NBER Reporter · 2003-04-21 · Preparation of the NBER Reporter is under the editorial supervision...

60
NBER Reporter Spring 2003 1. NATIONAL BUREAU OF ECONOMIC RESEARCH SPRING 2003 Program Report Monetary Economics Understanding what monetary policy can do to enhance economic performance (and, just as importantly, what it cannot do) is a continuing challenge for economic policymakers around the world. Researchers in the NBER’s Monetary Economics Program contribute to this effort with a combination of theoretical and empirical studies on the effects of cen- tral bank actions and the design of monetary policy. These studies are circulated as NBER Working Papers, presented and discussed at regular Program meetings, and subsequently published in NBER volumes and academic journals. Our regular program meetings also aim to facilitate interaction between researchers working in universities and those working in central banks. Much frontier research on monetary economics occurs within research staffs of the Federal Reserve System and other central banks around the world. These central bank researchers often are invited to present their work and to participate in the discussion of other recent studies. We are delighted that Ben S. Bernanke, one of our long-term members and program director for the past two years, was appointed by President George W. Bush in 2002 to become a Governor of the Federal Reserve. When he was confirmed, I returned to the role of Program Director, which I had held previously. In this report, I summarize a few of the strands of research on mon- etary economics that have engaged NBER researchers in recent years. None of these issues is fully settled, but significant progress has been made. In the process, I will offer a few of my own judgments about what we know and about where more research is still needed. The Dynamic Effects of Monetary Policy According to textbook theory, changes in monetary policy influence employment and production in the short run but, in the long run, affect Reporter OnLine at: www.nber.org/reporter NBER N. Gregory Mankiw* NBER Service Brings You New Data for Free A new, free NBER email serv- ice gives you daily email links to all U.S. government data releases, including unemployment, trade, interest rates, GDP, etc. We keep track of your preferences and email you the requested links when they are released. To sign up for any or all of the government releases, visit www.nber.org/releases and register your choices. IT’S FREE!! Reporter * Mankiw is Director of the NBER’s Program on Monetary Economics and the Allie S. Freed Professor of Economics at Harvard University (see page 24). The numbers in brackets throughout this report refer to NBER Working Papers. A com- plete list of NBER Monetary Economic Working Papers can be found at http://papers.nber.org/papersbyprog/ME.html IN THIS ISSUE Program Report: Monetary Economics 1 Research Summaries: Explaining Exchange Rate Behavior 5 Health, Income, and Inequality 9 Patents 12 NBER Profiles 15 Conferences 17 Bureau News 22 Bureau Books 48 Current Working Papers 49

Transcript of NBER Reporter · 2003-04-21 · Preparation of the NBER Reporter is under the editorial supervision...

NBER Reporter Spring 2003 1.

NATIONAL BUREAU OF ECONOMIC RESEARCH

SPRING 2003

Program Report

Monetary Economics

Understanding what monetary policy can do to enhance economicperformance (and, just as importantly, what it cannot do) is a continuingchallenge for economic policymakers around the world. Researchers inthe NBER’s Monetary Economics Program contribute to this effort witha combination of theoretical and empirical studies on the effects of cen-tral bank actions and the design of monetary policy. These studies arecirculated as NBER Working Papers, presented and discussed at regularProgram meetings, and subsequently published in NBER volumes andacademic journals.

Our regular program meetings also aim to facilitate interactionbetween researchers working in universities and those working in centralbanks. Much frontier research on monetary economics occurs withinresearch staffs of the Federal Reserve System and other central banksaround the world. These central bank researchers often are invited topresent their work and to participate in the discussion of other recentstudies. We are delighted that Ben S. Bernanke, one of our long-termmembers and program director for the past two years, was appointed byPresident George W. Bush in 2002 to become a Governor of the FederalReserve. When he was confirmed, I returned to the role of ProgramDirector, which I had held previously.

In this report, I summarize a few of the strands of research on mon-etary economics that have engaged NBER researchers in recent years.None of these issues is fully settled, but significant progress has beenmade. In the process, I will offer a few of my own judgments about whatwe know and about where more research is still needed.

The Dynamic Effects of Monetary Policy

According to textbook theory, changes in monetary policy influenceemployment and production in the short run but, in the long run, affect

Reporter OnLine at: www.nber.org/reporter

NBER

N. Gregory Mankiw*

NBER Service BringsYou New Data for Free

A new, free NBER email serv-ice gives you daily email links to allU.S. government data releases,including unemployment, trade,interest rates, GDP, etc. We keeptrack of your preferences andemail you the requested linkswhen they are released. To sign upfor any or all of the governmentreleases, visit www.nber.org/releasesand register your choices. IT’SFREE!!

Reporter

* Mankiw is Director of the NBER’s Program on Monetary Economics and theAllie S. Freed Professor of Economics at Harvard University (see page 24). Thenumbers in brackets throughout this report refer to NBER Working Papers. A com-plete list of NBER Monetary Economic Working Papers can be found athttp://papers.nber.org/papersbyprog/ME.html

IN THIS ISSUE

Program Report:Monetary Economics 1

Research Summaries:Explaining Exchange Rate Behavior 5

Health, Income, and Inequality 9Patents 12

NBER Profiles 15Conferences 17

Bureau News 22Bureau Books 48

Current Working Papers 49

2. NBER Reporter Spring 2003

The National Bureau of Economic Research is a private, nonprofit researchorganization founded in 1920 and devoted to objective quantitative analysis ofthe American economy. Its officers and board of directors are:

President and Chief Executive Officer — Martin FeldsteinVice President for Administration and Budget — Susan ColliganController — Kelly Horak

BOARD OF DIRECTORS

Chairman — Michael H. MoskowVice Chairman — Elizabeth E. BaileyTreasurer — Robert Mednick

DIRECTORS AT LARGE

Peter Aldrich Jacob A. Frenkel Michael H. MoskowElizabeth E. Bailey Judith M. Gueron Alicia MunnellJohn Herron Biggs Robert S. Hamada Rudolph A. OswaldAndrew Brimmer George Hatsopoulos Robert T. ParryJohn S. Clarkeson Karen N. Horn Peter G. PetersonDon R. Conlan Judy C. Lewent Richard N. RosettGeorge Eads John Lipsky Marina V. N. WhitmanMartin Feldstein Laurence H. Meyer Martin B. Zimmerman

DIRECTORS BY UNIVERSITY APPOINTMENT

George Akerlof, California, Berkeley Marjorie B. McElroy, DukeJagdish W. Bhagwati, Columbia Joel Mokyr, NorthwesternWilliam C. Brainard, Yale Andrew Postlewaite, PennsylvaniaMichael J. Brennan, California, Los Angeles Uwe Reinhardt, PrincetonGlen G. Cain, Wisconsin Nathan Rosenberg, StanfordFranklin Fisher, MIT Craig Swan, MinnesotaSaul H. Hymans, Michigan David B. Yoffie, Harvard

Arnold Zellner, Chicago

DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS

Mark Drabenstott, American Agricultural Economics AssociationGail Fosler, The Conference BoardA. Ronald Gallant, American Statistical AssociationRichard C. Green, American Finance AssociationRobert Mednick, American Institute of Certified Public AccountantsAngelo Melino, Canadian Economics AssociationRichard D. Rippe, National Association for Business EconomicsJohn J. Siegfried, American Economic AssociationDavid A. Smith, American Federation of Labor and Congress of

Industrial OrganizationsJosh S. Weston, Committee for Economic DevelopmentGavin Wright, Economic History Association

The NBER depends on funding from individuals, corporations, and privatefoundations to maintain its independence and its flexibility in choosing itsresearch activities. Inquiries concerning contributions may be addressed toMartin Feldstein, President & CEO, NBER 1050 Massachusetts Avenue,Cambridge, MA 02138-5398. All contributions to the NBER are taxdeductible.

The Reporter is issued for informational purposes and has not been reviewed bythe Board of Directors of the NBER. It is not copyrighted and can be freelyreproduced with appropriate attribution of source. Please provide the NBER’sPublic Information Department with copies of anything reproduced.

Preparation of the NBER Reporter is under the editorial supervision of DonnaZerwitz.

Requests for subscriptions, changes of address, and cancellations should besent to Reporter, National Bureau of Economic Research, Inc., 1050Massachusetts Avenue, Cambridge, MA 02138-5398. Please include the cur-rent mailing label.

NATIONAL BUREAU OF ECONOMIC RESEARCH

NBERonly prices and inflation rates. When a centralbank slows the rate of money growth, forinstance, the end result will be a lower rate ofinflation, but the transition to lower inflationcan take some time, and it often entails a peri-od of depressed economic activity, includinghigher unemployment. This short-run tradeoffbetween inflation and unemployment is oftencalled the Phillips curve, after the classic studyof this topic by A. W. Phillips in the 1950s.Much research in the NBER MonetaryEconomics Program has been devoted to doc-umenting and explaining these dynamicresponses to monetary policy.

One approach to this empirical issue is tostudy particular episodes of disinflationarypolicy. A classic study following this approachis that by Christina Romer and David Romer.[2966] Following in the footsteps of NBERresearchers Milton Friedman and AnnaSchwartz and their renowned Monetary Historyof the United States, the Romers read throughthe minutes of the meetings of the FederalOpen Market Committee to identify episodesin which Fed policy switched toward a tougherstance against inflation. They find that aftereach of these so-called Romer dates, the econ-omy experienced a substantial decline in pro-duction and employment. The Romers inter-pret their findings as strong evidence for theeffect of monetary policy on real economicactivity.

In another study, Laurence M. Ball lookedat data from OECD countries and found everyepisode in recent history during which theinflation rate experienced a significant and sus-tained decline. [4306] In almost every episodethat Ball identified, the country also experi-enced a period with output below trend. This isconsistent with a short-run tradeoff betweeninflation and real economic activity. Ballreports that the output effects are smaller (thatis, reducing inflation is less costly) when thedisinflation is rapid and when a country hasmore flexible labor contracts.

More recent studies on the dynamic effectsof monetary policy have taken a very differentapproach to the data, but they have reachedbroadly similar conclusions. A commonmethodology is to try to identify “monetarypolicy shocks” — movements in some measureof monetary policy that cannot be predicted orexplained contemporaneously with the eco-nomic variables that typically drive monetarypolicy. These random movements in policy areinterpreted as a natural experiment that can beused to determine the policy’s effect. Oncethese shocks are identified, statistical tech-niques can be used to trace their effects on

Reporter

NBER Reporter Spring 2003 3.

employment, production, inflation,and other variables of interest.

Lawrence J. Christiano, MartinEichenbaum, and Charles Evans sum-marize these studies as follows: “Theliterature has not yet converged on aparticular set of assumptions for iden-tifying the effects of an exogenousshock to monetary policy.Nevertheless, there is considerableagreement about the qualitative effectsof a monetary policy shock in thesense that inference is robust across alarge subset of the identificationschemes that have been considered inthe literature.” [6400] This robust con-clusion includes the classic textbookresult that monetary policy first influ-ences employment and production andonly later affects inflation.

The accumulation of many empiri-cal studies following varied strategieshas led to a consensus among econo-mists about how monetary policyaffects the key measures of macroeco-nomic performance. The exact timingis open to debate, but a rough rule ofthumb is that employment and pro-duction respond about six monthsafter a change in monetary policy,whereas it takes a year or more beforethere is any significant movement inthe inflation rate.

The Amazingly LowInflation andUnemployment of the1990s

During the late 1990s, the UnitedStates experienced an unusual combi-nation of low inflation and low unem-ployment. In 1999, for instance, theunemployment rate averaged 4.2 per-cent for the year, while the inflationrate as measured by the consumerprice index was a mere 2.2 percent. Tosome casual observers, these fortu-itous events suggested that the short-run tradeoff between inflation andunemployment no longer existed, orperhaps that it never existed at all.

Many NBER researchers haverejected this interpretation of therecent data. Indeed, the Phillips curveas an empirical phenomenon is stillvery much alive and well. For example,James Stock and Mark Watson have

examined the best methods for fore-casting inflation. They conclude that,“Inflation forecasts produced by thePhillips curve generally have beenmore accurate than forecasts based onother macroeconomic variables,including interest rates, money andcommodity prices.” [7023].

Why, then, did the U.S. economyexperience a rare combination of lowunemployment and low inflation dur-ing the late 1990s? Part of the answeris that the Fed had produced low infla-tion during the previous decade, whichin turn made credible monetary policy-makers’ claims that they were aimingfor low inflation in the future. Lowerexpectations of inflation shift theshort-run tradeoff between inflationand unemployment in a favorabledirection because these expectationsinfluence the behavior of wage andprice setters.

Yet lower expectations of inflationare not the whole story behind theimpressive macroeconomic perform-ance of the 1990s. Part of the answeralso lies in the widely noted accelera-tion in productivity growth thatoccurred in the second half of thedecade. As Robert J. Gordon puts it,“The post-1995 technological accelera-tion, particularly in information tech-nology and accompanying revival ofproductivity growth, directly con-tributed both to faster output growthand to holding down the inflationrate.” [8771].

The current state of the inflation-unemployment tradeoff is often sum-marized in a statistic called theNAIRU, an ugly acronym that standsfor the non-accelerating inflation rateof unemployment. The NAIRU is likea speed limit for the economy, for ifthe economy grows so fast that unem-ployment falls below the NAIRU,inflation tends to rise. Yet the NAIRUis not constant over time. [5735] Inparticular, several recent studies havesuggested that an acceleration of pro-ductivity growth will cause the NAIRUto fall, at least for a while. [8320, 8421,8614, 8940] The reason for the appar-ent link between productivity growthand the NAIRU is very much an openquestion that should lead to futureresearch. One possible explanation isthat workers are slow to adjust their

wage demands to changes in their pro-ductivity. Until workers adapt to thenew environment, a shift in productiv-ity growth may alter the economy’snormal level of unemployment.

Thus, shifts in productivity growthimpinge on the short-run tradeoffbetween inflation and unemploymentand, indirectly, on the choices facingmonetary policymakers. When produc-tivity slows down, as it did during the1970s, monetary policymakers face adeteriorating short-run tradeoffbetween inflation and unemployment.When productivity speeds up, as it didduring the 1990s, monetary policymak-ers face an improving tradeoffbetween these two measures of eco-nomic performance.

The Puzzle of SluggishInflation

Many empirical studies of the infla-tion process suggest that inflation issluggish. This sluggishness of inflationappears in various guises. In studies ofthe Phillips curve, inflation is found toexhibit substantial inertia; that is, infla-tion is strongly correlated with its ownlagged values. [5735] In studies of par-ticular disinflationary episodes, infla-tion is found to fall only gradually.[4306] In studies that use statisticaltechniques to identify monetary policyshocks and their effects, these shocksare found to have a gradual anddelayed effect on the inflation rate.[5146, 6400] Certainly, these conclu-sions are consistent with the conven-tional wisdom of central bankers, whobelieve they can influence the inflationrate only with a substantial lag. This lagbetween central bank actions and infla-tion is one reason why central banksthat have chosen to target inflationoften look at expected inflation a yearor two ahead when judging whetherthey are on target.

Why does monetary policy influ-ence inflation with such a long lag?The answer is not at all obvious.Standard theories of the real effects ofmonetary policy emphasize the sticki-ness of wages or prices. According tothese theories, monetary fluctuationshave real effects in the short runbecause wages and prices do not adjust

4. NBER Reporter Spring 2003

instantly. [For surveys of this topic, see2285, 4677.] Even if this line ofthought is accepted, however, it fails toexplain the sluggishness of inflation— the change in the price level. Thestickiness of prices can explain whythe price level does not jump to a levelensuring full employment, but theinflation rate is determined by thoseprices that are changing, and thoseprices could respond quite quickly tochanges in monetary policy. Yet forsome reason not found in standardtheories of price adjustment, theydon’t. The failure to explain sluggishinflation shows that the dynamicbehavior of prices remains a funda-mental puzzle for students of businesscycle theory. [7884]

There have, however, been severalrecent attempts to explain the slug-gishness of inflation. These all departsignificantly from standard models ofprice setting. But several of theattempts are similar in their underlyingassumptions, and this common struc-ture may well point the way toward afinal resolution of the sluggish-infla-tion puzzle. The common assumptionis that price setters are inattentive:prices keep rising after changes inmonetary policy because most pricesetters are not paying close attention tothe policy change and, therefore, keepmarking up prices as if no change hadoccurred.

One approach to modeling inatten-tive price setters, explored by MichaelWoodford, is to use the tools of infor-mation theory and to assume thathumans have a limited channel forabsorbing information. [8673] That is,the human brain is assumed to beimperfect in the same way as a com-puter with a slow internet connectionwould be. Woodford uses this assump-tion to build a model of inflationdynamics. His model can explain slug-gish inflation, as well as the persistenteffects of monetary policy on output.He emphasizes that because not every-one shares the same information, pricesetting depends on “higher orderexpectations.” That is, price setterscare about not only their own expecta-tions of monetary policy, but also theirexpectations of others’ expectations,and their expectations of others’expectations of still others’ expecta-

tions, and so on.Ball has proposed another

approach to this problem. [7988] Hesuggests that when forming expecta-tions of any variable, people optimallyuse all information in the past valuesof that variable, but fail to incorporateinformation from other variables. Thatis, expectations are based on optimalunivariate forecasts. He argues that thisapproach to forecasting is nearlyrational, as multivariate forecasts offeronly slight improvement over univari-ate forecasts. Ball shows that this “nearrational” approach to expectations canexplain why inflation appears so slug-gish in recent data, while it was lesssluggish in data from early in the twen-tieth century.

In work I have undertaken withRicardo Reis, the assumption ofimperfect information among pricesetters is explored from a differentangle. [8290, 8614] We assume thateach period there is a fixed probabilitythat a price setter updates his informa-tion set; otherwise, he continues to setprices based on old plans and outdatedinformation. We show that this modelof “sticky information” can explainwhy inflation is so sluggish, and that itproduces dynamic responses to mone-tary policy similar to those estimated inthe empirical literature.

In all three of these models, infla-tion is sluggish because inflationexpectations respond too slowly tochanging circumstances. ChristopherCarroll has written an intriguingempirical study that gives some sup-port to this prediction. [8695] Carrolluses survey data to compare the infla-tion expectations of the general publicto the inflation expectations of profes-sional forecasters. He reports threenotable findings. First, he confirmsthat professional forecasters are betterat forecasting inflation than is the gen-eral public. Second, he finds that thepublic responds to the professionals’expectations with a lag that averagesabout one year. Third, he reports thatwhen the news media are producingmore stories about inflation, the pub-lic’s expectations adjust more quicklyto the professional forecasts. Thesefindings do not prove that the slug-gishness of inflation is attributable tothe inattentiveness of price setters, but

they are certainly consistent with thathypothesis.

Rules for MonetaryPolicy

The study of monetary policy aimsnot only to understand the effects ofcentral bank actions but also to pro-duce better monetary policy. Towardthis end, a large literature has emergedthat studies monetary policy rules. Apolicy rule is a contingency plan thatspecifies how the central bank willrespond to varying economic condi-tions.

There are two reasons for interestin monetary policy rules. One reasonput forward by some economists isthat monetary policy might possibly bebetter if central banks did not havediscretion but were committed to fol-lowing a monetary rule. Some of theseeconomists have argued that centralbanks use discretion unwisely and endup being the cause, rather than cure,for the business cycle. Others arguethat discretionary monetary policy isinherently inflationary. Monetary poli-cymakers often claim that their aim isprice stability, but once expectationsare formed, they are tempted to renegeon this announcement and take advan-tage of the short-run tradeoff betweeninflation and unemployment. The onlyway to avoid this time-inconsistency, itis argued, is to commit the centralbank to a policy rule.

Even if these arguments againstdiscretionary policy are rejected, how-ever, there is another reason for inter-est in policy rules — as guidelines forpolicymakers with discretion.Monetary policymaking is a difficultbusiness, and policymakers are alwayseager to hear objective, reasonedadvice on how to respond to econom-ic conditions. A policy rule that per-forms well by some criterion can beviewed as such advice.

There is a large literature that usesthe tools of modern monetary theoryto derive optimal policy rules. Anexcellent introduction to this literatureis a paper by Richard Clarida, JordiGali, and Mark Gertler, with the allur-ing title “The Science of MonetaryPolicy: A New Keynesian Perspective.”

NBER Reporter Spring 2003 5.

Research Summaries

In an era characterized by increas-ingly integrated national economies,the exchange rate is the key relativeprice in open economies. As such, agreat deal of attention has beenfocused on characterizing its behavior.Unfortunately, it is unclear how muchsuccess there has been in predictingthis critical relative price. As recentlyremarked, “There may be more fore-casting of exchange rates, with lesssuccess, than almost any other eco-nomic variable.”1 While this characteri-zation may be quite apt — a point Iwill return to later — it should not pre-vent us from attempting to identify theempirical determinants of exchange

rates, an enterprise separate from fore-casting exchange rates.

The Impact ofProductivity Changes

The first major line of inquiry I’vefollowed links changes in productivityto changes in nominal and realexchange rates. There is a long andvenerable literature that links these twovariables theoretically, most notablyassociated with Balassa andSamuelson.2 In these models, differ-ences in productivity levels betweentraded and nontraded sectors affectthe relative prices of these goods.Further, with traded goods pricesequalized in common currency terms,real exchange rates — which incorpo-rate the prices of nontraded good —will be affected.

The post-War yen has been the tra-ditional candidate for explanation by

this type of model.3 In addition, themodel typically is applied to economiesexperiencing rapid growth, since suchgrowth often is associated with rapidproductivity change in the tradable(manufacturing) sector. Hence, a natu-ral application of the model is to theEast Asian countries. Unfortunately,the data necessary for a direct test ofthe model do not readily exist. Instead,most analyses rely on observations onrelative prices to infer the validity ofthe approach. In order to conduct adirect test, I compiled sector-specificemployment and output data forChina, Indonesia, Japan, Korea,Malaysia, Philippines, Singapore,Taiwan, and Thailand, and estimatedthe implied relationships. The time-series evidence did not support themodel except in a few cases. Usingpanel regression techniques adapted topersistent time series,4 I find that themodel applies to the set of countriesincluding Indonesia, Japan, Korea,

Explaining Exchange Rate Behavior

Menzie D. Chinn*

* Chinn is a Research Associate in theNBER’s Program on International Financeand Macroeconomics and an AssociateProfessor of Economics at the University ofCalifornia, Santa Cruz. His profile appearslater in this issue.

[7147] One conclusion from this liter-ature is that optimal policy can oftenbe written as a form of a “Taylor rule,”according to which the short-terminterest rate set by the central bankresponds to inflation and a measure ofreal economic activity, such as thedeviation of output from its potential.[For a recent example, see 9149; for anopposing view, see 9421.]

Another conclusion from this liter-ature is that optimal policy should

obey the “Taylor principle,” whichstates that the nominal interest rateshould rise more than one-for-onewith the inflation rate. In many stan-dard models of the business cycle, thisprinciple ensures that shocks to theeconomy do not induce inflation to getout of control. There is considerableevidence that the successes of mone-tary policy over the two decades, com-pared to the problems in the 1970s,can be explained by reference to the

Taylor principle. [6442, 6768, 8471,8800] That is, the Fed has respondedaggressively to changes in inflationwhen choosing its target interest rateduring the recent period, whereas theFed appears to have responded muchless to inflation in the earlier period.This insufficient response to inflationmay explain why inflation got out ofhand in the United States in the 1970s.

6. NBER Reporter Spring 2003

Malaysia, and the Philippines.5 Part ofthe reason for the limited extent of thefinding may be that measured tradedgoods prices do not appear to beequalized, especially when the pricespertain to bundles of goods that arechanging rapidly. After all, the compo-sition of exports of Malaysia todaybears little resemblance to that of fortyyears ago.

Interestingly, there is some evi-dence that the productivity effectapplies even for more developedeconomies. Louis D. Johnston and Iexamined sector-specific productivitylevels for 14 OECD countries. Usingpanel cointegration methods, wefound that productivity levels did mat-ter for dollar-based real exchange ratelevels in the long run, although otherfactors mattered as well. These otherfactors included government spendingand the terms of trade. In a closelyrelated paper, we found that the sameconclusions held for trade-weightedOECD real exchange rates.6

More recently, Ron Alquist and Ihave examined the behavior of theeuro/dollar exchange rate, drawinginspiration from the large literaturethat ascribed the strength of the dollarand the weakness of the euro to thediffering prospects for accelerated pro-ductivity growth rates associated withthe diffusion of the New Economy.Using aggregate productivity datafrom 1985 to 2001, we found that pro-ductivity was strongly related to theeuro/dollar rate. One of the paradox-es of the results is that according tothe estimates, each one percentagepoint increase in the productivity dif-ferential between the United States andthe eurozone economies results in areal dollar appreciation of between 2and 5 percent. While other studieshave detected effects of a similarnature, the magnitude is somewhatlarger than has been found previously.Furthermore, it is hard — althoughnot impossible — to rationalize themagnitude of the effect theoretically.A combination of demand side effectsand an increase in productivity, local-ized to the technologies used in theUnited States, is one interpretation.7

Overvaluation

Models of purchasing power parity,or the Balassa-Samuelson hypothesis,naturally lend themselves to the exer-cise of determining whether a curren-cy is “overvalued” or otherwise mis-aligned. Indeed, one implication of theBalassa-Samuleson hypothesis is thatthe standard practice of measuringmisalignments as deviations from lin-ear trends is likely to provide mislead-ing conclusions. In work I conductedin the wake of the crises of 1997-8, Iasked whether the East Asian curren-cies were overvalued, given the possi-bility that the standard practice wasapplied inappropriately.

A long-run relationship betweenexchange rates and relative prices existsfor all currencies, with respect to atleast one reference currency (dollar oryen) or price deflator (CPI or PPI). Myresults indicate that the Malaysian ring-git, Philippine peso, and Thai baht wereovervalued a month before the bahtdevaluation in July 1997. On the otherhand, the Indonesian rupiah, Koreanwon, and Singapore dollar appear tohave been undervalued. Of theseresults, the implied undervaluations ofthe rupiah and won are the most count-er-intuitive, since these two currenciessuffered precipitous declines in value.Consequently, the widely held view thatcurrency overvaluation was at the heartof each of the East Asian currencycrises lacks credibility (although over-valuation probably did play some role).

Real Exchange RateBehavior and MarketCharacteristics

A large body of work has sought tocharacterize the adjustment of the realexchange rate toward its long-runvalue. Often, the long-run realexchange rate is thought to be whatsets the price of identical baskets ofgoods to be equal, when expressed incommon currency terms; this condi-tion often is termed purchasing powerparity. The mystery arises from thestylized fact that the adjustment takeslonger than what can be rationalizedby sticky prices.8 Yin-Wong Cheung,Eiji Fujii, and I merge the literature on

real exchange rates with that on indus-trial organization factors suggested bythe New Keynesian literature.9 We cal-culate real exchange rates sector-by-sector (for example, for chemicals, orfor fabricated metal products), andrelate the pace at which these realexchange rates revert to their long-runvalues to the characteristics of thosesectors, including the amount of intra-industry trade, size of price-cost mar-gins (a proxy measure for the degree ofsubstitutability of goods) and otherfactors thought to be importantincluding distance, exchange ratevolatility and inflation rates.

