfried, American Economic Association rience similar wage offsets. Nor are there wage William W. Lewis, Committee for Economic Development offsets for other types of fringe benefits whose Gavin Wright, Economic History Association cost to the employer is less likely to be affected The NBER depends on funding from individuals, corporations, and private foun- by obesity. dations to maintain its independence and its flexibility in choosing its research activities. Inquiries concerning contributions may be addressed to Martin Feldstein, President & CEO, NBER 1050 Massachusetts Avenue, Cambridge, Medicare and Medicaid MA 02138-5398. All contributions to the NBER are tax deductible. Many Health Care Program members — in- 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- cluding Bhattacharya, Amitabh Chandra, David duced with appropriate attribution of source. Please provide the NBER’s Public M. Cutler, Mark Duggan, Amy Finkelstein, Information Department with copies of anything reproduced. Victor Fuchs, Dana Goldman, Frank Lichtenberg, Requests for subscriptions, changes of address, and cancellations should be sent Thomas MaCurdy, Mark McClellan, Jonathan to Reporter, National Bureau of Economic Research, Inc., 1050 Massachusetts Skinner, and I — have studied aspects of the Avenue, Cambridge, MA 02138-5398. Please include the current mailing label.

2 NBER Reporter Spring 2006 Medicare and Medicaid programs, such months of life. Using various analytical drove up costs, and vice versa. as the causes of growth in program expen- methods, they find that a large compo- Laurence C. Baker has developed ditures and the role of disability in future nent of Medicare expenditures — $26 bil- important evidence on the ways that expenditures. lion in 1996 dollars, or nearly 20 percent financial incentives, particularly those Amy Finkelstein has looked at the of total Medicare expenditures — appears associated with managed care organiza- introduction of Medicare to learn how to provide no survival benefit, nor is it tions, can influence technology diffusion. the introduction of universal insurance likely that this extra spending improves For example, his work on the diffusion of affects health spending and technology the quality of life. While secular trends mid-level neonatal intensive care facili- adoption.5 Medicare, she hypothesizes, in health care technology have delivered ties,11 like his earlier work examining should have had a greater effect in areas of large health benefits, variation in health other technologies such as magnetic res- the country in which relatively few of the care intensity at a point in time suggests onance imaging and mammography,12 elderly had health insurance than in areas that more is not better. shows that tightening financial incentives in which many of the elderly were insured limits the diffusion of new technologies. prior to the introduction of the pro- Pharmaceutical Markets, This can have important implications for gram. She examines an annual hospital- Innovation, And Technology the subsequent utilization of health ser- level dataset from 1948–75 for six hospi- Diffusion vices and spending. Work by Baker and tal outcomes: total expenditures, payroll colleagues on the effects of the diffusion expenditures, employment, beds, admis- Health Care Program members have of several technologies in the late 1990s sions, and patient-days. She finds that conducted a wide array of research address- demonstrated their importance in influ- the effects were indeed greater in areas in ing how technological innovation in medi- encing utilization and spending.13 Baker which health insurance was less common; cine affects both health expenditures and also has examined the effects of technol- prior to 1965, hospital admissions were health outcomes. They have also investi- ogy diffusion on well-being. Baker and C. growing more slowly in the low-insurance gated the factors that promote or impede it S. Phibbs show that, counter to expecta- areas than in the high-insurance areas, technological innovation in medical care. tion, slowing the diffusion of neonatal but after 1965 this pattern reversed, with In a series of studies, Cutler and his intensive care facilities can improve health admissions growing much more quickly collaborators have measured the benefits of outcomes, apparently by helping to direct in those areas most affected by Medicare’s specific medical innovations. For the most high-risk births to hospitals that are most introduction. Similar patterns are evident part, he has concluded, the benefits have successful at caring for such low birth in the other hospital outcome variables, been large and underappreciated. Much of weight babies. including expenditures. These results the work is summarized in his book, Your Government health insurance pro- suggest that the overall spread of health Money or Your Life.8 Extending work of grams directly affect diffusion and the insurance explains at least 40 percent of Cutler and Mark McClellan on the contri- incentives to innovate through their influ- the dramatic increase in health spending bution of medical innovations to changes ence on demand. Mark Duggan and Fiona in the United States between 1950 and in heart attack mortality among Medicare M. Scott Morton examine the effects on 1990. beneficiaries,9 Skinner, Douglas Staiger, drug pricing and innovation of what was, In a series of articles, Jon Skinner and Elliott Fisher examine Medicare at least before 2006, the largest govern- and colleagues have examined variation costs and outcomes for Acute Myocardial ment drug program.14 In 2003, Medicaid in expenditures for the care of Medicare Infarction (AMI) in the Medicare popu- provided prescription drug coverage to beneficiaries and their implications for lation during 1986–2002.10 They find more than 50 million people nationwide. the efficiency of Medicare. For example, that the gains in mortality that Cutler To determine the price that it will pay the range of variation in resources used and McClellan observed from 1986–98 for each drug, Medicaid uses the aver- for end-of-life care for Medicare beneficia- did not continue subsequently, and that age private sector price. When Medicaid ries in the United States is striking, even expenditures, after a brief pause, contin- is a large part of the demand for a drug, among top-rated hospitals.6 Skinner and ued to increase. In cross-sectional analy- this creates an incentive for its maker colleagues also use geographical variations ses, they find that regions experiencing to increase the prices it charges other in health care spending to measure the the greatest drop in mortality following health care consumers. Using drug utili- incremental value of health care intensity AMI were not those with the largest gains zation and expenditure data for the top among the elderly Medicare population.7 in expenditures. Patients living in regions 200 drugs in 1997 and in 2002, they To correct for the reverse causation prob- that had invested early in low-cost and investigate the relationship between the lem — that residents of “sicker” areas tend highly effective care, such as beta blockers, Medicaid market share (MMS) and the to require more health care — they use a experienced the largest declines in mor- average price of a prescription. Their esti- set of instruments characterizing health tality with no adverse impact on expen- mates imply that a 10 percentage-point care intensity either among hip fracture ditures. The factors yielding the greatest increase in the MMS is associated with patients or among patients in their last six benefits to health were not the factors that a 7 to 10 percent increase in the aver-

NBER Reporter Spring 2006 3 age price of a prescription. In addition, vator receives adequate profits, while their sible for racial disparities. How much can Medicaid rules increase a firm’s incentive patients also have adequate access to inno- be explained by differences in the medical to introduce new versions of a drug in vative products and technologies.18 They care that white and non-white patients order to raise price. These researchers find show that in health care markets char- receive for the same disease? that firms producing newer drugs with acterized by uncertainty and insurance, In a series of papers, Chandra, Skinner, larger sales to Medicaid are more likely to society may be able to ensure efficient Staiger, and their colleagues have demon- introduce new versions. Taken together, rewards for inventors and the efficient dis- strated the pivotal role that geography their findings suggest that government semination of inventions. Health insur- plays in racial disparities in health care. procurement rules can alter equilibrium ance resembles a two-part pricing con- Because their work is based on analyses of price and product proliferation in the pri- tract in which a group of consumers pay Medicare claims files, they do not directly vate sector. an up-front fee ex ante in exchange for assess the role of variation in insurance Frank Lichtenberg also has studied a relatively low fixed unit price ex post. coverage. They show that regions dem- the value of technological innovation This can allow innovators to extract suffi- onstrating a high level of racial dispar- in medical care. In a recent paper,15 he cient profits — from the ex ante payment ity in the use of one procedure adminis- has sought to understand the role of the — but still sell the good at marginal cost tered at the end of life are not especially introduction of new drugs on life expec- ex post. As a result, their research shows likely to exhibit similar disparities in the tancy changes over time and between that complete, efficient, and competitive use of unrelated procedures. Unusually nations. He estimates that about 40 per- health insurance markets lead to efficient large racial disparities in surgery are often cent of the two years added to the average innovation and utilization, even when the result of high white rates rather than life span between 1986 and 2000 can be moral hazard exists. Conversely, incom- low black rates. Differences in end-of-life traced to the introduction of new drugs. plete insurance markets lead to ineffi- care are driven more by residence than by He suggests that it may take several years ciently low levels of innovation. Second, race.20 for a new drug to be diffused to more con- an optimally designed public health A detailed picture of the importance sumers and have its full impact on survival insurance system can solve the innovation of geography emerges in a study of mor- rates, and that spending on new drugs problem by charging ex ante premiums tality after heart attack.21 In an analysis may be a cost-effective way to increase equal to consumer surplus, and ex post of fee-for-service Medicare patients hos- longevity. co-payments at or below marginal cost. pitalized for heart attack during 1997– Several groups of Health Care When these quantities are unknown, soci- 2001(with a sample greater than one mil- Program researchers have studied the ety almost always can improve static and lion), the researchers classified more than incentives for innovation. Tomas J. dynamic welfare by covering the unin- 4,000 hospitals into approximate deciles Philipson and Anupam B. Jena have esti- sured with contracts that mimic observed depending on the extent to which the mated the welfare gain resulting from the private insurance contracts. hospital served the African-American introduction of anti-HIV medications In other work, these researchers devel- population. Decile 1 (12.5 percent of and the profits that companies produc- op an economic framework to discuss the AMI patients) included hospitals without ing the drugs have earned from them.16 social insurance aspects of several innova- any African-American AMI admissions They report that profits only represent 5 tion policies including patents, research during 1997–2001. Decile 10 (10 per- percent of the consumer surplus attribut- subsidies, and pre-commitments to buy.19 cent of AMI patients) included hospitals able to the drugs, and suggest that policies They show that patents or rewards have an with the highest fraction of black AMI that seek to limit consumption based on advantage over research subsidies when a patients (33.6 percent). The main out- value to individual patients may provide new invention replaces an existing good come measures were 90-day and 30-day inadequate rewards to innovating firms. at lower cost. Research subsidies have mortality following AMI. Charles I. Jones, Paul M. Romer, and I also an advantage when inventions spawn an Patients admitted to hospitals dispro- examine the incentives to innovate, draw- entirely new product. portionately serving African-Americans ing a distinction between static efficien- experienced no greater level of morbidi- cy — when quantities correspond to the Racial Disparities In Health ties or severity of the infarction. Yet hos- competitive equilibrium, when the drug Care Delivery pitals in Decile 10 experienced a risk- already exists — and dynamic efficiency.17 adjusted 90-day mortality rate of 23.7 The latter condition requires that profits The federal government, and a grow- percent (the 95 percent confidence inter- equal surplus. We find that the rewards to ing research literature, has sought to val is 23.2–24.2) compared to 20.1 per- innovation can be too small or too large in understand the causes of differences in cent (the 95 percent confidence inter- the presence of health insurance, depend- health outcomes among different racial val is 19.7–20.4) in Decile 1 hospitals. ing on the shape of the demand curve. and ethnic groups within the United Differences in outcomes between hospi- Neeraj Sood and Darius Lakdawalla also States. Much of their effort is directed tals were not explained by income, hos- address the ways to ensure that the inno- toward finding the mechanisms respon- pital ownership status, hospital volume,

 NBER Reporter Spring 2006 Census region, urban status, or hospital likely to be uninsured than male work- are most likely to offer relatively unprofit- surgical treatment intensity. Thus, risk- ers. The same is not true for blacks and able services; nonprofits fall in the mid- adjusted mortality following AMI is sig- Hispanics: their rates of coverage from dle. For-profits are also more responsive nificantly higher in U.S. hospitals that dis- other sources are also lower than rates for to changes in service profitability than the proportionately serve African-Americans. whites, so that they are significantly more other two types. Policies that try to equalize racial dif- likely to be uninsured even after adjust- These different approaches to service ferences in outcomes within areas may ing for observables. Examining the period mixes should affect the financial perfor- have little effect on disparities overall, from 1980 to 2000, she finds that the mance of the different ownership types. while a reduction in overall mortality at adjusted gap in own-employer coverage Yu-Chu Shen and colleagues apply meta- these hospitals could reduce dramatically for women has been relatively flat over analytical methods to synthesize studies black-white disparities in health. Other this period and is about half as large as the that investigate the effect of ownership on studies by the same group draw similar male/female wage gap, so that measuring hospital financial performance, focusing conclusions.22 inequality in wages plus health insurance on two questions: 1) what is the magni- Peter W. Groeneveld and I, in papers would result in a smaller estimate of male/ tude of the ownership effect on financial with Paul A. Heidenreich and Sara Laufer, female compensation inequality than performance? 2) how do differences in have observed similar phenomena for measuring wages alone. The same is gener- analytic methods and other study fea- other cardiac treatments. For example, ally true for blacks, although their health tures affect the estimates of ownership in an analysis of Medicare claims files insurance gap is much closer in magni- effect?27 In a systematic review of 41 for the years 1990–2000, we found that tude to their wage gap. For Hispanics, the studies, they find that the diverse results black patients had higher mortality fol- health insurance gap is nearly identical in the hospital ownership literature can be lowing cardiac arrest than white patients, to the wage gap and both are increasing explained largely by differences in under- and that the difference in outcomes was over time. Thus, Levy finds no evidence lying theoretical frameworks, assumptions explained in part by the lower rate at that adding health insurance to estimates about the functional form of the depen- which black patients received implant- of labor market compensation inequality dent variables, and model specifications. able cardioverter-defibrillators.23 We would widen disparities for women versus Weaker methods and functional forms then examined whether similar dispari- men, blacks versus whites, or Hispanics tend to predict larger differences in finan- ties characterized the use of the devices versus whites. cial performance between not-for-profits for ventricular arrhythmias. In the early and for-profits. The combined estimates 1990s, black patients with ventricular Industrial Organization of across studies suggest little difference in arrhythmias were about half as likely as Medical Care cost among all three types of hospital whites to receive the devices, but by 1999, ownership, and that for-profit hospitals they received the devices at about two- NBER researchers have played an generate more revenue and greater prof- thirds the rate of whites. Declining geo- important part in the application of indus- its than not-for-profit hospitals, although graphic variation in device implantation trial organization approaches to medical the difference is only of modest economic explained about 20 percent of the reduc- care issues. There continues to be a great significance. There is little difference in tion in racial disparity.24 deal of interest in the behavior of for-prof- revenue or profits between government Hypothesizing that labor market it and not-for-profit institutions, which and not-for-profit hospitals. inequality may be reflected in differ- often compete with one another in, for To study the interaction between ences in non-wage compensation, Helen example, markets for hospital services. It competing not-for-profit and for-prof- Levy examines white-nonwhite and male- is widely believed that for-profit and not- it hospitals, Duggan asks whether the female differences in health insurance for-profit hospitals offer different service behavior of private not-for-profit hospi- coverage among full-time workers.25 She mixes, in part because for-profits more tals is systematically related to the share finds that two-thirds of the gap in insur- aggressively seek well-insured patients. of nearby hospitals organized as for-prof- ance coverage for blacks or Hispanics is In an analysis of American Hospital it firms.28 His findings show that the explained by differences in observable Association data for every U.S. urban, not-for-profit hospitals in areas with pre- characteristics (primarily education and acute care hospital (1988–2000), Jill R. dominantly for-profit hospitals are sig- occupation). The gap for women is not Horwitz asks how service profitability nificantly more responsive to a change in explained by controlling for observables. affects hospital specialization, comparing financial incentives than their counter- However, for women, coverage from other government, non-government nonprofit, parts in areas served by few for-profit pro- sources — primarily employer-sponsored and for-profit hospitals.26 She categorizes viders. Differences in financial constraints coverage as a dependent rather than as a more than 30 services as relatively profit- and other observable factors correlated policyholder — more than makes up for able, unprofitable, or variable. For-profits with for-profit hospital penetration do their lower rates of own-employer cover- are most likely to offer relatively profitable not explain the heterogeneous response. age. Consequently female workers are less medical services; government hospitals The findings suggest that not-for-profit

NBER Reporter Spring 2006 5 hospitals mimic the behavior of private porting the hypothesis that firms merge 11303, May 2005. for-profit providers when they actively in part to avoid having excess capacity 5 A. Finkelstein, “The Aggregate Effects compete with them. once their products lose patent protection of Health Insurance: Evidence from the Modifying physician behavior, and and/or their late-stage drugs fail in clini- Introduction of Medicare,” NBER Working understanding how physicians respond to cal trials. This analysis of post-merger per- Paper No. 11619, September 2005. incentives, is an increasing focus of policy formance strongly confirms the impor- 6 J. E. Wennberg, E. Fisher, T. Stukel, as interest in pay-for-performance and tance of controlling for pre-merger firm J. Skinner, S. Sharp, and K. Bronner, other programs to improve the quality of characteristics. Once the researchers con- “Variations in End of Life Care Health health care has grown. Physician incen- trol for a firm’s propensity to merge, they Care Among Highly Respected Hospitals in tives are controversial because they may find that mergers have very little effect the United States,” British Medical Journal induce doctors to make treatment deci- on a firm’s subsequent growth in market 328, March 13, 2004. sions that are not in the best interests of value, employees, R and D expenditures, 7 J. Skinner, E. Fisher, and J. E. their patients. Martin Gaynor, James B. and sales among large biotech/pharma- Wennberg, “The Efficiency of Medicare” in Rebitzer, and Lowell J. Taylor examine ceutical firms. For small firms, however, Analyses in the Economics of Aging, D. the effect of physician incentives in an mergers appear to be an effective growth Wise ed., Chicago: University of Chicago HMO network.29 They set out a theo- strategy, presumably because mergers pro- Press, 2005. retical framework for assessing the degree vide a source of funds for financially dis- 8 D. M. Cutler, Your Money or Your to which incentive contracts do in fact tressed firms. Life: Strong Medicine for America’s induce physicians to deviate from a stan- Health Care System, New York: Oxford dard guided only by patient interests and Other Health Care Program University Press, 1994. professional medical judgment. They ana- Research 9 D. M. Cutler and M. McClellan, “Is lyze details of an HMO’s incentive con- Technological Change in Medicine Worth tracts, along with internal expenditure Health Care Program research spans It?” Health Affairs (Millwood), 2001, records, in their empirical evaluation of a wide range of other areas. For exam- 20(5), pp. 11–29. the model. They estimate that the HMO’s ple, Christopher Ruhm has studied the 10 J.Skinner, D. Staiger, and E. Fisher, “Is incentive contract provides a typical phy- relationship between the macroecono- Technological Change in Medicine Always sician an increase, at the margin, of ten my and health, showing that mortality Worth It? The Case of Acute Myocardial cents in income for each dollar reduction increases during economic downturns.31 Infarction,” Health Affairs web exclusive, in medical utilization expenditures. The And, David Meltzer has examined how February 7, 2006. average response is a 5 percent reduction a change in the organization of hospi- 11 L.C. Baker and C.S. Phibbs, “Managed in medical expenditures. These researchers tal care — placing hospitalized patients in Care, Technology Adoption, and Health also find suggestive evidence that financial the hands of “hospitalist” physicians who Care: The Adoption of Neonatal Intensive incentives linked to commonly used qual- devote nearly all of their work time to Care,” Rand Journal of Economics 33:3, ity measures may stimulate an improve- inpatient care — can both lower costs and pp. 524–48, Autumn 2002. ment in measured quality. improve hospital outcomes. 12 L. C. Baker, “Managed Care and The economics of the pharmaceuti- Technology Adoption in Health Care: cal industry remains an important topic Evidence from Magnetic Resonance both because the industry is an impor- 1 M. Chernew, D. M. Cutler, and P. S. Imaging,” Journal of Health Economics tant source of major advances in care and Keenan, “Increasing Health Insurance Costs 20:3, May 2001, pp. 395–421. because the industry has undergone dra- and the Decline in Insurance Coverage”, 13 L.C. Baker, H. Birnbaum, J. Geppert, matic change in the past decade. Perhaps Health Services Research 40(4): pp.1021– D. Mishol, and E. Moyneur, “Relationship the most visible manifestation of change 39, 2005. between technology availability and health has been industry consolidation; the suc- 2 M. Chernew, D. M. Cutler, and P. S. care spending,” Health Affairs web exclu- cess of mergers is a matter of interest not Keenan, “Charity Care, Risk Pooling, and sive, W3, November 5, 2003, pp.537–51. only to investors but also to anyone con- the Decline in Private Health Insurance,” 14 M. Duggan and F. M. Scott Morton, cerned with pharmaceutical innovation. American Economic Review, Papers and “The Distortionary Effects of Government Patricia M. Danzon, Andrew Epstein, and Proceedings 95(2): pp.209–13, 2005. Procurement: Evidence From Medicaid Sean Nicholson examine the determi- 3 M. K. Bundorf and M. V. Pauly, Prescription Drug Purchasing,” NBER nants of merger and acquisition activity “Is Health Insurance Affordable for the Working Paper No.10930, November 2004, in the biotech and pharmaceutical indus- Uninsured?” NBER Working Paper and Quarterly Journal of Economics, try between 1985 and 2000, as well as No. 9282, October 2002. forthcoming. the impact of a merger on a firm’s market 4 J. Bhattacharya and M. K. Bundorf, 15 F. Lichtenberg, “The Impact of New value, employment, R and D expendi- “The Incidence of the Healthcare Costs Drug Launches on Longevity: Evidence tures, and sales.30 They find evidence sup- of Obesity,” NBER Working Paper No. from Longitudinal, Disease-Level Data

 NBER Reporter Spring 2006 from 52 Countries, 1982–2001,” NBER Acute Myocardial Infarction in Hospitals For-Profit, And Government Hospitals,” Working Paper No. 97543, June 2003, that Disproportionately Treat African- Health Affairs, Vol. 24, Issue 3, pp.790– and International Journal of Health Care Americans,” Circulation, October 25, 801. Finance and Economics, 5, pp. 47–73, 2005. 27 Y. Shen, K. Eggleston, C. Schmid, 2005. 22 A.Barnato, L. Lucas, D. Staiger, J. E. and J. Lau, “Hospital Ownership And 16 T. J. Philipson and A. B. Jena, “Surplus Wennberg, and A. Chandra, “Hospital- Financial Performance: A Quantitative Appropriation from R&D and Health Care level Racial Disparities in Acute Myocardial Research Review,” NBER Working Technology Assessment Procedures,” NBER Infarction Treatment and Outcomes,” Paper No. 11662, October 2005. Working Paper No. 12016, February 2006. Medical Care, 43(4), April 2005. Also 28 M. Duggan, “Hospital Market Structure 17 A. M. Garber, C. I. Jones, and P. M. K. Baicker, A.Chandra, and J. Skinner, and the Behavior of Not-For-Profit Romer, “Insurance and Incentives for “Geographic Variation and the Problem Hospitals,” RAND Journal of Economics, Medical Innovations,” NBER Working of Measuring Racial Disparities in 33(3): pp.433–46, Autumn 2002. Paper No. 12080, March 2006. Health Care,” and K.Baicker, A.Chandra, 29 M.Gaynor, J. B. Rebitzer, and L. J. 18 D. Lakdawalla and N. Sood, J. Skinner, and J. E. Wennberg, “Who Taylor, “Physician Incentives in Health “Insurance and Innovation in Health You Are and Where You Live: How Race Maintenance Organizations,” Journal of Care Markets,” NBER Working Paper No. and Geography Affect the Treatment of Political Economy 112, No. 4: pp.915–31, 11602, September 2005. Medicare Beneficiaries.” 2004. 19 D. Lakdawalla and N. Sood, “Social 23 P. W. Groeneveld, P. A. Heidenreich, 30 P. M. Danzon, A. Epstein, and S. Insurance and the Design of Innovation and A. M. Garber, “Racial Disparity in Nicholson, “Mergers and Acquisitions in Policy,” Economic Letters, Vol. 85, 2004, Cardiac Procedures and Mortality among the Pharmaceutical and Biotech Industries,” pp. 57–61. Long-Term Survivors of Cardiac Arrest,” Managerial and Decision Economics, 20 K. Baicker, A. Chandra, J. Skinner, and Circulation, 108(3), 2003, pp. 286–91. forthcoming. J. E. Wennberg, “Who You Are and Where 24 P. W. Groeneveld, P. A. Heidenreich, 31 C. Ruhm, “Mortality Increases During You Live: How Race and Geography Affect and A. M. Garber, “Trends in Implantable Economic Downturns”, International the Treatment of Medicare Beneficiaries,” Cardioverter-Defibrillator Racial Dispar­ Journal of Epidemiology, forthcom- Health Affairs, October 2004: Var33 – ity: The Importance of Geography,” Journal ing; “Macroeconomic Conditions, Health Var44; and K. Baicker, A. Chandra, and of the American College of Cardiology, and Mortality”, in A. M. Jones ed., Elgar J. Skinner, “Geographic Variation and the 45(1), 2005, pp. 72–8. Companion to Health Economics, Problem of Measuring Racial Disparities 25 H. Levy, “Health Insurance and the Cheltenham, UK: Edward Elgar in Health Care,” Perspectives in Biology Wage Gap,” NBER Working Paper No. Publishing, forthcoming; and “Healthy and Medicine 48(1), 2005. 11975, January 2006. Living in Hard Times”, Journal of Health 21 J. Skinner, A. Chandra, D. Staiger, J. 26 J. R. Horwitz, “Making Profits And Economics, 24(2), March 2005, pp. 341– Lee, and M. McClellan, “Mortality After Providing Care: Comparing Nonprofit, 63.

NBER Reporter Spring 2006 7 Research Summaries

Consumption and Saving: Theory and Evidence

Christopher D. Carroll*

Consumption and saving decisions mal behavior of consumers with standard recent paper provides the rigorous foun- are at the heart of both short- and long- attitudes toward risk (constant relative dations for the mathematical methods run macroeconomic analysis (as well as risk aversion) facing income uncertainty employed in my earlier work.4 Another much of microeconomics). In the short of the kind that appears to exist in house- paper with Miles Kimball5 explores the run, spending dynamics are of central hold-level data sources. The first paper theoretical implications of borrowing importance for business cycle analysis and found that target or “buffer-stock” saving limitations; and, a very short new paper the management of monetary policy. And may be optimal under some circumstanc- describes a conceptual trick that can be in the long run, aggregate saving deter- es; the second paper found that, depend- used to simplify and accelerate the solu- mines the size of the aggregate capital ing on households’ income profiles and tion of many kinds of optimal intertem- stock, with consequences for wages, inter- their degree of impatience, it can be opti- poral choice models.6 As an aid to other est rates, and the standard of living. mal for average household spending pat- researchers, I have posted on my web page Since the pioneering work of Fried- terns to mirror average household income computer software that implements this man and of Modigliani and Brumberg profiles over much of the life cycle. This trick to solve a variety of standard optimi- in the 1950s, the principal goal of the was surprising because, in models without zation problems. My web page also con- economic analysis of saving has been to uncertainty, optimizing consumers spend tains software that reproduces the compu- formulate mathematically rigorous theo- based on their expected lifetime resources tational and empirical results in most of ries of behavior. But that goal was dif- without regard to the expected timing my published papers, as well as a set of lec- ficult until recently because the optimal of income. That is, spending patterns by ture notes (and associated software) that response of saving to uncertainty was age are not intrinsically determined by provide a comprehensive treatment of the difficult to compute. Research was gener- income patterns by age. (This work, and methods for solving these models.7 ally carried out under the assumption that my subsequent related work, assumes that In the end, however, mathematical uncertainty might boost saving some- consumers have successfully solved any models are useful only insofar as they can what, but that behavior in the presence “self-control” problems of the type that be related to empirical evidence about of uncertainty was likely to be broadly David Laibson and others have so persua- the real world. Toward the end of match- similar to optimal behavior in a world in sively described). ing theory and data, Andrew Samwick which households had perfect foresight This paper was related to two other, and I wrote two papers8,9 whose goal was about their future circumstances. more abstract, papers. The more fun- to get a quantitative sense of the nature In two papers that grew out of my damental of these,2 written with Miles and magnitude of household responses 1990 dissertation,1 I showed that the Kimball, showed that in the presence of to uncertainty. The first of these papers presence of uncertainty could change uncertainty, households with low levels found that a standard source of microeco- the nature of optimal behavior in qual- of wealth will respond more to a windfall nomic data, the Panel Study of Income itatively and quantitatively important infusion of cash than households with Dynamics, implied that income uncer- ways. Specifically, I examined the - opti ample resources. The other paper3 dem- tainty was very large indeed. According onstrated that the logic of precautionary to the benchmark specification, a conser- saving undermines the standard “Euler vative estimate was that in any given year *Chris Carroll is a Research Associate in the equation” method of testing for optimiz- about a third of households could expect NBER’s Programs on Monetary Economics ing consumption behavior. their “permanent” income to rise or fall and Economic Fluctuations and Growth and Mathematical and computational by as much as 10 percent. (“Permanent” a Professor of Economics at the Johns Hopkins University. His profile appears later in this aspects of optimal behavior have remained changes in income here mean the kind issue. a theme in my research to the present. A of change associated with a promotion

 NBER Reporter Spring 2006 or being laid off and settling for a new vidual households have no control) to be borrowing to finance their current lower-paying job). The second paper with measure the size of uncertainty. Empirical expenditures, while people in slow-grow- Samwick estimated that as much as 40 results in that paper suggested that pre- ing economies should anticipate that they percent of the wealth held by the typical cautionary motives for saving were more may need to save a lot if they wish to household represented a response to the important for people in the upper half of maintain their current standard of living fact that some households face greater the income distribution, and that precau- in the future. The logic therefore suggests uncertainty than others. tionary behavior is manifested partly in a that we should expect to see a negative An important caveat about these reluctance to borrow against home equity association between saving and growth. results is that many of the wealthiest when unemployment is high, rather than One objection to this thread of rea- households are missing from the PSID an explicit accumulation of greater liquid soning might be that countries’ saving dataset on which the estimates are based. assets. rates differ partly for cultural reasons, and Since a large proportion of aggregate If uncertainty matters this much for it seems natural to expect that countries wealth is held by the richest few percent spending decisions on average, it seems whose saving rates are high because of a of households, these estimates very like- plausible that the changes in uncertainty cultural preference for saving would con- ly overstate the proportion of aggregate that accompany business cycles might be sequently exhibit high growth. Byung- wealth that can be attributed to precau- an important source of fluctuations in Kun Rhee and Changyong Rhee and I tionary motives. Indeed, another paper10 consumer spending. Wendy Dunn and used data on immigrants to Canada15 showed that the theoretical model used I showed13 that while there does not to investigate the possibility that cultur- in the first paper with Samwick severe- seem to be any systematic relationship al differences explain saving differences. ly underpredicts the wealth holdings of between spending and various measures Under the “cultural” theory of saving, one the wealthiest households in the United of households’ financial condition, mea- might expect that immigrants from high- States even if wealthy individuals are sures of consumers’ degree of uncertainty saving countries (for example, Japan) to assumed to be more patient than others. (especially their assessment of whether save more than immigrants from low- That paper argued that a bequest motive the unemployment rate is likely to rise) saving countries (for example, Sweden). in which bequests are a “luxury” good is have a powerful impact on spending (par- But we found no evidence of such a pat- essential to explaining why saving rates of ticularly purchases of big-ticket items like tern, either in Canada or in a subsequent wealthy households are so high. A subse- vehicles and houses). study using Census data from the United quent paper11 showed that the “bequests In fact, the model in that paper sug- States.16 as luxuries” model can also explain a vari- gested that, if anything, the mystery is why Furthermore, the evidence clearly ety of facts about the portfolio choices uncertainty-driven fluctions in expendi- suggests that the relationship between of wealthy households, particularly their tures on durable goods are not even larg- saving and growth is dynamic, not static: comparatively high tolerance for finan- er. According to the model, most of the countries that go through periods of pro- cial risk. people who were on the verge of buying longed growth tend to experience rising Another potential problem with my a car should be willing to postpone their saving rates, while countries that experi- work with Samwick is that we were forced purchase in response to even a very mod- ence sustained economic slowdowns tend by data limitations to make the assump- est increase in uncertainty. While the evi- to suffer declining saving rates. tion that income risk is something over dence confirms that durable goods spend- Both the sluggish response of spend- which people have no control. If instead ing is indeed more volatile than spending ing to uncertainty and the pattern in people make employment choices based on nondurables like food, the size of the which increases (or decreases) in growth partly on the riskiness of the different discrepancy is not as large as the rational produce increases (respectively, decreases) alternatives (for example, if risk-averse optimization model tends to suggest it in saving might be explained by a model in people seek civil service jobs while the should be. which spending “habits” exert a powerful risk-lovers become entrepreneurs), then This finding seems to fit with the influence on behavior. A paper with Jody the estimated effect of uncertainty on sav- results of an earlier paper with David N. Overland and David N. Weil17 explored ing might be incorrect. The likeliest effect Weil14 which found that, across coun- how the incorporation of spending hab- would be to underestimate the impor- tries, the relationship between aggregate its modifies the predictions of a model tance of precautionary behavior, since the saving and aggregate growth is not what of optimal spending behavior. A subse- theory tends to suggest that those who dis- would be expected from the standard quent paper18 incorporated both hab- like risk more will both avoid risky occu- framework in which spending depends its and uncertainty, and argued that the pations and save more. But in an attempt on expectations about future income. The broad patterns of saving and growth seen to get around this problem, Karen Dynan problem is that people living in fast- in the East Asian “tiger” economies could and Spencer Krane and I wrote a paper12 growing economies should expect their be explained in a model where both pre- that used temporary regional variations future incomes to be large relative to their cautionary motives and habit formation in unemployment risk (over which indi- current incomes, and should therefore were important. This work meshes with

NBER Reporter Spring 2006 9 a prominent strand of the macroeconom- nation of the epidemiological modeling 4 C. D. Carroll, “Theoretical Foundations ics and finance literatures over the past framework and its application).20 of Buffer Stock Saving,” NBER Working decade that has argued that habit forma- The paper with Slacalek proposes to Paper No. 10867, November 2004. tion can explain a wide range of empirical reconcile the microeconomic and macro- 5 C. D. Carroll and M. S. Kimball, observations that are difficult to reconcile economic evidence about consumption “Liquidity Constraints and Precautionary with standard models without habits. dynamics by applying the same model of Saving,” NBER Working Paper No. 8496, A new paper with Jirka Slacalek,19 sticky expectations. The essential idea is October 2001. however, casts doubt on the view that hab- that people have a very good understand- 6 C. D. Carroll, “The Method of its are the right explanation for the slug- ing of the circumstances they face in their Endogenous Gridpoints for Solving gishness of aggregate spending dynamics. own lives (for example, they know wheth- Dynamic Stochastic Optimization This paper points out that habits imply er they have been fired), but they do Problems,” NBER Technical Working that spending dynamics should be similar not pay as much attention to macroeco- Paper No. 309, June 2005, forthcoming in microeconomic and macroeconomic nomic developments (for example, they in Economics Letters. Paper and software data. Yet empirical studies using micro- may not know the latest aggregate unem- available at http://econ.jhu.edu/people/ economic data, using exactly the same ployment statistic). Since household-spe- ccarroll/EndogenousArchive.zip methods as applied to macroeconomic cific uncertainty is much greater than 7 C. D. Carroll, Lecture notes on solving data, find very different results. While the aggregate uncertainty (a rough estimate microeconomic dynamic stochastic optimi- data hint that there may be some modest is that household-specific risks are about zation problems, Archive, Johns Hopkins habit formation effects in a few categories 100 times larger than macroeconomic University, 2006. Available at http://econ. of spending, models in which habits are a risks), it makes sense for busy consumers jhu.edu/people/ccarroll/solvingmicrodsops. dominant force in microeconomic spend- to pay less than perfect attention to the zip ing decisions can be decisively rejected. macroeconomy. 8 C. D. Carroll and A. A. Samwick, “The The new paper relates to another Whether or not this particular expla- Nature of Precautionary Wealth,” NBER strand of my research, which argues that nation for the conflict between micro- Working Paper No. 5193, July 1995, and economists should pay more attention economic and macroeconomic data on Journal of Monetary Economics, 40(1): than has been customary to the evidence consumption dynamics is accepted, this pp. 41–71, 1997. provided by surveys of households. A conflict seems likely to be a topic of grow- 9 C.D. Carroll and A. A. Samwick, 2001 NBER working paper proposed ing attention over the next few years. “How Important Is Precautionary Saving?” modeling household survey data on infla- While great progress has been made in NBER Working Paper No. 5194, July 1995, tion expectations using a simple model of understanding the quantitative implica- and Review of Economics and Statistics, disease transmission. The idea is that rath- tions of alternative models of consump- 80(3): pp. 410–19, August 1998. er than forming their own independent tion and saving behavior, much remains 10 C. D. Carroll, “Why Do the Rich views of the likely future inflation rate, to be understood. Save So Much?” NBER Working Paper typical people’s views are formed by expo- No. 6549, May 1998, and in Does Atlas sure to the views of experts as represented 1 C. D. Carroll, “The Buffer-Stock Theory Shrug? The Economic Consequences in the news media. In this model, house- of Saving: Some Macroeconomic Evidence,” of Taxing the Rich, J. B. Slemrod, ed., holds’ forecasts of inflation, while not fully Brookings Papers on Economic Activity, Harvard University Press, 2000. “rational” in the economist’s usual strict 1992 (2): pp. 61–156; and “Buffer-Stock 11 C. D. Carroll, “Portfolios of the Rich,” sense of the term, do not deviate very long Saving and the Life Cycle/Permanent Income NBER Working Paper No. 7826, August or very far from the experts’ view. The Hypothesis,” NBER Working Paper No. 2001, and in Household Portfolios: paper presented empirical evidence that 5788, October 1996, and Quarterly Journal Theory and Evidence, MIT Press, information in newspaper reports about of Economics, CXII (1): pp. 1–56, 1997a. Cambridge, MA, 2002. inflation seems to filter out to the popula- Available at http://econ.jhu.edu/people/ 12 C.D. Carroll, K. E. Dynan, and tion gradually rather than instantly. The ccarroll/BSLCPIH.zip S. S. Krane, “Unemployment Risk and proposed model can be interpreted as 2 C. D. Carroll and M. S. Kimball, Precautionary Wealth: Evidence from providing a concrete theoretical justifica- “On the Concavity of the Consumption Households’ Balance Sheets,” Review of tion for the model of “sticky expectations” Function,” Econometrica, 64(4): pp. 981– Economics and Statistics, 85(3), August that has become increasingly popular in 92, 1996. 2003. the macroeconomics literature in recent 3 C. D. Carroll, “Death to the Log- 13 C.D. Carroll and W. E. Dunn, years. (The NBER working paper was Linearized Consumption Euler Equation! “Unemployment Expectations, Jumping subsequently split into two papers, one (And Very Poor Health to the Second-Order (S,s) Triggers, and Household Balance containing the empirical evidence and a Approximation),” NBER Working Paper Sheets,” NBER Working Paper No. 6081, stripped-down version of the model, and No. 6298, December 1997, and Advances July 1997, and NBER Macroeconomics the other examining a detailed exami- in Macroeconomics, 1(1): Article 6, 2001. Annual, 1997, B.S. Bernanke and J. J.