The econometric results revealconsiderable evidence for the hypothe-sis that market imperfection is associ-ated with high persistence in devia-tions from purchasing power parity. Ingeneral, the two measures of marketimperfection — a price-cost marginand an index of intra-industry trade —are significant across different specifi-cations and have a positive impact onreal exchange rate persistence. Therobustness of the market structureeffects stands in stark contrast with theresults pertaining to the macroeco-nomic variables, which can yield coef-ficient estimates that vary across modelspecifications, and occasionally have asign different from what the theorypredicts. Overall, our analysis uncoverspositive evidence of market structureeffects on real exchange rate persist-ence.

Interest Rates, ExchangeRates and Expectations

A common method of predictingasset prices uses market-based indica-tors. For instance, futures prices oftenare cited as forecasts of commodities.Forward rates — agreements set todayfor a trade of currencies in the future— would seem to be an ideal indicatorfor the future exchange rate.Equivalently, according to a no-arbi-trage profits condition, when financialcapital is free to move, the forward rateequals the current exchange rateadjusted by the interest differential. Inreality, forward rates for developedeconomy currencies typically arebiased predictors of future spot rates;

NBER Reporter Spring 2003 7.

indeed, when interest differentialspoint to a dollar depreciation, the dol-lar on average appreciates. This wellknown fact led Jeffrey Frankel andKenneth Froot to use survey data toassess whether, for the major curren-cies, this bias was attributable to theexistence of a risk premium or tobiased expectations. In two importantworks, they conclude that the expecta-tions of market participants werebiased, and further that there was littleevidence that the bias in the forwardrate was caused by the presence of anexchange risk premium.10 Frankel and Iexamined a larger number of curren-cies and once again found evidence ofbiased expectations for 25 currenciesover a three-year period.11 Interestingly,in examining forward rate bias in alarger set of currencies (17), we findsomewhat more evidence in favor ofan exchange risk premium.12 To theextent that one believes that such riskpremiums arise from the differentiatednature of the bonds issued by separategovernments, the result makes sense.For instance, U.S. and German bondsmay be more substitutable than U.S.and Swedish bonds.

Guy Meredith and I13 investigatewhether interest rate differentials pointin the wrong direction for exchangerate changes for horizons much longerthan typically studied: five and tenyears, versus the one month or oneyear used in earlier studies. We findthat at these horizons, this perversecorrelation largely disappears. Whilethis finding appears robust to a num-ber of variations, its statistical signifi-cance has been disputed, given thesmall number of independent obser-vations in the post-Bretton Woods era(for example, five non-overlappingfive-year horizons). Hence, we usepanel regressions and confirm thefinding.

The interpretation of these resultsis complicated by the lack of agree-ment on the origins of the forwardrate bias. We propose a model whereinshocks to the interest rate parity rela-tionship (perhaps because of noisetraders) spur a central bank reactionfunction that serves to make exchangechanges correlated negatively withinterest differentials. Because centralbanks only can control short-term

interest rates, the effect is most pro-nounced at short horizons. Since long-term interest rates are a weighted aver-age of short-term interest rates, theeffect is muted at longer horizons.Further research may illuminate alter-native explanations.

What Do MarketParticipants Think?

The work previously recounteduses empirical methods to discern thedeterminants of exchange rate move-ments. Taking a different tack, Cheungand I conduct a survey study of for-eign exchange traders in the UnitedStates.14 Our results indicate that: morethan half of market respondentsbelieve that large players dominate inthe dollar-pound and dollar-Swissfranc markets, and technical tradingbest characterizes about 30 percent oftraders, with this proportion risingfrom five years ago. The responses alsosuggest that news about macroeco-nomic variables is incorporated rapidlyinto exchange rates, although the rela-tive importance of individual macro-economic variables shifts over time.Finally, economic fundamentals appearto be more important at longer hori-zons, while short-run deviations ofexchange rates from their fundamen-tals are attributed to excess speculationand institutional customer/hedge fundmanipulation.

Perhaps, unsurprisingly given themixed findings regarding purchasingpower parity, traders do not view theparity condition as a useful concept,even though a significant proportionof them believe that it affectsexchange rates at horizons of over sixmonths. Interestingly, these particularfindings do not appear to be location-specific. Ian W. Marsh, Cheung, and Iconducted a similar survey of U.K.-based foreign exchange dealers in1998.15 We confirm that many of thesecharacteristics also pertain to that mar-ket. Moreover, we find that there isclear heterogeneity of traders’ beliefs,but it is not possible to explain thesource of these disagreements interms of institutional detail, rank, ortrading technique (for example, techni-cal analysts versus fundamentalists).

Are Exchange RatesPredictable?

One of the key issues dominatingthe empirical literature is whetherexchange rates can be predicted.Previous assessments of nominalexchange rate determination havefocused on a narrow set of modelstypically of the 1970s vintage. Thecanonical papers in this literature areby Meese and Rogoff, who examinedmonetary and portfolio balance mod-els.16 These papers established the styl-ized fact that it is extremely difficult tobeat a random walk on a consistentbasis. Succeeding works by Mark andby Chinn and Meese overturned theseresults, but only at long (three or fouryear) horizons.17

More recently, several studies havere-evaluated the long-horizon results.Faust, Rogers, and Wright argue thatthe success of long-horizon regres-sions is specific to the particular timeperiod examined by Mark and Chinnand Meese.18 In work co-authored withCheung and Pascual,19 I also re-assessexchange rate prediction. Using awider set of models that have beenproposed in the last decade — interestrate parity, productivity based models,and a composite specification incorpo-rating sticky-price, productivity, andportfolio balance models — we com-pare these models against a bench-mark, the Dornbusch-Frankel stickyprice monetary model.

We examine the model’s perform-ance at various forecast horizons (1quarter, 4 quarters, 20 quarters) usingdiffering metrics (mean squared error,direction of change), as well as the“consistency” test proposed in Cheungand Chinn.20 About half of the esti-mates are based upon specificationsthat use contemporaneous informa-tion (that is, the forecast of December2000 uses December 2000 data on theright-hand side variables), while halfuse only lagged information (that is theDecember 2000 forecast uses dataeither one quarter, one year, or 5 yearsprior.) Consequently, half of our spec-ifications are at a great informationaldisadvantage.

We find that no model consistentlyoutperforms a random walk, by the

8. NBER Reporter Spring 2003

conventionally adopted mean squarederror measure. However, along a direc-tion-of-change dimension, certainstructural models do outperform arandom walk with statistical signifi-cance. We also find that interest ratespredict quite well, although only at thelongest horizon.

These forecasts are tied to the actu-al values of exchange rates in the longrun, although in a large number ofcases the elasticity of the forecastswith respect to the actual values is dif-ferent from unity. Overall, we find thatmodel/specification/currency combi-nations that work well in one period donot necessarily work well in anotherperiod.21

While the results are not very posi-tive, they do suggest that along somedimensions, structural models havepredictive power. And, it is importantto recognize that we have stacked thedeck against these models having goodpredictive power, in that half of ourestimates do not rely upon contempo-raneous information. So, while usefulforecasting models remain elusive, theidentification of key empirical factorsremains a productive, albeit challeng-ing, enterprise.

1 Alan Greenspan, Testimony of the FederalReserve Board’s semiannual monetary policyreport to the Congress, before the Committeeon Banking, Housing, and Urban Affairs,U.S. Senate, July 16, 2002. 2 B. A. Balassa, “The Purchasing PowerParity Doctrine: A Reappraisal,” Journalof Political Economy, 72 (1964), pp.584-96; and P. A. Samuelson, “TheoreticalNotes on Trade Problems,” Review ofEconomics and Statistics, 46 (1964),pp. 145-54.3 See M. D. Chinn, “Whither the Yen?Implications of an Intertemporal Model ofthe Yen/Dollar Rate,” Journal of theJapanese and International Economies,11 (2) (June 1997), pp. 228-46.4 These are panel regression techniques adapt-ed to data that appear to follow unit rootprocesses. See for instance P. Pedroni,“Purchasing Power Parity in CointegratedPanels,” Review of Economics andStatistics, 83 (4) (2001), pp. 727-31.5 M. D. Chinn, “The Usual Suspects?Productivity and Demand Shocks and Asia-Pacific Real Exchange Rates,” NBER

Working Paper No. 6108, July 1997, andin Review of International Economics,8 (1) (February 2000), pp. 20-43.6 M. D. Chinn and L. D. Johnston, “RealExchange Rate Levels, Productivity andDemand Shocks: Evidence from a Panel of14 Countries,” NBER Working Paper No.5709, August 1996; and M. D. Chinn,“Sectoral Productivity, Government Spendingand Real Exchange Rates: EmpiricalEvidence for OECD Countries,” NBERWorking Paper No. 5943, February 1997,and in Equilibrium Exchange Rates, R.MacDonald and J. L. Stein, eds., Boston:Kluwer Academic Publishers, 1999, pp.163-90. 7 R. Alquist and M. D. Chinn, “Productivityand the Euro-Dollar Exchange RatePuzzle,” NBER Working Paper No. 8824,March 2002. 8 See K. A. Froot and K. S. Rogoff,“Perspectives on PPP and Long-Run RealExchange Rates,” in G. M. Grossman andK. S. Rogoff, eds., Handbook ofInternational Economics, Vol. III,Amsterdam: North-Holland, 1995.9 Y. Cheung, M. D. Chinn, and E. Fujii,“Market Structure and the Persistence ofSectoral Real Exchange Rates,” NBERWorking Paper No. 7408, October 1999;also “Market Structure and the Persistence ofSectoral Deviations from Purchasing PowerParity,” International Journal ofFinance and Economics, 6 (2) (April2001), pp. 95-114. 10 J. A. Frankel and K. A. Froot, “UsingSurvey Data to Test Standard PropositionsRegarding Exchange Rate Expectations,”American Economic Review, 77 (1)(March 1987), pp. 133-53; and “ForwardDiscount Bias: Is It an Exchange RiskPremium?” Quarterly Journal ofEconomics, 104 (1) (February 1989), pp.139-61.11 M. D. Chinn and J. A. Frankel, “Patternsin Exchange Rate Forecasts for 25Currencies,” NBER Working Paper No.3807, December 1994, and “Are ExchangeRate Expectations Biased? Tests for a Cross-Section of 25 Currencies,” Journal ofMoney, Credit and Banking, 26 (4)(November 1994), pp. 759-70.12 J. A. Frankel and M. D. Chinn,“Exchange Rate Expectations and the RiskPremium: Tests for a Cross-Section of 17Currencies,” NBER Working Paper No.3806, August 1991, and Review ofInternational Economics, 1 (2) (June

1993), pp. 136-44. Some of these results areupdated in M. D. Chinn and J. A. Frankel,“Survey Data on Exchange RateExpectations: More Currencies, MoreHorizons, More Tests,” in W. Allen and D.Dickinson, eds., Monetary Policy, CapitalFlows and Financial Market Develop-ments in the Era of FinancialGlobalisation: Essays in Honour ofMax Fry, London: Routledge, 2002, pp.145-67.13 G. Meredith and M. D. Chinn, “Long-Horizon Uncovered Interest Rate Parity,”NBER Working Paper No. 6797,November 1998.14 Y. Cheung and M. D. Chinn,“Macroeconomic Implications of the Beliefsand Behavior of Foreign Exchange Traders,”NBER Working Paper No. 7417,November 1999; and “Traders, MarketMicrostructure and Exchange RateDynamics,” NBER Working Paper No.7416, November 1999, published as“Traders and Exchange Rate Dynamics: ASurvey of the U.S. Market,” Journal ofInternational Money and Finance, 20(4) (August 2001), pp. 439-71.15 Y. Cheung, M. D. Chinn and I. W.Marsh, “How Do UK-Based ForeignExchange Dealers Think Their MarketOperates?” NBER Working Paper No.7524, February 2000. 16 R. Meese and K. S. Rogoff, “EmpiricalExchange Rate Models of the Seventies: DoThey Fit Out of Sample?” Journal ofInternational Economics, 14 (1983),pp. 3-24; and “The Out-of-Sample Failure ofEmpirical Exchange Rate Models: SamplingError or Misspecification?” in J. A. Frenkel,ed., Exchange Rates and InternationalMacroeconomics, Chicago: University ofChicago Press, 1983, pp. 67-105.17 N. C. Mark, “Exchange Rates andFundamentals: Evidence on Long HorizonPredictability,” American EconomicReview, 85 (1995), pp. 201-18; and M.Chinn and R.Meese, “Banking on CurrencyForecasts: How Predictable Is Change inMoney?” Journal of InternationalEconomics, 38 (1-2) (1995), pp. 161-78. 18 J. Faust, J. Rogers and J. Wright,“Exchange Rate Forecasting: The ErrorsWe’ve Really Made,” paper presented at con-ference on “Empirical Exchange RateModels,” University of Wisconsin,September 28-29, 2001. Forthcoming inJournal of International Economics.The long horizon finding has been re-estab-lished in a panel context; see N. C. Mark

NBER Reporter Spring 2003 9.

and D. Sul, “Nominal Exchange Rates andMonetary Fundamentals: Evidence from aSmall Post-Bretton Woods Panel,” Journalof International Economics, 53 (1)February 2001, pp. 29-52.19 Y. Cheung, M. D. Chinn and A. G.Pascual, “Empirical Exchange Rate Modelsof the Nineties: Are Any Fit to Survive?”NBER Working Paper No. 9393,

December 2002. 20 Y. Cheung and M. D. Chinn, “Integration,Cointegration, and the Forecast Consistencyof Structural Exchange Rate Models,”Journal of International Money andFinance, 17 (5) (1998), pp. 813-30.21 These results are confirmed in a relatedpaper which also assesses in-sample fit. See Y.Cheung, M. D. Chinn and A. G. Pascual,

“What Do We Know about RecentExchange Rate Models? In-Sample Fit andOut-of-Sample Performance Evaluated,”mimeo (October 2002), forthcoming in P.DeGrauwe, ed., Exchange Rate Econo-mics: Where Do We Stand? Cambridge:MIT Press for CESifo.

Richer, better-educated people livelonger than poorer, less-educated peo-ple. According to calculations from theNational Longitudinal MortalitySurvey which tracks the mortality ofpeople originally interviewed in theCPS and other surveys, people whosefamily income in 1980 was greater than$50,000, putting them in the top 5 per-cent of incomes, had a life-expectancyat all ages that was about 25 percentlonger than those in the bottom 5 per-cent, whose family income was lessthan $5,000. Lower mortality and mor-bidity is associated with almost anypositive indicator of socioeconomicstatus, a relationship that has come tobe known as “the gradient.” African-Americans have higher but HispanicAmericans lower mortality rates thanwhites; the latter is known as the“Hispanic paradox,” so strong is thepresumption that socioeconomic sta-tus is protective of health. Not onlyare wealth, income, education, andoccupational grade protective, but soare several more exotic indicators. Onestudy found that life-spans were longeron larger gravestones, another thatwinners of Oscars live nearly fouryears longer than those who were

nominated but did not win.Many economists have attributed

these correlations to the effects ofeducation, arguing that more educatedpeople are better able to understandand use health information, and arebetter placed to benefit from thehealthcare system. Economists alsohave emphasized the negative correla-tion between socioeconomic statusand various risky behaviors, such assmoking, binge drinking, obesity, andlack of exercise. They have also point-ed to mechanisms that run from healthto earnings, education, and labor forceparticipation, and to the role of poten-tial third factors, such as discountrates, that affect both education andhealth.

Epidemiologists argue that theeconomists’ explanations at best canexplain only a small part of the gradi-ent; they argue that socioeconomic sta-tus is a fundamental cause of health. Theyfrequently endorse measures toimprove health through manipulatingsocioeconomic status, not only byimproving education but also byincreasing or redistributing incomes.Fiscal policy is seen as an instrumentof public health, an argument that isreinforced by ideas, particularly associ-ated with Richard Wilkinson, thatincome inequality, like air pollution ortoxic radiation, is itself a health hazard.Even if economic policy has no directeffect on health, the positive correla-tion between health and economic sta-tus implies that social inequalities in

wellbeing are wider than would be rec-ognized by looking at income alone.1

Income and Educationamong Cohorts andIndividuals

Christina Paxson and I2 looked atthe relationship between health andeconomic status among Americanbirth cohorts. We focused on the ideathat health is determined by an indi-vidual’s income relative to other mem-bers of a reference group whose mem-bership typically is unobserved by theanalyst. Even if income inequality hasno direct effect on health, the fact thatthe reference groups are not observedmeans that the slope of the relation-ship between health and incomedepends on the ratio of the between-to-within group components ofincome inequality. For example, ifdoctors’ health depends on the incomeof other doctors, and economists’health on the income of other econo-mists, then the health-to-income rela-tionship in the pooled data will flattenif the average incomes of the twogroups pulls apart.3

Among birth cohorts there is astrong protective effect of income onmortality; the elasticity of mortalityrates with respect to income is approx-imately –0.5. These estimates are con-sistent with estimates from the individ-ual data in the National Longitudinal

Health, Income, and Inequality

Angus Deaton*

* Deaton is a Research Associate in theNBER’s Programs on EconomicFluctuations and Growth, and Health Care,and the Dwight D. Eisenhower Professor ofInternational Affairs at PrincetonUniversity. His profile appears later in thisissue.

10. NBER Reporter Spring 2003

Mortality Study (NLMS), and showmuch the same pattern over the lifecycle, with income most highly protec-tive against mortality in middle age, inthe mid-40s for women and the mid-50s for men. Although it is difficult totest for reverse causality in the cohortdata, we can experiment in the NLMSby looking at the effects of income atthe time of interview on the probabil-ity of death over an interval someyears later, thus eliminating or at leastreducing the effects of including in thesample people who are already sick,and whose income is already reducedby the illness that will later kill them.Somewhat surprisingly, there is only asmall reduction in the estimated pro-tective effects of income as we movethe death interval forward from thedate of interview.

Paxson and I also look at therespective roles of education andincome in protecting health. In bothcohort and individual data, income andeducation are protective when ana-lyzed separately. Taken together, thepicture depends on the level of aggre-gation. In the individual data, the effectof each is robust to allowing for theother, which is consistent with theview that both education and incomepromote health in different ways.Education makes it easier to use andbenefit from new health informationand technologies and income makeslife easier more generally, reducingstress and wear and tear, for exampleby having help to look after the chil-dren, or the money to buy first classtravel. In the aggregated cohort data,income and education are more highlycorrelated than in the individual data,so it is harder to distinguish theireffects. Nevertheless, we find that,conditional on education, increases incohort average income are hazardousto health, a finding that is consistentwith other evidence of hazardouseffects of income variation over thebusiness cycle.4,5 Parallel work onBritish birth cohorts also shows a pro-tective effect of education, althoughan additional year of schooling ismuch less protective in Britain than inthe United States.6 Still, cohort incomeis never estimated to be protective ofcohort mortality in Britain, whetheranalyzed in isolation or in competition

with education. Interestingly, analysisof MSA averages shows similar resultsto the American birth cohorts; citieswith higher average education or high-er average income have lower mortali-ty, but conditional on average income,the correlation between income andmortality is negative.7 The contrastbetween the effects of income in theindividual and aggregate data remainsan important unresolved puzzle.

Inequality, Race, andHealth

Why might income inequality be ahealth hazard, and what accounts forthe fact that people die earlier inAmerican states and cities whereincome inequality is higher? If incomeis protective of health, and the rela-tionship is concave, then redistributionfrom rich to poor will improve aggre-gate health, although this effectappears to be too small to explain thegeographical patterns in the UnitedStates. If health depends on others’incomes, for example if health islinked to relative deprivation, thenincome will be protective of health forindividuals, and income inequality willbe hazardous to health in the aggre-gate.8 But if the NLMS is used to lookat the probability of death as a func-tion of income for white males andfemales on a state by state basis, thereis no evidence of any link between theestimated coefficients and state-levelmeasures of income inequality.

Darren Lubotsky and I7 have inves-tigated the relationship betweenincome inequality, race, and mortalityat both the state and metropolitan sta-tistical area level. In both the state andthe city data, mortality is positively andsignificantly correlated with almost anymeasure of income inequality. Becausewhites have higher incomes and lowermortality rates than blacks, placeswhere the population has a large frac-tion of blacks are also places whereboth mortality and income inequalityare relatively high. However, the rela-tionship is robust to controlling foraverage income (or poverty rates) andalso holds, albeit less strongly, for blackand white mortality separately.Nevertheless, it turns out that race is

indeed the crucial omitted variable. Instates, cities, and counties with a high-er fraction of African-Americans,white incomes are higher and blackincomes are lower, so that incomeinequality (through its interracial com-ponent) is higher in places with a highfraction black. It is also true that bothwhite and black mortality rates arehigher in places with a higher fractionblack and that, once we control for thefraction black, income inequality hasno effect on mortality rates, a resultthat has been replicated by VictorFuchs, Mark McClellan, and JonathanSkinner9 using the Medicare recordsdata. This result is consistent with thelack of any relationship betweenincome inequality and mortality acrossCanadian or Australian provinces,where race does not have the samesalience. Our finding is robust; it holdsfor a wide range of inequality meas-ures; it holds for men and women sep-arately; it holds when we control foraverage education; and it holds oncewe abandon age-adjusted mortalityand look at mortality at specific ages.None of this tells us why the correla-tion exists, and what it is about citieswith substantial black populations thatcauses both whites and blacks to diesooner.

In a review of the literature oninequality and health, I note thatWilkinson’s original evidence, whichwas (and in many quarters is still) wide-ly accepted showed a negative cross-country relationship between lifeexpectancy and income inequality, notonly in levels but also, and moreimpressively, in changes. But subse-quent work has shown that these find-ings were the result of the use of unre-liable and outdated information onincome inequality, and that they do notappear if recent, high quality data areused. There are now also a large num-ber of individual level studies explor-ing the health consequences of ambi-ent income inequality and none ofthese provide any convincing evidencethat inequality is a health hazard.Indeed, the only robust correlationsappear to be those among U.S. citiesand states (discussed above) which, aswe have seen, vanish once we controlfor racial composition. I suggest thatinequality may indeed be important for

NBER Reporter Spring 2003 11.

health, but that income inequality is lessimportant than other dimensions, suchas political or gender inequality.10

Social versus MedicalDeterminants of Health

Most of the work on inequality,income, and health looks at cross-sec-tional or geographic data, with thetime-series relatively unexplored.Paxson and I 6 look at income, incomeinequality, and mortality over time inthe United States and the UnitedKingdom. The postwar period usefullycan be broken in two. In the quartercentury up to the early 1970s, therewas steady productivity growth, withmean and median income growing inparallel, and very little change inincome inequality. After 1970, in theUnited States, productivity growth wasmuch slower; although there was agood deal of income growth at the topof the income distribution, real medi-an family income stagnated or fell.Slow income growth was accompaniedby rapid growth in income inequality.The United Kingdom shared the risein income inequality, which was evenmore marked than in the UnitedStates, but did not experience the sameslowdown in the growth of realincomes. If income and incomeinequality are important determinantsof mortality decline, and even allowingfor some background trend decline inmortality, then the United States andthe United Kingdom should have sim-ilar patterns of mortality decline up tothe early 1970s, followed by slowerdecline after 1970, particularly in theUnited States which had an unfavor-able trend in both growth and inequal-ity. But the data show precisely thereverse. Mortality decline acceleratedin both countries after 1970, and thereis no obvious difference in the patternsin the two countries. Indeed, the mostobvious distinction between Britainand the United States is that changes intrends start a few years earlier in theUnited States. These findings suggestthat, as argued by Cutler and Meara, 11

changes in mortality over the last halfcentury in the two countries have beendriven, not by changes in income andincome inequality, but by changes in

risk factors or in medical technology,with the changes being adopted morerapidly in the United States.

The Origins of theGradient

The two way mechanism betweenincome and health is generally difficultto disentangle, but Anne Case,Lubotsky, and Paxson 12 eliminate thechannel that runs from health toincome by focusing on children wherethe correlation between their poorhealth and low family income cannotbe attributed to the lower earnings ofthe children. Using several large,nationally representative datasets, theyfind that children’s health is positivelyrelated to household income, and thatthe relationship between householdincome and children’s health statusbecomes more pronounced as childrengrow older. A large component of therelationship between income and chil-dren’s health can be explained by thearrival and impact of chronic healthconditions in childhood; children havemuch the same health status at birth,but adverse health shocks are moreeffectively reversed by children in bet-ter-off households. Children’s health isclosely associated with long-run aver-age household income, and theadverse health effects of lower perma-nent income accumulate over chil-dren’s lives, so that the children ofpoorer parents arrive at the thresholdof adulthood with lower health statusand educational attainment — the lat-ter, in part, as a consequence of poorhealth. Case, Lubotsky, and Paxsonspeculate that poorer health and con-sequent lower educational attainmentmay compromise poor children’s earn-ings ability in adulthood, and that thegradient in adults is likely a product ofpoor health status and low income inchildhood.

Health Status andEconomic Status inSouth Africa

In many ways, that income shouldbe an important determinant of healthis more plausible in poor countries

than in rich ones. When many peopledo not have enough money to buyfood, adults and children often sufferthe short and long-term effects of apoor diet, and parents who do nothave enough money to feed their chil-dren report severe consequences fortheir own wellbeing. Anne Case hasused data from a new integrated sur-vey of health and economic wellbeingin South Africa to examine the impactof the South African old age pensionon the health of pensioners, and ofthe prime aged adults and childrenwho live with them.13 Her work findsevidence of a large causal effect ofincome on health status — working atleast in part through sanitation and liv-ing standards, in part through nutri-tional status, and in part through thereduction of psychosocial stress. Thepension is used to upgrade householdfacilities, some of which have conse-quences for health. The household’swater source being on-site and thepresence of a flush toilet are both sig-nificantly more likely, the greater thenumber of years of pension receipt inthe household. In addition, the pres-ence of a pensioner in the householdon average reduces the probability ofan adult skipping a meal by 20 percent,and the presence of two pensionersreduces the probability by 40 percent.All adults in the survey were askedabout depression, which is inextricablylinked to stress and health status. Forhouseholds pooling income, the pres-ence of pensioners significantlyreduces reported depression, and theeffect is larger the greater the numberof pensioners. Governments interest-ed in improving health status may findthe provision of cash benefits to beone of the most effective policy toolsavailable to them. And cash provides ayardstick against which other healthinterventions can be measured.

Case finds that limitations in activi-ties of daily living (ADLs) are associat-ed with worse health status among theelderly and near elderly in SouthAfrica, and that limitations for womenare associated with larger erosions inhealth status than are those for men.Pensioners with limitations in ADLsreport better health status than doolder adults with the same limitationsbut who do not receive the pension. In

12. NBER Reporter Spring 2003

addition, older adults in larger house-holds report better health status withlimitations in ADLs than do otherolder adults. These results are consis-tent with a model in which money (inthe form of a pension) brings help(purchased or volunteered) whenrespondents cannot dress or bathe bythemselves.