10 NBER Reporter Spring 2006 Rotemberg, eds., MIT Press, Cambridge, Behavior? Evidence from Immigrants,” Expectations and Consumption Dynamics,” MA, 1997, pp. 165–229. NBER Working Paper No. 6568, May manuscript, Johns Hopkins University, 14 C. D. Carroll and D. N. Weil, “Saving 1998, and Economic Development and 2006. and Growth: A Reinterpretation,” NBER Cultural Change, 48(1): pp.33–50, 20 The basic model and evidence are pub- Working Paper No. 4470, September 1993, October 1999. lished in C. D. Carroll, “Macroeconomic and Carnegie-Rochester Conference 17 C. D. Carroll, J. R. Overland, and Expectations of Households and Series on Public Policy, 40: pp. 133–92, D. N. Weil, “Saving and Growth with Professional Forecasters,” Quarterly Journal June 1994. Habit Formation,” American Economic of Economics, 118(1): pp. 269–98, 2003; 15 C. D. Carroll, C. Rhee, and B. Rhee, Review, 90(3): pp. 341–55, June 2000. the full model and its extensions can be found “Are There Cultural Effects on Saving? 18 C. D. Carroll, “‘Risky Habits’ and the in “The Epidemiology of Macroeconomic Some Cross-Sectional Evidence,” The Marginal Propensity to Consume Out of Expectations,” in The Economy as an Quarterly Journal of Economics, CIX(3): Permanent Income,” NBER Working Paper Evolving Complex System, III, L. Blume pp. 685–700, August 1994. No. 7839, August 2000, and International and S. Durlauf, eds. Oxford University 16 C. D. Carroll, C. Rhee, and B. Rhee, Economic Journal, 14(4): pp.1–41, 2000. Press, 2006. “Does Cultural Origin Affect Saving 19 C. D. Carroll and J. Slacalek, “Sticky

Child Health and Human Capital

Janet Currie*

When economists use the phrase Intervention Programs exactly why they work. It is possible that “human capital,” it generally means “edu- much of the beneficial effect of Head cation.” But one’s health can also be For example, my work with Eliana Start is not through explicitly education- viewed as a form of capital. Both educa- Garces and Duncan Thomas shows that al interventions but rather through man- tion and health are strongly influenced by Head Start (a pre-school intervention dates to improve nutrition, link families “family background,” which is commonly for poor children) improves long-term with community services, and increase measured using parent’s education and outcomes for disadvantaged children, utilization of preventive health care.3 income. Much of my research over the although it does not bring these chil- Head Start’s emphasis on getting past decade seeks to evaluate the effect dren up to the level of their more advan- children into care remedies an impor- of public programs designed to improve taged peers. Using a special supplement tant limitation of programs that focus the outcomes of children from disadvan- to the Panel Study of Income Dynamics, primarily on extending health insurance taged backgrounds. In my forthcoming we ask whether children who attended via such programs as Medicaid or the book, The Invisible Safety Net: Protecting Head Start had better outcomes (on a State Child Health Insurance Program the Nation’s Poor Children and Families, I range of measures) than their own sib- (SCHIP). Lack of health insurance argue that while the cash welfare system lings who did not attend. We find that remains an important issue, but is not receives more attention, elements of a among whites, children who attended the major determinant of child health. largely invisible safety net of in-kind pro- Head Start were about 25 percent more One reason is that providing eligibil- grams have proven remarkably effective likely to have completed high school ity for health insurance does not always in improving the lives of poor children.1 than their siblings who did not. Among lead people to use care appropriately. African-Americans, the Head Start chil- In a broader review of the “take up” of dren were half as likely to have been social programs, I discuss the low take-up booked or charged with a crime. This is rate among individuals eligible for pub- * Currie is a Research Associate in the NBER’s the first study to show a lasting effect of lic health insurance; this is an important Programs on Labor Studies, Children, and 2 Education. She is also the Charles E. Davidson Head Start. social problem that reduces the use of chair and professor of economics at the University Still, programs like Head Start remain preventive care and may increase the use of California, Los Angeles. “black boxes,” in that we know little about of expensive palliative care.4

NBER Reporter Spring 2006  SES and Child Health in countries with universal health insur- carbon monoxide (CO) and particulates ance such as Canada. Using a panel of (PM10) that occurred over the 1990s Maternal education is one impor- Canadian children, Mark Stabile and I saved more than 1,000 infant lives in tant determinant of take-up and of other show that the health of poor children California.9 health behaviors. However, it has been relative to that of richer children wors- Nutrition is a key determinant of difficult to demonstrate this relation- ens with age, just as it does in the United health that is receiving increasing atten- ship empirically because maternal educa- States. We argue that this deterioration tion, given an “epidemic” of obesity and tion is a choice. To tackle this problem, may be related to a higher “arrival rate” obesity-related diseases such as diabetes. I Enrico Moretti and I compiled data on of negative health shocks among poor have examined the determinants of child openings of two- and four-year colleges children. For example, poor children are nutrition in a series of studies with Jayanta between 1940 and 1990. We used data more likely than richer ones to have new Bhattacharya, Steven Haider, and Thomas about the availability of colleges in the chronic conditions diagnosed at virtually Deliere. We find that poverty is an impor- woman’s county of residence in her sev- all ages, and they are also more likely to be tant predictor of nutritional outcomes enteenth year as an instrument for her hospitalized.7 Perhaps surprisingly, in our among preschool children, but not among education at the time of her child’s birth. data, both rich and poor children were school-aged children. However, “food We found that higher maternal education equally likely to recover from any given insecurity” (missing meals or being afraid does improve infant health, as measured health shock. Identifying the sources of that there will not be sufficient money by birth weight and gestational age. It these health shocks and policies that may to buy food) is not predictive of poorer also increases the probability that a new prevent them is an important avenue for nutritional outcomes among either group mother is married, reduces parity (birth future research. of children (although it could be viewed order), increases use of prenatal care, and as a bad outcome in itself ). Nevertheless, reduces smoking, thus suggesting that Threats to Child Health there are many children with nutritional these are important pathways for the ulti- deficiencies, even among those who con- mate effect on health.5 One example of a negative health sume too many calories.10 In subsequent work, Moretti and I shock not prevented by conventional Using data from the National Health created a unique longitudinal dataset of medical care is unintentional injuries. and Nutrition Examination Surveys, we California births from the 1960s to the Such injuries are a leading cause of death also find that poor families reduce expen- present in order to investigate the rela- among children over the age of one in ditures and calories consumed in response tionship between maternal income (mea- the United States. Joseph Hotz and I to cold weather shocks (a “heat or eat” sured at the time of the mother’s birth and show that accident rates are responsive to effect), although we find no evidence that at the time of the child’s birth), maternal child care policy — they are lower among this affects the quality of the diet. Despite birth weight, and the child’s birth weight. children in care when the care givers are recent concerns about inadequacies in We used names and birth dates to link the more educated — although stiffer child child nutrition programs, we find that records of mothers and children and also care regulations may also increase acci- the School Breakfast program improves identified mothers who were siblings. We dent rates among children pushed out of the quality of children’s diets.11 Taken as showed that there is a strong inter-genera- regulated care by higher prices.8 a whole, these studies suggest that there tional correlation in the birth weight of Pollution is another factor that affects is a link between poverty and poor child mothers and children, but that a measure disadvantaged children disproportion- nutrition that is mitigated by the food of household income at the time of the ately. In our study of the effects of air safety net that is in place, particularly for mother’s birth is also predictive of low pollution in California on infant health, school aged children. birth weight in her child. Our most inter- Matthew Neidell and I find that the most While most of the economic research esting finding is that there is an interac- polluted zip codes have 50 percent more on child health focuses on physical health, tion between maternal low birth weight mothers who are high school dropouts mental health may be much more impor- and maternal poverty in the production than the least polluted ones. This compli- tant. The majority of workdays lost among of child low birth weight. Together these cates our attempts to identify the causal adults are attributable to mental health findings suggest that inter-generational effect of pollution. We use individual- problems, and many such problems have correlations in health could play a role level vital statistics data to investigate their roots in childhood. The best avail- in the inter-generational transmission the effects of criterion air pollutants on able estimates suggest that mental health of income. Parent’s income affects child infant mortality, fetal deaths, low birth problems may be much more prevalent health, and health at birth affects future weight, and prematurity. Our models are among poor than among non-poor chil- income.6 identified using within-the-zip-code level dren, confounding attempts to measure The relationship between family variation in pollution levels that remains the effects of mental illness per se. Stabile income and child health starts at birth after controlling for seasonal patterns and and I use nationally representative sam- but grows stronger as children age, even weather. We find that the reductions in ples of U.S. and Canadian children to

12 NBER Reporter Spring 2006 examine the medium-term outcomes of been argued that managed care might the transmission of poverty from one children with symptoms of Attention have some offsetting benefits for patients. generation to the next could perhaps be Deficit Hyperactivity Disorder (ADHD), For instance, it would guarantee access to improved if we understood these linkages the most common child mental health providers who were contractually obli- better. problem.12 gated to treat Medicaid patients, whereas Rather than relying on diagnoses, under the fee-for-service system, many we use “screener” questions administered providers did not accept Medicaid. 1 Forthcoming from Princeton University to all children, and we use sibling fixed However, incentives facing provid- Press, Spring 2006. effects to control for omitted variables ers are complex and may result in many 2 E. Garces, D. Thomas, and J. Currie, such as poverty. We find large negative consequences that were not intended by “Longer Term Effects of Head Start,” NBER effects on test scores and schooling attain- legislators. John Fahr and I find that the Working Paper No. 8054, December 2000, ments, and positive effects on the prob- introduction of MMC in California was and American Economic Review, 92, 4, ability of being placed in special educa- accompanied by shifts in the composition September 2002, pp. 999–1012. tion. The effects are remarkably similar in of the Medicaid caseload away from black 3 J. Currie and M.Neidell, “Getting Inside Canada and the United States. Moreover, children, and that black children who lost the ‘Black Box’ of Head Start Quality: the effects are approximately linear, sug- coverage were subsequently more likely What Matters and What Doesn’t,” NBER gesting that even moderate symptoms to go without doctor visits. Using a panel Working Paper No. 10091, November have costs in terms of educational attain- of all California births among mothers in 2003. ment. In contrast, physical health prob- the 1990s, Anna Aizer, Moretti, and I are 4 J. Currie, “The Take-up of Social lems such as asthma are found to have able to follow mothers who were required Benefits,” NBER Working Paper No. insignificant effects. These results indicate to join MMC plans between births. We 10488, May 2004, forthcoming in A. that mental health conditions might well find that mothers forced to switch to Auerbach, D. Card, and J. Quigley, eds. prove to be a “missing link” between fam- MMC were more likely to delay prenatal Poverty, the Distribution of Income, and ily background, child health, and educa- care and to suffer adverse birth outcomes Public Policy, New York: Russell Sage. tional attainments. than other mothers.14 5 J. Currie and E. Moretti, “Mother’s Aizer and I also examine estimates Education and the Intergenerational The Role of Health Insurance of “network effects” in the utilization Transmission of Human Capital: Evidence of public prenatal care services provided from College Openings,” NBER Working Despite the key role of family back- under the Medicaid program. We find Paper No. 9360, December 2002, and ground and non-medical threats to child that these effects are similar for first-time Quarterly Journal of Economics, VCXVIII, health, most discussions of disparities in mothers and for second-time mothers 4, November 2003, pp.1495–532. child health focus not on more general who have already used prenatal care ser- 6 J. Currie and E. Moretti, “Biology as interventions, such as Head Start, but vices. This suggests that the measured Destiny? Short and Long-Run Determinants rather on the role of health insurance. I effects do not represent transmission of of Intergenerational Transmission of Birth have continued to study Medicaid, the information about the services between Weight,” NBER Working Paper No. 11567, main system of public health insurance network members, because mothers who August 2005. for poor women and children. Using indi- have already used the services presumably 7 J. Currie and M. Stabile, “Socioeconomic vidual-level vital statistics data, Jeffrey know about them. Moreover, the estimat- Status and Health: Why is the Relationship Grogger and I find that state welfare ed effects are much reduced when we con- Stronger for Older Children?” NBER reforms prior to 1996 were associated trol for the hospital of delivery. Perhaps Working Paper No. 9098, August 2002, with reductions in the use of prenatal surprisingly, women who live in the same and American Economic Review, 93, 5, care and with negative impacts on infant neighborhoods, but who are from differ- December 2003, pp.1813–23. health, presumably because women who ent ethnic groups, tend to deliver in dif- 8 J. Currie and J. V. Hotz, “Accidents Will went off the welfare rolls were no lon- ferent hospitals. These results suggest that Happen? Unintentional Injury, Maternal ger automatically eligible for Medicaid it is the enrollment services provided by Employment, and Child Care Policy,” coverage.13 hospitals, and not the woman’s “network,” NBER Working Paper No. 8090, January Over the 1990s, most states switched that facilitates access to Medicaid-spon- 2001, and Journal of Health Economics, their Medicaid caseloads from traditional sored prenatal care services.15 23, 1, January 2004, pp.25–59. fee-for-service to some form of Medicaid In summary, my research points to 9 J. Currie and M. Neidell, “Air Pollution managed care (MMC). Like the managed a holistic view of child human capital and Infant Health: What Can We Learn care programs that cover most privately development that encompasses educa- From California’s Recent Experience?” insured Americans, MMC restricts access tional attainment, physical, and mental NBER Working Paper No. 10251, January to services in order to reduce costs. In the health, and seeks to explore the feedbacks 2004, and Quarterly Journal of Economics, case of Medicaid patients, though, it has between them. Interventions to reduce CXX, 3, August 2005, pp.1003–30.

NBER Reporter Spring 2006 3 10 J. Currie, J. Bhattacharya, and S. 10608, July 2004, and Journal of Human NBER Working Paper No. 8812, February Haider, “Poverty, Food Insecurity, and Resources, forthcoming. 2002, Journal of Public Economics, 89,1, Nutritional Outcomes in Children and 12 J. Currie and M. Stabile, “Child Mental January 2005, pp. 85–108. A. Aizer, J. Adults,” Journal of Health Economics, 23, Health and Human Capital Accumulation: Currie and E. Moretti, “Competition in 2, July 2004, pp. 839–62. The Case of ADHD,” NBER Working Imperfect Markets: Does it Help California’s 11 J. Bhattacharya, J. Currie, T. DeLiere, Paper No. 10435, April 2004, forthcoming Medicaid Mothers?” NBER Working Paper and S. Haider, “Heat or Eat? Income in Journal of Health Economics. No.10430, April 2004, forthcoming in Shocks and the Allocation of Nutrition 13 J. Currie and J. Grogger, “Medicaid Review of Economics and Statistics. in American Families,” NBER Working Expansions and Welfare Contractions: 15 A. Aizer and J. Currie, “Networks or Paper No. 9004, June 2002, and American Offsetting Effects on Prenatal Care and Neighborhoods? Correlations in the Use Journal of Public Health 93(7), July 2003, Infant Health,” NBER Working Paper No. of Publicly-Funded Maternity Care in pp.1149–54. J. Bhattacharya, J. Currie, 7667, April 2000, and Journal of Health California,” NBER Working Paper No. and S. Haider, “Breakfast of Champions? Economics, 21, March 2002, pp.313–35. 9209, September 2002, Journal of Public The Effects of the School Breakfast Program 14 J. Currie and J. Fahr, “Medicaid Economics, 88, 12, December 2004, pp. on the Nutrition of Children and their Managed Care: Effects on Children’s 2573–85. Families,” NBER Working Paper No. Medicaid Coverage and Utilization of Care,”

Estate Taxation

Wojciech Kopczuk*

Taxation of estates and inheritances capital. It is heavily progressive, with U.S. economic decisions. Second, there is an is one of the most controversial issues in federal tax rates currently approaching 50 important theoretical question regarding tax policy. While this type of taxation is percent and exceeding 70 percent in the the role that this type of taxation should viewed by some as an integral part of a past. It is closely tied to the propagation play in the tax system. Third, the existing system that guarantees equality of oppor- of inequality and the impact of redistri- data on estate taxpayers provides a source tunities, others describe it as a “death bution. It affects the intergenerational of information that can shed light on tax” and argue that it is both inherently mobility of wealth. Its impact and its cost central economic, but non-tax, issues. In unfair to levy a tax at death and that it depend on the presence and nature of a my research, I have pursued each of these is particularly costly to do so, highlight- bequest motive. How individuals plan directions. ing its adverse effect on wealth accumula- for leaving an estate depends also on their In a few of my papers, I looked at tion, discrimination against savers, nega- acceptance and attitudes toward their own how transfer taxation affects economic tive consequences for the survival of small death, thus providing a natural place for decisions. The notion that the estate tax businesses, and a multitude of avoidance looking for examples of the importance forces people to make difficult late-in-life, opportunities. of psychological considerations. The U.S. even deathbed, decisions, has its place in From an economist’s point of view, estate tax is nominally a tax on individu- the political discourse about the tax, but estate taxation touches on a wide array of als, but its incidence depends on family is it really true? Using linked estate and important topics. It is a form of a tax on structure and interrelationships. The tax income tax data, I studied how estates of has been dubbed a “voluntary tax,” high- people who suffered from a lengthy ill- lighting that tax avoidance and adminis- ness differ from estates of those who died *Wojciech Kopczuk is a Faculty Research Fellow tration issues are also very important. instantaneously.1 I found that the size of in the NBER's Public Economics Program Estate taxation has figured in eco- the reported estate (of wealthy estate tax- and an associate professor of economics and international and public affairs at Columbia nomic research in three different ways. payers) is as much as 20 percent lower for University. His profile appears later in this First, one may be interested in under- decedents whose terminal illness lasted issue. standing how the actual estate tax affects months or more, but I also showed that

14 NBER Reporter Spring 2006 this effect is unlikely to be explained by a stand on the nature of intergeneration- planner should have a contingency plan in medical expenses or lost wages. Instead, I al links. Unfortunately, despite a lot of place. As mentioned earlier, my research found strong evidence pointing to a flurry research on this topic, there is no consen- shows that much estate planning takes of estate planning activity following the sus regarding the types of bequest motive place shortly before death, suggesting that onset of a terminal illness, that results in a or even the prevalence of any bequest procrastination in this context is plausi- reduction in the value of the reported tax- motive. Joseph Lupton and I revisit the ble. One possibility is that standard mod- able estate and therefore tax liability. influential work of Michael Hurd,6 who els do not accurately represent how people How strongly do estates respond to demonstrated that people with and with- incorporate mortality risk in their behav- estate taxation? By exploiting more than out children have similar consumption ior. Borrowing from the psychological lit- 80 years of IRS data covering multiple tax patterns in the old age, thereby putting erature on terror management theory, Joel regimes, and age variation of estate tax in question the possibility that they have Slemrod and I explore the consequences decedents, Joel Slemrod and I estimate an different bequest considerations.7 We of utility-reducing fear in acknowledging elasticity of reported estates with respect to relax the assumption that children are one’s own mortality.9 We equip agents the net-of-tax rate of about -0.16, suggest- a deterministic indicator of a bequest with a fear function that increases with ing that the estate tax does in fact reduce motive, and instead show that consump- subjective mortality risk and the ability reported estates, either because it curtails tion patterns of the elderly are explained to repress information. We conclude that wealth accumulation or induces tax avoid- by a parsimonious structural two-type such agents are “behavioral;” in particu- ance, or both.2 Note, though, that (as has model with both bequest-motive and no- lar, we find that such individuals have an been argued in the taxable income elas- bequest types. We estimate that the first incentive to behave in a time-inconsistent ticity literature, including my own work) group constitutes three-fourths of the fashion regardless of whether they actu- both avoidance and wealth accumulation population but that for most people the ally repress information or not. channels entail similar short-term effi- difference between the bequest and non- The U.S. federal estate tax was intro- ciency costs, although the longer-term bequest consumption patterns is small, duced in 1916. It has always applied to implications are likely very different.3 In and only at the very top of the wealth a relatively small group of the wealthi- another joint paper, we also show that distribution do these differences become est decedents; applying at the peak of the estate tax has important implications economically important. Overall, we find its coverage in the 1970s to over 7 per- for charitable contributions.4 Finally, we no evidence that having children is an cent of adult deaths, and at its mini- demonstrated that the reported timing of important indicator of the presence of a mum coverage to less than 0.5 percent. Its death is sensitive to tax considerations: in bequest motive. long history and its focus on the top of a four-week period surrounding estate tax One popular argument for taxa- the distribution make estate tax statistics reforms, more taxable deaths are observed tion of estates is that a tax on “acci- a natural source for studying long-term during the “low-tax” regime than during dental” bequests, that is, on savings of changes in wealth concentration. This is the “high-tax” regime.5 We were, unfor- non-annuitized individuals subject to what Emmanuel Saez and I have done.10 tunately, unable to conclude how much stochastic mortality, is particularly effi- We relied on (unfortunately confiden- of this response represents the strength cient because it does not stimulate any tial) IRS micro databases that include of willpower of tax-averse individuals behavioral response. I show that this rea- all of the estate tax returns filed between and how much reflects cheating by their soning is potentially misleading, because 1916 and 1945, samples for a few years beneficiaries, but this finding provides non-annuitization may be due to a mar- between 1962 and 1976, and annual sam- another example of the variety of behav- ket failure that can be addressed by gov- ples starting in 1982, and supplemented ioral responses that individuals pursue in ernment policy.8 Theoretically, such an this data with published tabulations for response to tax incentives. intervention could reduce or eliminate other years. We applied the estate-multi- From the theoretical point of view, accidental bequests altogether and would plier technique (that amounts to weight- one way of thinking about taxation of be preferred to their taxation. I also show ing individuals by the inverse of their estates is as a tax on capital. Under the that estate taxation may provide implicit mortality risk) and constructed estimates standard model of perfect altruism, the annuitization (the necessary and sufficient of wealth controlled by groups within the question of how to tax estates reduces to condition being that the present value of top 2 percent of the wealth distribution the question of how to tax capital with estate tax payments falls with age), and going back to 1916. Similar to findings infinitely lived agents, and there is a large argue that it does so in practice. In prin- from studies of the long-term evolution of and growing literature on the subject. ciple, such an annuity should be welcome income inequality for example, (Piketty However, the relationship between a tax by individuals who do not have access to and Saez, 2003),11 we find that wealth on estates and a tax on wealth or capital actuarially fair insurance markets. concentration decreased rapidly in the income depends crucially on the nature Although the timing of death is 1930 and 1940s but there is no evidence of the bequest motive. Any theoretical uncertain, it does occur eventually with of an increase in past 20 years. This lat- analysis of estate taxation requires taking probability one, and a forward-looking ter result is particularly puzzling in light