1 A. S. Deaton, “Policy Implications of theGradient of Health and Wealth,” HealthAffairs, 21 (March/April 2002), pp.13–30.2 A. S. Deaton and C. Paxson, “Mortality,Education, Income and Inequality amongAmerican Cohorts,” NBER Working PaperNo. 7140, May 1999, and in D. A. Wise,Themes in the Economics of Aging,Chicago: University of Chicago Press, 2001,pp. 129-65.3 A. S. Deaton, “Inequalities in Income andInequalities in Health,” NBER WorkingPaper No. 7141, May 1999, and in F.Welch, The Causes and Consequencesof Increasing Inequality, Chicago:University of Chicago Press, 2001, pp.

285–313.4 C. J. Ruhm, “Are Recessions Good for yourHealth?” NBER Working Paper No.5570, May 1996, and in QuarterlyJournal of Economics, 115 (2000), pp.617-50.5 U. Gerdtham and C. J. Ruhm, “DeathsRise in Good Economic Times: Evidencefrom the OECD,” NBER Working PaperNo. 9357, December 2002.6 A. S. Deaton and C. Paxson, “Mortality,Income, and Income Inequality among Britishand American Cohorts,” NBER WorkingPaper No. 8534, October 2001, and forth-coming in D. A. Wise, ed., Perspectiveson the Economics of Aging, Chicago:University of Chicago Press, forthcoming.7 A. S. Deaton and D. Lubotsky,“Mortality, Inequality and Race inAmerican Cities and States,” NBERWorking Paper No. 8370, July 2001, andSocial Science and Medicine, forthcom-ing.8 A. S. Deaton, “Relative Deprivation,Inequality, and Mortality,” NBERWorking Paper No. 8099, January 2001.9 V. R. Fuchs, M. McClellan, and J.

Skinner, “Area Differences in Utilization ofMedical Care and Mortality among U.S.Elderly,” NBER Working Paper No.8628, December 2001.10 A. S. Deaton, “Health, Inequality, andEconomic Development,” NBER WorkingPaper No. 8318, June 2001, and inJournal of Economic Literature, (41)(March 2003), pp. 113-58.11 D. M. Cutler and E. Meara, “Changes inthe Age Distribution of Mortality over the20th Century,” NBER Working Paper No.8556, October 2001.12 A. Case, D. Lubotsky, and C. Paxson,“Economic Status and Health in Childhood:the Origins of the Gradient,” NBERWorking Paper No. 8344, June 2001, andAmerican Economic Review, 92 (5)(2002), pp. 1308-34.13 A. Case, “Does Money Protect HealthStatus? Evidence from South AfricanPensions,” NBER Working Paper No.8495, October 2001, and in D. A. Wise,ed., Perspectives on the Economics ofAging, Chicago: University of ChicagoPress, forthcoming.

Patents

Jean O. Lanjouw*

The past twenty years have seenvery significant changes in U.S. patentlaw and policy, strengthening the rightsof inventors and significantly expand-ing those rights across the globe. TheUnited States has broadened areas inwhich patents can be received, notablysoftware, genetic information, andbusiness methods; has instituted a newunified Court of Appeals for theFederal Circuit to hear appeals onpatent cases from all district courts;

and has given universities and govern-ment laboratories the right to patentand license the output of publicly-funded research. Bilateral negotiationsand, more recently, internationaltreaties also have led other countries torevise their patent systems. In particu-lar, NAFTA and the intellectual prop-erty component of the treaty establish-ing the World Trade Organization -WTO - (known as TRIPS) extensivelyharmonized and extended patentrights internationally. Jaffe discussesthese changes and surveys relatedempirical studies.1

The domestic patent reforms havebeen driven by the emergence of newareas of research and commerce, andby the view that a healthy knowledge-

based economy requires strong protec-tion of intellectual property (IP). Atthe same time, however, serious con-cerns have arisen. My work on thepatent system as an institution hasfocused on two such areas of concern.The first is the large and growing costsassociated with litigating patent rights.The second is the extension of patentrights on pharmaceuticals to countriesin the developing world where drugaccess is already limited by extremepoverty.

Patent Enforcement

Dealing with patent disputes is partof business life for most firms, but it is

* Lanjouw is a Faculty Research Fellow inthe NBER’s Program on Productivity andan associate professor of economics at theUniversity of California, Berkeley. Her pro-file appears later in this issue.

NBER Reporter Spring 2003 13.

not a phenomenon well understood byeconomists. Josh Lerner and I2 haveprovided a survey of the small body ofempirical research on patent litigationthat was available in the mid-1990s3. Ina set of papers4 Mark Schankermanand I provide an empirical basis forevaluating what has happened inpatent litigation and its implicationsfor R and D incentives and patent pol-icy. We study the determinants ofpatent suits and settlements during1978-99 by linking detailed informa-tion from the U.S. patent office, thefederal courts, and industry sources. Inorder to characterize litigation risk, westart from a set of almost 10,000 liti-gated patents and then draw a match-ing random set from the universe ofpatents, controlling for technology andyear of application. This very compre-hensive database reveals that, contraryto popular perception, the incidence ofpatent suits has not been rising onceone controls for the rapid increase inpatenting itself and a shift towardmore litigious technology areas. How-ever, we also show that litigation isconcentrated and primarily involvesfirms and patents with particular char-acteristics. For example, the risk that agiven patent is subject to a suit is muchhigher if it is owned by an individualor a small firm than if it is owned by acorporation. Patentees with a largeportfolio of patents to trade, or othercharacteristics that facilitate “coopera-tive” resolution of disputes, are lesslikely to litigate. At the same time,post-suit outcomes — such as whethera case goes to trial and who wins — donot depend on these characteristics.Thus, small patentees appear to be at adisadvantage in enforcing their patentrights in that their greater litigation riskis not offset by a more rapid resolutionof their suits. Both the benefits ofpatent portfolio and company size insettling disputes, and the heterogeneityin litigation risk, point to the potentialvalue of patent litigation insurance.Currently there are a number ofproviders of litigation insurance in theUnited States and Europe. However,demand has been severely limited byhigh prices while, at the same time,profitability of insuring companies hasbeen undermined by the widespreaduse of pooled prices. Our empirical

analysis could be used to developinsurance pricing schemes that recog-nize the heterogeneity of litigationrisk.

One piece of evidence from ourstudies5 suggests that there is a threatvalue associated with having controlover many patents in an area: firmswith portfolios that are large relative tothe disputants they are likely toencounter are significantly less likely touse the courts. In a paper with JoshLerner, I explore a particular avenuefor strategic use of litigation.6 A plain-tiff may request, and be granted by thecourt, a preliminary injunction pre-venting an accused infringer fromusing the patented innovation duringthe time that a case is being decided. Itis claimed that requests for preliminaryinjunctions requests are used strategi-cally by financially secure plaintiffs togo beyond the avoidance of “irrepara-ble harm” and to extract even greaterprofit by raising the cost of legal dis-putes. In an earlier paper7, Lerner andI develop a model in which differencesin financing costs drive the use of pre-liminary injunctions, and we explorethe implications for efficiency andincentives. Detailed data compiled on250 patent suits and the financial statusof the litigants indicates that largercorporate plaintiffs with high levels ofcash and equivalents are significantlymore likely to request preliminaryinjunctions.

One of the difficulties facedunderstanding the effect of enforce-ment costs on the value of patent pro-tection is that much of the impact ofthose costs may not be in the form ofdirect expenditures on litigation. Mostobviously, 95 percent of suits settlewithout a trial, three-quarters beforeeven a pre-trial hearing.8 Thus evenwhen a suit is filed, legal costs will havemost of their effect on returns byaltering threat points and thus settle-ment terms. More subtly, enforcementcosts may induce indirect changes inbehavior. I develop and estimate amodel of patentee decisions to payannual renewal fees that incorporatesthe cost of enforcement as part of thedecision, in addition to the renewalfees themselves.9 The basic premise ofthe model is that patents lose theirvalue when they would not be

enforced, irrespective of whether oneobserves a suit. With the model esti-mates it is possible to simulate renewalbehavior, and thus patent value, withdifferent menus of legal costs and costallocation rules.10

The Global PatentSystem forPharmaceuticals

The global system of patent rightsalso is undergoing an unusually dra-matic period of evolution. One of themost important institutional changeshas been bringing intellectual propertywithin the system of rules governingthe world trading system. While in thepast the United States and other richcountry governments have used bilat-eral pressure to influence others’ intel-lectual property (IP) laws and enforce-ment, now the standards setting andthe process for resolving internationaldisputes over IP policies have moved,at least in part, to the WTO.

The inclusion of IP within theUruguay Round of the GATT negoti-ations, and the current standardsrequired for membership in the WTO,have been extremely controversial.The main issue has been the treatmentof pharmaceutical patents, includingthe allowed uses of compulsory licens-ing to deal with public health.Historically, many countries have limit-ed the protection of drug products.Concerned with the prospect of high-er prices, the developing countries andtheir advocates have resisted theexpansion of rights over pharmaceuti-cal innovations. On the other side,firms, together with rich country gov-ernments, have insisted on the impor-tance of the worldwide availability ofpatents to support R and D. The con-troversy over TRIPS began as a ratherlimited, though bitter, fight amongexperts at the trade round. With grow-ing awareness of HIV/AIDS and thediscovery of expensive new treat-ments, the question of the appropriateform of global pharmaceutical patentshas moved onto the editorial pages andinto the public eye.

In this polarized debate, there is animportant role for economists inunderstanding the implications of

14. NBER Reporter Spring 2003

global patents (or their restriction) andin using information about pricing andincentives to develop economicallywell-grounded policies. For some yearsI have pursued these twin objectives.In an initial period of informationgathering, I spent half a year in Indiainterviewing firms, non-governmentalorganizations, and government offi-cials about their views of the country’scommitment to introducing pharma-ceutical patents and what it wouldmean for local industry and con-sumers. I then outlined the possiblecosts and benefits of introducing pro-tection for drug products there,11 andpulled together available statistics andinterview data to consider what couldbe said about their magnitudes. Theresulting paper highlights just how lit-tle empirical evidence supports anyposition in this debate.

My on-going research attempts tofill in some of the gaps. In a recentpaper, Iain Cockburn and I12 do thegroundwork for a future assessment ofwhether newly-available patent protec-tion in the developing world increasesprivate R and D on diseases concen-trated in those countries. Theseinclude diseases such as malaria, schis-tosomiasis, and Chagas disease. Thissubclass of products is rarely brokenout in accounts or statistical sources.We put together tables on disease-spe-cific patenting, bibliometric citations,and NIH funding, and survey Indianpharmaceutical firms to determinetheir current focus on products specif-ic to developing country markets. Thepaper gives a baseline picture of pri-vate investment in products treatingtropical diseases before TRIPS; it willbe updated periodically to trackchanges as new patent laws are imple-mented. There are, of course, otherconstraints on investments for poormarkets besides weak patent protec-tion, and the paper relates someinsights on these problems drawnfrom interviews with U.S.-based phar-maceutical executives. (An empiricalassessment of the response of innova-tion to policy and price changes in theenvironmental area is in Lanjouw andMody, 1996.)13

Currently I am examining the rela-tionship between IP protection forpharmaceuticals and related policies

on the speed of new product launchesin a country. In another current proj-ect I am estimating differentiatedproduct demand models for India inorder to estimate the effect on localconsumer welfare of introducingpatent protection.

Part of my recent work has beendirected at using a basic understandingof the tradeoffs implicit in the patentsystem and some basic facts aboutglobal drug markets to devise an eco-nomically rational global patent sys-tem. I discuss various aspects of this ina series of papers.14 The main factualpoint is that global drug markets differenormously by disease. Some diseasesare concentrated almost exclusively inpoor countries. For these, any marketwill largely only serve patients there. Atthe same time, there are many “global”diseases which are just as important inthe poor countries but which haveworldwide markets. For these, spend-ing is heavily concentrated in richcountries. For example, I estimate thatalmost half of the world’s population— some 3 billion people — live incountries that together represent lessthan 2 percent of global spending oncardiovascular drugs.

Because the global markets are sovery different for these two types ofdiseases, the tradeoffs between pricingand incentives when adding protectionin new countries also differ. The opti-mal global patent system would reflectthis fact. It is not obvious, however,how to differentiate protection in afeasible way. In the work cited, I havedeveloped an unusual mechanism fordifferentiating protection across dis-eases. It uses the fact that inventorsmust request permission to file for apatent overseas (a foreign filinglicense). This requirement alreadyexists in the U.S. patent code and alsoin a number of other countries. Themechanism works by requiring inven-tors to make a particular “Declaration”to obtain permission to file overseas.

The system the mechanism wouldallow is one in which patent protectionin poor countries differed across dis-eases depending on the importance ofthose countries’ markets as a potentialsource of research incentives. No one,for example, would argue that Africannations are an important source of

incentives for doing cancer research.The search for a cure for cancer isdriven by demand from the big west-ern markets. However, patent protec-tion in Africa — together withincreased funding — might be animportant spur to malaria research.Thus, the mechanism would give a sys-tem with minimal patent protection inthe poorest countries, allowing them tobenefit from generic production. As acountry developed, protection wouldbroaden gradually to cover more dis-eases, starting with those, like malaria,of particular importance there. Theresult would be a patent system tai-lored to both development levels andto the characteristics of different drugmarkets.

1 A. B. Jaffe, “The U.S. Patent System inTransition: Policy Innovation and theInnovative Process,” NBER Working PaperNo. 7280, August 1999, and in ResearchPolicy, 2000.2 J. O. Lanjouw and J. Lerner, “TheEnforcement of Intellectual Property Rights:A Survey of the Empirical Literature,”NBER Working Paper No. 6296,December 1997, and in the Annalesd’Economie et de Statistique, (49/50)(July 1998), pp. 223-46.3 More recent work involving NBERresearchers includes, for example, Graham etal, 2002, and Hall and Ziedonis, 2001. SeeS. J. H. Graham, B. H. Hall, D. Harhoff,and D. C. Mowery, “Patent Quality Control:A Comparison of U.S. Patent Re-examina-tions and European Patent Opposition,”NBER Working Paper No. 8807,February 2002; and B. H. Hall and R. H.Ziedonis, “The Patent Paradox Revisited:An Empirical Study of Patenting in theSemiconductor Industry, 1979-1999,”NBER Working Paper No. 7062, March1999, and in RAND Journal ofEconomics, (32) (1) (2001), pp. 101-28.4 J. O. Lanjouw and M. Schankerman, “AnEmpirical Analysis of the Enforcement ofPatent Rights in the United States,” in W.M. Cohen and S. Merrill, eds., Patents inthe Knowledge-Based Economy,Washington, D.C.: National Academy Press,2003; J. O. Lanjouw and M. Schankerman,“Enforcing Intellectual Property,” NBERWorking Paper No. 8656, December 2001;and J. O. Lanjouw and M. Schankerman,“Characteristics of Patent Litigation: A

NBER Reporter Spring 2003 15.

Window on Competition,” NBER WorkingPaper No. 6297, December 1997, and inThe Rand Journal of Economics, (32)(1) (2001), pp. 129-51.5 See J. O. Lanjouw and M. Schankerman,“An Empirical Analysis of the Enforcementof Patent Rights in the United States.”6 J. O. Lanjouw and J. Lerner, “Tilting theTable? The Predatory Use of PreliminaryInjunctions, ” NBER Working Paper No.5689, July 1996, and in The Journal ofLaw and Economics, (XLIV) (2)(2001), pp. 573-603. 7 See J. O. Lanjouw and J. Lerner, “TheEnforcement of Intellectual Property Rights:A Survey of the Empirical Literature.”8 See also J. O. Lanjouw and M.Schankerman, “An Empirical Analysis ofthe Enforcement of Patent Rights in theUnited States.”9 J. O. Lanjouw, “Patent Protection in theShadow of Infringement: SimulationEstimations of Patent Value,” NBERWorking Paper No. 4475, September 1993,

and in The Review of EconomicStudies, (65) (1998), pp. 671-710.10 J. O. Lanjouw, “Beyond Lawyers’ Fees:Economic Consequences of a ChangingLitigation Environment,” NBER WorkingPaper No. 4835, August 1994. Revised ver-sion available. See also J. O. Lanjouw, A.Pakes, and J. Putnam, “How to CountPatents and Value Intellectual Property: Usesof Patent Renewal and Application Data,”NBER Working Paper No. 5741,September 1996, and in The Journal ofIndustrial Economics, (XLVI) (4)(December 1998), pp. 405-33.11 J. O. Lanjouw, “The Introduction ofProduct Patents in India: ‘HeartlessExploitation of the Poor and Suffering’?”NBER Working Paper No. 6366, January1998.12

J. O. Lanjouw and I. Cockburn, “NewPills for Poor People? Empirical EvidenceAfter GATT,” NBER Working PaperNo. 7495, January 2000, and in WorldDevelopment, (29) (2) (2001), pp. 265-

89.13 J. O. Lanjouw and A. Mody, “Innovationand the International Diffusion ofEnvironmentally Responsive Technology,”Research Policy, (25) (1996), pp. 549-71.14 J. O. Lanjouw, “Beyond TRIPS: A NewGlobal Patent Regime,” Policy Brief No. 3,The Center for Global Development, July2002, at http://www.cgdev.org/fellows/lan-jouw.html; J. O. Lanjouw, “A Patent Policyfor Global Diseases: U.S. and InternationalLegal Issues,” Harvard Journal of Law& Technology, (16) (1) (Fall 2002); J. O.Lanjouw, “A Patent Proposal for GlobalDiseases,” Policy Brief No. 84, TheBrookings Institution, June 2001; and J. O.Lanjouw, “Intellectual Property and theAvailability of Pharmaceuticals in PoorCountries,” Center for Global DevelopmentWorking Paper No. 5, April 2002, andforthcoming in Innovation Policy and theEconomy, (3).

NBER Profile: Menzie Chinn

Menzie Chinn is a Research Associatein the NBER’s International Finance andMacroeconomics Program and is aNational Fellow at the NBER during the2002-3 academic year. He is alsoProfessor of Economics at the Universityof California at Santa Cruz. During 2000-1, Chinn served as senior staff economistfor international finance on the Councilof Economic Advisers during the Clintonand Bush Administrations. His responsi-bilities included research on Japanesemacroeconomic policy issues and assess-ing the implications of energy marketdevelopments. He also has been a visitingscholar at the IMF, the Federal ReserveBoard, and Humboldt University, Berlin.

Chinn received his A.B. in economicsfrom Harvard University in 1984 and hisM.A. and Ph.D. in Economics from the

University of California, Berkeley. Hebegan teaching at UC Santa Cruz in 1991.His research focuses on the macroeco-nomic interaction between countries,using econometric methods. His earlierwork examined the determinants ofexchange rate behavior among developedcountries, with an emphasis on the role ofproductivity differentials. His laterresearch examines the empirical factorsunderlying currency crises. Recently, hehas examined the interaction betweencapital controls, institutional characteris-tics of economies, and the financial devel-opment of credit and equity markets.

Chinn resides in Santa Cruz, California,and is married to Laura Schwendinger, acomposer at the University of Illinois,Chicago.

16. NBER Reporter Spring 2003

NBER Profile: Angus S. Deaton

Angus S. Deaton is a ResearchAssociate in the NBER’s Programs onEconomic Fluctuations and Growthand Health Care and is the Dwight D.Eisenhower Professor of InternationalAffairs at Princeton University’sWoodrow Wilson School. His currentresearch interests include the determi-nants of health in rich and poor coun-tries and the measurement of povertyand inequality around the world.

Deaton received his B.A., M.A., andPh.D. degrees from CambridgeUniversity in England, where he hasalso taught. A British citizen, he wasProfessor of Econometrics at the

University of Bristol from 1976 to1983.

Deaton is a Fellow of the BritishAcademy, the American Academy ofArts and Sciences, and the Economet-ric Society. In 1978 he was the firstrecipient of the Econometric Society’sFrisch Medal.

He lives in Princeton with his wife,the economist Anne Case. When theyare not working, they like to cook, andthey maintain homes away from homeat the Metropolitan Opera in NewYork, on the Madison River in Ennis,Montana, and in the British AirwaysLounge at Heathrow.

NBER Profile: Jean Olson Lanjouw

Jean O. Lanjouw is a Faculty ResearchFellow in the NBER’s Program onProductivity. She is also a Senior Fellow inEconomic Studies and GovernanceStudies at the Brookings Institution, aSenior Fellow at the Center for GlobalDevelopment, Washington, DC, and anAssociate Professor of Economics in theDepartment of Agricultural and ResourceEconomics, University of California atBerkeley.

Lanjouw obtained her A.B. inMathematics and Economics from MiamiUniversity and her M.Sc. and Ph.D. inEconomics from the London School of

Economics, UK. Her current projectsfocus on domestic and international prop-erty rights issues. Her research has beenpublished in a wide variety of academicjournals including the Review of EconomicStudies, Econometrica, the Journal ofDevelopment Economics, and the Journal ofIndustrial Economics. She also has organizedseveral conferences on patent reform andstatistics and has consulted for the WorldBank, the UNDP, and statistical organiza-tions in South Africa and Brazil.

She and her husband, Peter Lanjouw,currently reside in Washington, D.C. withtheir two children, Max and Else.

NBER Reporter Spring 2003 17.

Conferences

New Development in Empirical International Trade

The 15th annual TRIO confer-ence — one in a series sponsored bythe NBER, CEPR, and TCER tobring together economists from theUnited States, Europe, and Japan —on “New Development in EmpiricalInternational Trade” took place inTokyo on December 10 and 11,2002. This year the JapaneseResearch Institute of Economy,Trade, and Industry was also a co-sponsor of the event. Kyoji Fukao,Hitotsubashi University; TakeoHoshi, University of California, SanDiego; Sadao Nagaoka and DavidWeinstein, NBER and ColumbiaUniversity, organized this program:

Stephen Redding and AnthonyVenables, London School ofEconomics, “Explaining Cross-Country Export Performance:International Linkages and InternalGeography”Discussants: Peter Debaere,University of Texas, and Shumpei

Takemori, Keio University

Donald Davis and DavidWeinstein, NBER and ColumbiaUniversity, “Why Countries Trade:Insights from Firm-Level Data”Discussants: Motoshige Itoh,University of Tokyo, and StephenRedding

Kyoji Fukao; Hikari Ishido,Institute of Developing Economies;and Keiko Ito, ICSEAD, “VerticalIntra-Industry Trade and ForeignDirect Investment in East Asia”Discussants: James Harrigan,Federal Reserve Bank of New York,and Shujiro Urata, WasedaUniversity

James Harrigan and RohitVanjani, Federal Reserve Bank ofNew York, “Is Japan’s Trade (Still)Different?”Discussants: Eiichi Tomiura, KobeUniversity, and Ryuhei Wakasugi,

Yokohama National University

Keith Head and John Ries,University of British Columbia,“Foreign Direct Investment versusExports: A Test of the SelectionHypothesis”Discussants: Kyoji Fukao and JotaIshikawa, Hitotsubashi University

Simon Evenett, World TradeInstitute, “Do All NetworksFacilitate International Commerce?U.S. Law Firms and theInternational Market for CorporateControl”Discussants: Takeo Hoshi andSadao Nagaoka

Takamune Fujii, Aichi University,and Fukunari Kimura, KeioUniversity, “Globalizing Activitiesand the Rate of Survival: PanelData Analysis on Japanese Firms”Discussants: David Weinstein, andLaixun Zhao, Hokkaido University

Redding and Venables investigatethe role of the international productmarket in determining export per-formance. They seek to explain thewide variations in countries’ exportperformance over the last quarter cen-tury. For example, East Asian coun-tries have seen real exports increase bymore than 800 percent since the early1970s, while those of Sub-SaharanAfrica have increased by just 70 per-cent. The authors decompose thegrowth in countries’ exports into thecontribution from increases in externaldemand versus those from improvedinternal supply-side conditions.Building on the result of this decom-position, Redding and Venables ana-lyze the determinants of export per-formance. They find that poor externalgeography, poor internal geography,

and poor institutional quality con-tribute in approximately equal parts toexplain Sub-Saharan Africa’s poorexport performance.

Davis and Weinstein examine thedeterminants of exporting byJapanese. They begin by documentingthe tremendous concentration inJapanese exporting: four firms accountfor 20 percent of all Japanese exports,and 12 firms account for 40 percent ofexports. Japan is a particularly interest-ing laboratory for studying exportbehavior because Japanese dataenables researchers to estimate pro-ductivity, returns, and capital stocksusing more sophisticated techniquesthan they can with U.S. data. Theauthors use the Japanese estimates tolink firm-level exports to threeexplanatory variables: relative produc-

tivity differences; differences in pro-duction techniques; and increasingreturns. Their results suggest that stan-dard models of comparative advantageseem to be quite relevant for under-standing specialization and exportbehavior. In particular, there seems tobe a very robust relationship betweenexporting and firm-level total factorproductivity. Moreover, firms with thehighest growth in capital intensity alsohave the highest growth in exportshares and the propensity to export.Interestingly, the authors find very lit-tle support for the monopolistic com-petition model.

In a related paper, Head and Riesinvestigate whether productivity differ-ences explain why some Japanese man-ufacturers sell only to the domesticmarket while others serve foreign mar-

18. NBER Reporter Spring 2003

kets through exports or foreign directinvestment (FDI). They show thatfirms that export and do FDI are moreproductive than firms that only export.However, they fail to find that higherproductivity is associated with eitherthe export decision or the ratio of FDIto exports.

Fukao, Ishido, and Ito show thatalthough intra-industry trade (IIT) inEast Asia is significantly less than inEurope, this form of trade has grownat a very rapid rate. Moreover, whilethe share of IIT remained almost con-stant from 1996 to 2000 for most EUcountries, it increased rapidly amongEast Asian countries. According tothe authors’ calculations, the share ofvertical IIT in total intra-East Asiantrade grew from 16.6 percent in 1996to 23.7 percent in 2000, while the sharein total intra-EU trade increased onlyslightly, from 37.5 percent to 40 per-cent during the same period. Mostinterestingly, vertical intra-industrytrade (VIIT), that is trade of goodswith very different prices within verynarrow product categories, has riseneven faster. The increase in VIIT in

these countries seems to have a strongpositive correlation with the extent ofthe activities by Japanese electricalmachinery MNEs.

Harrigan and Vanjani explorewhether Japanese trade in manufac-tured goods differs from the rest-of-the world average and from that of theUnited States. Using a simple industry-level gravity model, they construct ameasure of normalized imports bydividing bilateral industry-levelimports by the importer’s aggregateabsorption and the exporter’s industryoutput. They find that Japan importsless than one might expect, but alsoexports less. Moreover, they find thatrelative to the United States, Japaneseexport performance is half as strongtoday as it was in the mid-1980s, andJapan is more open to imports fromthe United States than the UnitedStates is to imports from Japan.

Evenett estimates the impact ofglobal networks of American lawfirms on overseas mergers and acquisi-tions (M&A) by U.S. corporations.Now that many nations can reviewproposed mergers, U.S. law firms can

help clients overcome regulatory hur-dles abroad, effectively greasing themarket for corporate control.However, they can also oppose trans-actions that are inimical to their clients’interests. His findings suggest thatBaker & McKenzie — the U.S. lawfirm with the most overseas offices —has facilitated such transactions,whereas the combined effect of fivesmaller global U.S. law firms has tend-ed to reduce outward U.S. M&A.