NBER Reporter Spring 2006 5 of the sharp increase in income concen- in the 1930s and 1940s, and has been and the Elasticity of Taxable Income,” tration over this period. However, these increasing since the 1970s. Reconciling it NBER Working Paper No. 10044, October findings are consistent with the Survey of with the flat wealth concentration series 2003, and Journal of Public Economics, Consumer Finances (for broader wealth in the past 20 years therefore requires that 89 (2005), pp. 2093–119; J. Slemrod and categories), as documented by John Karl the relative wealth from inheritances has W. Kopczuk, “The Optimal Elasticity of Scholz.12 been declining, while self-made wealth Taxable Income,” NBER Working Paper One potential explanation that we has been increasing. These findings are 7922, September 2000, and Journal of offer for the lack of an increase in wealth consistent with the pattern observed in Public Economics, 84 (2002), pp. 91– 112. concentration is that increases in income the Forbes list, where the fraction of peo- 4 concentration were driven by labor rather ple classified as deriving their wealth from W. Kopczuk and J. Slemrod, “Tax than capital incomes, so it may be that not inheritance halved over the past 20 years, Consequences on Wealth Accumulation enough time has passed for the increase and with Census data about self-employ- and Transfers of the Rich,” Death and in wealth accumulation concentration ment and the number of employers. The Dollars: The Role of Gifts and Bequests to materialize. Another potential factor results also provide an important qualifi- in America, A. H. Munnell and A. is changing income mobility. However, cation to the interpretation of the drop Sunden, eds., Washington, D.C.: Brookings Institution Press (2003), pp. 213–249. Emmanuel Saez, Jae Song, and I study in income and wealth concentration in 5 longitudinal Social Security earnings data the 1930s and 1940s: our findings sug- W. Kopczuk and J. Slemrod, “Dying that allow us to trace the same individual gests that entrepreneurial wealth declined to Save Taxes: Evidence from Estate Tax over long periods of time and therefore during that period more than inherited Returns on the Death Elasticity,” NBER to understand how income mobility has wealth did. This is further supported by Working Paper No. 8158, March 2001, evolved over the past 50 years (and with historical lists of the wealthy, which show and Review of Economics and Statistics, 85 (2003), pp. 256–65. less detail since 1937). In our still pre- that the importance of inherited wealth at 6 liminary work, we find no evidence that the top of the wealth distribution peaked M. Hurd, “Savings and Bequests,” mobility of earnings has changed much after World War II. NBER Working Paper No. 1826, November over time.13 Whether the estate tax in the United 1989, and “Mortality Risk and Bequests,” Econometrica, 57 (1989), pp. 360–99. My work with Lena Edlund provides States will remain an important issue 7 a different perspective for thinking about depends on the fate of its ongoing phase- W. Kopczuk and J. Lupton, “To Leave long-term changes in wealth concentra- out that culminates in complete repeal or not To Leave: The Distribution of tion.14 We observe that the gender dis- scheduled to occur in 2010. As is well Bequest Motives,” NBER Working Paper tribution of estate taxpayers evolved over known, the repeal is part of a set of No. 11767, November 2005, forthcoming in the Review of Economic Studies. time. In particular, the number of women provisions that sunset in 2011, so that 8 among the very wealthy estate taxpayers current law specifies that in 2011 the W. Kopczuk, “The Trick is to Live: (top 0.01 percent) rose until the 1960s, estate tax will revert to its 2001 version. Is the Estate Tax Social Security for the but has been declining since the 1970s. Policymakers have provided researchers Rich?” NBER Working Paper No. 9188, We argue that the gender distribution with a rich set of experiments that will September 2002, and Journal of Political Economy, 111 (2003), pp. 1318–41. of the wealthy group reveals the relative help in years to come in understanding 9 importance of self-made and inherited the effect of estate taxation itself and, per- W. Kopczuk and J. Slemrod, “Denial wealth. While women and men inherit haps more importantly, other economic of Death and Economic Behavior,” NBER from their parents about equally, entre- decisions related to death and intergen- Working Paper No. 11485, July 2005, and preneurship remains predominantly the erational transfers. Advances in Theoretical Economics, The domain of men. This notion is strongly B.E. Journals of Theoretical Economics, 5 1 (2005). supported by the Forbes 400 list of the W. Kopczuk, “Bequest and Tax 10 richest Americans. There are of course Planning: Evidence from Estate Tax W. Kopczuk and E. Saez, “Top Wealth many potentially confounding factors that Returns,” NBER Working Paper, forthcom- Shares in the United States, 1916–2000: can affect the number of women at the ing. Evidence from Estate Tax Returns,” NBER very top of the wealth distribution, such 2 W. Kopczuk and J. Slemrod, “The Impact Working Paper No. 10399, March 2004, as bequests to widows, changes in gender- of the Estate Tax on Wealth Accumulation and National Tax Journal, 57 (2004), pp. 445–88. specific mortality and the age gap between and Avoidance Behavior of Donors,” NBER 11 spouses, community property rules and Working Paper No. 7960, October 2000 and T. Piketty and E. Saez, “Income tax treatment of married couples that we Rethinking Estate and Gift Taxation, W. Inequality in the United States: 1913– discuss in detail. We reach the conclusion G. Gale, J. R. Hines Jr., and J. B. Slemrod, 1998,” NBER Working Paper No. 8467, that the relative importance of self-made eds, Washington, D.C.: Brookings Institution September 2001, and Quarterly Journal of wealth in the twentieth century indeed Press (2001), pp. 299–343. Economics, 118 (2003), pp. 1–39. followed a U-shaped pattern: it decreased 3 W. Kopczuk, “Tax Bases, Tax Rates, 12 J. K. Scholz, “Wealth Inequality and the

16 NBER Reporter Spring 2006 14 Wealth of Cohorts,” May 2003, University “Earnings Mobility in the United States, L. Edlund and W. Kopczuk, “Women, of Wiconsin, mimeo. 1937–2004: Evidence from Social Security Wealth and Mobility,” NBER Working 13 W. Kopczuk, E. Saez, and J. Song, Administration Data,” work in progress. Paper forthcoming.

Rules for Monetary Policy

Michael Woodford*

Much of my recent research has modern macroeconomic analysis to be familiar classes of sticky-price dynamic sought to use economic analysis to deter- brought to bear on the evaluation of stochastic general equilibrium (DSGE) mine the consequences of alternative stabilization policies, in the context of models — models that incorporate key rules for the conduct of monetary policy, models with sufficient claim to quanti- elements of the current generation of and to formulate rules that will be desir- tative realism to be of interest to policy- empirical models of the monetary trans- able from the standpoint of individual making institutions. My own work has mission mechanism, and even some rela- welfare. Interest in the study of mon- sought to extend the analysis of optimal tively small complete macro models — it etary rules has increased over the past monetary policy rules in directions that is possible to show that the expected util- decade, for reasons having to do with bring the theoretical literature into clos- ity of the representative household varies progress in central banking and prog- er contact with the practical concerns of inversely with the expected discounted ress in macroeconomic theory. On the modern central bankers.1 value of a quadratic loss function, the one hand, many central banks — most arguments of which are measures of price notably, but not only, the “inflation tar- Inflation Stabilization and and wage inflation on the one hand and geting” banks — have increasingly come Welfare measures of real activity relative to a to organize their policy deliberations (time-varying) target level of activity on around an attempt to conform to specific One goal of my research has been to the other.2 Thus, it makes sense to rank targets or objectives, sometimes explic- clarify which kinds of macroeconomic alternative monetary policies according it quantitative targets. Moreover, cen- stabilization objectives best serve eco- to how well they stabilize (an appropri- tral banks worldwide have increased the nomic welfare. Grounding the objec- ate measure of) inflation on the one degree to which they discuss their deci- tives of policy in consumer welfare has hand, and how well they stabilize (an sions with financial market participants a number of advantages: one avoids the appropriate measure of ) the output gap and the general public, and this too has arbitrariness otherwise attendant upon on the other. The theory clarifies both increased the importance that the banks the choice of a particular definition of the appropriate definition of these stabi- assign to having a clear framework to “price stability,” “full employment” or lization objectives, and the appropriate guide their deliberations. At the same other conventional objectives. And, it relative weights to assign to them when a time, the development of a new gen- also makes possible a natural integration choice must be made between them. eration of quantitative macroeconomic of the theory of optimal monetary poli- The answer obtained depends, of models — that can be estimated using cy with the theory of optimal taxation. course, on the structure of the econo- macroeconomic time series and have Yet it is not immediately obvious what my.3 In particular, inflation variability optimizing foundations that allow an the conventional goals of monetary sta- reduces welfare because of the presence explicit evaluation of outcomes in terms bilization policy — especially the nearly of nominal rigidities; the precise nature of individual welfare — has allowed universal emphasis that central banks of these rigidities determines the appro- place on maintaining a low and stable priate form of the inflation-stabiliza- inflation rate — have to do with con- tion objective. For example, if wages are * Woodford is a Research Associate in the sumer welfare; after all, the arguments flexible (or there are efficient contracts NBER’s Programs on Monetary Economics of household utility functions gener- in the labor market), and price adjust- and Economic Fluctuations and Growth and the John Bates Clark Professor of Political ally are assumed to be the quantities of ments are staggered in the way assumed Economy at Columbia University. His profile various goods and services, but not their in the popular specification proposed by appears later in this issue. prices. Nonetheless, I have shown that in Guillermo Calvo4 (with an equal prob-

NBER Reporter Spring 2006 7 ability of any given price being revised for seeking to stabilize an appropriately ond-order perturbation expansions as an in any time period), then inflation varia- defined measure of “core” inflation rath- approximate characterization of equilib- tion results in distortions caused by the er than an equally weighted price index. rium outcomes under a given rule. misalignment of prices that are adjusted Pierpaolo Benigno has used reasoning of But Benigno and I have shown that at different times. The resulting welfare this kind to argue that a monetary union it is possible, in the case of quite a losses are proportional to the expected would maximize welfare by seeking to broad class of optimal policy problems discounted sum of squared deviations stabilize an index that does not weight in DSGE models, to find a quadratic of the inflation rate from zero. Other the different countries’ inflation rates loss function that correctly approximates assumptions about the timing of price strictly in proportion to the size of their expected utility under any policy, yet adjustments also imply that inflation economies,11 as is true of the inflation involves no non-zero linear terms. In variations reduce welfare, but with a dif- measure used in the European Central that way, welfare can be evaluated to ferent form of loss function, and thus a Bank’s definition of its price stability second order using only a first-order different ranking of equilibria in which objective. Similarly, if wages are sticky (log-linear) solution for the equilibrium prices are not completely constant. For as are goods prices, as implied by many under a candidate policy.14 Essentially, example, if the probability of adjust- empirical DSGE models, then instabil- our method incorporates into the loss ment of an individual price is increas- ity in the rate of growth of a broad index function itself the second-order effects ing in the time since that price was last of nominal wages results in distortions of stabilization policy on the average reviewed — a specification that is both similar to those created by variations in levels of endogenous variables in a sec- intuitively plausible and more consistent goods price inflation. If wages are stag- ond-order perturbation solution of the than the simple Calvo specification with gered in accordance with the Calvo spec- model. This allows us to consider how empirical models of inflation dynam- ification, then the welfare-theoretic loss the existence of steady-state distortions ics5 — then welfare losses are proportion- function includes a term proportional to (attributable either to market power or, al to a discounted sum of squared devia- the squared rate of goods price inflation more importantly, to taxes) affects the tions of the current inflation rate from and another term proportional to the relative weights that should be placed a moving average of recent past infla- squared rate of wage inflation each peri- on alternative stabilization objectives. tion rates, rather than deviations from od. In this case, optimal policy involves a Under the specifications that we regard zero.6 The goal of policy then should tradeoff between inflation stabilization, as most empirically realistic, the impor- be to keep inflation from differing too nominal wage growth stabilization, and tance of inflation stabilization relative to greatly from the current “inertial” rate output-gap stabilization, as first shown output-gap stabilization is increased the of inflation, which implies that inflation by Chris Erceg, Dale Henderson, and more distorted is the economy’s steady- should not be reduced too abruptly if it Andy Levin.12 state level of output; this is because has been allowed to exceed its optimal Analysis of these questions has stabilization of inflation does more to long-run level.7 A similar conclusion is required careful consideration of the increase the average level of output than obtained if prices are assumed to be auto- conditions under which a linear-qua- does stabilization of output, and this matically indexed to a lagged price index, dratic (LQ) stabilization policy problem consideration is more important for wel- as in the well-known empirical model (minimization of a quadratic loss func- fare the more sub-optimal is the steady- of Christiano, Eichenbaum, and Evans8 tion subject to constraints that repre- state level of output.15 and related studies, or if some prices are sent the log-linearized structural rela- adjusted in accordance with a backward- tions of a DSGE model) yields a correct Expectations and looking “rule of thumb,” as proposed in local approximation to optimal policy Optimal Policy the empirical model of inflation dynam- in the exact DSGE model. In fact, it ics of Jordi Gali and Mark Gertler.9 is not generally sufficient that the loss My research has emphasized that, The theory also provides important function be a correct quadratic local when choosing a policy to best serve the insights into the question of which price approximation to household utility — if goal of stabilization, it is crucial to take index or indexes it is more important that local approximation involves non- account of the effects of the policy’s sys- to stabilize. Again, the answer depends zero linear terms, then a correct second- tematic component on people’s expecta- on the nature of the nominal rigidities. order approximation to utility cannot be tions of future policy. For this reason, If prices are adjusted more frequently obtained by substituting into the approx- my work has focused largely on the study in some sectors of the economy than in imate objective a solution for the equi- of policy rules: this forces one to think others, then the welfare-theoretic loss librium under a given policy that is accu- about the systematic patterns that one function puts more weight on varia- rate only to first order.13 For this reason, can expect to be anticipated by sufficient- tions in prices in the sectors where prices much of the recent literature seeking to ly sophisticated market participants. are stickier, as first shown by Kosuke evaluate policy rules in DSGE models Taking account of the effects of sys- Aoki.10 This provides a theoretical basis has found it necessary to compute sec- tematic policy on policy anticipations

18 NBER Reporter Spring 2006 has important consequences for the con- variables often are directed to choose bility of a similar situation arose in the clusions one reaches about optimal poli- among alternative possible scenarios on United States, the Fed undertook a bold cy, some of which are counter-intuitive. the basis of a purely forward-looking cri- experiment with policy signaling, com- One fairly general result is that optimal terion. But such an approach may lead mitting to maintain a low federal funds policy will not be purely forward-look- to time-inconsistent choices, and even rate “for a considerable period” as a sub- ing; that is, it will not depend solely upon when it does not, it will almost inevitably stitute for further interest-rate cuts. This what can be achieved with respect to the lead to policy that is insufficiently iner- seems to have dissipated the market anxi- stabilization objectives now, or in the tial.18 An optimal outcome can in fact ety about premature tightening that had future, but also on past conditions that often be achieved through a procedure threatened to derail the U.S. recovery.22 no longer affect what is currently possible focused on ensuring that projections sat- A possible objection to advice of this to achieve. A history-dependent policy isfy an appropriate target criterion at all kind is that theoretical analyses of opti- can improve stabilization outcomes, to times, but the criterion should be histo- mal policy that assume a rational expec- the extent that it is correctly anticipat- ry-dependent. The acceptable transition tations equilibrium consistent with what- ed, by changing people’s expectations path along which the inflation rate and ever kind of systematic policy is adopted about subsequent policy at the time that output gap should be projected to return exaggerate the degree of precision with economic disturbances occur. And, an to their medium-term target levels will which a central bank can expect to con- appropriate shift in expectations often depend on recent past conditions.19 trol the expectations of market partici- can mitigate the degree to which the dis- Purely forward-looking policy can be pants simply by disciplining its own pro- turbances interfere with macroeconomic especially harmful when the zero lower cedures. In recent work, I have sought to stability.16 bound on short-term nominal interest relax this assumption by assuming instead For example, I have shown that when rates is reached, as in Japan for the past only that the central bank should expect one takes account of forward-looking several years, and as some feared could that private-sector expectations about behavior, it can be desirable for a central occur in the United States in 2003. When the future evolution of the economy will bank to only gradually adjust its oper- the zero bound is reached, further mon- not be too far from model-consistency, as ating target for overnight interest rates etary stimulus is possible only by shifting measured by a relative-entropy criterion when underlying fundamentals change, expectations about future policy. But if (which ensures that the public will not rather than jumping immediately to a policy is expected to be conducted in a believe in patterns that they should be new level that depends only on current purely forward-looking way in the future, able to reject on the basis of even short conditions. This kind of policy iner- then there will be no reason for the pub- time series). One can then characterize tia — often argued to characterize actu- lic to expect looser policy in the future the optimal policy decision of the cen- al central bank behavior, but frequently simply because the zero bound currently tral bank if it wishes to choose a robust assumed to indicate a failure of central prevents interest rates from being cut policy — one that is not too bad even bankers to fully optimize — can reduce as sharply as would be needed to create under the worst of the outcomes that can the amplitude of the swings in short-term demand in line with the economy’s pro- occur under “near-rational expectations.” interest rates required to stabilize infla- ductive capacity. Gauti Eggertsson and My analysis shows that the qualitative tion and real activity in response to real I have shown that this can result in a conclusions of the rational-expectations disturbances. It allows market partici- protracted and severe deflationary con- analysis of optimal policy continue to pants to anticipate that the movements traction, even when the same real funda- apply. For example, policy commitment in short rates that occur will be more mentals would be consistent with a much continues to be important — indeed, the persistent, resulting in a larger effect on more benign outcome in the case of alter- losses resulting from discretionary pol- long rates and other asset prices, which native policy expectations. A desirable icy are even greater in the case of allow- are what matter for the effect of policy on outcome requires advance commitment ance for near-rational expectations; and aggregate demand.17 Hence calls for cen- to a history-dependent policy, under optimal policy continues to be history- tral bankers to respond more promptly to which rates will be kept unusually low for dependent — in fact, even more history- changes in conditions in order to avoid a period of time even after fundamentals dependent than if the central bank could “getting behind the curve” may actually have recovered, even though higher rates count on the public’s having precisely be counter-productive. would be called for under the latter con- model-consistent expectations.23 Prescriptions for purely forward- ditions if one were determined to avoid looking policy in the name of optimiza- generating inflationary pressures.20 It is Optimal Target Criteria for tion also characterize many normative arguable that the Bank of Japan’s empha- Policy discussions of inflation-forecast target- sis (prior to 2001) on its determination ing. Central banks that base their inter- to end loose monetary policy as quickly One way of specifying a rule for the est-rate decision on projections of the as possible prolonged the Japanese defla- conduct of policy that has both practical future evolution of inflation and other tion unnecessarily.21 When the possi- and normative relevance is in terms of a

NBER Reporter Spring 2006 9 “target criterion” that the central bank amount that inflation should be allowed mal target criterion on model structure, is committed to ensure is satisfied (or at to increase will be strictly limited by the research of this kind cannot hope to least, projected to be satisfied) each time required proportionality between the derive a single rule that would represent its instrument setting is reviewed.24 The inflation projection and the projected a universally optimal policy prescription. criteria used by inflation-forecast target- output-gap change. After the real effects And in any event, even a minimally real- ing central banks, such as the Bank of of the disturbance dissipate, the rate at istic degree of complexity in one’s model England (which seeks to ensure that CPI which the output gap should be returned implies that a fully optimal criterion will inflation is always projected to reach to zero will be determined by the neces- be more complex than any principle for its target level of 2 percent per year at a sity of programming lower-than-average guiding policy deliberations that one horizon two to three years in the future), inflation during a period of output-gap can imagine actually being adopted at are an example of commitments of this growth. Anticipation of this kind of his- a central bank.28 Nonetheless, I believe kind. They represent the closest approxi- tory-dependent policy should restrain that the study of optimal target criteria mation to the ideal of rule-based policy- price increases during the period of high for fairly simple environments that cap- making yet observed. At the same time, costs, mitigating the temporary effect ture important features of more realistic target criteria often provide an especially of the shock on the available inflation/ models can suggest qualitative features convenient way of characterizing opti- output tradeoff at the cost of a slower of desirable target criteria. For example, mal policy. For example, it may be pos- recovery.26 And, because the projected one important conclusion from my study sible to specify optimal policy in this medium-term growth rate of the output of this topic is that an optimal target cri- way independently of the parameters gap will always be zero, a credible com- terion almost surely will not be focused governing the statistical properties of mitment to such a criterion would never so exclusively on projected outcomes the economic disturbances affecting the allow ambiguity about the medium-term two or more years in the future, as are economy; the target criterion is then a outlook for inflation, despite the exis- the criteria that currently are used at the particularly robust characterization of tence of transitory variations in the infla- leading inflation-targeting central banks, optimal policy. tion rate in response to shocks. at least according to their official rheto- Marc Giannoni and I have shown More complex (and realistic) eco- ric. In a realistic model, a commitment that in the case of a very general class nomic models imply that a more com- of this form is unlikely even to suffice of linear-quadratic policy problems, it plex target criterion would be needed to determine an appropriate short-term is possible to derive a target criterion to implement a fully optimal policy. For policy stance, in the absence of auxiliary that is robustly optimal in the sense example, if the likelihood of a price revi- assumptions such as a constant interest just described: a credible commitment sion increases with the time since the rate over the projection horizon, while to ensure that the criterion holds at all last revision, then the optimal target the forecast-targeting exercise is likely to times will implement an optimal equilib- criterion allows the short-run inflation be time-inconsistent with the addition of rium, regardless of the statistical prop- projection to be an increasing function such an assumption.29 An approach that erties of the various types of exogenous of recent past inflation Thus tempo- is both coherent and transparent would disturbances, as long as they are all addi- rary increases in inflation should not be instead require central banks to commit tive, mean-zero disturbances.25 The pre- immediately reversed. (Other sources of themselves in advance to clear criteria cise form of the optimal target criterion intrinsic inflation inertia, such as the for judging the acceptability of the tran- depends, however, on the non-stochastic kind of indexation commonly assumed sition paths along which an economy is part of the structural equations of one’s in current-vintage empirical DSGE mod- expected to return to its normal state fol- model of the transmission mechanism. In els, lead to a similar conclusion.) If wages lowing a disturbance. the case of a canonical “New Keynesian” and prices are sticky, then the optimal model, with an aggregate-supply relation target criterion involves projected nomi- 1 These developments are described of the kind implied by flexible wages and nal wage growth as well as projected in more detail in Interest and Prices: Calvo-style staggered pricing, the opti- goods price inflation. Foundations of a Theory of Monetary mal target criterion is a “flexible inflation Moreover, if a binding lower bound Policy, Princeton University Press, 2003. target,” under which short-run depar- on interest rates sometimes forces targets 2 This approach was first illustrated in tures of the inflation rate from a con- to be missed, then the target criterion in J. J. Rotemberg and M. Woodford, “An stant long-run target level should vary subsequent periods should be adjusted Optimization-Based Framework for the inversely with the projected growth in in proportion to the size of the target- Evaluation of Monetary Policy,” NBER the output gap. Such a criterion would ing errors. This would create the kind of Macroeconomics Annual 12: pp. 297– allow inflation to increase temporari- anticipations of history-dependent pol- 346 (1997). The general method is dis- ly in response to a positive cost-push icy that mitigate the distortions created cussed in “Inflation Stabilization and shock, for example, given the expected by the lower-bound constraint.27 Welfare,” NBER Working Paper No. 8071, decline in the output gap, although the Given the dependence of the opti- January 2001, and Contributions to