Kimura and Fujii present evidenceon how globalizing activities affect therate of survival of Japanese firms.They show a positive correlationbetween survival and the share of for-eign sales after controlling for a seriesof standard variables. They also pres-ent evidence that small firms that aremembers of Japanese corporategroups tend to fail at even higher ratesthan unaffiliated firms do.

These papers will be published in aspecial edition of the Journal ofJapanese and International Economiesscheduled for December 2003.

Fourth Annual Conference in India

On January 13-15, 2003 the NBERand India’s National Council forApplied Economic Research (NCAER)again brought together a group ofNBER economists and about twodozen economists from Indian univer-sities, research institutions, and gov-ernment departments for their fourthannual conference in India. RaghuramG. Rajan, NBER and University ofChicago, and Martin S. Feldstein,NBER and Harvard University, organ-ized the conference jointly withSuman Bery and Shashanka Bide ofNCAER.

The U.S. participants were: Isher J.Ahluwalia, University of Maryland;John H. Cochrane, NBER andUniversity of Chicago; David M.Cutler, Mihir A. Desai, Martin S.Feldstein, and Jeffrey A. Frankel,NBER and Harvard University; AnneO. Krueger, on leave from the NBERat the IMF; Michael Grossman andRobert E. Lipsey, NBER and CityUniversity of New York; and MichaelH. Moskow, NBER Director andPresident of the Federal Reserve Bankof Chicago.

After introductory remarks about

the U.S. and Indian economies byNBER President Feldstein and SumanBery of NCAER, the participants dis-cussed: trade policy and foreign directinvestment; fiscal deficits and how todeal with them; monetary policy andfinancial sector reforms; economicreforms (including privatization), em-ployment, and poverty; and healthissues.

A summary of the conference dis-cussion will be available on the NBERweb site at www.nber.org/india.

NBER Reporter Spring 2003 19.

An NBER Conference on Infla-tion Targeting, organized by MichaelWoodford of NBER and PrincetonUniversity, took place on January 23-25. The conference agenda was:

Mervyn A. King, Bank of England,“What has Inflation TargetingAchieved?”

Lars E. O. Svensson and MichaelWoodford, NBER and PrincetonUniversity, “Implementing OptimalPolicy through Inflation-ForecastTargeting”Discussant: Bennett T. McCallum,NBER and Carnegie-MellonUniversity

Laurence M. Ball, NBER and JohnHopkins University, and NiamhSheridan, International MonetaryFund, “Does Inflation TargetingMatter?”Discussant: Mark Gertler, NBERand New York University

Stephen G. Cecchetti, NBER and

Ohio State University, and JunhanKim, Ohio State University,“Inflation Targeting, Price LevelTargeting, and Output Variability”Discussant: N. Gregory Mankiw,NBER and Harvard University

Marc P. Giannoni, ColumbiaUniversity, and Michael Woodford,“Optimal Inflation Targeting Rules”Discussant: Edward Nelson, Bankof England

Athanasios Orphanides, FederalReserve Board, and John C.Williams, Federal Reserve Bank ofSan Francisco, “ImperfectKnowledge, Inflation Expectations,and Monetary Policy”Discussants: George W. Evans,University of Oregon

Christopher A. Sims, NBER andPrinceton University, “FiscalImplications of Inflation Targeting”Discussant: Stephanie Schmitt-Grohe, NBER and RutgersUniversity

Marvin Goodfriend, FederalReserve Bank of Richmond,“Inflation Targeting in the UnitedStates”Discussant: Donald Kohn, FederalReserve Board

Jiri Jonas, International MonetaryFund, and Frederic S. Mishkin,NBER and Columbia University,“Inflation Targeting in TransitionEconomies”Discussant: Olivier J. Blanchard,NBER and MIT

Ricardo J. Caballero, NBER andMIT, and Arvind Krishnamurthy,Northwestern University, “InflationTargeting and Sudden Stops”Discussant: Ben S. Bernanke,Federal Reserve Board

Panel Discussion: “Is Deflation aThreat for the U.S.?”Martin Feldstein, NBER andHarvard University, MichaelWoodford, and Ben S. Bernanke

King spoke about the experienceof inflation targeting in the UnitedKingdom. Following its exit from theExchange Rate Mechanism inSeptember 1992, the UK adoptedinflation targeting as a means of pro-viding a direct and transparent frame-work for monetary policy. A key com-ponent in the successful implementa-tion of the framework was a willing-ness to explain how the economy wasevolving, particularly in the Bank ofEngland’s Inflation Report. In May 1997,the framework was sharpened by mak-ing the target symmetric, and by set-ting up an independent committee atthe Bank of England to set monetarypolicy. The Committee structure hasproved invaluable in introducing arange of accountable views into thediscussion. Many issues still remain tobe resolved. For example, what typesof central banks benefit from an infla-tion targeting framework? And, overwhat horizon should the central banksbring inflation back to target?

Svensson and Woodford considerthe way in which inflation-forecast tar-geting should be conducted in order tobring about a socially optimal equilibri-um, with an optimal long-run averageinflation rate and optimal transitoryresponses to disturbances. They evalu-ate variant schemes from the point ofview of avoiding stabilization bias,incorporating the kind of historydependence that is needed to causeexpectations to respond optimally toshocks, and achieving determinancy ofequilibrium. They also evaluate thesevariants in terms of the transparencyof the connection with the ultimatepolicy goals and the robustness of thepolicy rule to perturbations of themodel. A suitably designed inflation-forecast targeting rule can implementan optimal equilibrium, at the sametime have a more transparent connec-tion to policy goals, and be morerobust than competing instrumentrules, they conclude.

Ball and Sheridan examine the

effects of inflation targeting on eco-nomic performance, as measured bythe behavior of inflation, output, andinterest rates. They compare sevenOECD countries that adopted infla-tion targeting in the early 1990s to thir-teen that did not. During the targetingperiod, performance improved alongmany this effect, there is no evidencethat inflation targeting is beneficial.

The dramatic improvement inmacroeconomic outcomes during the1990s — stable, low inflation and high,stable growth — can be ascribed atleast partly to improved monetary pol-icy. Central banks became more inde-pendent and many of them adoptedinflation targeting. Cecchetti and Kimexamine the potential for furtherimprovements by refining the conceptof inflation targeting. They construct ageneral model that encompasses abroad array of possible target regimesand apply it to the data. Their resultssuggest that the vast majority of coun-tries could benefit from moving to

Inflation Targeting

20. NBER Reporter Spring 2003

price-path targeting, in which the cen-tral bank makes up for periods ofabove (below) target inflation withlater periods of below (above) targetinflation.

Giannoni and Woodford charac-terize optimal monetary policy for arange of alternative economic models,applying the general theory developedin an earlier paper (NBER WorkingPaper No. 9419). The rules they com-pute have the advantage of being opti-mal regardless of the assumed charac-ter of exogenous additive distur-bances, although other aspects of theirmodel specification do affect the formof the optimal rule. In each case thatthey consider, optimal policy can beimplemented through a flexible infla-tion targeting rule, under which thecentral bank is committed to adjustingits interest-rate instrument so as toensure that projections of inflationand other variables satisfy a target cri-terion. The authors show which addi-tional variables should be taken intoaccount, beyond the inflation projec-tion, and how much weight the vari-able should get for any given parame-terization of the structural equations.They also explain what relative weightsshould be placed on projections fordifferent horizons in the target criteri-on, and the manner and degree towhich the target criterion should behistory-dependent. The authors thenassess the likely quantitative signifi-cance of the various factors consid-ered in the general discussion by esti-mating a small, structural model ofU.S. monetary transmission withexplicit optimizing foundations. Theycompute an optimal policy rule for theestimated model, and it corresponds toa multi-stage inflation-forecast target-ing procedure. Finally, they considerthe degree to which actual U.S. policyover the past two decades has con-formed to the optimal target criteria.

Orphanides and Williams investi-gate the role of imperfect knowledgeabout the structure of the economy inthe formation of expectations, macro-economic dynamics, and the efficientformulation of monetary policy.Economic agents rely on an adaptivelearning technology to form expecta-tions and continuously update theirbeliefs about the dynamic structure of

the economy based on incoming data.The process of perpetual learningintroduces an additional layer ofdynamic interactions between mone-tary policy and economic outcomes.The authors find that policies thatwould be efficient under rationalexpectations can perform poorly whenknowledge is imperfect. In particular,policies that fail to maintain tight con-trol over inflation are prone toepisodes in which the public’s expecta-tions of inflation become uncoupledfrom the policy objective and stagfla-tion results, in a pattern similar to thatexperienced in the United States dur-ing the 1970s. More generally, theauthors show that policy shouldrespond more aggressively to inflationunder imperfect knowledge than underperfect knowledge.

According to Sims, inflation tar-geting may do more harm than good ifthere is a substantial chance that thecentral bank in fact cannot controlinflation. A prerequisite for centralbank control of inflation is appropri-ate coordination with, or backup by,fiscal policy; the nature of the requiredcoordination will depend on whetherand how central bank independencefrom the fiscal authority has beenimplemented. These considerationssuggest that in those countries whereinflation control has been most diffi-cult in the past, inflation targeting maybe least useful. Where inflation controlhas been successful in the past, thebenefits of inflation targeting mayhave more to do with the associatedchanges in the policy process and inthe central bank’s communication withthe public than with the inflation targetitself.

Goodfriend begins by tracing theorigins of the case for inflation target-ing in postwar U.S. monetary history.He describes five aspects of inflationtargeting practiced implicitly by theGreenspan Fed. He argues that lowlong-run inflation should be an explic-it priority for monetary policy. Further,as a practical matter, it is not desirablefor the Fed to vary its short-run infla-tion target. Strict inflation targetingcan be regarded as efficient, con-strained, countercyclical stabilizationpolicy. Finally, Goodfriend suggeststhat the Fed publicly acknowledge its

implicit priority for low long-run infla-tion, that Congress recognize that pri-ority, and that representatives of theFederal Open Market Committee inreturn consider participating in a mon-etary policy forum to better inform thepublic and Congressional oversightcommittees about current monetarypolicy.

Jonas and Mishkin examine theinflation targeting experience in threetransition countries: the Czech Republic,Poland, and Hungary. While thesecountries have missed inflation targets,often by a large margin, they neverthe-less progressed well with disinflation.A key lesson from the experience ofthe inflation targeting transition coun-tries is that economic performance willimprove, and support for the centralbank will be higher, if the centralbanks emphasize not undershootingthe inflation target as much as theyavoid overshooting. Also, economicperformance will be enhanced if infla-tion targeting central banks in transi-tion countries do not engage in activemanipulation of the exchange rate.The relationship between the centralbank and the government in thesecountries has been quite difficult, butthis can be alleviated by having directgovernment involvement in setting theinflation target and with a more activerole for the central bank in communi-cating with the government and thepublic. In addition, having technocratsrather than politicians appointed tohead the central bank may help indepersonalizing the conduct of mone-tary policy and increasing support forthe independence of the central bank.The authors also address the futureperspective of monetary policy in thetransition economies and concludethat, even after the EU accession,inflation targeting can remain the mainpillar of monetary strategy in the threeexamined accession countries beforethey join the EMU.

Emerging economies experiencesudden stops in capital inflows.During these sudden stops, havingaccess to monetary policy is useful,but mostly for “insurance” rather thanaggregate demand reasons. In thisenvironment, a central bank that can-not commit to monetary policy choic-es will ignore the insurance aspect and

NBER Reporter Spring 2003 21.

follow a procyclical rather than theoptimal countercyclical policy. Theinefficiency is exacerbated by the pres-ence of an expansionary bias. In orderto solve these problems, Caballeroand Krishnamurthy propose modify-ing the central bank’s objective toinclude state-contingent inflation tar-gets and to target a measure of infla-tion that overweights non-tradableinflation.

In the panel discussion, Feldsteinemphasized that deflation is particular-ly damaging if the rate of deflation isso great that nominal wages must falland short-term interest rates areapproximately zero. Businesses andhouseholds suffer because the value ofreal assets falls relative to the value ofnominal debts. Although the UnitedStates is not close to that situationnow, a failure to achieve a strongenough recovery might cause the lowinflation rate to shift over time todeflation. In addition to a more expan-sionary monetary policy that includesaction with longer-term securities, theUnited States could reverse deflationby a fiscal stimulus. It is possible tohave a fiscal stimulus without increas-ing the budget deficit by combining atemporary investment tax credit with atemporary increase in the tax on busi-ness income.

Bernanke noted that deflationoften is associated with bad macroeco-nomic outcomes (as in the 1930s, or inJapan today) but not always (Chinatoday). A key issue is whether the zero

bound on nominal interest rates isbinding at full employment; in a rapid-ly growing economy, in which the realinterest rate is high, the bound isunlikely to be relevant, but in a slackeconomy the bound may pose a crucialconstraint. Bernanke argued that thechances of protracted deflation in theUnited States are remote, as the finan-cial system and inflation expectationsare both stable, and policymakers areprepared to act aggressively to resistdeflationary shocks. Should deflationoccur, even if the nominal interest ratereached zero, the Federal Reservecould respond by purchasing assets(including Treasury securities, foreignsecurities, and securities issued by gov-ernment-sponsored enterprises), bylending aggressively through the dis-count window, and by cooperatingwith fiscal authorities. Bernanke notedthat in a monetarist framework, inwhich money is a substitute for a vari-ety of assets, increasing the moneysupply has the potential to affect aspectrum of asset yields. He conclud-ed by suggesting that central bank anti-deflationary policies might be mademore precise by targeting asset pricesrather than by stipulating quantities ofassets to be purchased.

Woodford agreed that the risk ofdeflation in the United States seemsminimal, but argued that the ability ofexpansion of the monetary basethrough open-market operations toincrease nominal aggregate demand isrelatively limited once the zero lower

bound on short-term nominal interestrates is reached, as it has been in Japan.Substitution of money for other short-maturity, riskless nominal assets (alsowith interest yields near zero) in theportfolios of private agents makesvery little difference to the situation ofthose agents and hence their behavior;and the idea that open-market pur-chases of longer-maturity debt orother assets can substantially changerelative prices of alternative financialassets ignores the failure of central-bank attempts to manipulate the termstructure of interest rates in the past.Nonetheless, warnings that deflation islike a “black hole” from which aneconomy may be unable to leave oncethe state is entered are too pessimistic,because they assume a completelymechanical, backward-looking modelof expectations formation. A publiccommitment to a price-level targetshould instead create a situation inwhich deflation will not create expecta-tions of deflation but rather increasedexpectations of inflation, which willtend to automatically limit the extentof the deflation.

These papers and their discussionwill be published by the University ofChicago Press in an NBER ConferenceVolume, Inflation Targeting. Its availabili-ty will be announced in a future issueof the NBER Reporter. They are alsoavailable at “Books in Progress” on theNBER’s website.

*

22. NBER Reporter Spring 2003

Mankiw to Chair CEA

NBER Research Associate N. GregoryMankiw, Director of the NationalBureau’s Program of Research onMonetary Economics, has been nomi-nated by President Bush to becomeChairman of the Council of EconomicAdvisers. His appointment requiresSenate confirmation. Mankiw replacesNBER Research Associate R. GlennHubbard, who will be returning toColumbia University.

Mankiw is the Allie S. FreedProfessor of Economics at HarvardUniversity. He received his A.B. fromPrinceton University and his Ph.D.from MIT. His NBER affiliation beganin 1985, and he has twice served asDirector of the NBER’s MonetaryEconomics Program. Mankiw alsoedited the NBER volume MonetaryPolicy, which was published by theUniversity of Chicago Press in 1994.

Four of the recent Chairs of thePresident’s Council of EconomicAdvisers were also NBER ResearchAssociates at the time of their nomina-tions: Martin Feldstein, appointed byRonald Reagan; Michael Boskin, underPresident George H.W. Bush; JosephStiglitz, President William J. Clinton;and R. Glenn Hubbard, PresidentGeorge W. Bush.

Bureau News

Romers To Direct NBER’s Monetary Economics Program

NBER Research Associates ChristinaD. Romer and David H. Romer willsucceed N. Gregory Mankiw as co-directors of the NBER’s Program onMonetary Economics.

Christina Romer is currently theClass of 1957 Professor of Economicsat the University of California at

Berkeley. She is a specialist in monetaryeconomics and economic history, andhas studied changes in American busi-ness cycles over time and the causes ofthe Great Depression. David Romer isthe Herman Royer Professor inPolitical Economy at the University ofCalifornia at Berkeley. He is a specialistin monetary economics and macroeco-nomic theory. He has conductedresearch on New Keynesian microeco-nomic foundations, inflation, and thedeterminants of cross-country incomedifferences. Together the Romers haveconducted a number of studies on theeffects and determinants of Americanmonetary policy. They are co-editorsof the NBER volume Reducing Inflation:Motivation and Strategy.

Christina Romer received her B.A.from the College of William and Maryin 1981 and her Ph.D in Economicsfrom MIT in 1985. She has receivedthe NSF Presidential YoungInvestigator Award and fellowshipsfrom the Sloan Foundation and theGuggenheim Foundation. DavidRomer received his A.B. fromPrinceton University in 1980 and his

Ph.D. from MIT in 1985. He is theauthor of the leading graduate text-book in macroeconomics, AdvancedMacroeconomics. Both taught at Princetonfrom 1985-8 before joining theBerkeley economics faculty. They aremarried and have 3 children.

NBER Reporter Spring 2003 23.

Ruth P. Mack Dies at Age 99

Ruth Prince Mack, a member ofthe NBER’s research staff in NewYork City from the 1940s through the1960s, died in early March. She was 99years old.

Mack was the author of a Technical

Paper in 1954 titled Factors InfluencingConsumption: An Experimental Analysis ofShoe Buying. Her other work at theNBER was the two Studies in BusinessCycles that she wrote and which werepublished in 1956 and 1967, respective-

ly. Consumption and Business Fluctuations:A Case Study of the Shoe, Leather, HideSequence and Information, Expectations, andInventory Fluctuations: A Study of MaterialsStock on Hand and on Order.

Insurance ProjectThe NBER’s Insurance Project,

directed by Kenneth A. Froot, NBERand MIT, and Howard Kunreuther,NBER and University ofPennsylvania, met in Cambridge onJanuary 31 and February 1. The fol-lowing papers were discussed:

J. David Cummins, University ofPennsylvania, and Christopher M.Lewis, Fitch Risk Management,“Catastrophic Events, ParameterUncertainty and the Breakdown ofImplicit Long-Term Contracting inthe Insurance Market: The Case ofTerrorism Insurance”Discussant: Joan Lamm-Tenant,General Cologne Re CapitalConsultants

Gordon Woo, Risk ManagementSolutions, “Insuring Against Al-Qaeda”Discussant: John Major, GuyCarpenter & Company, Inc.

Thomas Russell, Santa ClaraUniversity, “The Costs and Benefitsof the Terrorism Risk Insurance Act:A First Look”

Discussant: Paul Freeman, Universityof Denver

Darius Lakdawalla, NBER andRAND, and George Zanjani, FederalReserve Bank of New York,“Insurance, Self-Protection, and theEconomics of Terrorism”Discussant: Howard Kunreuther

Alma Cohen, Harvard University,“Asymmetric Information andLearning: Evidence from theAutomobile Insurance Market”Discussant: Gordon Stewart,Insurance Information Institute

Bradley Herring, Emory University,and Mark Pauly, University ofPennsylvania, “Incentive-CompatibleGuaranteed Renewable HealthInsurance Premiums”Discussant: Thomas McGuire,Harvard University

Thomas Davidoff, University ofCalifornia, Berkeley; Jeffrey Brown,NBER and University of Illinois atUrbana-Champaign; and PeterDiamond, NBER and MIT,

“Annuities and Individual Welfare”Discussant: Amy Finkelstein, NBER

Olivier Mahul, INRA, “EfficientRisk Sharing within a CatastropheInsurance Pool”Discussant: Kenneth A. Froot

Alex Boulatov and Dwight Jaffee,University of California, Berkeley,“The Response of CatastropheInsurance Markets to ExtremeEvents: A Real Option Approach”Discussant: Kent Smetters, NBERand University of Pennsylvania

Neil Doherty and Paul Kleindorfer,University of Pennsylvania, “Market,Cotnract Designs, and Mutualization:Insuring Catastrophic Losses”Discussant: David Durbin, Swiss Re

Geoffrey Heal, Columbia University,and Howard Kunreuther, “You CanOnly Die Once: Managing DiscreteInterdependent Risks”Discussant: Nathaniel Keohane, YaleUniversity

Holtz-Eakin to Head CBO

Douglas J. Holtz-Eakin, who hasbeen affiliated with the NBER since1985, has been chosen to head theCongressional Budget Office. On leaveas a professor of economics at SyracuseUniversity, he most recently had beenChief Economist of the President’s

Council of Economic Advisers (CEA).In the past, Holtz-Eakin held aca-

demic appointments at Columbia Uni-versity and Princeton University. Healso served as Senior Staff Economistof the CEA in 1989-90.

Holtz-Eakin’s long-standing interest

is in the economics of public policy.His most recent NBER Working Paper(No. 8261), analyzed the distortionresulting from income versus estatetaxation; in Working Paper No. 7980 hediscussed personal income taxes andthe growth of small firms.

Cummins and Lewis examinethe reaction of the stock prices of U.S.property-casualty insurers to theWorld Trade Center (WTC) terrorist

attack of September 11, 2001.Theories of insurance market equilib-rium and theories of long-term con-tracting predict that large loss events

that deplete capital and increase uncer-tainty will affect weakly capitalizedinsurers more significantly thanstronger firms. The results here are

24. NBER Reporter Spring 2003

consistent with this prediction.Insurance stock prices generallydeclined following the WTC attack.However, the stock prices of insurerswith strong financial ratings rebound-ed after the first post-event week,while those of weaker insurers did not,consistent with the flight-to-qualityhypothesis.

The broad spectrum of domesticterrorism in the United States tradi-tionally has been underwritten as aninsurance risk without the need forelaborate risk management tools. Thedisparate aims, origins, and domicilesof the various U.S. terrorist groupsprovide an intrinsic degree of portfo-lio diversification, and risk accumula-tions are bounded by the limited dam-age objectives and capabilities of disaf-fected U.S. citizens. By comparisonwith the sizeable database of domesticterrorist acts, the record of foreignattacks is barren: until 9/11, the U.S.mainland had not been attacked sincethe British burned the White House in1812. Against a background of al-Qaeda threats to “destroy” America,insurers are obliged by the TerrorismRisk Insurance Act of 2002 to offerinsurance coverage against foreignattacks. Woo outlines steps by whichthis coverage may be managed in arisk-informed manner, using knowl-edge of al-Qaeda modus operandi,expressed in a logical form suitable fordecisionmaking on pricing and accu-mulating terrorism risk.

Russell provides estimates of thecosts and benefits of the TerrorismRisk Insurance Act of 2002. He esti-mates the expected costs to the tax-payer of the federal insurance back-stop to be around $6 billion. The chiefbenefit of the Act is the increase inconstruction jobs made possible byconstruction loans now protectedagainst terrorism risk by the resuscita-tion of the private terrorism insurancemarket. Russell estimates the numberof these jobs at around 65,000, sub-stantially less than the administrationclaim of 300,000. Other benefits arelikely to be small, suggesting that (atthe rate of $10,000 per job created) thecosts of the Act may substantiallyexceed the benefits.

Lakdawalla and Zanjani investi-gate the rationale for public interven-

tion in the terrorism insurance market.They argue that government subsidiesfor terror insurance are aimed, in part,at discouraging self-protection andlimiting the negative externalities asso-ciated with self-protection. Cautiousself-protective behavior by a target canhurt public goods, including nationalprestige, if it is seen as “giving in” tothe terrorists, and may increase theloss probabilities faced by others if itencourages terrorists to substitutetoward more vulnerable targets. Theauthors argue that these externalitiesdistinguish the terrorism insurancemarket and help to explain why avail-ability problems in this market haveengendered much stronger govern-ment responses than similar problemsin other catastrophe markets.

Cohen tests the predictions ofadverse selection models using datafrom the automobile insurance market.In contrast to what recent research hassuggested, she finds that the evidenceis consistent with the presence ofinformational asymmetries in this mar-ket: higher insurance coverage is corre-lated with more accidents. Consistentwith the presence of learning by poli-cyholders about their risk type, such acoverage correlation exists only forpolicyholders who have had three ormore years of driving experience priorto joining their insurer. Consistentwith the presence of learning by insur-ers about repeat customers, Cohenfinds that, as the experience of theinsurer with a group of policyholdersincreases, the coverage-accidents cor-relation declines in magnitude andeventually disappears. Finally, consis-tent with insurers having more infor-mation about their repeat customersthan would be available to other insur-ers, she finds that policyholders under-report their past claims when joiningthe insurer and that policyholders wholeave the insurer are disproportionate-ly those with a poor claims historywith the insurer.

In practice, an age profile of pre-miums that decreases with age mightresult in such high premiums foryounger individuals that insurancemight be considered unaffordable.Herring and Pauly use medicalexpenditure data to estimate an opti-mal competitive age-based premium

schedule for a benchmark renewablehealth insurance policy. They find thatthe amount of prepayment by youngerindividuals necessary to cover futureclaims is mitigated by three factors:high-risk individuals either will recoveror die; low-risk expected expenseincreases with age; and the likelihoodof developing a high-risk conditionincreases with age. The resulting opti-mal premium path generally increaseswith age. In addition, the authors findthat actual premium paths exhibited bypurchasers of individual insurancewith guaranteed renewability are closeto an optimal schedule.

Davidoff, Brown, andDiamond advance the theory ofannuity demand in several new direc-tions. First, they derive sufficient con-ditions under which complete annuiti-zation is optimal, showing that thiswell-known result holds true in a moregeneral setting than in Yaari (1965).Specifically, when markets are com-plete, sufficient conditions need notimpose exponential discounting,intertemporal separability, or obedi-ence of expected utility axioms onpreferences; nor do annuities need beactuarially fair, or longevity risk theonly source of consumption uncer-tainty. All that is required is that con-sumers have no bequest motive andthat annuities pay a rate of returngreater than that of otherwise match-ing conventional assets, net of admin-istrative costs. Second, the authorsshow that full annuitization may not beoptimal when markets are incomplete.Some annuitization is optimal as longas conventional markets are complete.The incompleteness of markets canlead to zero annuitization, but the con-ditions on both annuity and bond mar-kets are stringent. Third, the authorsextend the simulation literature thatcalculates the utility gains from annu-itization by considering consumerswhose utility depends both on presentconsumption and on “standard-of-liv-ing” to which they have become accus-tomed. The value of annuitizationhinges critically on the size of the ini-tial standard-of-living relative towealth.

Mahul examines optimal cata-strophic risk sharing arrangementswithin a pool whose financial

NBER Reporter Spring 2003 25.

resources may be insufficient to pay allvalid claims in full. Under this threat ofdefault on payment, the mutualityprinciple and the standard allocationof aggregate risk based on individualrisk tolerances hold within a sub-poolof agents who decide to stay in thepool after a cataclysmic event. In thecontext of the insurance market, thisconstraint precludes obtaining a first-best optimal participating policy withfull insurance and a variable premium.The second-best optimal insurancecontract provides full (marginal) cover-age above an ex post variabledeductible. This deductible is such thatthe total claims paid to the policyhold-ers and the financial resources of thepool are equalized. This innovativecontract contrasts with current catas-trophe insurance schemes based onpro rated indemnification should theinsured losses exceed the capital avail-able.