20 NBER Reporter Spring 2006 Macroeconomics 2(1), article 1 (2002). in J. Steinsson, “Optimal Monetary Policy 15 Benigno and Woodford, “Inflation The results of the latter paper are gener- in an Economy with Inflation Persistence,” Stabilization” [cited footnote 2]. alized in P. Benigno and M. Woodford, Journal of Monetary Economics 50: pp. 16 “Pitfalls of Forward-Looking Monetary “Inflation Stabilization and Welfare: The 1425–56 (2003). Policy,” American Economic Review Case of a Distorted Steady State,” NBER 10 K. Aoki, “Optimal Monetary Policy 90(2): pp. 100–4 (2000); and Interest Working Paper No. 10838, October 2004, Responses to Relative Price Changes,” and Prices [cited footnote 1], chapter 7. and Journal of the European Economics Journal of Monetary Economics 48: pp. 17 “Optimal Monetary Policy Inertia,” Association 3: pp.1185–236 (2005). 55–80 (2001). Aoki’s analysis is general- NBER Working Paper No. 7261, July 3 The results summarized here are dis- ized in Interest and Prices [cited footnote 1999; parts of this paper appear in cussed further in Interest and Prices [cited 1], section 4.3. revised form in “Optimal Interest-Rate footnote 1], chapter 6. 11 P. Benigno, “Optimal Monetary Smoothing,” Review of Economic Studies 4 G. A. Calvo, “Staggered Prices in a Policy in a Currency Area,” International 70:pp. 861–86 (2003). The desirability Utility-Maximizing Framework,” Journal Economic Review 44: pp. 195–222 of policy inertia is also analyzed in more of Monetary Economics 12: pp. 383–98 (1999). complex models in J. J. Rotemberg and (1983). 12 C. J. Erceg, D. W. Henderson, and M. Woodford, “Interest-Rate Rules in an 5 A. Wolman, “Sticky Prices, Marginal A. T. Levin, “Optimal Monetary Policy Estimated Sticky-Price Model,” NBER Cost, and the Behavior of Inflation,” Federal with Staggered Wage and Price Contracts,” Working Paper No. 6618, June 1998, Reserve Bank of Richmond Economic Journal of Monetary Economics 46: and in J. B. Taylor, ed., Monetary Policy Quarterly 85: pp. 29–48 (1999); R. pp. 281–313 (2000). This derivation Rules, University of Chicago Press for Mash, “Optimizing Microfoundations for is generalized in P. Benigno and M. NBER, 1999; and in M. P. Giannoni and Inflation Persistence,” Oxford University Woodford, “Optimal Stabilization Policy M. Woodford, “How Forward-Looking Department of Economics discussion paper when Wages and Prices are Sticky: The is Optimal Monetary Policy?” Journal of no. 183, January 2004; K. D. Sheedy, Case of a Distorted Steady State,” NBER Money, Credit and Banking 35(6–2): “Structural Inflation Persistence,” working Working Paper No. 10839, October 2004, pp. 1425–69 (2003). paper, Cambridge University, November and in J. Faust, A. Orphanides, and D. 18 “Commentary: How Should Monetary 2005. Riefschneider, eds., Models and Monetary Policy be Conducted in an Era of Price 6 K. D. Sheedy, “Resistance to Persistence: Policy, Federal Reserve Board, 2005. Stability?” in Federal Reserve Bank of Optimal Monetary Policy Commitment,” 13 For further discussion of the general Kansas City, New Challenges for working paper, Cambridge University, problem and a demonstration of the pit- Monetary Policy, 1999. November 2005. falls of “naïve” LQ approximation, see 19 L. E. O. Svensson and M. Woodford, 7 This is not the conclusion that Sheedy Interest and Prices [cited footnote 1], sec- “Implementing Optimal Policy through draws from his loss-function derivation in tion 6.1, and P. Benigno and M. Woodford, Inflation-Forecast Targeting,” NBER the paper cited in footnote 6. For my own “Optimal Taxation in an RBC Model: Working Paper No. 9747, June 2003, analysis of the consequences of intrinsic A Linear-quadratic Approach,” NBER and in B.S. Bernanke and M. Woodford, inflation inertia, see Interest and Prices Working Paper No. 11029, January eds., The Inflation Targeting Debate, [cited footnote 1], section 7.1; and M. P. 2005. University of Chicago Press for NBER, Giannoni and M. Woodford, “Optimal 14 Our method was first introduced in 2005. Inflation Targeting Rules,” NBER Working P. Benigno and M. Woodford, “Optimal 20 G. B. Eggertsson and M. Woodford, Paper No. 9939, September 2003, and Monetary and Fiscal Policy: A Linear- “The Zero Bound on Interest Rates and in B. S. Bernanke and M. Woodford, Quadratic Approach,” NBER Working Optimal Monetary Policy,” Brookings eds., The Inflation Targeting Debate, Paper No. 9905, August 2003, and Papers on Economic Activity 2003–1: University of Chicago Press for NBER, NBER Macroeconomics Annual 18: pp. pp. 139–211. 2005. 271–333 (2003). It is also illustrated 21 More recently, the BOJ has consciously 8 L. J. Christiano, M. Eichenbaum, and in the Benigno and Woodford papers sought to signal an intention not to tight- C. Evans, “Nominal Rigidities and the cited in footnotes 2, 12, and 13 above. en prematurely, at least partially along Dynamic Effects of a Shock to Monetary A general algorithm for the application the lines argued for by Eggertsson and Policy,” NBER Working Paper No. 8403, of this method to the derivation of LQ myself; see, for example, K. Ueda, “The July 2001, and Journal of Political approximations to policy problems is dis- Bank of Japan’s Struggle with the Zero Economy 113: pp. 1–45 (2005). cussed and illustrated in F. Altissimo, V. Lower Bound on Nominal Interest Rates: 9 J. Gali and M. Gertler, “Inflation Curdia, and D. Rodriguez Palenzuela, Exercises in Expectations Management,” Dynamics: A Structural Econometric “Linear-Quadratic Approximation to CIRJE discussion paper, University of Analysis,” Journal of Monetary Economics Optimal Policy: An Algorithm and Two Tokyo, September 2005. 44: pp. 195–222 (1999). A welfare-theo- Applications,” working paper, European 22 For discussions of this episode, see retic loss function is derived for this model Central Bank, September 2005. my “Central-Bank Communication and

NBER Reporter Spring 2006 21 Policy Effectiveness,” NBER Working 24 An important early discussion is by L. [cited footnote 7]. Paper No. 11898, December 2005; and E. O. Svensson, “Inflation Forecast Tar- 27 Eggertsson and Woodford [cited foot- B. S. Bernanke, V. R. Reinhart, and B. P. geting: Implementing and Monitoring note 20]. Sack, “Monetary Policy Alternatives at the Inflation Targets,” European Economic 28 See, for example, the optimal target cri- Zero Bound: An Empirical Assessment,” Review 41:pp. 1111–46 (1997). terion derived for a small empirical model Brookings Papers on Economic Activity 25 M. P. Giannoni and M. Woodford, in Giannoni and Woodford [cited footnote 2004–1: pp. 1–78. “Optimal Interest-Rate Rules: I. General 7]. 23 “Robustly Optimal Monetary Policy Theory,” NBER Working Paper No. 9419, 29 “Inflation Targeting and Optimal under Near-Rational Expectations,” January 2003. Monetary Policy,” Federal Reserve Bank NBER Working Paper No. 11896, 26 Both this paragraph and the following of St. Louis Economic Review, July/ December 2005. one are based on Giannoni and Woodford August 2004, pp. 15–41.

NBER Profile: Christopher Carroll Christopher Carroll is an NBER Research In addition to articles in economics jour- Associate in the Programs on Monetary nals, he has written Encyclopedia Britannica Economics and Economic Fluctuations and articles on consumption-related topics. He Growth and Professor of Economics at the was also the recipient of an Alfred P. Sloan Johns Hopkins University in Baltimore. Foundation early career fellowship and of Originally from Knoxville, Tennessee, he the TIAA-CREF/Samuelson Certificate of received his A.B. in Economics from Harvard Excellence for his work on precautionary sav- University in 1986 and his Ph.D. from the ing behavior over the life cycle. MIT in 1990. Carroll’s recent research has focused on After graduating from MIT, he worked models of the dynamics of expectations for- at the Federal Reserve Board in Washington, mation, particularly on how expectations where his responsibilities included prepa- reflect households’ learning from each other ration of the forecast for consumer expen- and from experts. This focus flows from a diture. He moved to the Johns Hopkins career-long interest in consumer sentiment University in 1995, but spent 1997–8 work- and its determinants. He is also an associ- ing at the President’s Council of Economic ate editor at the Review of Economics and Advisers in Washington; there, his respon- Statistics and the Journal of Business and sibilities included analysis of Social Security Economic Statistics. reform proposals, tax and pension policy, and Carroll lives in Columbia, Maryland bankruptcy reform. with his wife, Jennifer Manning. His inter- Carroll’s research has focused primarily est in economics dates from early in his life: on consumption and saving behavior, with his father recently retired from a career as an emphasis on reconciling the empirical evi- Professor of Economics at the University dence from both microeconomic and mac- of Tennessee, where he taught Industrial roeconomic sources with theoretical models. Organization for many years.

22 NBER Reporter Spring 2006 NBER Profile: Wojciech Kopczuk

Wojciech Kopczuk is a Faculty Research and — somewhat less proudly, but jointly Fellow in the NBER’s Public Economics with Joel Slemrod — of the 2001 IgNobel Program. He is also an Associate Professor Prize in Economics. His research interests of Economics and International and Public include behavioral responses to income and Affairs at Columbia University. estate taxation, tax avoidance and administra- Kopczuk received his M.Sc. in Computer tion, optimal taxation and tax systems, evolu- Science from Warsaw University (Poland) in tion of wealth and income inequality, and the 1996 and his Ph.D. in Economics from the nature of intergenerational linkages. University of Michigan in 2001. Before mov- Kopczuk lives in New York City with his ing to Columbia University in 2003, he was wife, Joanna, and son, Kuba. When he used to an assistant professor at the University of have free time, he wasted it playing the game British Columbia. of “Go,” tinkering with his Microsoft-free Kopczuk was the recipient of an computer, and listening to music described by Alfred P. Sloan Research Fellowship (2005) one of his colleagues as “head-banging”.

NBER Profile: Michael Woodford

Michael Woodford is a Research Central do Brasil, and as Professorial Fellow Associate in the NBER’s Programs on in residence at the Reserve Bank of New Economic Fluctuations and Growth and Zealand. He is a Fellow of the American Monetary Economics and the John Bates Academy of Arts and Sciences and of the Clark Professor of Political Economy at Econometric Society. Columbia University. He received his A.B. Woodford’s research has been pub- from the University of Chicago, his J.D. from lished in all of the major economic journals. Yale Law School, and his Ph.D. in economics His 2003 Interest and Prices: Foundations from MIT. of a Theory of Monetary Policy, published Woodford began his academic career by Princeton University Press, won the at Columbia in 1984 and moved to the Association of American Publishers Award University of Chicago in 1986. He became for Best Professional/Scholarly Book in a full professor at Chicago in 1992. In 1995, Economics. His most recent book, The Woodford left for Princeton, where he served Inflation Targeting Debate, co-edited with as the Harold H. Helm ’20 Professor of Federal Reserve Chairman Ben S. Bernanke, Economics and Banking until 2004. He has was published by the University of Chicago been at Columbia since 2004. Press in 2005. Woodford has been a Visiting Professor Woodford lives in Manhattan with his in universities throughout the United States wife, Argia Sbordone, an economist at the and in several foreign countries. He is a Federal Reserve Bank of New York, and their Consultant to the Federal Reserve Bank of daughter, Antonia. He enjoys running, mov- New York, and also has served as a Consultant ies, and off-Broadway theater. to the European Central Bank and the Banco

NBER Reporter Spring 2006 23 Conferences

Seventh Annual Conference in India On January 15-17, 2006 the NBER The U.S. participants were: David Maryland; and David Weil, NBER and and India’s National Council for Applied Autor, Esther Duflo, and Glenn Ellison, Brown University. Economic Research (NCAER) again NBER and MIT; Jagdish Bhagwati, After introductory remarks about the brought together a group of NBER econ- Columbia University; Mihir A. Desai, U.S. and Indian economies by NBER omists and about thirty economists from Martin S. Feldstein, and Robert T. Jensen, President Feldstein and Bimal Jalan of Indian universities, research institutions, NBER and Harvard University; Daniel NCAER, the participants discussed: glo- and government departments for their Kessler, NBER and Stanford University; balization; growth; trade; financial poli- seventh annual conference in India. Mihir Anne O. Krueger and Raghuram Rajan, cies; social infrastructure; energy and reg- A. Desai and Martin S. Feldstein, NBER on leave from the NBER at the IMF; ulation; and taxation and public finance. and Harvard University, organized the Arvind Subramanian, IMF; Richard H. Partial support for this project conference jointly with Suman Bery of Thaler, NBER and University of Chicago; comes from the Institute for Financial NCAER. John Wallis, NBER and University of Management and Research in India.

Bureau News

NBER Researchers Head to DC Edward P. Lazear, an NBER Research of Business and a Senior Fellow at the Associate in the NBER’s Programs on Associate in the Labor Studies, Aging, and Hoover Institution. The other two cur- Corporate Finance and the Development Education Programs who has been affiliat- rent members of the President’s Council of the American Economy. Kroszner ed with the NBER since 1978, is the new of Economic Advisers are also former will serve under Bernanke, who is a for- Chairman of the President’s Council of NBER researchers: Matthew Slaughter mer Director of the NBER’s Program on Economic Advisers. As such, he succeeds and Katharine Baicker. Monetary Economics and was a professor several NBER Research Associates who In a related development, Randall S. at Princeton. Bernanke was confirmed by have held that position: Martin Feldstein, Kroszner is a new member of the Federal the Senate on January 31 and sworn in Michael Boskin, Joseph Stiglitz, R. Glenn Reserve Board of Governors. A former the following day; his four-year term as Hubbard, N. Gregory Mankiw, and Ben professor of economics at the University Fed Chairman ends January 31, 2010, and S. Bernanke. of Chicago’s Graduate School of Business, his 14-year term as a Fed member ends Lazear is a Professor of Economics Kroszner has been affiliated with the January 31, 2020. at Stanford University’s Graduate School NBER since 1998. He was a Research

Velasco to Become Chile’s Finance Minister Andres Velasco, a Research Associate School of Government, has been picked ments of Chile’s new President, Michelle in the NBER’s Program on International as Chile’s new finance minister. Velasco, Bachelet. Finance and Macroeconomics and Profes­ who has been affiliated with the NBER sor of Economics at Harvard’s Kennedy since 1994, was among the first appoint-

24 NBER Reporter Spring 2006 NBER Announces 2006 Nonprofit Fellowships

NBER Research Associate James M. ed the award recipients from a pool of 24 will examine the inter-year dynamics of Poterba has announced the recipients applicants. household charitable giving. of NBER Fellowships for the Study of The four graduate students who will Faculty grants will go to: Ginger Nonprofit Institutions for the 2006– receive fellowship support are: Jillian Jin, University of Maryland, for work 7 academic year. This NBER fellow- Berk, Brown University, who will ana- on “College and University Reactions ship program is designed to encourage lyze non-profit organizations that man- to the Publication of School Rankings,” research on nonprofit institutions by age prisons; Brianna Briggs, University a project on how colleges adapt their NBER Research Associates and Faculty of California, Santa Barbara, who will behavior in response to published data Research Fellows, and by graduate stu- study faith-based organizations pro- that ranks schools on various dimen- dents in economics working closely with viding welfare services in Los Angeles; sions; and Sendhil Mullainathan, them through support of dissertation Yee Wai Chong, Stanford University, Harvard University, for “Psychological research on the same subject. A selec- who will investigate the business activi- Drivers of Giving,” a project on how tion committee consisting of Charles ties of non-profit organizations in the marketing strategies, framing, and other Clotfelter, David Cutler, Steven Levitt, arts and other fields; and Marit Rehavi, factors affect the success of charitable A. Mitchell Polinsky, and Poterba select- University of California, Berkeley, who fund-raising.

Indian Economy The NBER’s Working Group on the Evidence from India” Eric Edmonds and Nina Pavcnik, Indian Economy met in Cambridge on Kalpana Kochhar, Utsav Kumar, Dartmouth College and NBER, and December 16, 2005. Working Group Arvind Subramanian, and Petia Topalova, Yale University, “ Director Mihir A. Desai of Harvard Ioannis Tokatlidis, International Trade Liberalization, Child Labor, and University, organized this program: Monetary Fund, and Raghuram Schooling: Evidence from India” Rajan, International Monetary Abhijit Banerjee, MIT; Shawn Cole, Fund and NBER, “India’s Pattern of Kaivan Munshi, Brown University and Harvard University; Esther Duflo, Development: What Happened, What NBER, and Mark Rosenzweig, Yale MIT and NBER; and Leigh Linden, Follows” University, “Why is Mobility in India Columbia University, “Remedying so Low? Social Insurance, Inequality, Education: Evidence from Two Siddhartha G. Dastidar, Columbia and Growth” Randomized Experiments in India” University; Raymond Fisman, Columbia University and NBER; and Marianne Bertrand, University of Esther Duflo, and Rema Hanna, New Tarun Khanna, Harvard University, “ Chicago and NBER; Simeon Djankov, York University, “Monitoring Works: Limits to Policy Reversal: Privatization The World Bank; Rema Hanna; and Getting Teachers to Come to School” in India” Sendhil Mullainathan, MIT and NBER, “ Is Corruption Efficient? Jagadeesh Sivadasan, University I. Serdar Dinc, University of Getting a Driving License in New of Michigan, and Joel Slemrod, Michigan, and Nandini Gupta, Indiana Delhi” University of Michigan and NBER, University, “ The Decision to Privatize: “Tax Law Changes, Income Shifting, The Role of Political Competition and and Measured Wage Inequality: Patronage”

Many efforts to improve school tion in urban slums. A remedial educa- of all children in treatment schools by quality by adding school resources have tion program hired young women from 0.14 standard deviations in the first year, proven to be ineffective.Banerjee, Cole, the community to teach basic litera- and 0.28 in the second year, relative to Duflo, and Linden present the results of cy and numeracy skills to children lag- comparison schools. A computer-assist- two experiments conducted in Mumbai ging behind in government schools. The ed learning program provided each child and Vadodara, India, designed to evalu- authors find the program to be very in the fourth standard with two hours of ate ways to improve the quality of educa- effective: it increased average test scores shared computer time per week, in which

NBER Reporter Spring 2006 25 students played educational games that higher than in the comparison schools rather than labor-intensive manufactur- reinforced mathematics skills. The pro- and children were 40 percent more likely ing, and industries with typically higher gram was also very effective, increasing to be admitted into regular schools. average scale. Kochhar, Kumar, Rajan, math scores by 0.36 standard deviations Changes to the income tax law relat- Subramanian, and Tokatlidis show that the first year, and 0.54 the second year. ing to wages and profits could provide some of these distinctive patterns exist- These results were not limited to the incentives for shifting income from wages ed even prior to the beginning of eco- period in which students received assis- to profits (or vice versa), with important nomic reforms in the 1980s, and argue tance, but persisted for at least one year implications for measured wage inequal- that they stem from the idiosyncratic after leaving the program. Two instru- ity. Sivadasan and Slemrod use a large policies adopted soon after India’s inde- mental variable strategies suggest that dataset covering all registered plants in pendence. The authors then look to the while remedial education benefited the the manufacturing sector in India (over future, using the growth of fast-moving children who attended the remedial the period 1986 to 1995) to examine Indian states as a guide. Despite recent classes, their classmates, who did not the effects of an income tax law change reforms that have removed some of the attend the remedial courses but did expe- (effective from1 992) that eliminated the policy impediments that might have sent rience smaller classes, did not post gains, double taxation of wages paid to partners India down its distinctive path, it appears confirming that resources alone may not in partnership firms. They find an imme- unlikely that India will revert to the pat- be sufficient to improve outcomes. diate and pervasive response by partner- tern followed by other countries. In the rural areas of developing ship firms to the tax law change, reflect- Dastidar, Fisman, and Khanna countries, teacher absence is a wide- ed in a significant shifting of income examine the effect of regime change on spread problem. Duflo and Hanna test from profits to managerial wages. This privatization using the 2004 election sur- whether a simple incentive program shift has important implications for mea- prise in India. In that election, the pro- based on teacher presence can reduce sured wage inequality, since about half reform BJP was unexpectedly defeated teacher absence, and whether it has the of registered manufacturing plants are by a less reformist coalition. Government potential to lead to more teaching activi- incorporated in the form of partnerships controlled companies that were being ties and better learning. In sixty infor- (including most family-run business- studied for complete privatization by the mal one-teacher schools in rural India, es). They find a significant jump in the BJP dropped by 7.5 percent relative to randomly chosen out of 120 (the treat- mean and median relative wage of skilled private firms. By contrast, government ment schools), a financial incentive pro- workers (which includes managers and controlled firms that were not being con- gram was initiated to reduce absentee- partners) following the tax law change in sidered for privatization, or firms that ism. Teachers were given a camera with 1992. This sudden increase in measured had already been fully privatized, did a tamper-proof date and time function, wage inequality follows major trade lib- not experience a significant drop relative along with instructions to have one of eralization and deregulation reforms to private firms. Firms that the BJP had the children photograph the teacher and announced in July 1991. They also find slated for definite future privatization other students at the beginning and end that the income shifting induced by the experienced intermediate declines of of the school day. The time and date tax law change explains almost all of approximately 3.5 percent. The authors stamps on the photographs were used the observed increase in measured wage interpret this as evidence consistent with to track teacher attendance. A teach- inequality; this finding is robust to inclu- investor belief of policy irreversibility in er’s salary was a direct function of his sion of controls for a number of other privatization, where reforms may reach attendance. The remaining sixty schools potential sources of post-liberalization a “point of no return” beyond which served as comparison schools. The intro- increases in wage inequality. Once they future regimes have difficulty reversing duction of the program resulted in an control for income shifting, they find a those policies. Taking advantage of an immediate decline in teacher absence. much more moderate increase in wage “intermediate event” where policies were The absence rate (measured using unan- inequality, in line with the pre-1991 expected to be more heavily influenced nounced visits both in treatment and trends. These results show that income by the communist party, the authors comparison schools) changed from an shifting responds strongly to tax incen- still find evidence consistent with policy average of 42 percent in the comparison tives, highlighting the need to control irreversibility. schools to 22 percent in the treatment for the potential effects of tax incentives Dinc and Gupta investigate the role schools. When the schools were open, in studies of wage inequality. of political competition and patron- teachers were as likely to be teaching in India seems to have followed an age in the privatization of government- both types of schools, and the number of idiosyncratic pattern of development, owned enterprises by using a unique students present was roughly the same. certainly compared to other fast-growing firm-level dataset from India. They find The program positively affected child Asian economies. While the emphasis on that the government is reluctant to priva- achievement levels: a year after the start services rather than manufacturing has tize firms located in regions where the of the program, test scores in program been widely noted, within manufactur- ruling party faces more political compe- schools were 0.17 standard deviations ing India has emphasized skill-intensive tition from parties in opposition. They

26 NBER Reporter Spring 2006 also find that no government-owned trate how even temporary adjustment In the speed group, individuals were enterprise located in the home state of costs may have long-term effects on the given bonuses for getting their licenses the politician in charge of that enterprise affected. faster, while in the lesson group, they is ever privatized. These results are robust Munshi and Rosenzweig examine were given free driving licenses. Moreover, to firm-level characteristics such as prof- the hypothesis that the persistence of the authors performed a surprise driving itability, size, and the influence of labor low spatial and marital mobility in rural test at the end of the study to produce a groups; to industry and time effects; and India, despite increased growth rates and reliable measure of skill. They document to state-level differences in income, edu- rising inequality in recent years, is attrib- several revealing facts about corruption. cation, and urbanization. utable to the existence of sub-caste net- First, consistent with theories of endoge- Few issues are more controversial in works that provide mutual insurance to nous red tape, the bureaucratic system the contemporary globalization debate their members. Using unique panel data shows a great deal of arbitrariness. While than the effects of trade liberalization on providing information on caste loans the average person took 30 days to get poverty and well-being in low-income and sub-caste identification, the authors the license, others took four times as countries. The question of how changes show that households that out-marry or long. Similarly, bureaucrats automatical- in trade policy affect child labor and migrate lose the services of these net- ly failed some on their driving exam, schooling is particularly contentious. works, which dampens mobility when many of whom were skilled drivers, in a Edmonds, Pavcnik, and Topalova alternative sources of insurance or finance way that was unrelated to driving skills. study the relationship between changes of comparable quality are unavailable. At Second, the system was very responsive in trade policy and schooling and child the aggregate level, the networks appear to individual needs. Those who wanted labor using detailed household level data to have coped successfully with the rising the license faster (for example, the speed from the Indian National Sample Survey inequality within sub-castes that accom- group), were able to get it faster. Third, (NSS) spanning the period of trade liber- panied the Green Revolution. Indeed, the system was quite insensitive to social alization initiated in 1991. They explore this increase in inequality lowered over- needs. Sixty percent of those who got the the causal link between liberalization and all mobility, which was low to begin license were rated as “automatic failures” changes in child labor by relating child with, even further. These results suggest on the authors’ driving test. Tellingly, the labor to district and intertemporal varia- that caste networks will continue to group that was given lessons was only tion in exposure to tariff cuts. During smooth consumption in rural India for slightly more likely to be able to get the the time period of this study, India expe- the foreseeable future, as they have for license than the control group and less rienced dramatic declines in child labor centuries. likely than the speed group. Similarly, and increases in schooling attendance. To better understand corruption, the speed group got their license faster However, the authors find that children Bertrand, Djankov, Hanna, and by simply paying, not by becoming bet- living in districts more exposed to tariff Mullainathan carefully follow about ter drivers. These findings reject the cuts observed smaller declines in child 800 individuals through the process of “grease-the-wheels” view of corruption, labor and smaller increases in school getting a driving license in New Delhi, where bribes enhance efficiency. In short, attendance. They believe the findings India. They also generate exogenous vari- corruption produces bad drivers. reflect some of the adjustment costs asso- ation by randomly assigning people into ciated with trade liberalization and illus- three groups: control, speed, and lesson.

NBER Reporter Spring 2006 27 Economic Fluctuations and Growth The NBER’s Program on Economic University of Oslo; and Giovanni Home Bias Puzzle” Fluctuations and Growth met at the L. Violante, New York University, Discussant: Pierre-Olivier Gourinchas, Federal Reserve Bank of San Francisco “Insurance and Opportunities: The University of California, Berkeley and on February 3. NBER Research Welfare Implications of Rising Wage NBER Associates Steven J. Davis, University of Dispersion” Chicago, and Charles I. Jones, University Discussant: Orazio Attanasio, Lars Ljungqvist, Stockholm School of of California, Berkeley, organized this University College London and NBER Economics, and Thomas J. Sargent, program: New York University and NBER, “Jobs Emmanuel Farhi, MIT, and Ivan and Unemployment in Macroeconomic Mark Aguiar, Federal Reserve Bank of Werning, MIT and NBER, “Inequality, Theory:A Turbulence Laboratory” Boston, and Erik Hurst, University of Social Discounting, and Progressive Discussant: Robert E. Hall, Stanford Chicago and NBER, “Lifecycle Prices Estate Taxation” University and NBER and Production” Discussant: Emmanuel Saez, University Discussant: Richard Rogerson, Arizona of California, Berkeley and NBER Robert Shimer, University of Chicago State University and NBER and NBER, “Mismatch” Stijn Van Nieuwerburgh and Laura Discussant: Ricardo Lagos, New York Jonathan Heathcote, Georgetown Veldkamp, New York University, University University; Kjetil Storesletten, “Information Immobility and the

Using scanner data and time diaries, empirical patterns quite well. Taken a calibration exercise, the authors find Aguiar and Hurst document how house- together, these results highlight the dan- that the observed rise in wage dispersion holds substitute time for money through ger of interpreting lifecycle expenditure in the United States over the past three shopping and home production. They without acknowledging the changing decades implies a welfare loss roughly find that there is substantial heteroge- demands on time and the available mar- equivalent to a 2.5 percent decline in neity in prices paid across households gins of substituting time for money. lifetime consumption. This number is for identical consumption goods in the Heathcote, Storesletten, and the net effect of a welfare gain of around same metro area at any given point in Violante provide an analytical charac- 5 percent from an endogenous increase time. For identical goods, prices paid are terization of the welfare effects of chang- in labor productivity, coupled with a loss highest for middle aged, rich, and large es in cross-sectional wage dispersion, of around 7.5 percent associated with households, consistent with the hypoth- using a class of tractable heterogeneous- greater dispersion in consumption and esis that shopping intensity is low when agent economies with various insurance leisure. the cost of time is high. The data suggest market structures. They express welfare To what degree should societies that a doubling of shopping frequency effects both in terms of changes in the allow inequality to be inherited? What lowers the price paid for a given good observable joint distribution over indi- role should estate taxation play in shap- by approximately 10 percent. From this vidual wages, consumption and hours, ing the intergenerational transmission elasticity and observed shopping inten- and in terms of the underlying param- of welfare? Farhi and Werning explore sity, the authors impute the opportunity eters defining preferences and the chang- these questions by modeling altruisti- cost of time for the shopper that peaks ing nature of wage risk. The analysis cally-linked individuals who experience in middle age at a level roughly 40 per- reveals an important tradeoff for welfare privately observed taste or productivity cent higher than that of retirees. Using calculations. On the one hand, as wage shocks. They also study allocations where this measure of the price of time and uncertainty rises, so does the cost associ- the social welfare criterion values future observed time spent in home produc- ated with missing insurance markets. On generations directly, placing a positive tion, they estimate the parameters of a the other hand, greater wage inequal- weight on their welfare so that the effec- home production function. They find an ity presents opportunities for increasing tive social discount rate is lower than elasticity of substitution between time aggregate productivity by concentrating the private one. For any such difference and market goods in home production market work among more productive in social and private discounting, con- of close to two. Finally, they use the esti- workers. This productivity gain means sumption exhibits mean-reversion and mated elasticities for shopping and home that improving the degree of insurance a steady-state, cross-sectional distribu- production to calibrate an augmented against wage risk offers larger welfare tion for consumption and welfare exists, lifecycle consumption model. The aug- gains than redistributive policies that where no one is trapped at “misery”. The mented model predicts the observed reduce individual wage variability. In optimal allocation can then be imple-

28 NBER Reporter Spring 2006 mented by a combination of income and ize in home assets, remain uninformed representative family model implies that estate taxation. The authors find that the about foreign assets, and amplify their when generous government-supplied UI optimal estate tax is progressive: fortu- initial information asymmetry. is included, the unrealistic result obtains nate parents face higher average marginal Ljungqvist and Sargent use three that economic activity will collapse. tax rates on their bequests. general equilibrium models with jobs Shimer develops a dynamic model Van Nieuwerburgh and Veldkamp and unemployed workers to study the of mismatch. Workers and jobs are ran- note that many explanations for home or effects of government mandated unem- domly assigned to labor markets. Each local bias rely on information asymmetry: ployment insurance (UI) and employ- labor market clears at each instant but investors know more about their home ment protections (EP). To illuminate some labor markets have more work- assets than other assets. One criticism of the forces in these models, they study ers than jobs, hence unemployment, these theories is that asymmetry should how UI and EP affect outcomes when and some have more jobs than work- disappear when information is tradable. there is higher “turbulence” in the sense ers, hence vacancies. As workers and But if investors have asymmetric prior of worse skill transition probabilities jobs move between labor markets, some beliefs, yet choose how to allocate lim- for workers who suffer involuntary lay- unemployed workers find vacant jobs ited learning capacity before investing, offs. Matching and search-island models and some employed workers lose or leave they will not necessarily obtain foreign have labor market frictions and incom- their job and become unemployed. The information. Investors want to exploit plete markets. The representative family model is quantitatively consistent with increasing returns to specialization: the model with employment lotteries has no the comovement of unemployment, job bigger the home-information advantage, labor market frictions and complete mar- vacancies, and the rate at which unem- the more desirable are home-assets; but kets. The adverse welfare state dynamics ployed workers find jobs over the business the more home-assets investors expect to coming from high UI indexed to past cycle. It can also address a variety of labor own, the higher the value of additional earnings are so strong that they deter- market phenomena, including duration home information. Even with a small mine outcomes in all three models. The dependence in the job finding probabili- home information advantage, and even high aggregate labor supply elasticity ty and employer-to-employer transitions, when foreign information is not harder that emerges from employment lotteries and it helps explain the cyclical volatility to obtain, many investors will special- and complete insurance markets in the of vacancies and unemployment.

✱

NBER Reporter Spring 2006 29 Insurance Workshop The NBER’s Working Group on Discussant: Stewart C. Myers, MIT and Jamin, University de Los Andes, Insurance, directed by Kenneth A. Froot NBER “Earthquake Vulnerability Reduction of Harvard Business School and Howard Program in Colombia: A Probabilistic Kunreuther, University of Pennsylvania, Rustam Ibragimov, Harvard Cost-Benefit Analysis” met in Cambridge on February 10 and University, and Dwight Jaffee Discussant: Neil Doherty, University of 11. Their agenda was: and Johan Walden, University Pennsylvania of California, Berkeley, “Non- David M. Cutler, Harvard University Diversification Traps in Markets for Panel Discussion: Terrorism Risk and NBER; Amy Finkelstein, Catastrophic Risk?” Insurance MIT and NBER; and Kathleen M. Discussant: David Durbin, Swiss RE Christopher Lewis, The Hartford McGarry, University of California, Financial Services Group; Erwann Los Angeles and NBER, “Preference Panel Discussion: Natural Disaster Michel-Kerjan, University of Heterogeneity and Insurance Markets” Insurance Pennsylvania; Lloyd Dixon, RAND Discussant: Alexander Muermann, Karen Clark, Applied Insurance Corporation; and Gordon Woo, Risk University of Pennsylvania Research; Gordon Stewart, Insurance Management Solutions Information Institute; David Moss, Moderator: Howard Kunreuther Silke Brandts and Christian Laux, Harvard University; and Scott Goethe University, Frankfurt, “ART Harrington, University of Pennsylvania David Cummins, University of versus Reinsurance: The Disciplining Moderator: Kenneth A. Froot Pennsylvania, and Yijia Lin and Effect of Information Insensitivity” Richard Phillips, Georgia State Discussant: Morton Lane, Lane Martin Grace, Robert Klein, and University, “An Empirical Investigation Financial Zhiyong Liu, Georgia State University, of the Pricing of Financially “Mother Nature on the Rampage: Intermediated Risks with Costly Darius Lakdawalla, RAND Implications for Insurance Markets” External Finance” Corporation and NBER, and George Discussant: Robert J. Shiller, Yale Discussant: Anne Gron, Northwestern Zanjani, Federal Reserve Bank of University and NBER University New York, “Catastrophic Bonds, Reinsurance, and the Optimal Francis Ghesquiere and Olivier Collateralization of Risk Transfer “ Mahul, The World Bank, and Luis

Standard theories of insurance, dating ric information effect, the authors may surance offer less aggressively. Outsiders from Rothschild and Stiglitz (1976), stress see higher or lower risk individuals pur- are subject to adverse selection as only a the role of adverse selection in explain- chasing insurance in equilibrium. They high-risk insurer might find it optimal ing the decision to purchase insurance. provide an empirical example from the to change reinsurers. This creates a hold- In these models, higher risk people buy U.S. life insurance industry in which pat- up problem that allows the incumbent to full or near-full insurance, while lower terns of insurance coverage are consistent extract an information rent. An informa- risk people buy less complete coverage, with the preference heterogeneity model. tion-insensitive ART product with a para- if they buy at all. While this prediction Specifically, they show that individuals metric or index trigger is not subject to appears to hold in some real world insur- who engage in risky behavior are both sys- adverse selection. It can therefore be used ance markets, including health insurance, tematically higher risk (higher mortality) to compete against an informed reinsurer, in many others, such as life insurance, it is and less likely to purchase insurance. They thereby reducing the premium that a low- the lower risk individuals who have more discuss the implications of preference het- risk insurer has to pay for the indemnity insurance coverage. Cutler, Finkelstein, erogeneity for the existence, nature, and contract. However, ART products exhibit and McGarry consider a simple extension empirical detection of inefficiencies in an interesting fate in this model as they to the standard model in which individu- insurance markets. are useful, but not used in equilibrium als vary in their risk tolerance as well as Brandts and Laux demonstrate because of basis-risk. risk type, with those with lower tolerance a novel benefit of “Alternative Risk Catastrophe bonds feature full col- for risk becoming endogenously lower risk Transfer” (ART) products with paramet- lateralization of a specific risk, and thus through investment in risk reducing activi- ric or index triggers: when a reinsurer has appear to be inefficient risk transfer instru- ties. Depending on whether this behavioral private information about his client’s risk, ments — ones that completely abandon effect dominates the standard asymmet- outside reinsurers will price their rein- the modern insurance principle of econo-

30 NBER Reporter Spring 2006 mizing on collateral through diversifica- prompting insurers to reassess their risk plete picture of the efficiency of the risk tion. Lakdawalla and Zanjani confirm and business strategies in Southeastern mitigation project under consideration. this paradox in an idealized world of com- states. Grace, Klein, and Liu examine Ghesquiere, Jamin, and Mahul present a plete insurance contracts, where catastro- recent trends in the affected homeown- probabilistic cost-benefit analysis relying phe bonds indeed have no role to play. ers insurance markets, analyze factors on a catastrophe risk model. It produces However, the real world admits a poten- affecting insurers’ adjustments in these risk metrics, thus providing the decision- tially important role. Insurers may find it markets, and discuss how these markets maker with a more complete risk analysis difficult to contract completely over the are likely to further change in response of the net benefits of the project. This is division of assets in the event of insol- to the reassessment of hurricane risk and illustrated with the earthquake vulner- vency, and, more generally, difficult to regulatory reactions and related policy ability reduction project in Colombia write contracts with a full menu of state and regulatory issues. They conduct an financed by the World Bank. contingent payments. Instead, the basic econometric analysis of insurers’ relative Under perfect market conditions, contract promises to indemnify with ref- shares of homeowners insurance expo- standard capital budgeting theory pre- erence to a particular customer’s loss expe- sures in Florida statewide and by county dicts that the hurdle rates on financial- rience only. In this environment, custom- over the period 1996-2005. Their analy- ly intermediated risk products should ers of the insurer will have different levels sis indicates several factors that appear reflect only non-diversifiable risk and be of exposure to default. When contract- to be associated with changes in insurers’ constant across firms. However, theoreti- ing constraints limit the insurer’s ability share of exposures. Among these factors, cal research by Froot and Stein (1998), to smooth out such differences, there is publicly traded stock insurers and mutu- among others, suggests that when firms a potential niche for catastrophe bonds al insurers have tended to increase their invest in non-hedgeable assets under in serving those who would be heav- presence in the Florida market, although conditions where capital is costly, proj- ily exposed to default. The authors show publicly traded insurers have retrenched ect pricing should reflect the covariabil- that catastrophe bonds may be useful in from the highest-risk areas. Overall, this ity of the project with the firm’s existing mitigating differences in default exposure, analysis and the examination of other portfolio, even if this covariability repre- which arise with contractual incomplete- market data suggest that significant mar- sents non-systematic risk. These research- ness and heterogeneity among insureds. ket restructuring has occurred and is like- ers argue that their theory is especially Heterogeneity is required, because it ly to continue as some insurers with large applicable to financial institutions pric- undermines the efficiency of a mechanical numbers of exposures will retrench and ing intermediated risks. The theoretical pro rata division of assets that takes place other insurers step in to fill the gap. The research also suggests that the prices of in the event of insurer insolvency. price of insurance also will increase sig- intermediated risks will reflect the capital Ibragimov, Jaffee, and Walden nificantly in high-risk areas, although strain that such risks place on the inter- develop a simple model for markets for the magnitude and pace of rate hikes will mediary and hence reflect implicit alloca- catastrophic risk. The model explains depend on insurers’ risk assessments and tions of capital to the intermediary’s busi- why insurance providers may choose not regulatory approvals. The supply of insur- ness lines (Myers and Read 2001, Zanjani to offer insurance for catastrophic risks ance and market restructuring will be 2002). Cummins, Lin, and Phillips test and not to participate in reinsurance affected greatly by insurers’ reassessments these theoretical predictions by analyz- markets, even though there is enough of hurricane risk, the supply and cost of ing the prices of insurance risks for U.S. market capacity to reach full risk sharing reinsurance, and regulatory policies. property-liability insurers over the peri- through diversification in a reinsurance Cost-benefit analysis is a standard od 1997-2004. Specifically, they regress market. This is a nondiversification trap. tool for determining the costs and ben- insurance price variables on capital allo- The authors show that nondiversification efits, and thus the efficiency, of planned cations by line, measures of insurer insol- traps may arise when risk distributions projects. However, one of the major dif- vency risk, and other risk and control have heavy left tails and liability is limit- ficulties in risk mitigation investments variables. Their results provide strong ed. When they are present, there may be a is that benefits are by nature uncertain support for theoretical predictions that coordination role for a centralized agency as they depend on the occurrence of a prices of intermediated risks vary across to ensure that risk sharing takes place. natural disaster. In this context, the stan- firms to reflect insolvency risk, marginal The severe hurricane seasons of 2004 dard approach relying on the average capital allocations, and non-systematic and 2005 and the resulting losses are value of benefits may provide an incom- covariability.