Boulatov and Jaffee connect thetraditional financial and insurance liter-atures in the context of real optiontheory. They use the analogy betweeninsurance and investment under uncer-tainty in order to study the generalequilibrium in an insurance industry

with heterogeneous competing firms.They show that even under theassumption of rationality (defined inthe traditional sense), insurance com-panies should be treated as strategicagents when catastrophic events arepossible. The authors derive the equi-librium investment (insurance policies)using a generalization of the standardmodel of partially reversible invest-ment under uncertainty.

Doherty and Kleindorfer askwhether market insurance can occurnaturally under conditions of ambigu-ity and, if so, what contractual andmarket structure it should assume.They find that, far from impedinginsurance contracting, ambiguity alonecan provide a sufficient basis for risksharing of catastrophic losses. Thegains from risk sharing between theparties cover a wide range of parame-ters and derive from two mechanisms.The first is differences in ambiguityaversion between the primary insurerand the reinsurer. Greater ambiguityaversion on the part of the latter moti-vates reinsurance. This gain is quiteintuitive. If the reinsurer dislikes ambi-guity less than the insurers, then a pricecan be found such that both of them

gain from the transfer of risk. The sec-ond mechanism comes from the cata-strophic nature of the loss. Unless thecatastrophic event hits all primaryinsurers to the same degree, the rein-surance mechanism can be used todiversify risk within the catastrophiclass state.

Heal and Kunreuther extendtheir earlier analysis of interdependentsecurity issues to a general class ofproblems involving discrete interde-pendent risks. There is a threat of anevent that can only happen once, andthe risk depends on actions taken byothers. Any agent’s incentive to investin managing the risk depends on theactions of others. Security problems atairlines and in computer networkscome into this category, as do prob-lems of risk management at organiza-tions facing the possibility of bank-ruptcy, and individuals’ choices aboutwhether to be vaccinated against aninfectious disease. Surprisingly theframework also covers certain aspectsof investment in R and D. Here, theauthors extend their earlier analysis tocover heterogeneous agents and tocharacterize the tipping phenomenon.

The NBER’s new WorkingGroup on Entrepreneurship met inCambridge on February 1. JoshLerner, NBER and HarvardUniversity, organized this program:

Naomi R. Lamoreaux andKenneth L. Sokoloff, NBER andUniversity of California, LosAngeles, “The Decline of theIndependent Inventor: ASchumpeterian Story?”Discussant: Rebecca Henderson,NBER and MIT

Edward Lazear, NBER and

Stanford University,“Entrepreneurship”Discussant: Diane Burton, MIT

Marianne P. Bittler, RAND;Tobias J. Moskowitz, NBER andUniversity of Chicago; and AnnetteVissing-Jorgensen, NBER andNorthwestern University, “Why DoEntrepreneurs Hold LargeOwnership Shares? Testing AgencyTheory Using Entrepreneur Effortand Wealth”Discussant: Antoinette Schoar,NBER and MIT

Joshua Gans, Melbourne BusinessSchool; David Hsu, the WhartonSchool; and Scott Stern, NBER andNorthwestern University, “UncertainIntellectual Property Rights andStart-Up CommercializationStrategy: The Strategic Impact ofPatent Grants Lags”Discussant: Manju Puri, NBER andStanford University

Steven Klepper, Carnegie-MellonUniversity, “The Geography ofOrganizational Knowledge”Discussant: Boyan Jovanovic, NBERand New York University

Joseph Schumpeter argued inCapitalism, Socialism and Democracy thatthe rise of large firms’ investments inin-house R and D spelled the doom ofthe entrepreneur. Lamoreaux andSokoloff explore this idea by analyzing

the career patterns of three cohorts ofinventors from the late nineteenth andearly twentieth century. They find thatover time highly productive inventorswere increasingly likely to form long-term attachments with firms. In the

Northeast, these attachments seem tohave taken the form of employmentpositions within large firms, but in theMidwest inventors were more likely tobecome principals in firms bearingtheir names. Entrepreneurship, there-

Entrepreneurship

26. NBER Reporter Spring 2003

fore, was by no means dead, but theincreasing capital requirements — bothfinancial and human — for effectiveinvention, and the need for inventorsto establish a reputation before theycould attract support, made it more dif-ficult for creative people to pursuecareers as inventors. The relative num-bers of highly productive inventors inthe population correspondingly de-creased, as did patenting rates per capita.

The theory Lazear proposes is thatentrepreneurs are jacks-of-all-tradeswho may not excel in any one skill, butare competent in many. He presents amodel of the choice to become anentrepreneur, the primary implicationof which is that individuals with bal-anced skills are more likely than othersto become entrepreneurs. The modelhas implications for the proportion ofentrepreneurs by occupation and byincome and yields a number of predic-tions for the distribution of income byentrepreneurial status. Using a datasetof Stanford alumni, Lazear tests thepredictions and finds that they hold. Inparticular, by far the most importantdeterminant of entrepreneurship ishaving background in a large numberof different roles. Further, income dis-tribution predictions, for example, thatthere are a disproportionate number ofentrepreneurs in the upper tail of thedistribution, are borne out.

Bittler, Moskowitz, and Vissing-Jorgensen augment the standard prin-cipal-agent model to accommodate anentrepreneurial setting, in which effort,ownership, and firm size are deter-

mined endogenously. They test themodel’s predictions (some novel) usingnew data on entrepreneurial effort andwealth. Accounting for unobservedfirm heterogeneity using instrumentalvariables, they find that entrepreneurialownership shares increase with outsidewealth, decrease with firm risk, anddecrease with firm size. Effort increas-es with ownership and size, and bothownership and effort increase firmperformance. The magnitude of theeffects in the cross-section of firmssuggests that agency theory is impor-tant for explaining the large averageownership shares of entrepreneurs.

Gans, Hsu, and Stern consider theimpact of the intellectual property (IP)system on the timing of coopera-tion/licensing by start-up technologyentrepreneurs. While productive effi-ciency considerations argue in favor ofearly licensing, delays in the granting ofpatent rights induce an informationalasymmetry between the inventor andpotential licensees. Employing adataset combining information aboutthe timing of patent grants and coop-erative licensing, the authors establishthree key findings: 1) pre-grant licens-ing is quite common, occurring inmore than 50 percent of their sample;2) the prevalence of pre-grant licens-ing varies across different economicenvironments; and 3) the hazard ratefor achieving a cooperative licensingagreement nearly doubles with thegranting of formal IP rights. Thoughadditional work remains to be done,these findings suggest that uncertain-

ties in the patent system may delay effi-cient bargaining by serving as a sourceof informational asymmetry betweenstart-up innovators and other firms cru-cial to the commercialization process.

Klepper analyzes the evolution ofthe geographic distribution of produc-ers in the television receiver and auto-mobile industries. Both industries expe-rienced sharp shakeouts and evolved tobe oligopolies, suggestive of increasingreturns. The television receiver indus-try initially was concentrated regionallybut evolved to be more dispersed overtime. In contrast, the automobile indus-try initially was dispersed but evolved tobe heavily concentrated around onecity, Detroit, MI, which initially had noproducers. Neither pattern conformsto theories that portray agglomerationsas being beneficial to the firms thatpopulate them. Klepper develops andtests an alternative theory to explainthe geographic evolution of the twoindustries. Firms are assumed to differin terms of their initial competence atthe time of entry, which shapes theirlong-term performance. They acquiretheir competence from firms in relatedindustries and prior entrants into thenew industry. He analyzes the locationand performance of entrants in bothindustries and shows how differentialimportance in the two sources ofcompetence plays a critical role inexplaining the contrasting evolutionof the geographic structure of thetwo industries.

NBER Reporter Spring 2003 27.

Economic Fluctuations and GrowthThe NBER’s Program on

Economic Fluctuations and Growthmet in San Francisco on February 7.Fernando Alvarez, NBER andUniversity of Chicago, and RobertKing, NBER and Boston University,organized this program:

Andrew Ang, NBER and ColumbiaUniversity; Monika Piazzesi,NBER and University of California,Los Angeles; and Min Wei,Columbia University, “What doesthe Yield Curve Tell us about GDPGrowth?”Discussant: Urban Jermann, NBERand University of Pennsylvania

Susan Athey, NBER and StanfordUniversity; Andrew Atkeson,NBER and University of California,

Los Angeles; and Patrick J. Kehoe,Federal Reserve Bank ofMinneapolis, “The Optimal Degreeof Discretion in Monetary Policy”Discussant: Lawrence Christiano,NBER and Northwestern University

Boyan Jovanovic, NBER and NewYork University, and Peter L.Rousseau, NBER and VanderbiltUniversity, “Mergers asReallocation” (NBER WorkingPaper No. 9279)Discussant: Andrew Atkeson

Marcelo Veracierto, FederalReserve Bank of Chicago, “On theCyclical Behavior of Employment,Unemployment and Labor ForceParticipation”Discussant: Robert E. Hall, NBER

and Stanford University

Yongsung Chang, University ofPennsylvania, and Sun-Bin Kim,Concordia University, “FromIndividual to Aggregate LaborSupply: A Quantitative AnalysisBased on a Heterogeneous AgentMacroeconomy”Discussant: Thomas E. MaCurdy,NBER and Stanford University

Aubhik Khan, Federal ReserveBank of Philadelphia, and Julia K.Thomas, University of Minnesota,“Inventories and the Business Cycle:An Equilibrium Analysis of (S,s)Policies”Discussant: Valerie A. Ramey,NBER and University of California,San Diego.

Ang, Piazzesi, and Wei build adynamic model for GDP growth andyields that completely characterizesexpectations of GDP. The model doesnot permit arbitrage. Contrary to pre-vious studies, this paper concludes thatthe short rate has more predictivepower than any term spread. Theauthors confirm this finding by fore-casting GDP out-of-sample. Themodel also recommends the use oflagged GDP and the longest maturityyield to measure slope. Greater effi-ciency enables the yield-curve modelto produce superior out-of-sampleGDP forecasts than unconstrainedordinary least squares at all horizons.

Athey, Atkeson, and Kehoe ana-lyze monetary policy design in aneconomy with an agreed-upon socialwelfare function that depends on therandomly fluctuating state of theeconomy. The monetary authority hasprivate information about that state. Inthe model, well-designed rules tradeoff society’s desire to give the mone-tary authority flexibility to react to itsprivate information against society’sneed to guard against the standardtime-inconsistency problem arisingfrom the temptation to stimulate theeconomy with unexpected inflation.The authors find that the optimaldegree of monetary policy discretion is

decreasing in the severity of the time-inconsistency problem. As this prob-lem becomes sufficiently severe, theoptimal degree of discretion is zero.They also find that, despite the appar-ent complexity of this dynamic mech-anism design problem, society canimplement the optimal policy simply bylegislating an inflation cap that specifiesthe highest allowable inflation rate.

Jovanovic and Rousseau arguethat takeovers have played a major rolein speeding up the diffusion of newtechnology. The role of takeovers issimilar to that of entry and exit offirms. The authors focus on and com-pare two periods: 1890-1930, duringwhich electricity and the internal com-bustion engine spread through the U.S.economy, and 1971-2001, the Infor-mation Age.

Veracierto evaluates how well areal business cycle (RBC) model thatincorporates search and leisure deci-sions simultaneously can account forthe observed behavior of employ-ment, unemployment, and being out ofthe labor force. This work contrastswith the previous RBC literature,which analyzed employment or hoursfluctuations either by lumping togetherunemployment and out-of-the-labor-force into a single non-employmentstate or by assuming fixed labor force

participation. Once the three employ-ment states are introduced explicitly,Veracierto finds that the RBC modelgenerates highly counterfactual labormarket dynamics.

Chang and Kim investigate themapping from individual to aggregatelabor supply using a general equilibri-um heterogeneous-agent model withan incomplete market. They calibratethe nature of heterogeneity amongworkers using wage data from thePanel Survey of Income Dynamics.The gross worker flow betweenemployment and nonemployment,and the cross-sectional earnings andwealth distributions in the model, arecomparable to those in the micro data.The authors find that the aggregatelabor supply elasticity of such an econ-omy is around one, bigger than microestimates but smaller than those oftenassumed in aggregate models.

Khan and Thomas develop anequilibrium business cycle model inwhich final goods’ producers pursuegeneralized inventory policies withrespect to intermediate goods, a conse-quence of nonconvex factor adjust-ment costs. Calibrating the model toreproduce the average inventory-to-sales ratio in postwar U.S. data, theauthors find that it explains half of thecyclical variability of inventory invest-

28. NBER Reporter Spring 2003

ment. Moreover, inventory accumula-tion is strongly procyclical, and pro-duction is more volatile than sales, asin the data. The model economyexhibits a business cycle similar to thatof a comparable benchmark without

inventories, although the authors doobserve somewhat higher variability inemployment and lower variability inconsumption and investment. Thus,equilibrium analysis, which necessarilyendogenizes final sales, alters our

understanding of the role of invento-ry accumulation for cyclical move-ments in GDP. The presence of inven-tories does not substantially raise thevariability of production, because itdampens movements in final sales.

Industrial Organization

The NBER’s Program onIndustrial Organization met at theBureau’s California Office onFebruary 7 and 8. Dennis W. Carltonand Austan Goolsbee, both of NBERand University of Chicago, organizedthe program meeting, at which thesepapers were discussed:

Shane Greenstein, NBER andNorthwestern University, andMichael Mazzeo, NorthwesternUniversity, “Differentiation Strategyand Market Deregulation: LocalTelecommunication Entry in theLate 1990s”Discussant: Glenn Woroch,University of California, Berkeley

Steven Berry, NBER and YaleUniversity, and Joel Waldfogel,NBER and University ofPennsylvania, “Product Quality andMarket Size”

Discussant: Tim Bresnahan, NBERand Stanford University

Gustavo E. Bamberger andLynette R. Neumann, Lexecon,and Dennis W. Carlton, “AnEmpirical Investigation of theCompetitive Effects of DomesticAirline Alliances”Discussant: Severin Borenstein,NBER and University of California,Berkeley

Amil Petrin, NBER and Universityof Chicago, and Kenneth Train,University of California, Berkeley,“Omitted Product Attributes inDiscrete Choice Models”Discussant: Frank Wolak, NBERand Stanford University

Ali Hortacsu and Chad Syverson,University of Chicago, “SearchCosts, Product Differentiation, and

the Welfare Impact of Entry: ACase Study of S & P 500 IndexFunds”Discussant: Alan Sorenson, NBERand Stanford University

Luis Garicano, University ofChicago, and Thomas Hubbard,NBER and University of Chicago,“Specialization, Firms, and Markets:The Division of Labor Within andBetween Law Firms”Discussant: Jonathan Levin,Stanford University

Toshi Iizuka, Vanderbilt University,and Ginger Z. Jin, University ofMaryland, “The Effects of Direct toConsumer Advertising in thePrescription Drug Market”Discussant: Scott Stern, NBER andNorthwestern University

Greenstein and Mazzeo examinethe role of differentiation strategies inthe development of markets for localtelecommunication services in the late1990s. The prior literature has usedmodels of interaction among homoge-nous firms, but this paper is motivatedby the claim that entrants differ sub-stantially in their product offerings andbusiness strategies. Exploiting a new,detailed dataset of CLEC (CompetitiveLocal Exchange Carriers) entry intoover 700 U.S. cities, the authors takeadvantage of recent developments inthe analysis of entry and competitionamong differentiated firms. They findstrong evidence that CLECs takeaccount of both potential marketdemand and the business strategies ofcompetitors when making their entrydecisions. This suggests that firms’incentives to differentiate their servic-

es should shape the policy debate forcompetitive local telecommunications.

When quality is produced withfixed costs, a high-quality firm canundercut its rivals’ prices and may findit profitable to invest more in quality asmarket size grows. As a result, a mar-ket can remain concentrated even as itgrows large. By contrast, when qualityis produced with variable costs, a widerange of product qualities can coexistin the market because they are offeredat different prices. Larger markets willfragment and offer products with awider range of qualities. Using U.S.urban areas as markets, Perry andWaldfogel examine the relationshipsbetween market size and product qual-ity — and between market size andproduct concentration — for twoindustries that differ in their qualityproduction process. The authors docu-

ment that in the restaurant industry,where quality is produced largely withvariable costs, the range of qualitiesincreases with market size, with each“product” maintaining a small marketshare. In daily newspapers, where qual-ity is produced with fixed costs, theaverage quality of products increaseswith market size, and the market doesnot fragment as it grows.

Bamberger, Carleton, andNeumann empirically investigate theeffect of two recent domestic airlinealliances. They find that both alliancesbenefited consumers: average fares fellby about 5 to 7 percent after the cre-ation of the alliances on those citypairs affected by the alliances. Theyalso find that total traffic increased 6percent after the creation of at leastone of the alliances. The average fareand traffic effects arise in part because

NBER Reporter Spring 2003 29.

the alliance partners’ rivals respond tothe increased competition from analliance. Finally, the authors find thatthe size of the alliance effect on aver-age fares depends on the pre-alliancelevel of competition on a city pair,with the effect being larger on thosecity pairs where the level of competi-tion was initially relatively low.

Petrin and Train describe twomethods for correcting an omittedvariables problem in discrete choicemodels: a fixed effects approach and acontrol function approach. The con-trol function approach is easier toimplement and applicable in situationsfor which the fixed effects approach isnot. The authors apply both methodsto a cross-section of disaggregateddata on customer’s choice among tele-vision options including cable, satel-lite, and antenna. As theory predicts,the estimated price response rises sub-stantially when either correction isapplied. All of the estimated parame-ters and the implied price elasticitiesare very similar for both methods.

Two salient features of the compet-itive structure of the U.S. mutual fundindustry are the large number of fundsand the sizeable dispersion in the feesfunds charge investors, even withinnarrow asset classes. Differences inportfolio financial performance alonedo not seem able to fully explain thesefeatures. Hortaçsu and Syversonfocus on the retail S&P 500 indexfunds sector, where they find similarpatterns of fund proliferation andprice dispersion. This suggests thatcostly investor search and non-portfo-lio fund differentiation may play animportant role in the mutual fundindustry. To quantify the welfareimpact of such factors in the marketfor mutual funds, the authors con-struct a model of industry equilibriumin which consumers conduct costlysearch over differentiated products.

Using panel data on fund fees andmarket shares in the retail S&P 500index fund sector, the authors find thatfairly small search costs can explain theconsiderable price dispersion in thesector. Further, consumers valuefunds’ observable non-portfolio attrib-utes — such as fund age and the num-ber of other funds in the same fundfamily — in largely plausible ways.Finally, the authors investigate the pos-sibility that there are too many funds inthe sector from a social welfare stand-point. They quantify the welfareimpact of a counterfactual sectorstructure where entry is restricted to asingle fund; they find that restrictingentry would yield nontrivial gains fromreduced search costs and productivitygains from scale economies. However,these may be counterbalanced by size-able losses from monopoly marketpower and reduced product variety.

What is the role of firms and mar-kets in mediating the division of labor?Garicano and Hubbard use confi-dential microdata from the Census ofServices to examine law firms’ bound-aries. First they examine how the spe-cialization of lawyers and firmsincreases as lawyers’ returns to special-ization increase. In fields wherelawyers increasingly specialize withmarket size, the relationship betweenthe share of lawyers who work in afield-specialized firm and market sizeindicates whether firms or marketsmore efficiently mediate relationshipsbetween lawyers in this and otherfields. The authors then examinewhich pairs of specialists tend to workin the same versus different firms; thisprovides evidence on the scope offirms that are not field-specialized.They find that whether firms or mar-kets mediate the division of laborvaries across fields in a way that corre-sponds to differences in the value ofcross-field referrals, consistent with

Garicano and Santos’s (2001) proposi-tion that firms facilitate specializationby mediating exchanges of economicopportunities more efficiently thanmarkets.

In 1997 there was an importantchange in direct-to-consumer (DTC)advertising of ethical drugs. For thefirst time, the Food and DrugAdministration (FDA) permittedbrand-specific DTC ads on TV with-out a “brief summary” of comprehen-sive risk information. This led to athree-fold growth in DTC advertisingexpenditure over four years, followedby an intensive debate about theeffects of DTC advertising on patientand doctor behaviors. Iizuka and Jinempirically examine the effects ofDTC ads on ethical drugs by combin-ing 1996-9 DTC advertising data withthe annual National AmbulatoryMedical Care Survey (NAMCS). Theauthors find that DTC advertisingleads to a large increase in the numberof outpatient drug visits, a moderateincrease in the time spent with doctors,but has no effect on doctors’ specificchoice among prescription drugs with-in a therapeutic class. Consistent withthe proponents’ claim, this findingsuggests that DTC ads encouragepatient visits but do not challenge doc-tors’ authority in the specific choice ofprescription drugs. The authors cannotrule out the possibilities, however, thatDTC ads may induce doctors to useprescription drugs over alternativetreatments, and that doctors mayspend extra time clarifying DTC ads ifthey do not prescribe the most adver-tised drug(s). The results suggest thatthe effect of DTC advertising is pri-marily market-expanding rather thanbusiness-stealing, and therefore DTCadvertising is a public good for alldrugs in the same therapeutic class.

30. NBER Reporter Spring 2003

A new NBER Working Group onPersonnel Economics met inCambridge on March 6 and 7.Edward P. Lazear, NBER andStanford University, organized themeeting, at which these papers werediscussed:

Edward P. Lazear and Paul Oyer,Stanford University, “Ports ofEntry”

Audra Bowlus, University ofWestern Ontario, and LarsVilhuber, Cornell University,“Displaced Workers, Early Leavers,and Re-Employment Wages”

Steffen Huck, University CollegeLondon; Dorothea Kübler,Humboldt University Berlin; andJörgen Weibull, Boston University,“Social Norms and Economic

Incentives in Firms”Illoong Kwon, University ofMichigan, and Eva MeyerssonMilgrom, Stanford University,“Occupation Structure andBoundaries of Internal LaborMarkets”

Christian Grund, University ofBonn, “The Wage Policy of Firms:Comparative Evidence for the U.S.and Germany from Personnel Data”

Anders Frederiksen and NielsWestergaard-Nielsen, AarhusSchool of Business, “Where DidThey Go?”

Ann Bartel and Casey Ichniowski,NBER and Columbia University;Richard Freeman, NBER andHarvard University; and MorrisKleiner, NBER and University of

Minnesota, “The Effects ofEmployee Attitudes about theirWorkplaces on Turnover andProductivity: An Analysis of RetailCommercial Banking”

Paul A. Lengermann, FederalReserve Board, “Is it Who You Are,Where You Work, or With WhomYou Work? Reassessing theRelationship Between SkillSegregation and Wage Inequality”

Tor Eriksson, Aarhus School ofBusiness, and Jaime Ortega,Universidad Carlos III de Madrid,“The Adoption of Job Rotation:Testing the Theories”

Amos Golan, American University,and Julia Lane, Urban Institute,“The Dynamics of WorkerReallocation: A Markov Approach”

Early statements on internal labormarkets view firms as consisting ofports-of-entry jobs and other jobs.Workers are hired into the former andpromoted to the latter. In the strictestform, external hiring only takes placeat certain job levels and thereafterworkers are insulated from the forcesof market competition. Lazear andOyer use data from the SwedishEmployers’ Confederation to deter-mine the existence of ports of entry infirms that represent a large part of theSwedish economy. Although there is agreat deal of promotion from within,at every level there remains significanthiring from the outside. The data aremore consistent with tournament the-ory, or with theories of firm-specifichuman capital, than they are with themore rigid institutional views.

Bowlus and Vilhuber lay out asearch model that explicitly takes intoaccount the information flow prior toa mass layoff. Using universal wagedata files that allow them to identifyindividuals working with healthy anddisplacing firms, both at the time ofdisplacement and at any other timeperiod, the authors test the predictionsof the model about re-employmentwage differentials. Workers leaving a

“distressed” firm have higher re-employment wages than workers whostay with the distressed firm until dis-placement. This result is robust to theinclusion of controls for worker qualityand unobservable firm characteristics.

Huck, Kübler and Weibull studythe interplay between economic incen-tives and social norms in firms. Theyoutline a simple model of team pro-duction, in which workers’ efforts aresubstitutes, and analyze this situationunder different social norms. Themain focus is on “efficiency norms,”that is, norms that arise from workers’desire for, or peer pressure towards,social efficiency for the team as awhole. The authors examine the possi-bility of multiple equilibriums and theeffect of economic incentives on theset of equilibriums, in particularwhether economic incentives that aretoo strong may knock out efficientequilibriums. Partnerships, comple-mentarity in production, stochasticproduction, and alternative incentiveschemes also are considered.

Kwon and Milgrom use data fromthe private sector in Sweden to look athow institutional settings influencefirms’ recruitment strategies, and theinteraction between occupation bound-

aries and firm boundaries. The Swedishdata encompass entire populations ofestablishments in the private sector,including characteristics of employeesand occupations and informationabout wages and work hours. The dataalso cover entire subpopulations ofworkers in the private sector for a 20-year period. The authors ask: Whatwere the hiring patterns of Swedishprivate employers in 1978 and 1988?And, how did these patterns affect theindividual white-collar worker pay?They find that employers commonlyhire from both within and outside thefirm to all the different ranks in thefirm, contradicting simple theories thathigher ranks are filled by promotionfrom within the firm. Smaller firmstend to hire more from outside, and tohigher ranks, than the larger firms.Large firms tend to fill job slots fromdifferent occupations but from withinthe firm. Filling jobs with outside hires(irrespective of within or outsideoccupation) is most common for thetop ranks. White-collar workers’ wagesincrease with occupation tenure andgeneral labor market tenure, possiblyreflecting the accumulation of occupa-tion-specific and general human capi-tal, but wages decrease with firm tenure.

Personnel Economics

NBER Reporter Spring 2003 31.

Generally, hiring appears restricted asmuch by occupation-specific and gen-eral human capital as by the bound-aries of the firm.

Grund compares the wage policyof a German and a U.S. firm, focusingon the relationship between wages andhierarchies. Prior studies examinedonly one particular firm, but in thispaper two plants with the same ownersand similar production processes indifferent institutional environmentsare inspected. Grund finds convexwage profiles over the hierarchy levelsof both plants. The U.S. plant showsconsiderably higher intensity of intra-firm competition in terms of higherintra-level wage inequality and yearlypromotion rate. In contrast, wages aremore distinctly attached to hierarchylevels in the German firm, as the wageregressions show.

Frederiksen and Westergaard-Nielsen study individual job separa-tions and their associated destinationstates for all individuals in the privatesector in Denmark during 1980-95,accounting for their magnitude andcyclical flows. The authors find thatindividual and workplace characteris-tics as well as business cycle effects areimportant in explaining individualbehavior. They find that structural andgrowth policies reduce transitions intounemployment but, in general, havedifferent implications for the economy.Policy interventions targeting dis-placed workers coming from plant clo-sures are inefficient, the authors argue.