NBER Reporter Spring 2006 31 Industrial Organization The NBER’s Program on Industrial Daron Acemoglu and Amy Discussant: C. Lanier Benkard, Organization, directed by Nancy L. Rose Finkelstein, MIT and NBER, “Input Stanford University and NBER of MIT, met in California on February and Technology Choices in Regulated 24 and 25. Liran Einav and Jon D. Industries: Evidence from the Health Duncan Simester, MIT; Yu (Jeffrey) Levin, both of NBER and Stanford Care Sector” Hu, Purdue University; Erik University, organized this program: Discussant: Catherine Wolfram, Brynjolfsson, MIT; and Eric T. University of California, Berkeley and Anderson, Northwestern University, Silke Januszewski Forbes, University NBER “Does Current Advertising Cause of California, San Diego, and Mara Future Sales? Evidence from the Direct Lederman, University of Toronto, Justine Hastings, Yale University and Mail Industry” “Control Rights, Network Structure, NBER; Thomas J. Kane, Harvard Discussant: Jean Pierre Dube, and Vertical Integration: Evidence from University and NBER;and Douglas University of Chicago Regional Airlines” O. Staiger, Dartmouth College Discussant: Severin Borenstein, and NBER, “Public School Choice Ryan M. Johnson, University of University of California, Berkeley and and Student Outcomes: Choices, Arizona; David H. Reiley, University NBER Preferences, and Heterogeneous of Arizona and NBER; and Juan Treatment Effects in a School Choice Carlos Munoz, Pontifica Universidad Pat Bajari, University of Michigan Lottery” Catolica, Chile, “The War for the Fare: and NBER; Stephanie Houghton, Discussant: Eric A. Hanushek, How Driver Compensation Affects Bus Duke University; and Steve Tadelis, Stanford University and NBER System Performance” University of California, Berkeley, Discussant: Phillip Leslie, Stanford “Bidding for Incomplete Contracts: An Jaap H. Abbring, Free University, University and NBER Empirical Analysis” Amsterdam, and Jeffrey Campbell, Discussant: Ken Hendricks, University Federal Reserve Bank of Chicago, “A of Texas at Austin Firm’s First Year”

Forbes and Lederman investigate force are used. However, by bringing the and may also involve significant adap- the relationship between vertical inte- regional’s labor force “in-house,” owner- tation and renegotiation costs. Bajari, gration and the importance of control ship of a regional may erode some of the Houghton, and Tadelis propose a styl- rights under incomplete contracts in the labor cost savings that are very reason ized model of bidding for incomplete U.S. regional airline industry. Regional why majors subcontract certain flights contracts and apply it to data from high- airlines operate flights for major carri- to regionals.Using data on majors’ use of way paving contracts. In the model, bid- ers under the major’s brand. The majors regionals in the second quarter of 2000, ders respond strategically to contractual market the regionals’ flights as their own. the authors test whether majors’ choice of incompleteness; adaptation costs, broadly There is substantial variation in wheth- organizational form reflects this tradeoff defined, are an important determinant er regionals are owned by the major for between greater control and lower labor of the observed bids. The authors then which they operate. Furthermore, several costs. They find that owned regionals are estimate the costs of adaptation and bid- majors own some of their regional part- more likely to serve city pairs that are der markups using a structural auction ners but also contract with others. Forbes more integrated into the major’s network, model. Their estimates suggest that adap- and Lederman illustrate the benefits and where externalities not internalized by tation costs on average account for about costs of vertical integration between a the regional will be the greatest. Further, 10 percent of the winning bid. The dis- major and regional. They argue that when owned regionals are more likely to serve tortions from private information and unforeseen disruptions create the need city pairs with adverse weather condi- local market power, which are the focus for schedule adjustments — as frequently tions, where unforeseen schedule disrup- of much of the literature on optimal pro- occurs in the airline industry — the major tions will be more common. curement mechanisms, are much smaller will internalize the impact of the dis- Procurement contracts are often by comparison. ruption on its entire network, while the incomplete because the initial plans and Acemoglu and Finkelstein examine regional will not. Ownership of a regional specifications are changed and refined the implications of regulatory change for mitigates this incentive problem by giv- after the contract is awarded to the low- the input mix and technology choices of ing the major rights of control over how est bidder. This results in a final cost to regulated industries. They present a sim- the regional’s physical assets and labor the buyer that differs from the low bid, ple framework that emphasizes changes

32 NBER Reporter Spring 2006 in relative factor prices faced by regulated have significant gains in test scores as suggest a way to distinguish between the firms under different regimes, and inves- a result of attending their first choice informative and persuasive roles of adver- tigate how this might affect their tech- school. However, students who placed a tising, providing insight into the mecha- nology choices through substitution of high value on academics experience sig- nism by which advertising differentially (capital embodied) technologies for tasks nificant gains in test scores. These- find affects various customer subsets. previously performed by labor. Then, ings are consistent with the conclusion Two systems of bus driver compensa- they empirically study the change from that parents may trade-off expected gains tion exist in Santiago, Chile. Most drivers full-cost to partial-cost reimbursement to academic achievement for gains along are paid per passenger transported, but under the Medicare Prospective Payment other dimensions, such as proximity or a second system compensates other driv- System (PPS) reform, which increased the racial composition. ers with a fixed wage. The per-passenger relative price of labor faced by U.S. hos- Abbring and Campbell deter- drivers have incentives to engage in “ La pitals. Using the interaction of hospitals’ mine the structural shocks that shape a Guerra por el Boleto” (“The War for the pre-PPS Medicare share of patient days firm’s first year by estimating a structural Fare”), in which drivers change their driv- with the introduction of these regulatory model of firm growth, learning, and sur- ing patterns to compete for passengers. changes, they can document a substantial vival using monthly sales histories from Johnson, Reiley, and Muñoz take advan- increase in capital-labor ratios and a large 305 Texas bars. They find that hetero- tage of a natural experiment provided by decline in labor inputs associated with geneity in firms’ pre-entry scale deci- the coexistence of these two compensa- PPS. Most interestingly, they find that the sions accounts for about 40 percent of tion schemes on similar routes in the same PPS reform seems to have encouraged the their sales variance; persistent post-entry city. Using data on intervals between bus adoption of a range of new medical tech- shocks account for most of the remainder. arrivals, they find that the fixed-wage con- nologies. They also show that the reform They find no evidence of entrepreneur- tract leads to more bunching of buses, was associated with an increase in the skill ial learning. Variation of the firms’ fixed and hence longer average passenger wait composition of these hospitals, which is costs consistent with an annual lease cycle times. The per-passenger drivers are assist- consistent with technology-skill or capi- explains their exit rates. The authors use ed by a fascinating group of independent tal-skill complementarities. the estimated model to price a new bar’s information intermediaries called sapos Hastings, Kane, and Staiger use data option to exit, which accounts for 124 who earn their living by standing at bus surrounding implementation of a pub- percent of its value. stops, recording arrival times, and sell- lic school choice program with school Simester, Hu, Brynjolfsson, and ing the information to subsequent driv- assignment by lottery to estimate the Anderson present findings from a large- ers who drive past. This bus-bunching effect of attending a first choice school scale field experiment that allows them phenomenon has frustrated passengers in on academic outcomes. They show that to study whether there is a causal rela- cities around the world, so it is exciting students who place a high value on aca- tionship between current advertising and to see evidence that contract design can demic achievement in their school-choice future sales. The experimental design improve performance in this dimension. decision should have significant gains in overcomes limitations that have affected According to the results here, a typical bus their own academic outcomes as a result previous investigations of this issue. They passenger in Santiago waits roughly 10 of winning the lottery to attend their first- find that current advertising does affect percent longer for a bus on a fixed-wage choice school. Using random assignment future sales, but the sign of the effect var- route relative to an incentive-contract to schools generated by the lottery, they ies depending on the customers targeted. route. However, the improved wait times estimate the impact of winning the school For the firm’s best customers, the long-run on the incentive contract routes come at choice lottery on academic achievement, effect of increases in current advertising is a cost. The incentives lead drivers to drive allowing the treatment effect of attend- actually negative, while for other custom- noticeably more aggressively, causing at ing a first choice school to vary with the ers the effect is positive. They argue that least 67 percent more accidents per kilo- estimated preference for academics that these outcomes reflect two competing meter driven. Most people in Santiago generated the choice. Their results indi- effects: brand-switching and inter-tem- blame “La Guerra por el Boleto” for the cate that, on average, students do not poral substitution. Furthermore, the data poor service provided by buses.

NBER Reporter Spring 2006 33 Development of the American Economy The NBER’s Program on the America’s Gilded Age” Leah Platt Boustan, Harvard Development of the American Economy, University; Price V. Fishback, directed by Claudia Goldin of Harvard Paul W. Rhode, University of North University of Arizona and NBER; University, met in Cambridge on March Carolina and NBER, and Koleman S. and Shawn E. Kantor, University of 4. The following papers were discussed: Strumpf, University of North Carolina, California, Merced and NBER, “The “Manipulating Political Stock Markets: Effect of Internal Migration on Local Farley Grubb, University of Delware A Field Experiment and a Century of Labor Markets:American Cities During and NBER, “The Net Asset Position of Observational Data” the Great Depression” the U.S. National Government, 1784– 1802: Hamilton’s Blessing or the Spoils Daron Acemoglu and Simon Johnson, Joseph P. Ferrie, Northwestern of War?” (NBER Working Paper No. MIT and NBER, and James Robinson, University and NBER, “Wealth During 11868) Harvard University and NBER, Wartime: Mobility in Real and Personal “Liberty, Fraternity, Equality, and Property Ownership in the U.S., Douglas A. Irwin, Dartmouth College Industry: The Economic Consequences 1860–1870” and NBER, “Tariff Incidence in of the French Revolution”

The War for Independence left the of non-traded goods, the degree of pro- kets. They first investigate the speculative National Government deeply in debt, but tection was much less than indicated by attacks on TradeSports market in 2004 among the spoils of winning that war was nominal rates of protection; the results of when a single trader made a series of large an empire of land. So, was the National Irwin’s work suggest that the 30 percent investments in an apparent attempt to Government solvent after 1783? Was its average tariff on imports yielded just a 12 make one candidate appear stronger. Next net asset position — land assets minus debt percent implicit subsidy to import-com- they examine the historical Wall Street liabilities — positive or negative? Grubb peting producers while effectively tax- markets where political operatives from gathers evidence to answer this question ing exporters at a rate of 14 percent. He the contending parties actively and open- by constructing a yearly time series of also finds that tariff policy redistributed ly bet on city, state, and national races; the the U.S. net asset position, including the large amounts of income (about 9 per- record is rife with accusations that parties subcomponents of that position, from cent of GDP) across groups, although the tried to boost their candidates through 1784 through 1802. The results indicate impact on consumers was roughly neutral investments and wash or phantom bets. that the National Government was sol- because they devoted a sizeable share of Finally, they report the results of a field vent, had more than enough land assets their expenditures to exportable goods. experiment involving a series of planned, to cover its debt liabilities in this period, These findings may explain why import- random investments — accounting for 2 but only if it maintained the default on competing producers pressed for even percent of total market volume — in the the Continental Dollar (its non-interest- greater protection in the face of already Iowa Electronic Market in 2000. In every bearing debt). To do this and not ruin its high tariffs and why consumers (as voters) speculative attack that they study, there creditworthiness, it had to distinguish, did not strongly oppose the policy. were measurable initial changes in prices. legally and in the marketplace, between Political stock markets have a long However, these were quickly undone and its interest-bearing and its non-interest- history in the United States. Organized prices returned close to their previous lev- bearing debt. It did this, in part, by only prediction markets for Presidential elec- els. They find little evidence that politi- paying interest and no principal on its tions have operated on Wall Street (1880- cal stock markets can be systematically debts and by curtailing direct swaps of 1944), the Iowa Electronic Market (1988- manipulated beyond short time periods. land for debt. present), and TradeSports (2001-present). The French Revolution of 1789 In the late nineteenth century, the Proponents claim that such markets effi- had a momentous impact on neighbor- United States imposed high tariffs to pro- ciently aggregate information and provide ing countries. The French Revolutionary tect domestic manufacturers from foreign superior forecasts to polls. An important armies during the 1790s invaded and competition. Irwin examines the magni- counterclaim is that such markets may controlled Belgium, the Netherlands, tude of protection given to import-com- be subject to manipulation by interested Italy, Switzerland, and parts of Germany. peting producers and the costs imposed parties. Rhode and Strumpf analyze this Napoleon in the early 1800s extended on export-oriented producers by focusing argument by studying alleged and actual the French control over these territories on changes in the domestic prices of trad- speculative attacks — that is, large trades, and also conquered Spain. Together with ed goods relative to non-traded goods. uninformed by fundamentals, intended invasion came various radical institu- Because the tariff increased the prices to change prices — in these three mar- tional changes. Most notably, the French

34 NBER Reporter Spring 2006 Revolution removed the legal and eco- Revolution destroyed (the institutional wealth of information about the economic nomic barriers that had protected the underpinnings of ) the power of oligar- environment in their home market. They privileged (nobility, clergy, and urban chies and elites opposed to economic use data on these “push” factors — includ- oligarchies), established the principle change, and combined with the arrival of ing weather conditions and the generosity of equality before the law, and led to a new economic and industrial opportuni- of New Deal policies — in “sending areas” reorganization of the state. Acemoglu, ties in the second half of the nineteenth to develop an instrument for in-migration Johnson, and Robinson argue that the century, helped to pave the way for future for their sample of large U.S. cities. With French Revolution can be considered an economic growth. this approach, they can address the endog- exogenously imposed institutional change Debates over U.S. immigration pol- enous location choices of migrants, who on neighboring countries and they study icy have prompted numerous studies of tend to be attracted to cities with high its impact on economic outcomes. They the impact of immigrant arrivals on local wages or strong wage growth, thus obscur- also use this experience to evaluate several labor markets in the United States. Yet ing any negative relationship between in- central theses about institutions: 1) that immigration from abroad comprises a migration and wages. they efficiently adapt to a society’s- char small share of total flows into local labor Using a new sample of 8,195 white, acteristics; 2) that evolved institutions markets. Since 1940, new foreign entry native-born American males located in are inherently superior to those ratio- has accounted for less than 11 percent of both the 1860 and 1870 U.S. federal cen- nally designed; 3) that institutions must cross-county moves and less than 20 per- sus manuscript schedules, Ferrie assess- be “appropriate” and cannot be “trans- cent of moves across state lines. To cap- es wealth mobility over the decade that planted”; and 4) that the Civil Code ture this central feature of competition in included the Civil War. He explores the has adverse economic effects. Both at local labor markets, Boustan, Fishback, impact of Emancipation on the wealth of the country and the city level, the evi- and Kantor examine the impact of inter- those who owned slaves in 1860 (when dence suggests that areas that were occu- nal migration on annual earnings and slaves were a component of personal pied by the French and that underwent employment in major U.S. cities in the wealth) with a new sub-sample of 401 radical institutional reform experienced late 1930s. They also explore the effect Southerners linked to the slave schedules somewhat more rapid economic growth, of these labor supply shocks on the out- of the census. Previous studies (Steckel, urbanization and industrialization, espe- migration of existing workers and the 1990) that have examined mid-19th cen- cially after 1850, although in some speci- in-migration of firms. These worker-firm tury real estate wealth mobility have been fications the results are not statistically adjustments, which often go unobserved, unable to assess the extent of mobility in significant and in some others, there were may account for the limited relationship total wealth (as respondents were asked slight pre-existing trends. On the whole, between immigration and wages found in to report both real wealth and personal the evidence here is more consistent with previous studies. Internal migration rep- wealth only in 1860 and 1870, two cen- the view that the institutional reforms resented the vast majority of population suses that until now had not been linked brought about by the French Revolution flows to and from local labor markets dur- on a large scale). Ferrie also compares had long-run beneficial effects; this mili- ing the 1930s. Because internal migrants wealth mobility in the 1860s to wealth tates against all four of the above the- originate their moves from within the mobility a century later in the 1960s and ses. The interpretation here is that the United States, the authors have access to a 1970s (using the NLS).

NBER Reporter Spring 2006 35 Bureau Books

NBER Macroeconomics Annual 2005 NBER Macroeconomics Annual 2005, specific real-world examples and prob- Monetary Economics and International edited by Mark Gertler and Kenneth S. lems. This year’s twentieth edition of the Finance and Macroeconomics, respec- Rogoff, is available from the MIT Press Annual covers such issues as: the sources tively. Gertler is the Henry and Lucey for $35.00 in paperback and $70.00 of European unemployment and churn Moses Professor of Economics at New clothbound. The NBER Macroeconomics in U.S. labor markets; the causes of the York University; Rogoff is the Thomas Annual presents, extends, and applies pio- 20-year trend decline in volatility in U.S. D. Cabot Professor of Public Policy and neering work in macroeconomics and output; and the appropriate targets for Economics at Harvard University. stimulates research on important pol- monetary policy. E-mail orders for this volume to: mit- icy issues. The papers here link theo- Gertler and Rogoff are NBER [email protected]. retical and empirical developments with Research Associates in the Programs on

A New Architecture for the U.S. National Accounts

A New Architecture for the U.S. grated set of U.S. national accounts. The the Department of Labor, the Census National Accounts, edited by Dale W. purpose of that new architecture is not Bureau, and the Governors of the Federal Jorgenson, J. Steven Landefeld, and only to integrate the existing systems of Reserve System. One important part of William D. Nordhaus, is available from accounts, but also to identify gaps and the motivation for the new architecture is the University of Chicago Press this inconsistencies and to expand and incor- to integrate the different components and spring for $99.00. To order this volume, porate systems of non-market accounts to make them consistent. This volume is contact: University of Chicago Press, with the core system. the first step toward achieving that goal. Order Department, 11030 South Langley Because the U.S. economy represents Jorgenson is the Samuel W. Morris Avenue, Chicago, IL 60628-2215, or 1- almost 30 percent of the world econ- University Professor of Economics at 800-621-2736. Academic discounts are omy, it is not surprising that account- Harvard University. Landefeld is the available. ing for this huge and diverse set of eco- Director of the Bureau of Economic This NBER volume, part of the series nomic activities requires a decentralized Analysis. Nordhaus is an NBER Research of Studies in Income and Wealth (No. statistical system. This volume outlines Associate in the Program on Productivity 66), brings together a distinguished group the major assignments among institu- and the Sterling Professor of Economics of contributors to initiate the develop- tions, including the Bureau of Economic at Yale University. ment of a comprehensive and fully inte- Analysis, the Bureau of Labor Statistics,

36 NBER Reporter Spring 2006 NBER Working Papers On-Line

A complete list of all NBER Working Papers with searchable abstracts, and the full texts of Working Papers (issued since November 1994) are available at http://www.nber.org/wwp.html to anyone located at a university or other organization that subscribes to the (hard copy) Working Paper series. If you believe that your organization subscribes, but you cannot access the online Working Paper service, please e-mail the NBER at [email protected] for more information and assistance. ✱

Individual copies of NBER Working Papers, Historical Factors in Long-Run Growth Papers, and Technical Papers are available free of charge to Corporate Associates. For all others, there is a charge of $10.00 per hardcopy or $5.00 per down- loaded paper. (Outside the United States, add $10.00 per order for postage and handling.) Advance payment is required on all orders. To order, call the Publications Department at (617) 868-3900 or visit www.nber.org/papers. Please have ready the number(s) of any Working Paper(s) you wish to order.

Subscriptions to the full NBER Working Paper series include all 700 or more papers published each year. Subscriptions are free to Corporate Associates. For others within the United States, the standard rate for a full subscription is $3200; for academic libraries and faculty members, $2300. Higher rates apply for foreign orders. The on-line standard rate for a full sub- scription is $1750 and the on-line academic rate is $725. Partial Working Paper subscriptions, delineated by program, are also available. For further information, see our Web site, or please write: National Bureau of Economic Research, 1050 Massachusetts Avenue, Cambridge, MA 02138-5398. ✱

Titles of all papers issued since January 2006 are presented below. For previous papers, see past issues of the NBER Reporter. Working Papers are intended to make results of NBER research available to other economists in preliminary form to encourage discussion and suggestions for revision before final publication. They are not reviewed by the Board of Directors of the NBER.