How much do responses to firmopinion surveys vary among work-places as opposed to among workers?Is there a genuine “workplace effect”in employee opinion surveys? To theextent that systematic differences inattitudes exist across workplaces, dothese differences help predict work-place economic outcomes, such asproductivity or turnover? Bartel,Freeman, Ichniowski, and Kleinerexamine these questions acrossbranches of a large commercial bankin the New York metropolitan area.The bank provided files from its 1994and 1996 employee opinion surveysunder the condition that the authorsnot use its name in the publication oftheir results. The sample contains dataon 2245 employees working in 193

New York area bank branches in 1994,and 1439 employees working in 142branches in 1996. The smaller samplesizes for 1996 are attributable to clos-ings of 51 branches between 1994 and1996. The authors supplement thebank data with information from theNational Longitudinal Survey ofYouth (NLSY) on job satisfaction forthe same individual over time. Finally,they combine the employee attitudedata with information on the branches’financial performance, characteristicsof the branches’ local markets, andcharacteristics of the branch employ-ees, to see whether branch level atti-tudes help predict employee turnoverand productivity in the two cross sec-tions and over time. They concludethat there is a genuine branch, orworkplace, effect on how workers viewtheir workplace. Differences in atti-tudes are highly positively correlatedamong branches over time; this sug-gests that workplace effects are strong-ly persistent. The NLSY data show ahigher correlation of attitudes for thesame worker at a given workplace thanat different workplaces; this also pointsto the existence of a genuine work-place effect. Further, branches whereworkers have more favorable attitudestoward the firm have lower turnoverand higher productivity.

Lengermann argues that relyingon wages as a proxy for skill may beproblematic. Using a newly developedlongitudinal dataset linking virtuallythe entire universe of workers in thestate of Illinois to their employers, hedecomposes wages into componentsdue, not only to person and firm het-erogeneity, but also to the characteris-tics of their co-workers. Such “co-worker effects” capture the impact ofa weighted sum of the characteristicsof all workers in a firm on each indi-vidual employee’s wage. Lengermannrelies on the person-specific compo-nent of wages to proxy for co-worker“skills.” Because these person effectsare unknown ex ante, he first obtainsthem from a preliminary regressionthat excludes any role for co-workers.His estimates imply that a single stan-dard deviation increase in both a firm’saverage person effect and experiencelevel is associated, on average, withwage increases of 3 percent to 5 per-

cent. Firms that increase the wage pre-miums they pay workers appear to doso in conjunction with upgradingworker quality. Interestingly, the aver-age effect masks considerable variationin the relative importance of co-work-ers across industries. After allowing theco-worker parameters to vary across 2digit industries, he finds that industryaverage co-worker effects explain 26percent of observed inter-industrywage differentials. Finally, he decom-poses the overall distribution of wagesinto components due to persons,firms, and co-workers. While co-work-er effects indeed serve to exacerbatewage inequality, the tendency for highand low skilled workers to sort non-randomly into firms plays a consider-ably more prominent role.

Eriksson and Ortega test threetheories for why firms introduce jobrotation schemes: employee learning,employer learning, and employee moti-vation. The earlier literature used eitherinformation about establishment char-acteristics or data coming from person-nel records of a single firm. In order toimprove upon this, the authors use aunique dataset constructed by merginginformation from a fairly detailed sur-vey directed at Danish private sectorfirms with linked employer-employeepanel data. This allows them to includefirm and workforce characteristics aswell as firms’ human resource manage-ment practices as explanatory variables,and hence to carry out a more compre-hensive analysis.

The dynamism of the U.S. econo-my is particularly apparent in the real-location of workers into, out of, andwithin the labor market. Knowing theorder of magnitude of this allocationis important for a variety of empiricalexercises — not least of which is theempirical matching function. Golanand Lane analyze the dynamics ofworker allocation for the state ofIllinois — both for the entire work-force and for demographic subgroups.They perform their estimation withminimal distributional assumptions.The main results are that even amongthe more stable groups of workers,there is a great deal of reallocation;and that the reallocation behavior ofdifferent groups (gender, low/highwage, age) is significantly different.

32. NBER Reporter Spring 2003

Economic and social theorists havemodeled race and ethnicity as oneform of personal identity produced inresponse to the costliness of adoptingand maintaining a specific identity.Bodenhorn and Ruebeck look at thefree African-American population inthe mid-nineteenth century to investi-gate the costs and benefits of adoptingalternative racial identities. During thisperiod light-skinned African-Americanscould, and often did, choose to differ-entiate themselves from dark-skinnedAfrican-Americans. The authors modelthe choice as an extensive-form game,in which whites choose whether toaccept a separate mulatto identity andmixed-race individuals then choosewhether to adopt a mulatto identity.Adopting a mulatto identity generatespecuniary gains, but imposes psychiccosts. The authors quantify “the com-plexion gap” and find that mulattoesheld significantly more wealth thanblacks. Finally, they relate the complex-ion gap to community factors and findthat the benefits of adopting a mulattoidentity increased with the absolutesize of the mulatto community, butdecreased as the mulatto percentage ofthe African-American populationincreased at the neighborhood and citylevel. Thus, mulattoes benefited fromwhite preferences when they repre-sented a modest share of the African-American population. Yet if mostAfrican-Americans in a city were light-skinned, they became black in the eyes ofwhites and received no special treatment.

Costa notes that differentialsbetween blacks and whites in bothbirth weights and prematurity and still-birth rates have been persistent over theentire twentieth century. Differences inprematurity rates explain a large pro-portion of the black-white gap in birthweights, she finds, both among babiesborn under Johns Hopkins physiciansin the early twentieth century andbabies in the 1988 National Maternaland Infant Health Survey. In the earlytwentieth century, untreated syphiliswas the primary observable explainingdifferences in black-white prematurityand stillbirth rates. Today the primaryobservable explaining differences inprematurity rates is the low marriagerate of black women. Maternal birthweight accounts for 5-8 percent of thegap in black-white birth weights in therecent data, suggesting a role for inter-generational factors. The Johns Hopkinsdata also illustrate the value of breast-feeding in the early twentieth century:black babies fared better than whitebabies in terms of mortality andweight gain during the first ten days oflife spent in the hospital largelybecause they were more likely to bebreast-fed.

Moriguchi studies the dynamicevolution of the human resources man-agement (HRM) practices of Americancorporations during the 1920s and1930s. It has been claimed that privatewelfare capitalism — employers’ provi-sion of non-wage benefits, greateremployment security, and employee

representation to their blue-collarworkers — collapsed during the GreatDepression and was replaced by thewelfare state and industrial unionismunder the New Deal regime. However,the recent literature reveals consider-able differences among firms. Usingdata from 14 elite manufacturingfirms, Moriguchi tests the implicationsof implicit contract theory and investi-gates the effect of the Depression onwelfare capitalism and the subsequentdevelopment of corporate HRM prac-tices. He identifies positive relation-ships between the severity of theDepression and the degree of repudia-tion, but also finds that firms with ahigher commitment to workers wereless likely to repudiate and more likelyto remain unorganized and retainimplicit contractual relations.

It is often argued that branchingstabilizes banking systems by facilitat-ing diversification of bank portfolios;however, previous empirical researchon the Great Depression cannot bereconciled with this view. Analysesusing state-level data find that statesallowing branch banking had lowerfailure rates, while those examiningindividual banks find that branchbanks were more likely to fail. Carlsonand Mitchener argue that an alterna-tive hypothesis can reconcile theseseemingly disparate findings. Usingdata on national banks from the 1920sand 1930s, the authors show thatbranch banking increases competitionand forces weak banks to exit the

The NBER’s Program onDevelopment of the AmericanEconomy met in Cambridge onMarch 8. Program Director ClaudiaGoldin of Harvard University organ-ized the meeting. The followingpapers were discussed:

Howard Bodenhorn andChristopher S. Ruebeck, NBERand Lafayette College, “TheEconomics of Identity and theEndogeneity of Race”

Dora L. Costa, NBER and MIT,

“Race and Pregnancy Outcomes inthe Twentieth Century: A Long-TermComparison”

Chiaki Moriguchi, NBER andNorthwestern University, “DidAmerican Welfare Capitalists Breachtheir Implicit Contracts? PreliminaryFindings from Company-Level Data,1920-1940”

Mark Carlson, Federal ReserveBoard, and Kris J. Mitchener,NBER and Santa Clara University,“Branch Banking, Bank Competition,

and Financial Stability”

Alan L. Olmstead, University ofCalifornia, Davis, and Paul W.Rhode, NBER and University ofNorth Carolina, “An ImpossibleUndertaking: The Eradication ofBovine Tuberculosis in the UnitedStates”

Boyan Jovanovic, NBER and NewYork University, and PeterRousseau, NBER and VanderbiltUniversity, “General PurposeTechnologies”

Development of the American Economy

NBER Reporter Spring 2003 33.

banking system. This consolidationstrengthens the system as a whole with-out necessarily strengthening thebranch banks themselves.

In 1900 bovine tuberculosis repre-sented a growing threat to both animaland human health. In 1917, shortlyafter a series of scientific break-throughs allowed the early detection ofTB in cattle, the USDA embarked on anational campaign to eradicate the dis-ease. This was a wholly unprecedentedand highly controversial effort, withstate and federal agents inspectingnearly every cattle farm in the country,testing the animals, and condemning

nearly 4 million reactors to slaughterwithout full compensation. Olmsteadand Rhode analyze how the eradica-tion program functioned, how incen-tives were aligned to insure widespreadparticipation without excessive moralhazard problems, and why the UnitedStates led most European nations incontrolling the disease. The U.S. cam-paign was a spectacular success. Forthe farm sector alone, the annual ben-efits ranged between five and twelvetimes the annual costs. This represent-ed a small part of the story, becausethe most important benefit was reduc-ing human death and suffering.

Electricity and information tech-nology (IT) are perhaps the two mostimportant general-purpose technolo-gies (GPTs) to date. Jovanovic andRousseau analyze how the U.S. econ-omy reacted to them. The Electricityand IT eras are similar, but they alsodiffer in important ways. Electrificationwas adopted more broadly, whereas ITseems to be technologically more revo-lutionary. The productivity slowdownis stronger in the IT era, but the ongo-ing spread of IT and its continuingprecipitous price decline are reasonsfor optimism about growth in thecoming decades.

Health EconomicsThe NBER’s Program on Health

Economics met in Cambridge onMarch 14. Program Director MichaelGrossman organized the meeting, atwhich these papers were discussed:

David E. Bloom, NBER andHarvard University, and DavidCanning and Jaypee Sevilla,Harvard University, “Health, WorkerProductivity, and EconomicGrowth”

Jay Bhattacharya, NBER andStanford University, and DariusLakdawalla, NBER and RAND,“Time-Consistency and Addiction:

An Empirical Test”

Avi Dor, NBER and Case WesternReserve University, and WilliamEncinosa, Agency for Health CareResearch and Quality, “Cost-Sharingand Non-Compliance withPrescription Drugs”

Robert Kaestner, NBER andUniversity of Illinois, Chicago; TedJoyce and Sanders Korenman,NBER and Baruch College; andStanley Henshaw, AlanGuttmacher Institute, “Changes inBirths and Abortions FollowingWelfare Reform”

Nancy E. Reichman, ColumbiaUniversity; Hope Corman, NBERand Rider University; and KellyNoonan, Rider University, “Effectsof Child Health on Parents’Relationship Status”

Anthony T. Lo Sasso,Northwestern University, andThomas C. Buchmueller, NBERand University of California, Irvine,“The Effect of the State Children’sHealth Insurance Program onHealth Insurance Coverage” (NBERWorking Paper No. 9405)

Microeconomic analyses typicallysuggest that worker health makes animportant contribution to productivityand wages. Weil (2001) uses estimatesof the individual-level relationshipbetween health and wages to calibratean aggregate production function andsuggests that differences in health areroughly as important as differences ineducation in explaining cross-countrydifferences in gross domestic productper worker. Bloom, Canning, andSevilla directly estimate the effect ofhealth on worker productivity usingcross-country macroeconomic data.They find a positive and significanteffect. In addition, the estimated effectof health on aggregate output is con-sistent with the size of the effectfound in microeconomic studies.

In the past, many economists have

treated smokers and addicts as ration-al, time-consistent utility-maximizers.In recent years, that view has comeunder attack by those who argue thatsmokers and other addicts exhibittime-inconsistency and problems ofself-control. This conflict is significant,but intractable, because these two the-ories have very similar positive implica-tions but wildly different normativeones. However, while the positiveimplications are qualitatively similar,they differ quantitatively. Bhattacharyaand Lakdawalla use this fact and askwhich model better fits the actual life-time decision making patterns ofsmokers. Using repeated cross-section-al data from the National HealthInterview Surveys, they estimate struc-tural models of rational addiction andtime-inconsistency. Their results reveal

surprisingly little evidence in favor oftime-inconsistency.

Compliance with anti-diabeticmedications is crucial to reducingcomplications such as blindness,amputations, heart disease, and strokeamong diabetics. Dor and Encinosaexamine compliance within 90 daysafter the completion of anti-diabeticdrug prescriptions. About a third ofthe population never complies, a thirdalways complies, and the remainingthird partially complies. The authorsfind that the drug coinsurance rate hasthe effect of reducing compliance,after they control for chronic condi-tions, number of previous refills, anddemographic characteristics. An in-crease from 20 percent to 75 percentcoinsurance results in the share ofthose who never comply increasing by

34. NBER Reporter Spring 2003

27 percent and reduces the share offully compliant persons by almost 11percent. An increase in the copaymentfrom $6 to $10 results in a 13 percentincrease in the share of non-compliantpersons, and a nearly 11 percent reduc-tion in the share of fully compliantpersons. This same increase in copay-ment would reduce annual drug costsnationally by $177 million, simply byincreasing non-compliance. But, thisincrease in non-compliance also wouldincrease the rate of diabetic complica-tions, resulting in an additional $433.5million in costs annually.

Joyce, Kaestner, Korenman, andHeushaw analyze the associationbetween state and federal welfarereform and births and abortions. Statereform consists of a series of waiversfrom rules governing the Aid toFamilies with Dependent Children(AFDC) prior to the Welfare ReformAct (PRWORA). The authors alsoexamine the association between imple-mentation of Temporary Assistance toNeedy Families (TANF), the programthat replaced AFDC following PRWO-RA, and births and abortions. Theythen look more closely at one aspect ofreform, the family cap, and its associa-tion with reproductive choices. In

order to carry out these analyses, theycollected and analyzed individual abor-tion records from 21 states, the largestcompilation of such data ever attempt-ed. There is some evidence of anincrease in abortion associated withTANF. Among blacks, abortions riseafter TANF among older women, butthere is no decline in births. There isalso a rise in the abortion ratio amongblack women with two or more livebirths in states with family caps.Overall, there is at best only a modestchange in births and abortions associ-ated with TANF.

Reichman, Corman, andNoonan use data from the nationallongitudinal Fragile Families and ChildWellbeing Study of mostly unwed par-ents to estimate how poor child healthaffects one potential human resourceavailable to that child: the presence ofa father. The authors look at whetherparents are living in the same house-hold one year after the child’s birth andalso, more generally, at how their rela-tionships changed along a continuum(married, cohabiting, romanticallyinvolved, friends, or not involved) dur-ing the same one-year period. Since itmay not be easy to characterize poorchild health as a random event, the

authors account for the potentialendogeneity of child health in theirmodels. They find that having aninfant in poor health reduces the likeli-hood that parents will live together andincreases the likelihood that they willbecome less committed to their rela-tionship.

LoSasso and Buchmueller pres-ent the first national estimates of theeffects of the SCHIP expansions oninsurance coverage. Using CPS dataon insurance coverage during theyears 1996 through 2000, they findthat SCHIP had a small, but statisti-cally significant positive effect oninsurance coverage. Between 4 per-cent and 10 percent of children whomeet income eligibility standards forthe new program gained public insur-ance. These estimates indicate thatstates were more successful inenrolling children in SCHIP than theywere with prior Medicaid expansionsfocused on children just above thepoverty line. Crowd-out of privatehealth insurance was in line with esti-mates for the Medicaid expansions ofthe early 1990s, between 18 percentand 50 percent.

NBER Reporter Spring 2003 35.

Productivity Program MeetingThe NBER’s Program on

Productivity met in Cambridge onMarch 14. Shane Greenstein, NBERand Northwestern University, organ-ized the meeting. These papers werediscussed:

Barbara K. Atrostic and SangNguyen, Center for EconomicStudies, “IT and Productivity in IT-Using and IT-Producing Industries:New Micro Data Evidence”Discussant: Christopher Forman,Carnegie-Mellon University

James Bessen, Research onInnovation, “Technology Adoption

Costs and Productivity Growth: TheTransition to InformationTechnology”Discussant: Shane Greenstein

Dan Elfenbein, Harvard University,and Josh Lerner, NBER andHarvard University, “DesigningAlliance Contracts: Exclusivity andContingencies in Internet PortalAlliances”Discussant: Iain Cockburn, NBERand Boston University

Kenneth Flamm, University ofTexas, Austin, “Moore’s Law and theEconomics of Semiconductor Price

Trends”Discussant: Rebecca Henderson,NBER and MIT

Minjae Song, Harvard University,“Measuring Consumer Welfare inthe CPU Market”

Sean M. Dougherty, Robert H.McGuckin, and Bart Van Ark, TheConference Board,“Internationalization and theChanging Structure of BusinessR&D: Recent Trends andMeasurement Implications”

Atrostic and Nguyen use newplant-level data on information tech-nology (IT) collected by the U.S.Census Bureau to provide evidence onthe labor productivity impact of ITacross U.S. manufacturing plants in IT-producing and IT-using industries(defined under the North AmericanIndustrial Classification System).Previous plant-level studies examiningthe link between productivity andcomputers or other IT in the UnitedStates typically focused on the pres-ence of computers, either using dataon the stock of computer capital or oncurrent IT or computer investment asproxies for the computer stock.Studies that have detailed informationon IT generally have a relatively smallsample, do not include smaller plants,or are limited to specific manufactur-ing industries. The data for this study,in contrast, are collected from about30,000 plants across the U.S. manufac-turing sector. The authors use a directmeasure of IT: information on thepresence of computer networks. Theyfind that computer networks have apositive and significant effect on laborproductivity after they control forother important factors, such as capitalintensity and other plant characteris-tics, and even after taking account ofpossible endogeneity of the computernetwork variable. Also, small plantsappear to use computer networks

more efficiently than large plants.Using two panels of U.S. manufac-

turing industries, Bessen estimatescapital adjustment costs from 1961 to1996. He finds that adjustment costsrose sharply from 1974-83: they morethan doubled, from about 3 percent ofoutput to around 7 percent. Moreover,this increase is specifically associatedwith a shift to investment in IT. Butsuch large adoption costs imply thatthe Solow residual mismeasures pro-ductivity growth: adoption costs areresource costs that represent anunmeasured investment. Bessen findsthat when this investment is included,productivity grew about 0.5 percent peryear faster than official measures dur-ing the 1970s and early 1980s, reducingthe size of the productivity “slow-down.” Indeed, estimated productivitygrowth rates were roughly the samefrom 1974-88 as from 1949-73. Thustechnology transitions critically affectproductivity growth measurement.

Elfenbein and Lerner test theo-retical propositions from the technol-ogy licensing literature and from theliterature on information and controlin alliances, using a sample of over 100Internet portal alliance contracts. Thetechnology licensing literature suggeststhat one of the major factors drivingfirms’ decisions to transact exclusivelyfor an innovation is the magnitude orimportance of the innovation. The

authors find some support for thishypothesis, but in this setting exclusiv-ity decisions also relate to other fac-tors. The literature on information andcontrol in alliances suggests that theuse of verifiable performance meas-ures to allocate state contingent deci-sion rights depends on the level ofinformation asymmetry between thetwo parties and on the precision of theinformation. The authors test thesepropositions by looking at how thetiming of agreements (a proxy forenvironmental uncertainty) and exclu-sivity restrictions (a proxy for incentiveconflict) affect the use of a subset ofavailable performance measures.Consistent with the literature, they findthat contracts involve fewer contingen-cies as industries have matured. Whereincentive conflicts are potentiallygreater, more contingencies are used.

Flamm starts by describing thehistory of Moore’s Law, and explainswhy it has such potentially wide-rang-ing consequences. He then shows howa Moore’s Law prediction must be cou-pled with other assumptions in orderto produce an economically meaning-ful link to what is the key economicvariable of the information age: thecost or price of electronic functionali-ty, as implemented in a semiconductorintegrated circuit. Flamm then relatesthe historical evolution of semicon-ductor prices through the mid-1990s

36. NBER Reporter Spring 2003

to developments in several key param-eters over this historical period. Hesurveys the evidence on acceleration inthe rate of decline in leading edgesemiconductor prices in the mid-1990sand suggests that measured increasesin historical rates of decline seemunlikely to persist. Finally, he exploresthe nature of the man-made historicaland institutional economic processesthat made these technical accomplish-ments possible, and argues that theirconsequence has been an unappreciat-ed but radical transformation of theindustrial framework in which R and Dis undertaken within the global semi-conductor industry.

The personal computer CentralProcessing Unit (CPU) has undergonea dramatic improvement in quality,accompanied by an equally remarkabledrop in prices in the 1990s. How have

these developments in the CPU mar-ket affected consumer welfare? Songestimates demand for CPUs and meas-ures consumer welfare. The welfarecalculations show that consumer sur-plus makes up approximately 90 per-cent of the total social surplus and thata large part of the welfare gains comesfrom the introduction of new prod-ucts. Simulation results show how“quality competition” among firmscan generate a large gap between thereservation and actual prices.

Dougherty, Inklaar, McGuckin,and Van Ark examine the structure ofR and D internationally using informa-tion collected in interviews of 25multinational companies in four high-tech industries from the United States,European Union, and Japan. Theylook at the composition of R and Dactivities within the firm, how they are

structured internationally, and howthey are changing. The focus is on howthe production of research and devel-opment differ, based on the relativeuncertainty of research output as com-pared to development. The economicdistinctions between research anddevelopment have broad implicationsfor how companies allocate theirresources, and they help to explainrecent trends, including researchbecoming more concentrated in theUnited States and the development ofcommercial products more dispersedworldwide. The authors examinechanges in international R and D costsusing new purchasing power estimatesfor R and D and compare these tochanges in the trends of R and Dinvestments internationally from 1987to 1997.

International Finance and MacroeconomicsThe NBER’s Program on

International Finance and Macro-economics met in Cambridge onMarch 21. Richard Lyons, NBER andUniversity of California, Berkeley,and Andres Velasco, NBER andHarvard University, organized thisprogram:

Andrew Atkeson, University ofCalifornia, Los Angeles, and PatrickKehoe, NBER and University ofMinneapolis, “On the Benefits toTransparency in a Monetary PolicyInstrument”Discussant: Jon Faust, FederalReserve Board

Aaron Tornell, NBER andUniversity of California, LosAngeles, and Frank Westermann,

University of Munich, “The CreditChannel in Middle IncomeCountries”Discussant: Fernando Broner,University of Maryland

Cedric Tille, Federal Reserve Bankof New York, “How Valuable isExchange Rate Flexibility? OptimalMonetary Policy under SectoralShocks?”Discussant: Charles Engel, NBERand University of Wisconsin

Kathryn M.E. Dominguez andLinda L. Tesar, NBER andUniversity of Michigan; andSebastian Auguste and HermanKamil, University of Michigan,“Cross-Border Trading as aMechanism for Capital Flight:

ADRs, CEDEARS and theArgentine Crisis”Discussant: Sergio Schmukler, TheWorld Bank

Mark Aguiar and Gita Gopinath,University of Chicago, “Fire-SaleFDI and Liquidity Crises”Discussant: Bernard Dumas, NBERand INSEAD

Helene Rey, NBER and PrincetonUniversity; and Jean Imbs, HaroonMumtaz, and Morten Ravn,London School of Business, “PPPStrikes Back: Aggregation and theReal Exchange Rate” (NBERWorking Paper No. 9372)Discussant: Shang-Jin Wei, NBERand Harvard University

Monetary instruments differ intheir transparency — how easy it is forthe public to monitor the instrument— and their tightness — how closelylinked they are to inflation. Tightnessis always desirable in a monetary poli-cy instrument. Atkeson and Kehoeshow that transparency is desirable

when there is a credibility problem, inthat the government cannot commit toits policy. They illustrate their argu-ment by considering a classic questionin international economics: is theexchange rate or the money growthrate the better instrument of monetarypolicy? Their analysis suggests that the

greater transparency of exchange ratesmeans that if both instruments areequally tight, the exchange rate is pre-ferred.

With inflation under control inmany middle income countries (MICs),swings in credit, investment, and assetprices now have the most effect on these

NBER Reporter Spring 2003 37.

countries. Tornell and Westermannpresent a framework for analyzing howcredit market shocks are propagatedand amplified in MICs. In their modelthe strength of the credit channelderives from two key characteristics ofMICs: a sharp asymmetry across thetradables (T) sector and the morebank-dependent nontradables (N) sec-tor; and a significant degree of curren-cy mismatch in the N-sector. Thismakes movements in the real exchangerate the driving element in the amplifi-cation of shocks. Using quarterly datafor a group of MICs, the authors findevidence for a strong credit channel,for a balance sheet effect, and forasymmetric sectorial responses. Theirfindings indicate that inflation target-ing is not sufficient to guarantee eco-nomic stability, because such a policymight overlook the development oflending booms and associated sectorialasymmetries.

Tille explores the optimal mone-tary policy reaction to productivityshocks in an open economy. Earlierstudies assumed that countries special-ize in producing particular goods, buthe enriches the analysis by allowing forincomplete specialization. He con-firms the finding of Obstfeld andRogoff (2000) — who build onFriedman (1953) — that a flexibleexchange rate is highly valuable indelivering the optimal response tocountry-specific shocks. However, itsvalue is much smaller when shocks aresector-specific, because exchange ratefluctuations then lead to misallocations

between different firms within a sec-tor. The limitation on the value offlexibility is sizable even when special-ization is high.

Auguste, Dominguez, Kamil,and Tesar examine the surprising per-formance of the Argentine stock mar-ket in the midst of the country’s mostrecent financial crisis as well as the roleof cross-listed stocks in Argentinecapital flight. Although Argentineinvestors were subject to capital con-trols, they were able to purchase cross-listed stocks for pesos in Argentina,convert them into dollar-denominatedshares, re-sell them in New York, anddeposit the dollar proceeds in U.S.bank accounts. The authors show that:1) ADR discounts went as high as 45percent (indicating that Argentineinvestors were willing to pay significantamounts in order to legally move theirfunds abroad); 2) the implicit peso-dol-lar exchange rate on the eve of thedevaluation anticipated a 42 percentfall in the value of the peso relative tothe dollar; 3) local market factors inArgentina became more important inpricing peso denominated stocks withassociated ADRs, while the samestocks in New York were mainly pricedbased on global factors; and 4) capitaloutflow using the ADR and CEDEARmarkets was substantial (their estimatefor ADRs is between $835 million and$3.4 billion).

Aguiar and Gopinath use a firm-level dataset to show that foreignacquisitions increased by 91 percent inEast Asia between 1996 and 1998,

while intra-national merger activitydeclined. Firm liquidity plays a signifi-cant and sizeable role in explainingboth the increase in foreign acquisi-tions and the decline in the price ofacquisitions during the crisis. This con-trasts with the role of liquidity in non-crisis years and in non-crisis economiesin the region. This effect is also mostprominent in the tradable sector.Quantitatively, the observed decline inliquidity can explain 25 percent of theincrease in foreign acquisition activityin the tradable sectors. The nature ofM and A activity supports liquidity-based explanations of the East Asiancrisis and provides an explanation forthe puzzling stability of FDI inflowsduring the crises.

Imbs, Mumtaz, Ravn, and Reyshow the importance of a dynamicaggregation bias in accounting for thePurchasing Power Parity (PPP) puzzle.They prove that established time seriesand panel methods substantially exag-gerate the persistence of real exchangerates because of heterogeneity in thedynamics of disaggregated relativeprices. When heterogeneity is properlytaken into account, estimates of thereal exchange rate half-life fall dramat-ically, to little more than one year, orsignificantly below Rogoff ’s “consen-sus view” of three to five years. Theauthors show that corrected estimatesare consistent with plausible nominalrigidities, thus, arguably, solving thePPP puzzle.

*

38. NBER Reporter Spring 2003

International Trade and InvestmentThe NBER’s Program on Inter-

national Trade and Investment met inCambridge on March 28 and 29.Program Director Robert C. Feenstra,NBER and University of California,Davis, organized this program:

Andrew K. Rose, NBER andUniversity of California, Berkeley,“Do We Really Know that the WTOIncreases Trade?”

Bruce A. Blonigen, NBER andUniversity of Oregon, “EvolvingDiscretionary Practices of U.S.Antidumping Activity”

Andrew B. Bernard, NBER and

Dartmouth College; J. BrandfordJensen, Institute for InternationalEconomics; and Peter K. Schott,NBER and Yale University, “FallingTrade Costs, Heterogeneous Firms,and Industry Dynamics”

Wolfgang Keller, NBER andUniversity of Texas, Austin, andStephen R. Yeaple, University ofPennsylvania, “MultinationalEnterprises, International Trade, andProductivity Growth: Firm-LevelEvidence from the United States”(NBER Working Paper No. 9504)

Joshua Aizenman, NBER andUniversity of California, Santa Cruz,

and Mark M. Spiegel, FederalReserve Bank of San Francisco,“Institutional Efficiency, MonitoringCosts, and the Investment Share ofFDI”

Donald R. Davis and DavidWeinstein, NBER and ColumbiaUniversity, “A Search for MultipleEquilibria in Urban IndustrialStructure”

Pinelopi K. Goldberg, NBER andYale University, and Nina Pavcnik,NBER and Dartmouth College,“The Responses of the InformalSector to Trade Liberalization”(NBER Working Paper No. 9443)

Rose estimates the effect on inter-national trade of multilateral tradeagreements: the World Trade Organi-zation (WTO); its predecessor, theGeneralized Agreement on Tariffs andTrade (GATT); and the GeneralizedSystem of Preferences (GSP) extend-ed from rich countries to developingcountries. He uses a standard “gravi-ty”model of bilateral merchandisetrade and a large panel data set cover-ing over 50 years and 175 countries.An extensive search reveals little evi-dence that countries joining orbelonging to the GATT/WTO havedifferent trade patterns from out-siders. The GSP does seem to have astrong effect, and is associated with anapproximate doubling of trade.

Previous literature has discussedthe procedural biases that exist in U.S.Department of Commerce (USDOC)calculations of dumping margins.Blonigen examines the evolution ofdiscretionary practices and their role inthe rapid increase in average USDOCdumping margins since 1980. He findsthat USDOC discretionary practices,including use of “facts available” andcost of production tests, have played amajor role in rising dumping margins,with little evidence that changes in U.S.antidumping law or composition ofinvestigated products and countrieshave had much effect. Importantly, the

evolving effect of discretionary prac-tices is attributable not only to increas-ing use of these practices over time,but also to apparent changes in imple-mentation of these practices whichsignal a higher increase in the dumpingmargin whenever they are applied.

Bernard, Jensen, and Schottexamine the response of industriesand firms to changes in trade costs.They test the predictions of recentequilibrium models of internationaltrade with heterogeneous firms. Usingdisaggregated U.S. import data, theauthors create a new measure of tradecosts over time and industries. As themodels predict, productivity growth isfaster in industries with falling tradecosts. The authors also find evidencesupporting the major hypotheses ofthe heterogeneous firm models. Firmsin industries with falling (relative) tradecosts are more likely to die or becomeexporters. Existing exporters increasetheir shipments abroad. The results arestrongest for industries most likely tobe producing horizontally differentiat-ed tradeable goods.

Keller and Yeaple estimate inter-national technology spillovers to U.S.manufacturing firms via imports andforeign direct investment (FDI) be-tween 1987 and 1996. In contrast toearlier work, their results suggest thatFDI leads to significant productivity

gains for domestic firms. The size ofFDI spillovers is economically impor-tant, accounting for about 14 percentof productivity growth in U.S. firmsbetween 1987 and 1996. In addition,there is some evidence of imports-related spillovers, but it is weaker thanfor FDI. The authors also give a detailedaccount of why their study leads to dif-ferent results from those found in previ-ous work. Their analysis indicates thattheir results are likely to generalize toother countries and periods.

Aizenman and Spiegel study theimplications of institutional efficiencyon the pattern of foreign direct invest-ment (FDI). They posit that domesticagents have a comparative advantageover foreign agents in overcomingsome of the obstacles associated withcorruption and weak institutions. Theymodel these circumstances in a princi-pal-agent framework with costly ex-post monitoring and enforcement ofan ex-ante labor contract. Ex-postmonitoring and enforcement costs areassumed to be lower for domesticentrepreneurs than for foreign ones,but foreign producers enjoy a counter-vailing productivity advantage. Underthese asymmetries, multinationals payhigher wages than domestic producers,in line with the insight of efficiencywages and with the evidence about the“multinationals wage premium.” FDI

NBER Reporter Spring 2003 39.

also is more sensitive to increases inenforcement costs. The authors com-pare institutional efficiency levels for alarge cross section of countries in1989 to subsequent FDI flows from1990 to 1999. They find that institu-tional efficiency is associated positive-ly with the ratio of subsequent FDIflows to gross fixed capital formationand to private investment. This is truefor both simple cross-sections and forcross-sections weighted by countrysize.

In the context of the Allied bomb-ing of Japanese cities and industries inWWII, Davis and Weinstein developa new empirical test for multiple equi-

libriums and then apply it to data for114 Japanese cities in eight manufac-turing industries. The data reject theexistence of multiple equilibriums. Inthe aftermath of even gargantuanshocks, a city typically recovers notonly its population and its share ofaggregate manufacturing, but even thespecific industries it had before.

Goldberg and Pavcnik study therelationship between trade liberaliza-tion and informality. It is often claimedthat increased foreign competition indeveloping countries leads to anexpansion of the informal sector,defined as the sector that does notcomply with labor market legislation.

Using data from two countries thatexperienced large trade barrier reduc-tions in the 1980s and 1990s, Braziland Colombia, the authors examinethe response of the informal sector toliberalization. They find no evidenceof a relationship between trade policyand informality in Brazil. In Colombia,though, there is evidence of such arelationship, but only for the periodpreceding a major labor market reformwhich increased the flexibility of thelabor market. These results point tothe significance of labor market insti-tutions in assessing the effects of tradepolicy on the labor market.

Program on ChildrenThe NBER’s Program on Chil-

dren, directed by Jonathan Gruber ofMIT, met in Cambridge on April 3.Members and guests discussed thesepapers:

Anna Aizer, Princeton University,“Got Health? Advertising, Medicaid,and Child Health”

Karen Norberg, NBER, “Dads andCads: Parental Cohabitation and theHuman Sex Ratio at Birth”

Douglas Almond, NBER, andKenneth Chay, NBER andUniversity of California, Berkeley,“The Long-Run andIntergenerational Impact of PoorInfant Health: Evidence fromCohorts Born During the CivilRights Era”

David Figlio, NBER and Universityof Florida, “Testing, Crime, andPunishment”

Thomas S. Dee, NBER andSwarthmore College, “Are ThereCivic Returns to Education?”

Abhijit Banerjee, Shawn Cole, andLeigh Linden, MIT; and EstherDuflo, NBER and MIT,“Improving the Quality ofEducation in India: Evidence fromThree Randomized Experiments”

Of the ten million uninsured chil-dren in 1996, nearly half were eligiblefor the public health insurance pro-gram, Medicaid, but not enrolled.Little is known about the reasons low-income families fail to use public pro-grams or the consequences of failingto use them. Using detailed informa-tion on Medicaid outreach, enrollment,and hospitalization rates in California,Aizer finds that information andadministrative costs are significantdeterrents to program take-up. Con-trolling for selection into Medicaid,enrolling children early in Medicaidleads to a more efficient allocation ofhealth care resources by promotingprimary ambulatory care over moreexpensive hospital-based care resultingin fewer avoidable hospitalizations, shefinds.

Modern theory on sex allocation

predicts that parents may be able tovary the sex of their offspring accord-ing to the prospects for two-parentcare. Using data pooled from fourpublicly available longitudinal studies,Norberg finds that parents who wereliving with an opposite-sex spouse orpartner before the child’s conceptionor birth were significantly more likelyto have a male child than parents whowere living apart. This effect is observ-able even when the comparisons aremade between siblings, and even whenthose comparisons are made beforethe children’s conceptions. This “part-nership status effect” may be the resultof modern reproductive exposures,but a paternal investment effect wouldfit closely with the predictions ofadaptive sex allocation theory.

Almond and Chay use the sub-stantial improvements in health among

the cohorts of black infants born dur-ing the 1960s to estimate the long-runeffects of early life health conditions.The microdata contained in the annualNatality Detail files provides informa-tion on the demographic and socioe-conomic characteristics and health riskfactors of mothers giving birth, as wellas the health outcomes of the infant.These data can be linked to the infanthealth conditions that prevailed in thestate and year in which the mother wasborn. The evidence suggests a stronglink between infant health and bothadult health and infant health of thesubsequent generation. For example,the dramatic improvements in healthamong black infants born inMississippi during the 1960s are mir-rored by improved health among blackwomen born in Mississippi during the1960s who gave birth during the 1980s

40. NBER Reporter Spring 2003

and 1990s. A black adult female bornafter 1964 has lower rates of diabetesand other risk factors and is less likelyto give birth to a low birth weightinfant than a black woman born before1964. This pattern does not hold forwhite, Mississippi-born women, whoexperienced much smaller infanthealth improvements during the CivilRights Era. The authors find similarassociations in other states that wereaffected by the social programs of the1960s.

Figlio shows that schools respondto high-stakes testing by selectively dis-ciplining their students. Schools havean incentive to keep high-performingstudents in school and low-performingstudents out of school during the test-ing window in order to maximizeaggregate test scores. The evidencesupports this hypothesis — these pat-terns are precisely what are observedin the data, but only for students ingrades that are tested with highstakes for the school. Since studentssuspended during the testing windoware significantly more likely to miss theexamination, this result suggests thatschools may be deliberately attemptingto reshape the testing pool in responseto high-stakes testing. On the otherhand, schools have an incentive tokeep marginal students in school dur-ing the pre-testing preparation period,and have less of an incentive to keepvery high or very low-performing stu-dents in school during this so-called

“cram” period. Again, this pattern isobserved in the data. Taken together,these results indicate that schools maybe using student discipline as a tool tomanipulate aggregate test scores.

The hypothesized effects of educa-tional attainment on adult civicengagement and attitudes providesome of the most important justifica-tions for government intervention inthe market for education. In this study,Dee presents evidence on whetherthese externalities exist. He assessesand implements two strategies foridentifying the effects of educationalattainment. One is based on the avail-ability of junior and community col-leges; the other on changes in teenexposure to child labor laws. Theresults suggest that educational attain-ment has large and statistically signifi-cant effects on subsequent voter par-ticipation and support for free speech.Dee also finds that additional school-ing appears to increase the quality ofcivic knowledge as measured by thefrequency of newspaper readership.

Banerjee, Cole, Duflo, andLinden present the results of a two-year randomized evaluation of a large-scale remedial education program,conducted in Mumbai and Vadodara,India, along with the preliminaryresults of a randomized evaluation ofa computer-assisted learning programin Vadodara. The remedial educationprogram hires young women from thecommunity to teach basic literacy and

numeracy to children who reach “stan-dard three or four” without havingmastered these competencies. Theprogram, implemented by a non-gov-ernmental organization in collabora-tion with the government, is extremelycheap (it cost 5 dollars per child peryear) and is easily replicable: it is nowimplemented in 20 Indian cities, andreaches tens of thousands of children.The authors find the program to bevery effective: on average, it increasedlearning by 0.15 standard deviations inthe first year, and 0.39 in the secondyear. The gains are the largest for chil-dren at the bottom of the distribution:children in the bottom third gain 0.18standard deviations in the first year,and 0.59 in the second year. Theresults are very similar in the two stan-dards, and in the two cities. At the mar-gin, extending this program would be4.5 to 6 times more cost effective thanhiring new teachers. The preliminaryresults of the computer assisted learn-ing program, which is planned to bewidely implemented in India, are lessimpressive: on average, the programincreases test scores by an insignificant0.10 standard deviations. The effect ishigher (and significant) in schoolswhere the remedial education programis also present. On the basis of theseestimates, extending the computerassisted learning program wouldappear less cost effective than hiringnew teachers.

NBER Reporter Spring 2003 41.

Public EconomicsThe NBER’s Program on Public

Economics met in Cambridge onApril 10 and 11. Program DirectorJames M. Poterba, NBER and MIT,organized the meeting. These paperswere discussed:

Brian G. Knight, NBER andBrown University, “ParochialInterests and the CentralizedProvision of Local Public Goods:Evidence from CongressionalVoting on Transportation Projects”Discussants: Robert P. Inman,NBER and University ofPennsylvania

Stephen Coate, NBER and CornellUniversity, “Pareto ImprovingCampaign Finance Policy”

Discussant: Marco Battaglini,Princeton University

Jeffrey R. Brown, NBER andUniversity of Illinois, and AmyFinkelstein, NBER and HarvardUniversity, “Why is the Market forPrivate Long-Term Care InsuranceSo Small? The Roll of Pricing andMedicaid Crowd-Out”Discussant: David M. Cutler, NBERand Harvard University

Julie Berry Cullen, NBER andUniversity of Michigan, and RogerH. Gordon, NBER and Universityof California, San Diego, “Taxesand Entrepreneurial Activity:Theory and Evidence for the U.S.”(NBER Working Paper No. 9015)

Discussant: Austan Goolsbee,NBER and University of Chicago

Jagadeesh Gokhale, FederalReserve Bank of Cleveland;Laurence J. Kotlikoff, NBER andBoston University; and AlexiSluchynsky, ESPlanner, “Does itPay to Work?”Discussant: Robert A. Moffitt,NBER and Johns HopkinsUniversity

Jonathan Gruber, NBER and MIT,“Pay or Pray: The Impact ofCharitable Subsidies on ReligiousParticipation”Discussant: Laurence Iannaccone,George Mason University

Local public goods financed from anational tax base provide concentratedbenefits to recipient jurisdictions butdispersed costs, creating incentives forlegislators to increase own-districtspending but to restrain aggregatespending because of the associated taxcosts. Theoretically, therefore, onewould predict inefficiencies in the allo-cation of public goods, but there is lit-tle direct evidence that individual legis-lators respond to these incentives.Knight analyzes 1998 Congressionalvotes on transportation project fund-ing. He shows that legislators respondto common pool incentives: the prob-ability of supporting the projectsincreases with own-district spendingand decreases with the tax burdenassociated with aggregate spending.Having found that legislators dorespond to such incentives, Knight cal-culates the efficient level of publicgoods. The results suggest over-spend-ing in the aggregate, especially in polit-ically powerful districts, and a largeassociated deadweight loss.

Coate argues that campaign financepolicy, in the form of contribution lim-its and matching public financing, canbe Pareto-improving even under themost optimistic assumptions concern-ing the role of campaign advertisingand the rationality of voters. The argu-

ment assumes that candidates use cam-paign contributions to convey truthfulinformation to voters about their qual-ifications for office and that votersupdate their beliefs rationally on thebasis of the information they haveseen. It also assumes that campaigncontributions are provided by interestgroups and that candidates can offer toprovide policy favors for their interestgroups to attract higher contributions.

Long-term care represents one ofthe largest uninsured financial risksfacing the elderly in the United States,and yet we have virtually no evidenceto help explain the extremely limitednature of the private market for long-term care insurance. Brown andFinkelstein develop two complemen-tary analytical tools to begin filling thisvoid: a framework for assessing the“money’s worth” of private long-termcare insurance policies and a model ofthe insurance value of a long-term careinsurance contract for a risk averse,life-cycle consumer. Using state-of-the-art actuarial data on long-term careutilization probabilities and compre-hensive market data on insurance poli-cy characteristics and premiums, theyfind that private long-term care insur-ance contracts are priced lower than isactuarially fair for women and higherfor men. For a policy that covers all

types of paid care, a 65-year old malecan expect to receive 57 to 73 cents inpresent discounted value of benefitsfor every dollar paid in expected pres-ent value of premiums; by contrast, a65-year old woman can expect toreceive about $1.12 to $1.42 in benefitsfor every dollar paid in premiums.These results suggest that, given exist-ing market prices and the presence ofMedicaid as a payer-of-last-resort, pri-vate long-term care insurance is notvalued by individuals throughout sub-stantial portions of the wealth distri-bution. The very presence of Medicaidcrowds out the purchase of privateinsurance for well over half of house-holds, and significantly reduces thevalue of private insurance for the rest.In addition, marginal decreases in the“quality” of Medicaid-financed relativeto privately-financed care substantiallyincrease the value of private long-termcare insurance. By contrast, even sub-stantial reductions in premiums —such as those that might be achievedby the new federal tax subsidies forlong-term care insurance — would beinsufficient to make long-term careinsurance attractive to the median indi-vidual.

Entrepreneurial activity is pre-sumed to generate important spillovers,potentially justifying tax subsidies.

42. NBER Reporter Spring 2003

How does the tax law affect individualincentives? How much of an impacthas it had in practice? Cullen andGordon first show theoretically thattaxes can affect the incentives to be anentrepreneur simply because of differ-ences in tax rates on business versuswage and salary income; differences inthe tax treatment of losses versusprofits through a progressive ratestructure and through the option toincorporate; and risk-sharing with thegovernment. They then provide empir-ical evidence — using U.S. individualtax return data — that these aspects ofthe tax law have had large effects onactual behavior.

Gokhale, Kotlikoff, and Sluchynskyuse ESPlanner, a financial planning soft-ware program, to study the net worktax levied on workers with differentearnings capacities. The authors focuson lifetime average and marginal network tax rates, which are measured bycomparing the present values of life-time spending from working throughretirement, both in the presence and inthe absence of all tax-transfer pro-grams. They report eight findings: 1)The fiscal system is highly progressive;couples working full time and earningthe minimum wage receive 32 cents inbenefits, net of taxes, for every dollarthey earn. In contrast, households withmillion dollar salaries pay 51 cents in

taxes, net of benefits, per dollarearned. 2) Net subsidies are providedonly at the very bottom end of theincome distribution. Average net worktax rates of couples earning 1.5 timesthe minimum wage ($32,100 per year)are 14 percent. For working couplesearning 5 times the minimum wage($107,000), the net tax rate is 38 per-cent. 3) While the poor face negativeaverage taxes, like the middle class andthe rich, they face positive marginal nettaxes on working that exceed 50 per-cent. Moreover, certain low- and mod-erate-income households face substan-tially higher marginal net work taxrates than those faced by the rich. 4)Low-wage workers face confiscatorytax rates on switching from part-timeto full-time work. 5) The same is trueof secondary earning spouses in low-wage households. 6) The marginal nettax on working is particularly high foryoung households with low incomes.7) Average and marginal net work taxrates are relatively insensitive to theassumed rate of real wage growth andthe discount rate. 8) Major tax reforms,such as switching from income to con-sumption taxation, can have a signifi-cant effect on the fiscal system’s overallprogressivity.

The economic argument for subsi-dizing charitable giving relies on thepositive externalities of charitable

activities, particularly from the religiousinstitutions that are the largest recipi-ents of giving. But the net externaleffects of subsidies to religious givingwill depend also on their potentiallyimportant indirect effect on religiousparticipation. Religious participationcan be a complement to, or a substitutefor, the level of charitable giving.Understanding these spillover effectsof charitable giving may be quiteimportant, given the existing observa-tional literature suggesting that religios-ity is a major determinant of well-beingamong Americans. In his paper,Gruber investigates the impact of char-itable subsidies on religious participa-tion by using data over three decadesfrom the General Social Survey; he alsoconfirms the impact of such subsidieson religious giving using the ConsumerExpenditure Survey. He finds strongevidence that religious giving and reli-gious participation are substitutes: larg-er subsidies to charitable giving lead tomore religious giving, but less religiousattendance, with an implied elasticity ofattendance with respect to religiousgiving of –0.92. These results haveimportant implications for the debateover charitable subsidies. They alsoserve to validate economic models ofreligious participation.

NBER Reporter Spring 2003 43.

Asset PricingThe NBER’s Program on Asset

Pricing met in Chicago on April 11.Program Director John H. Cochraneand Lubos Pastor, both of NBERand University of Chicago, organizedthe meeting. These papers were dis-cussed:

Antonios Sangvinatsos, New YorkUniversity, and Jessica Wachter,NBER and New York University,“Does the Failure of theExpectations Hypothesis Matter forLong-Term Investors?”Discussant: Kenneth J. Singleton,NBER and Stanford University

Jun Pan, MIT, and Allen M.Poteshman, University of Illinois,“The Information in OptionVolume for Stock Prices”Discussant: Michael W. Brandt,NBER and University ofPennsylvania

Robert F. Stambaugh, NBER andUniversity of Pennsylvania,“Inference About Survivors”Discussant: Christopher Jones,University of Southern California

Maria Vassalou and Yuhang Xing,Columbia University, “Default Riskin Equity Returns”

Discussant: Ravi Jagannathan,NBER and Northwestern University

Robert F. Dittmar and ChristianT. Lundblad, Indiana University,and Ravi Bansal, Duke University,“Interpreting Risk Premia AcrossSize, Value and Industry Portfolios”Discussant: Lars P. Hansen, NBERand University of Chicago

Owen A. Lamont, NBER andUniversity of Chicago, “Go DownFighting: Short Sellers vs. Firms”Discussant: William N. Goetzmann,NBER and Yale University

Sangvinatsos and Wachter con-sider the consumption and portfoliochoice problem of a long-run investorwho has access to nominal bonds anda stock portfolio. In the presence ofunhedgeable inflation risk, there aremultiple pricing kernels that producethe same bond prices, but a uniquepricing kernel that equals the marginalutility of the investor. The authorsextend their model to account fortime-varying expected inflation andestimate it with data on inflation andterm structure. The estimates implythat the bond portfolio for the long-run investor looks very different fromthe portfolio of a mean-variance opti-mizer. In particular, the desire to hedgechanges in term premiums generateslarge hedging demands for long-termbonds.

Pan and Poteshman find strongevidence of information transmissionfrom the options market to underlyingstock prices. Taking advantage of aunique dataset from the ChicagoBoard of Options Exchange, theauthors construct put-to-call volumeratios for underlying stocks, using onlyvolume initiated by buyers to opennew option positions. Performing dailycross-sectional analyses from 1990 to2001, they find that buying stocks withlow put/call ratios and selling stockswith high put/call ratios generates anexpected return of 40 basis points perday and 1 percent per week. This resultoccurs during each year of the sample

period, and is not affected by theexclusion of earnings announcementwindows. Moreover, the result isstronger for smaller stocks, indicatingthat the options market may be a moreimportant avenue for informationtransmission for stocks with less effi-cient information flow. This analysisalso sheds light on the type ofinvestors behind the informed optiontrading. Specifically, option tradingfrom customers of full service brokersprovides the strongest predictability.The authors further show that whilepublic customers on average trade inthe options market as contrarians —buying fresh new puts on stocks thathave done well and calls on stocks thathave done poorly — firm proprietarytraders exhibit the opposite behavior.Finally, in contrast to the equityoptions market, there is no evidence inthe index options market of informedtrading.

Stambaugh explores the problemof making inferences about an assetthat has passed a survival test failed byother assets with lower realizedreturns. The more commonality thereis across assets in one’s prior uncer-tainty about unknown parameters, thegreater is the extent to which infer-ences about an asset’s expected return(or its alpha) are affected by its havingsurvived. In the absence of common-ality, a sample average can possess sub-stantial survival bias but can still equalthe appropriate inference about an

asset’s expected return. Various formsof commonality in returns acrossassets also play key roles. Conditioningon survival usually lowers, but some-times can raise, a surviving asset’sinferred alpha. Survival bias, as typical-ly computed, generally gives too severean adjustment for survival unless oneassumes that expected returns on allassets, dead and alive, are equal to acommon value that is completelyunknown.

Vassalou and Xing use Merton’s(1974) option pricing model to com-pute default measures for individualfirms and to assess the effect ofdefault risk on equity returns. The sizeeffect is a default effect, and this is alsolargely true for the book-to-market(BM) effect. Both exist only in seg-ments of the market with high defaultrisk. Default risk is systematic risk. TheFama-French (FF) factors containsome default-related information, butthis is not the main reason that the FFmodel can explain the cross-section ofequity returns.

Bansal, Dittmar, and Lundbladmodel dividend and consumptiongrowth rates as a vector-autoregression(VAR), from which they measure thelong-run response of dividend growthrates to consumption shocks. Theyfind that this long-run cash flow betacan justify well over 50 percent of thedifference in risk premiums across size,book-to-market, and industry sortedportfolios. Interestingly, the long-run

44. NBER Reporter Spring 2003

cash flow betas explain about half ofthe dispersion in the standard CapitalAsset Pricing Mode- based portfoliobetas for these assets. The authors’model highlights the reasons for thefailure of the market beta to justify thecross-section of risk premiums. Themarket beta itself is a weighted combi-nation of cash flow betas and addition-al priced sources of risk. Each risksource’s beta may be significant; how-

ever, a weighted combination of thebetas may not be significant in explain-ing the cross-section of risk premiums,as each source of risk carries a distinctprice. The results indicate that the size,book-to-market, and industry spreadsare not puzzling from the perspectiveof economic models.

Lamont studies battles betweenshort sellers and firms. Firms use avariety of means to impede short sell-

ing, including explicit or implicit legalthreats, investigations, lawsuits, andvarious technical actions intended tocreate a short squeeze. Firms that takethese actions create short-sale con-straints. Consistent with the hypothe-sis that short-sale constraints allowstocks to be overpriced, firms takinganti-shorting actions in the next yearhave very low abnormal returns ofabout negative 2 percent per month.