NBER Working Papers Paper Author(s) Title

11940 Daron Acemoglu Modeling Inefficient Institutions 11941 Lubos Pastor Estimating the Intertemporal Risk-Return Tradeoff Using Meenakshi Sinha the Implied Cost of Capital Bhaskaran Swaminthan 11942 Marc P. Giannoni Robust Optimal Policy in a Forward-Looking Model with Parameter and Shock Uncertainty 11943 Steven Haider Life-Cycle Variation in the Association Between Current Gary Solon and Lifetime Earnings 11944 Josh Lerner Innovation and Incentives: Evidence from Corporate R&D Julie Wolf 11945 Jonathan Gruber A Tax-Based Estimate of the Elasticity of Intertemporal Substitution 11946 Stephen G. Cecchetti Assessing the Sources of Changes in the Volatility of Real Alfonso Flores-Lagunes Growth Stefan Krause

NBER Reporter Spring 2006 37 Paper Author(s) Title

11947 Dani Rodrik What’s So Special About China’s Exports? 11948 Carol A. Corrado Intangible Capital and Economic Growth Charles R. Hulten Daniel E. Sichel 11949 Daron Acemoglu Did Medicare Induce Pharmaceutical Innovation? David Cutler Amy Finkelstein Joshua Linn 11950 Kevin F. Hallock The Value of Stock Options to Non-Executive Employees Craig A. Olsen 11951 Thomas C. Buchmueller Health Insurance Take-Up by the Near Elderly Sabina Ohri 11952 Dani Rodrik The Social Cost of Foreign Exchange Reserves 11953 Claudia Goldin The Quiet Revolution that Transformed Women’s Employment, Education, and Family 11954 Edward Miguel The Long-Run Impact of Bombing Vietnam Gerard Roland 11955 Thomas Piketty The Evolution of Top Incomes: A Historical and Emmanuel Saez International Perspective 11956 Jayachandran N. Variyam Nutrition Labels and Obesity John Cawley 11957 Inas Rashad Structural Estimation of Caloric Intake, Exercise, Smoking And Obesity 11958 Guillermo Caruana Production Targets Liran Einav 11959 Alan Auerbach Dividend Taxes and Firm Valuation: New Evidence Kevin Hassett 11960 Eric Hilt The Negative Trade-Off Between Risk and Incentives: Evidence from the American Whaling Industry 11961 Sanjiv Das Common Failings: How Corporate Defaults Are Correlated Darrell Duffie Nikunj Kapadia Leandro Saita 11962 Darrell Duffie Multi-Period Corporate Default Prediction with Stochastic Ke Wang Covariates Leandro Saita 11963 David M. Cutler The Determinants of Mortality Angus S. Deaton Adriana Lleras-Muney 11964 Jeffrey Brown Household Ownership of Variable Annuities James Poterba

38 NBER Reporter Spring 2006 Paper Author(s) Title

11965 Atila Abdulkadiroglu Changing the Boston School Choice Mechanism Parag A. Pathak Alvin E. Roth Tayfun Sonmez 11966 Enrique G. Mendoza Lessons From the Debt-Deflation Theory of Sudden Stops 11967 John Kennan Private Information, Wage Bargaining, and Employment Fluctuations 11968 V. Joseph Hotz Examining the Effect of the Earned Income Tax Credit Charles H. Mullin on the Labor Market Participation of Families on Welfare John Karl Scholz 11969 Victor Lavy From Forced Busing to Free Choice in Public Schools: Quasi-Experimental Evidence of Individual and General Effects 11970 Christiana Stoddard Incentives and Effort in the Public Sector: Have U.S. Peter Kuhn Education Reforms Increased Teachers’ Work Hours? 11971 Mark Gertler A Phillips Curve With an Ss Foundation John Leahy 11972 Joel Waldfogel The Median Voter and the Median Consumer: Local Private Goods and Residential Sorting 11973 Nicolas Magud Capital Controls: An Evaluation Carmen R. Reinhart 11974 James Poterba Lifecycle Asset Allocation Strategies and the Distribution Joshua Rauh of 401(k) Retirement Wealth Steven Venti David Wise 11975 Helen Levy Health Insurance and the Wage Gap 11976 Bruce D. Meyer Consumption, Income, and Material Well-Being After James X. Sullivan Welfare Reform 11977 Anthony T. Lo Sasso The Health Care Safety Net and Crowd-Out of Private Bruce D. Meyer Health Insurance 11978 Joseph Engelberg Comparing the Point Predictions and Subjective Probability Charles F. Manski Distributions of Professional Forecasters Jared Williams 11979 James J. Choi Reducing the Complexity Costs of 401(k) Participation Brigitte C. Madrian Through Quick Enrollment David Laibson 11980 Brigitte C. Madrian The U.S. Health Care System and Labor Markets 11981 Larry Chavis Consumer Boycotts: The Impact of the Iraq War Phillip Leslie on French Wine Sales in the U.S. 11982 John J. Donohue III Uses and Abuses of Empirical Evidence Justin Wolfers in the Death Penalty Debate

NBER Reporter Spring 2006 39 Paper Author(s) Title

11983 Elena Loutskina Securitization and the Declining Impact of Bank Finance Philip E. Strahan on Loan Supply: Evidence from Mortgage Acceptance Rates 11984 Leora Friedberg Life is Cheap: Using Mortality Bonds to Anthony Webb Hedge Aggregate Mortality Risk 11985 Lex Borghans People People: Social Capital and the Labor Market Bas ter Weel Outcomes of Underrepresented Groups Bruce A. Weinberg 11986 David H. Autor The Polarization of the U.S. Labor Market Lawrence F. Katz Melissa S. Kearney 11987 John J. Donohue Measurement Error, Legalized Abortion, and the Steven D. Levitt Decline in Crime: A Response to Foote and Goetz (2005) 11988 John F. Helliwell How’s Your Government? International Evidence Haifang Huang Linking Good Government and Well-Being 11989 Justin Wolfers Diagnosing Discrimination: Stock Returns and CEO Gender 11990 Donald S. Kenkel The Roles of High School Completion and GED Dean R. Dillard Receipt in Smoking and Obesity Alan D. Mathios 11991 Assaf Razin Vying for Foreign Direct Investment: A EU-type Efraim Sadka Model of Tax Competition 11992 Joseph Stiglitz Growth, Initial Conditions, Law, and Speed of Privatization in Transition Countries: 11 Years Later 11993 Torsten Persson Democracy and Development: The Devil in the Details Guido Tabellini 11994 Austan Goolsbee The Value of Broadband and the Deadweight Loss of Taxing New Technology 11995 Austan Goolsbee Valuing Consumer Products by the Time Spent Using Them: Peter J. Klenow An Application to the Internet 11996 Ricardo J. Caballero An Equilibrium Model of “Global Imbalances” Emmanuel Farhi and Low Interest Rates Pierre-Olivier Gourinchas 11997 James D. Adams How Rapidly Does Science Leak Out? J. Roger Clemmons Paula E. Stephan 11998 Phil Oreopoulos Short, Medium, and Long-Term Consequences of Poor Mark Stabile Infant Health: An Analysis Using Siblings and Twins Randy Walld Leslie Roos

40 NBER Reporter Spring 2006 Paper Author(s) Title

11999 Luigi Guiso Does Culture Affect Economic Outcomes? Paola Sapienza Luigi Zingales 12000 Murray Carlson Equilibrium Exhaustible Resource Price Dynamics Zeigham Khoker Sheridan Titman 12001 Rosemary J. Avery Regulating Advertisements: The Case of Donald S. Kenkel Smoking Cessation Products Dean R. Lillard Alan D. Mathios 12002 Daniel S. Hamermesh Time to Eat: Household Production Under Increasing Income Inequality 12003 Jeffrey R. Kling Incarceration Length, Employment, and Earnings 12004 Nancy E. Reichman Typically Unobserved Variables (TUVs) and Selection Hope Corman into Prenatal Inputs: Implications for Estimating Infant Kelly Noonan Health Production Functions Dhaval Dave 12005 Joshua Angrist Is Spanish-Only Schooling Responsible for Aimee Chin the Puerto Rican Language Gap? Ricardo Godoy 12006 James J. Heckman The Effects of Cognitive and Noncognitive Abilities on Jora Stixrud Labor Market Outcomes and Social Behavior Sergio Urzua 12007 Chul-In Lee Trends in Intergenerational Income Mobility Gary Solon 12008 David McAdams Perverse Incentives in the Medicare Michael Schwarz Prescription Drug Benefit 12009 John Beshears The Importance of Default Options for Retirement James J. Choi Savings Outcomes: Evidence from the United States David Laibson Brigitte C. Madrian 12010 Rafael Rob Piracy on the Silver Screen Joel Waldfogel 12011 J. Bradford DeLong A Short Note on the Size of the Dot-Com Bubble Konstantin Magin 12012 Judith Chevalier State Casket Sales and Restrictions: Fiona Scott Morton A Pointless Undertaking? 12013 Patrick Bajari Estimating Static Models of Strategic Interaction Han Hong John Krainer Denis Nekipelov 12014 Erica X. N. Li Optimal Market Timing Dmitry Livdan Lu Zhang

NBER Reporter Spring 2006 41 Paper Author(s) Title

12015 Ravi Jagannathan Do Hot Hands Persist Among Hedge Fund Managers? Alexey Malakhov An Empirical Evaluation Dmitry Novikov 12016 Thomas J. Philipson Surplus Appropriation from R&D and Health Care Anupam B. Jena Technology Assessment Procedures 12017 Jakub W. Jarek Optimal Value and Growth Tilts in Luis M. Viceira Long-Horizon Portfolios 12018 James J. Heckman Bias Corrected Estimates of GED Returns Paul A. LaFontaine 12019 Naci Mocan Ugly Criminals Erdal Tekin 12020 Darrell Duffie Valuation in Over-the-Counter Markets Nicolae Garleanu Lasse Heje Pedersen 12021 Matthew Gentzkow Does Television Rot Your Brain? Jesse M. Shapiro New Evidence from the Coleman Study 12022 Alejandro Justiniano The Time Varying Volatility of Giorgio E. Primiceri Macroeconomic Fluctuations 12023 Kalpana Kochhar India’s Patterns of Development: Utsav Kumar What Happened, What Follows Raghuram Rajan Arvind Krishnamurthy Ioannis Tokatlidis 12024 N. Gregory Mankiw Pervasive Stickiness (Expanded Version) Ricardo Reis 12025 Thomas J. Holmes Geographic Spillover of Unionism 12026 John H. Cochrane The Dog that Did Not Bark: A Defense of Return Predictability 12027 Philippe Aghion The Effects of Entry on Incumbent Innovation Richard Blundell and Productivity Rachel Griffith Peter Howitt Susanne Prantl 12028 Christopher L. House Durable Goods and Conformity Emre Ozdenoren 12029 Christopher Avery Cost Should be No Barrier: An Evaluation of the Caroline Hoxby First Year of Harvard’s Financial Aid Initiative Clement Jackson Kaitlin Burek Glenn Pope Mridula Raman 12030 Laurent E. Calvet Down or Out: Assessing the Welfare Costs of John Y. Campbell Household Investment Mistakes Paolo Sodini

42 NBER Reporter Spring 2006 Paper Author(s) Title

12031 Philippe Aghion The Unequal Effects of Liberalization: Robin Burgess Evidence from Dismantling the License Raj in India Stephen Redding Fabrizio Zilibotti 12032 Diego Comin Turbulent Firms, Turbulent Wages? Erica L. Groshen Bess Rabin 12033 Christian Broda Optimal Tariffs: The Evidence Nuno Limao David E. Weinstein 12034 David W. Galenson And Now for Something Completely Different: The Versatility of Conceptual Innovators 12035 Michael D. Bordo Legal-Political Factors and the Historical Evolution Peter L. Rousseau of the Finance-Growth Link 12036 Monika Piazzesi Housing, Consumption, and Asset Pricing Martin Schneider Selale Tuzel 12037 Monica Singhal Special Interest Groups and the Allocation of Public Funds 12038 Marcos Chamon Economic Transformation, Population Growth, Michael Kremer and the Long-Run World Income Distribution 12039 Pierre Azoulay PublicationHarvester: An Open-Source Software Andrew Stellman Tool for Science Policy Research Joshua Graff Zivin 12040 David M. Cutler How Do the Better Educated Do It? Socioeconomic Status Mary Beth Landrum and the Ability to Cope with Underlying Impairment Kate A. Stewart 12041 Edward Lazear Sorting in Experiments with Application to Ulrike Malmendier Social Preferences Roberto Weber 12042 Steven R. Grenadier Investment Under Uncertainty and Neng Wang Time-Inconsistent Preferences 12043 Howard Bodenhorn Urban Poverty, School Attendance, and Adolescent Labor Force Attachment: Some Historical Evidence 12044 Andrew K. Rose Offshore Financial Centers: Parasites or Symbionts? Mark M. Spiegel 12045 Peter C. Mancall Exports and Slow Economic Growth in the Lower Joshua L. Rosenbloom South Region, 1720-1800 Thomas Weiss 12046 Ellen McGrattan Why Did U.S. Market Hours Boom in the 1990s? Edward Prescott 12047 Nora Gordon The Causes of Political Integration: An Application to Brian Knight School Districts

NBER Reporter Spring 2006 43 Paper Author(s) Title

12048 Joanna Lahey State Age Protection Laws and the Age Discrimination in Employment Act 12049 Alberto Alesina Who Adjusts and When? Silvia Ardagna On the Political Economy of Reforms Francesco Trebbi 12050 Benjamin E. Hermalin A Framework for Assessing Corporate Governance Reform Michael S. Weisbach 12051 Patrick Bajari Bidding for Incomplete Contracts: An Empirical Analysis Stephanie Houghton Steve Tadelis 12052 Bruce A. Blonigen New Measures of Port Efficiency Using International Wesley W. Wilson Trade Data 12053 Andrew Leigh Competing Approaches to Forecasting Elections: Justin Wolfers Economic Models, Opinion Polling, and Prediction Markets 12054 Linda Bilmes The Economic Costs of the Iraq War Joseph E. Stiglitz An Appraisal Three Years after the Beginning of the Conflict 12055 Jay Shanken Estimating and Testing Beta Pricing Models: Guofo Zhou Alternative Methods and Their Performance in Simulations 12056 Howard Bodenhorn Single Parenthood and Childhood Outcomes 12057 Michael D. Hurd Some Answers to the Retirement-Consumption Puzzle Susann Rohwedder 12058 David W. Galenson The Most Important Works of Art of the Twentieth Century 12059 Paul Oyer The Making of an Investment Banker: Macroeconomic Shocks, Career Choice, and Lifetime Income 12060 Justin Wolfers Five Open Questions About Prediction Markets Eric Zitzewitz 12061 Louis Kaplow Taxation 12062 Miles Kimball Unhappiness After Hurricane Katrina Helen Levy Fumio Ohtake Yoshiro Tsutsui 12063 Uri Gneezy Putting Behavioral Economics to Work: Testing for John A. List Gift Exchange in Labor Markets Using Field Experiments 12064 Jeffrey Groen Program Design and Student Outcomes in George H. Jakubson Graduate Education Ronald G. Ehrenberg Scott Condie Albert Yung-Hsu

44 NBER Reporter Spring 2006 Paper Author(s) Title

12065 Ronald G. Ehrenberg Inside the Black Box of Doctoral Education: George H. Jakubson What Program Characteristics Influence Doctoral Jeffrey Groen Students’ Attrition and Graduation Probabilities? Eric So Joseph Price 12066 Roland G. Fryer, Jr. Testing for Racial Differences in the Mental Ability Steven D. Levitt of Young Children 12067 Raj Chetty A Bound on Risk Aversion Using Labor Supply Elasticities 12068 Dirk Jenter CEO Turnover and Relative Performance Evaluation Fadi Kanaan 12069 Howard Kunreuther Looking Beyond TRIA: A Clinical Examination of Erwann Michel-Kerjan Potential Terrorism Loss Sharing 12070 Don Fullerton Environmental Investment and Policy with Seung-Rae Kim Distortionary Taxes and Endogenous Growth 12071 Fali Huang Employee Screening: Theory and Evidence Peter Cappelli 12072 Agata Antkiewicz Recent Chinese Buyout Activity and Implications John Whalley for Global Architecture 12073 Erik Snowberg Partisan Impacts on the Economy: Evidence from Justin Wolfers Prediction Markets and Close Elections Eric Zitzewitz 12074 Kris James Mitchener Are Prudential Supervision and Regulation Pillars of Financial Stability? Evidence from the Great Depression 12075 Gary Gorton Asset Prices When Agents are Marked-to-Market Ping He Lixin Huang 12076 Robert G. King Discretionary Policy and Multiple Equilibria 12077 Thomas Lemieux Post-Secondary Education and Increasing Wage Inequality 12078 David Card Racial Segregation and the Black-White Test Score Gap Jesse Rothstein 12079 Andres Arias Why Have Business Cycle Fluctuations Gary D. Hansen Become Less Volatile? Lee E. Ohanian 12080 Alan M. Garber Insurance and Incentives for Medical Innovation Charles I. Jones Paul M. Romer 12081 Fumio Hayashi The Depressing Effect of Agricultural Institution Edward C. Prescott on the Prewar Japanese Economy 12082 Mark Aguiar Measuring Trends in Leisure: The Allocation of Time Erik Hurst Over Five Decades

NBER Reporter Spring 2006 45 Paper Author(s) Title

12083 Justin Wolfers Prediction Markets in Theory and Practice Eric Zitzewitz 12084 Gary Gorton Agency-Based Asset Pricing Ping He 12085 George J. Borjas Immigration in High-Skill Labor Markets: The Impact of Foreign Students on the Earnings of Doctorates 12086 Patricia Born The Effects of Tort Reform on Medical Malpractice W. Kip Viscusi Insurers’ Ultimate Losses Tom Baker 12087 Heitor Almeida Financial Constraints, Asset Tangibility, Murillo Campello and Corporate Investment 12088 George J. Borjas Making it in America: Social Mobility in the Immigrant Population 12089 Bennett T. McCallum Monetary and Fiscal Theories of the Price Level: Edward Nelson The Irreconcilable Differences 12090 Nicole M. Boyson Is There Hedge Fund Contagion? Christof W. Stahel René M. Stulz 12091 Elhanan Helpman Trade, FDI, and the Organization of Firms 12092 Steven J. Davis War in Iraq versus Containment Kevin M. Murphy Robert H. Topel 12093 Raghuram G. Rajan The Persistence of Underdevelopment: Institutions, Luigi Zingales Human Capital, or Constituencies? 12094 Michael R. Darby Innovation, Competition, and Welfare-Enhancing Monopoly Lynne G. Zucker 12095 Gita Gopinath Sticky Borders Roberta Rigobon 12096 Genia Long The Impact of Antihypertensive Drugs on the Number David Cutler and Risk of Death, Stroke, and Myocardial Infarction Ernst Berndt in the United States Jimmy Royer Andree-Anne Fournier Alicia Sasser Pierre Cremieux 12097 John A. List Friend or Foe? A Natural Experiment of the Prisoner’s Dilemma 12098 Wayne E. Ferson Testing Portfolio Efficiency with Conditioning Information Andrew F. Siegel 12099 Sebastian L. Mazzuca Political Conflict and Power-sharing in the Origins James A. Robinson of Modern Colombia

46 NBER Reporter Spring 2006 Paper Author(s) Title

12100 Marco Battaglini A Dynamic Theory of Public Spending, Taxation, and Debt Stephen Coate, 12101 Guillermo A. Calvo Phoenix Miracles in Emerging Markets: Recovering Alejandro Izquierdo without Credit from System Financial Crises 12102 Christopher J. Ruhm A Healthy Economy Can Break Your Heart 12103 Eric M. Leeper Dynamic Scoring: Alternative Financing Schemes Shu-Chun Susan Yang 12104 Chang-Tai Hsieh Taxes and Growth in a Financially Underdeveloped Jonathan A. Parker Country: Evidence from the Children 12105 Anna Aizer Public Health Insurance, Program Take-up, and Child Health 12106 Patrick Boyer Migration and Hedonic Valuation: The Case of Air Quality 12107 Patrick Bolton Pay for Short-Term Performance: Executive Compensation Jose Scheinkman in Speculative Markets Wei Xiong 12108 Daron Acemoglu Persistence of Power, Elites, and Institutions James A. Robinson 12109 Martin Lettau Reconciling the Return Predictability Evidence Stijn Van Nieuwerburgh 12110 J. Vernon Henderson “Sick of Local Government Corruption? Vote Islamic” Ari Kuncoro TECHNICAL PAPERS 320 Patrick Bajari Semiparametric Estimation of a Dynamic Game Han Hong of Incomplete Information 321 Jesus Fernandez-Villaverde Estimating Macroeconomic Models: A Likelihood Approach Juan F. Rubio-Ramirez 322 David S. Lee Regression Discontinuity Inference with Specification Error David Card

NBER Reporter Spring 2006 47 Nonprofit Org. U.S. Postage NBERReporter PAID National Bureau of NATIONAL BUREAU OF ECONOMIC RESEARCH Economic Research

1050 Massachusetts Avenue Cambridge, Massachusetts 02138-5398 (617) 868-3900

Change Service Requested