Corporate Finance

Axelson argues that an importantfriction in the issuance of financialsecurities is the fact that potentialinvestors may be privately informedabout the value of the underlyingassets. He shows how security designcan help to overcome this friction. Inthe single asset case, debt is often anoptimal security when the number ofpotential investors is small, but equitybecomes optimal as the degree ofcompetition increases. In the multipleasset case, debt backed by a pool ofassets is optimal if the number ofassets is large relative to the degree ofcompetition, but equity backed byindividual assets is optimal when thenumber of assets is small relative tothe degree of competition. Axelsonuses the theory to interpret security

design choices in financial markets.Gatev and Strahan argue that

banks have a unique ability to hedgeagainst market-wide liquidity shocks.Deposit inflows provide a hedge forloan demand shocks that follow declinesin market liquidity. Consequently, onedimension of bank “specialness” is thatbanks can insure firms against system-atic declines in market liquidity atlower cost than other financial institu-tions. The authors provide supportingevidence from the commercial paper(CP) market. When market liquiditydries up and CP spreads increase,banks experience funding inflows.These inflows allow banks to meetincreased loan demand from borrow-ers drawing funds from pre-existingCP backup lines, without running down

their holdings of liquid assets. More-over, the supply of cheap funds is largeenough that pricing on new lines ofcredit actually falls as market spreadswiden.

Kaplan and Schoar investigateindividual fund returns in the privateequity industry using a unique datasetcollected by Venture Economics. Theyfind a large degree of heterogeneityamong fund returns. Those returnspersist strongly across funds by privateequity firms. The returns also improvewith firm experience. Better perform-ing funds are more likely to raise fol-low-on funds and raise larger fundsthan more poorly performing firms.This relationship is concave, so thattop performing funds grow moreslowly than the market average. Finally,

The NBER’s Program on Cor-porate Finance met in Chicago onApril 11. Program Director RaghuramRajan and Per Strömberg, both ofNBER and University of Chicago,organized this program:

Ulf Axelson, University ofChicago, “Security Design withInvestor Private Information”Discussant: Philip Bond,Northwestern University

Evan Gatev and Philip E. Strahan,Boston College, “Banks’ Advantagein Hedging Liquidity Risk: Theoryand Evidence from the CommercialPaper Market”Discussant: Jeremy C. Stein, NBERand Harvard University

Steven N. Kaplan, NBER andUniversity of Chicago, andAntoinette Schoar, NBER and MIT,“Private Equity Returns: Persistenceand Capital Flows”Discussant: Alexander Ljunqvist,New York University

Atif Mian, University of Chicago,“Incentives, Supervision, andOrganizational Hierarchy: A Loan-Level Investigation of Banking”Discussant: Paola Sapienza,Northwestern University

Rafael La Porta and AndreiShleifer, NBER and HarvardUniversity, and Florencio Lopez-de-Silanes, NBER and YaleUniversity, “What Works inSecurities Laws?”

Discussant: Doug Baird, Universityof Chicago

Roman Inderst, London School ofEconomics, and Holger M. Müller,New York University, “The Effectof Capital Market Characteristics onthe Value of Start-Up Firms”Discussant: Stewart C. Myers,NBER and MIT

Marianne P. Bitler, RANDCorporation; Tobias J. Moskowitz,NBER and University of Chicago;and Annette Vissing-Jorgensen,NBER and NorthwesternUniversity, “Why Do EntrepreneursHold Large Ownership Shares?Testing Agency Theory UsingEntrepreneur Effort and Wealth”Discussant: Dirk Jenter, MIT

NBER Reporter Spring 2003 45.

the authors find that funds that areraised in boom times (and firms thatare started in boom times) are less like-ly to raise a follow-on fund, suggestingthat these funds perform worse.Several of these results differ substan-tially from those for mutual funds.

Mian uses a unique dataset thatcontains detailed information on everycorporate loan outstanding in thebanking sector of Pakistan during2001-2 (52,000 loans). Using a simpleempirical methodology that allows himto measure separately ex-ante monitor-ing, ex-post monitoring, “softness”,renegotiation, recovery, and litigationof loans, Mian presents a number ofnew findings. First, concerns aboutweak external supervision can be over-come by market-discipline coupledwith private incentives for banks.Second, government involvement isthe key to poor banking. Third, evenwith private markets, the organization-al hierarchy of banks can seriouslylimit their ability to lend to “soft infor-

mation” firms — firms in greatestneed of intermediation.

La Porta, Lopez-de-Silanes, andShleifer examine the effect of securi-ties laws on market development in 49countries. They find that publicenforcement of laws benefits securi-ties markets, especially in countrieswith efficient government bureaucra-cies. They also find that organizationof private enforcement through dis-closure and liability rules benefits secu-rities markets in countries with bothefficient and inefficient governmentbureaucracies.

Inderst and Müller show how thevalue, valuation, and success probabili-ty of start-ups depend on characteris-tics of the capital market. Market char-acteristics, including the return oninvestments, entry costs, and capitalmarket transparency, affect the relativesupply and demand for capital, andthus the relative bargaining power ofentrepreneurs and venture capitalists.Relative bargaining power, in turn,

determines ownership shares andincentives in start-up firms. In charac-terizing the short- and long-rundynamics of the venture capital mar-ket, the authors’ model sheds light onthe Internet boom and bust periods.

Bitler, Moskowitz, and Vissing-Jörgensen augment the standard prin-cipal-agent model to accommodate anentrepreneurial setting, where effort,ownership, and firm size are deter-mined endogenously. They test themodel’s predictions (some novel) usingnew data on entrepreneurial effort andwealth. They find that entrepreneurialownership shares increase with outsidewealth, decrease with firm risk, anddecrease with firm size; effort increas-es with ownership and size; and bothownership and effort increase firmperformance. The magnitude of theeffects in the cross-section of firmssuggests that agency theory is impor-tant for explaining the large averageownership shares of entrepreneurs.

Behavioral FinanceThe NBER’s Working Group on

Behavioral Finance met in Chicagoon April 12. Nicholas Barberis andRichard H. Thaler, both of NBERand University of Chicago, organizedthe meeting. These papers were dis-cussed:

Alan B. Krueger, NBER andPrinceton University, and KennethN. Fortson, Princeton University,“Do Markets Respond More toMore Reliable Labor Market Data?”Discussant: Anil K Kashyap, NBERand University of Chicago

Ming Dong, York University;David Hirshleifer and Siew HongTeoh, Ohio State University; and

Scott Richardson, University ofPennsylvania, “Does InvestorMisvaluation Drive the TakeoverMarket?”

Matthew Rhodes-Kropf andDavid T. Robinson, ColumbiaUniversity, and S. Viswanathan,Duke University, “Valuation Wavesand Merger Activity: The EmpiricalEvidence”Discussant: Steven N. Kaplan,NBER and University of Chicago

Markus K. Brunnermeier,Princeton University, and JonathanA. Parker, NBER and PrincetonUniversity, “Optimal Expectations”Discussant: Sendhil Mullainathan,

NBER and MIT

Louis K.C. Chan, University ofIllinois; Jason Karceski, Universityof Florida; and Josef Lakonishok,NBER and University of Illinois,“Analysts’ Conflict of Interest andBiases in Earnings Forecasts”Discussant: Russell Fuller, Fuller andThaler Asset Management

Asim Ijaz Khwaja, HarvardUniversity, and Atif Mian,University of Chicago, “PriceManipulation and Phantom Markets:An In-depth Exploration of a StockMarket”Discussant: Pete Kyle, DukeUniversity

Krueger and Fortson note thatsince 1979, the Bureau of LaborStatistics (BLS) has nearly quadrupledthe size of the sample used to estimatemonthly employment changes. Al-though first-reported employmentestimates are still noisy, the magnitudeof sampling variability has declined in

proportion to the increase in the sam-ple size. Still, a regression analysis ofchanges in interest rates on the day theemployment data are released finds noevidence that the bond market’s reac-tion to employment news intensified inthe late 1980s or 1990s; indeed, in thelate 1990s and early 2000s the bond

markets hardly reacted to unexpectedemployment news. For the time periodas a whole, an unexpected increase of200,000 jobs is associated with about a6 basis point increase in the interestrate on 30-year Treasury bonds, and an8 basis point increase in the interestrate on 3-month bills, all else equal.

46. NBER Reporter Spring 2003

Additionally, unexpected changes inthe unemployment rate and revisionsto past months’ employment estimateshave statistically insignificant effectson long-term interest rates.

Dong, Hirshleifer, Richardson,and Hong Teoh show that irrationalmarket misvaluation, at both the trans-action and aggregate levels, affects thevolume and character of takeoveractivity. The authors examine pre-takeover book/price ratios and pre-takeover ratios of residual-income-model-valuation-to-price for bidders,targets, and the aggregate stock marketas proxies for market misvaluation.They find that misvaluation of bidders,targets, and the aggregate stock marketinfluences the aggregate volume oftakeovers, the means of payment cho-sen, the premiums paid, target hostilityto the offer, the likelihood of offer suc-cess, bidder and target announcementperiod stock returns, post-takeoverlong-run returns, and the returns fromdiversifying transactions.

Merger intensity spikes in times ofhigh market valuations (that is, whenaverage market-to-book, M/B, ratiosare at their highest). To explorewhether this is the result of correlatedvaluation errors or behavioral mispric-ing, Rhodes-Kropf, Robinson, andViswanathan decompose M/B intothree components: firm-specific devia-tion from short-run industry valua-tions; short-run industry deviationsfrom long-run values, and long-runvalue to book. The fact that high M/Bbuys lower M/B is driven mostly byfirm-specific deviations from short-run industry average pricing. However,both targets and “acquires” are pricedabove their long-run industry average.When the authors find differencesbetween bidders and targets in long-run value-to-book, they find that lowbuys high. They also find that theindustry-specific component of M/B

is highly positively correlated withmerger intensity, and with the use ofstock. However, long-run value-to-book is not correlated with cash merg-er intensity and is negatively correlatedwith stock merger intensity, leading tolittle overall correlation between long-run value-to-book and merger activity.

Brunnermeier and Parker intro-duce a tractable, structural model ofsubjective beliefs. Since agents whoplan for the future care about expectedfuture utility flows, current felicity canbe increased by believing that betteroutcomes are more likely. On the otherhand, expectations that are biasedtowards optimism worsen decision-making, leading to poorer realizedoutcomes on average. Optimal expec-tations balance these forces by maxi-mizing the total well-being of an agentover time. The authors apply theirframework of optimal expectations tothree different economic settings. In aportfolio choice problem, agents over-estimate the return of their investmentand underdiversify. In general equilibri-um, agents’ prior beliefs are endoge-nously heterogeneous, leading to gam-bling. Second, in a consumption-savingproblem with stochastic income,agents are both overconfident andoveroptimistic, and consume morethan implied by rational beliefs early inlife. Third, in choosing when to under-take a single task with an uncertaincost, agents exhibit several features ofprocrastination, including regret,intertemporal preference reversal, anda greater readiness to accept commit-ment.

Analysts’ earnings forecasts areinfluenced by their desire to wininvestment banking clients. Chan,Karceski, and Lakonishok hypothe-size that the equity bull market of the1990s, along with the boom in invest-ment banking business, exacerbatedanalysts’ conflict of interest and their

incentives to strategically adjust fore-casts in order to avoid earnings disap-pointments. The authors documentshifts in: the distribution of earningssurprises; the market’s response to sur-prises and forecast revisions; and inthe predictability of non-negative sur-prises. Further confirmation is basedon samples with conflicts of interestthat are higher (including growthstocks and stocks with consecutivenon-negative surprises) or lower (suchas foreign markets).

Khwaja and Mian analyze aunique dataset containing all dailyfirm-level trades of every broker trad-ing on the stock exchange in Pakistanover a 32-month period. Examiningbroker behavior reveals that many bro-kers choose stocks in which they onlytrade for themselves rather than actingas intermediaries for outside investors.The authors find that when brokerstrade on their own behalf in a stock —act as “principals” — they earn 4 per-cent to 8 percent higher annual rates ofreturn. While broker “ability” does notexplain this effect, anecdotes suggest itis caused by direct price manipulationby brokers. The authors find strongevidence for such manipulation: whenprices are low, colluding brokers tradeamongst themselves to artificially raiseprices and attract naïve positive-feed-back traders. Once prices have risen,the former exit, leaving the latter tosuffer the ensuing price fall. Suchmanipulation of stock prices occurs inall types of stocks. However, the effectis larger in stocks of smaller firms, andfor firms with less concentrated owner-ship. Finally, while the higher prof-itability of principals is not attributableto inherent broker attributes, theauthors do find that more “able” bro-kers earn higher returns when theytrade as a principal in a stock.

NBER Reporter Spring 2003 47.

Environmental EconomicsThe NBER’s Working Group on

Environmental Economics met inCambridge on April 12. Don Fullerton,NBER and University of Texas,Austin, organized the meeting. Thesepapers were discussed:

David F. Bradford, NBER andPrinceton University, “Improving onKyoto: Greenhouse Gas Control asthe Purchase of a Global PublicGood”Discussant: Peter Wilcoxen,Syracuse University

Scott Barrett, Johns HopkinsUniversity, “Global DiseaseEradication”Discussant: Joshua Graff Zivin,Columbia University

Brian R. Copeland, University ofBritish Columbia, and M. ScottTaylor, NBER and University ofWisconsin, “Trade, Tragedy, and theCommons”Discussant: Istvan Konya, BostonCollege

Mustafa H. Babiker and JohnReilly, MIT; and Gilbert E.Metcalf, NBER and TuftsUniversity, “Tax Distortions andGlobal Climate Policy” (NBERWorking Paper No. 9136)Discussant: Lawrence H. Goulder,NBER and Stanford University

W. Michael Hanemann, Universityof California, Berkeley; Sheila M.

Olmstead, Yale University; andRobert N. Stavins, HarvardUniversity; “Does Price StructureMatter? Household Water DemandUnder Increasing-Block andUniform Prices”Discussant: Li Gan, University ofTexas at Austin

John A. List, University ofMaryland; Michael Margolis,Resources for the Future; andDaniel E. Osgood, University ofArizona, “Measuring thePreemption of Regulatory Takingsin the U.S. Endangered Species Act:Evidence from a NaturalExperiment”Discussant: Steve Polasky, Universityof Minnesota

One way to obtain a global publicgood is to set up an institution to buyit, with the nations of the world con-tributing to the cost according to what-ever sharing arrangements make politi-cal sense. Bradford suggests a way toexploit this approach in order to limitthe accumulations of greenhouse gasesin the atmosphere. The “service” thatproduces the control is the reduction inthe levels of emissions over time fromwhat the nations otherwise wouldchoose, also known as the “business asusual” emissions path. In the scheme asenvisioned, which could be used in asuccessor agreement to the KyotoProtocol, the fact that all nations aresellers of reductions ameliorates theenforcement problems typical of com-mitments to particular emission paths.Another difference from the Kyoto-style system: in the scheme sketchedhere, the distribution of burdens isexplicit, rather than implicit, in theallowable emission amounts. An infec-tious disease only can be eradicatedglobally if it is eliminated in everycountry. But does this simply requireinternational coordination, or does italso require cooperation? Using amodel that blends epidemiology, eco-nomics, and game theory, Barrettshows that coordination will not alwayssuffice, even when the global benefitsof eradication exceed the costs. In gen-

eral, eradication will require stronginternational institutions.

Copeland and Taylor develop atheory of resource managementwhereby the degree to which countriesescape “the tragedy of the commons”is determined endogenously and linkedexplicitly to changes in world pricesand other possible effects of marketintegration. The authors show howchanges in world prices can move somecountries from de facto open accesssituations to ones in which manage-ment replicates an unconstrained socialplanner. Not all countries can followthis path of institutional reform andthe authors identify key country char-acteristics (mortality rates, resourcegrowth rates, technology) that dividethe world’s set of resource rich coun-tries into three categories. Category Icountries will never be able to effec-tively manage their renewable re-sources. Category II countries exhibitde facto open access for low resourceprices, but can maintain a limited formof resource management at higherprices. Category III countries can fullyimplement efficient management andcan obtain the unconstrained first bestoutcome for some range of resourceprices. For Category III countries, defacto open access and limited manage-ment are but transitory phases.

Babiker, Metcalf, and Reilly con-

sider the efficiency implications ofpolicies to reduce global carbon emis-sions in a world with pre-existing taxdistortions. They first note that theweak double dividend — the proposi-tion that the welfare improvementfrom a tax reform in which environ-mental taxes are used to lower distort-ing taxes must be greater than the wel-fare improvement from a reform inwhich the environmental taxes arereturned in a lump sum fashion —need not hold in a world with multipledistortions. They then present a large-scale computable general equilibriummodel of the world economy with dis-tortionary taxation. They use thismodel to evaluate a number of policiesto reduce carbon emissions. They findthat the weak double dividend is notobtained in a number of Europeancountries. Their results also demon-strate that the interplay between car-bon policies and pre-existing taxes candiffer markedly across countries. Thus,one must be cautious in extrapolatingthe results from a country specificanalysis to other countries.

Olmstead, Hanemann, andStavins analyze the influence of theprice of water and the structure ofwater prices on residential waterdemand. They adapt a model from thelabor economics literature — theHausman model of labor supply under

48. NBER Reporter Spring 2003

progressive income taxation — to esti-mate water demand under increasing-block prices. They apply this structuralmodel to the most price-diverse,detailed, household-level water demanddata now available to estimate the priceelasticity of residential water demand.Their results indicate that the sensitiv-ity of residential water demand toprice is quite low, in contrast withresults of previous studies using simi-lar models to account for the piece-wise-linear budget constraint of blockprices. They also find, however, that

price elasticity is higher and demand islower among households facing blockprices than among households facinguniform marginal prices. The impactof the price structure on demandappears to be greater than the impactof marginal price itself.

Does the endangered species actendanger species? Margolis, Osgood,and List derive two empirical meas-ures to estimate the extent of preemp-tive habitat destruction. They illustratethe use of such measures by examiningdevelopment decisions on more than

70,000 plots of land that are potential-ly Pygmy Owl-critical habit areas inPima County, Arizona. Their modelsprovide direct measures of the scale ofpreemptive habitat destruction in unitsthat can indicate whether it is signifi-cant as an obstacle to conservation, orhow much it adds to the social cost ofachieving conservation goals. Thepreliminary findings suggest that pre-emption is occurring at rates that arestatistically significant.

Globalization inHistoricalPerspective

Globalization in Historical Perspective,edited by Michael D. Bordo, Alan M.Taylor, and Jeffrey G. Williamson, willbe available from the University ofChicago Press this spring for $95. Thisvolume collects eleven papers andcomments and a panel discussion, andpresents a comprehensive view ofglobalization. The volume is dividedinto three parts: the first explores howthe progress of globalization shouldbe measured in terms of the integra-tion of various markets; the secondplaces this knowledge in a wider con-text, examining why divergence ratherthan convergence has been the morepredominant outcome of globalizationefforts, with special attention paid tothe effects of technology and geogra-phy; the third studies the role of thefinancial sector, including exchangerate regimes, financial development,financial crises, and the architecture ofthe international financial system.Overall, the contributors show how anearlier global system was established

during the 19th century, and how anew system has evolved following thedisruptions of World War I, the GreatDepression, and World War II.

Bordo, Taylor, and Williamson areall NBER Research Associates. Bordois also a professor of economics atRutgers University. Taylor is associateprofessor of economics at theUniversity of California, Davis.Williamson is the Laird Bell Professorof Economics at Harvard University.

Health and LaborForce Participationover the Life Cycle:Evidence from thePast

Health and Labor Force Participationover the Life Cycle: Evidence from the Past,edited by Dora L. Costa, is availablefrom the University of Chicago Pressthis spring for $75.

The rise in life expectancy andretirement rates in the twentieth centu-ry has had dramatic effects.Forecasting future trends in health and

retirement rates, as we must now do,requires that we investigate underlyinglong-term trends and their causes. Tothat end, this volume draws on newdata — an extensive survey of UnionArmy veterans who were bornbetween 1820 and 1850 — to examinethe factors that affected health andlabor force participation in the UnitedStates in the nineteenth century. Thecontributors consider the impact of avariety of conditions — social class,wealth, occupation, family, and com-munity — on the morbidity and mor-tality of the group, and also addressseveral more specialized topics.

Costa is an NBER ResearchAssociate and the Ford CareerDevelopment Associate Professor ofEconomics at MIT. She is also theauthor of The Evolution of Retirement: AnAmerican Economic History, 1880-1990.

The Economics ofSchool Choice

The Economics of School Choice, editedby Caroline M. Hoxby, is available nowfrom the University of Chicago Pressfor $75. Since the U.S. Supreme Court

The following volumes may be ordered directly from the University of Chicago Press, Order Department, 11030 SouthLangley Avenue, Chicago, IL 60628-2215; 1-800-621-2736. Academic discounts of 10 percent for individual volumes and20 percent for standing orders for all NBER books published by the University of Chicago Press are available to universi-ty faculty; orders must be sent on university stationery.

Bureau Books

NBER Reporter Spring 2003 49.

has declared school voucher programsconstitutional, the many unansweredquestions about the potential effects ofschool choice will become especiallypressing. Contributors to this volumeinvestigate the ways in which schoolchoice affects a wide range of issues,presenting evidence on the impact ofschool choice on student achievement,school productivity, teachers, and spe-cial education. They also tackle suchdifficult questions as whether schoolchoice affects where people decide to

live, and how choice can be integratedinto a system of school financing thatgives children from different back-grounds equal access to resources.There is a discussion of the latest find-ings on Florida’s school choice program,as well as on voucher programs andcharter schools in several other states.

The resulting volume not onlyreveals the promise of school choice,but examines its pitfalls as well, show-ing how programs can be designedthat exploit the idea’s potential but

avoid its worst effects. With schoolchoice programs gradually becomingboth more possible and more popular,this book stands out as an essentialexploration of the effects such pro-grams will have, and a necessaryresource for anyone interested in theidea of school choice.

Hoxby directs the Economics ofEducation Program at the NationalBureau of Economic Research and is aprofessor of economics at HarvardUniversity.

NBER Working Papers On-Line

A complete list of all NBER Working Papers with searchable abstracts, and the full texts of Working Papers (issued sinceNovember 1994) are available at http://www.nber.org/wwp.html to anyone located at a university or other organization that sub-scribes 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 theNBER 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 avail-

able free of charge to Corporate Associates. For all others, there is a charge of $10.00 per hardcopy or $5.00 per downloadedpaper. (Outside the United States, add $10.00 per order for postage and handling.) Advance payment is required on allorders. To order, call the Publications Department at (617)868-3900 or visit www.nber.org/papers. Please have ready the num-ber(s) of any Working Paper(s) you wish to order.

Subscriptions to the full NBER Working Paper series include all 500 or more papers published each year. Subscriptions arefree to Corporate Associates. For others within the United States, the standard rate for a full subscription is $2200; for academiclibraries and faculty members, $1275. Higher rates apply for foreign orders. The on-line standard rate for a full subscription is $1560and the on-line academic rate is $630. 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 MassachusettsAvenue, Cambridge, MA 02138-5398.

*Titles of all papers issued since January 2003 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 dis-cussion and suggestions for revision before final publication. They are not reviewed by the Board of Directors of the NBER.

Current Working Papers

9408 Dirk Krueger Skill-specific rather than General Education: A Reason for Krishna B. Kumar US-Europe Growth Differences?

9409 Stephen Ansolobehere Why Is There So Little Money in Politics?John M. de FigueredoJames M. Snyder

9410 Dirk Krueger Pareto Improving Social Security Reform when Financial Felix Kubler Markets are Incomplete?

Paper Author(s) Title

NBER Working Papers

50. NBER Reporter Spring 2003

Paper Author(s) Title

9411 Alberto Alesina FractionalizationArnaud DevleeschauwerWilliam EasterlySergio KurlatRomain Wacziarg

9412 Kent A. Smetters Trading with the Unborn: A New Perspective on Capital Income Taxation

9413 Brian Jacob Rotten Apples: An Investigation of the Prevalence of TeacherSteven D. Levitt Cheating

9414 Brian Jacob Catching Cheating Teachers: The Results of an Unusual Steven D. Levitt Experiment in Implementing Theory

9415 Gilbert E. Metcalf Oligopoly Deregulation and the Taxation of CommoditiesGeorge Norman

9416 Gilbert E. Metcalf Oligopoly Deregulation in General Equilibrium: A TaxGeorge Norman Neutralization Result

9417 Robert E. Lipsey Foreign Firms and Indonesian Manufacturing Wages: An Fredrik Sjöholm Analysis With Panel Data

9418 Alan B. Krueger Another Look at the New York City School VoucherPei Zhu Experiment

9419 Marc P. Giannoni Optimal Interest-Rate Rules: I. General TheoryMichael Woodford

9420 Marc P. Giannoni Optimal Interest-Rate Rules: II. ApplicationsMichael Woodford

9421 Lars E. O. Svensson What is Wrong with Taylor Rules? Using Judgment InMonetary Policy through Targeting Rules

9422 Steven D. Levitt How Do Markets Function? An Empirical Analysis ofGambling on the National Football League

9423 Eli Ofek Limited Arbitrage and Short Sales Restrictions: Evidence fromMatthew Richardson the Options Markets Robert F. Whitelaw

9424 Ricardo Hausmann An Alternative Interpretation of the “Resource Curse”: TheoryRoberto Rigobon Policy Implications

9425 David H. Autor The Costs of Wrongful-Discharge LawsJohn J. Donohue IIIStewart J. Schwab

9426 Itay Goldstein An Information-Based Trade Off Between Foreign DirectAssaf Razin Investment and Foreign Portfolio Investment: Volatility,

Transparency, and Welfare

9427 Barry Eichengreen Capital Account Liberalization and Growth: Was Mr. MahathirDavid Leblang Right?

9428 Janet Currie Cut to the Bone? Hospital Takeovers and Nurse EmploymentMehdi Farsi ContractsW. Bentley MacLeod

9429 Bradley T. Heim Work Costs and Nonconvex Preferences in the Estimation ofBruce D. Meyer Labor Supply Models

9430 Lars E. O. Svensson Optimal Policy with Partial Information in a Forward-LookingMichael Woodford Model

9431 Ashish Arora R&D and the Patent PremiumMarco CeccagnoliWesley M. Cohen

NBER Reporter Spring 2003 51.

Paper Author(s) Title

9432 Gayané Hovakimian Corporate Investment with Financial Constraints: Sensitivity ofSheridan Titman Investment to Funds from Voluntary Asset Sales

9433 Martin D. D. Evans How is Macro News Transmitted to Exchange Rates? Richard K. Lyons

9434 Leonid Kogan The Price Impact and Survival of Irrational TradersStephen RossJiang WangMark Westerfield

9435 Silvana Tenreyro Economic Effects of Currency UnionsRobert J. Barro

9436 Kenn Ariga Mismeasurement of the CPIKenji Matsui

9437 Robert J. Gordon Hi-tech Innovation and Productivity Growth: Does Supply Create Its Own Demand?

9438 Lee Branstetter Is Japan’s Innovative Capacity in Decline?Yoshiaki Nakamura

9439 Elhanan Helpman Export versus FDIMarc J. MelitzStephen R. Yeaple

9440 Daniel S. Hamermesh Routine

9441 Wayne E. Ferson Tests of Multifactor Pricing Models, Volatility Bounds, andPortfolio Performance

9442 Francesco Caselli Dynastic ManagementNicola Gennaioli

9443 Pinelopi K. Goldberg The Response of the Informal Sector to Trade LiberalizationNina Pavcnik

9444 Albert Ando Inefficiency of Corporate Investment and Distortion ofDimitrios Christelis Savings Behavior in JapanTsutomu Miyagawa

9445 Kris James Mitchener The Productivity of the U.S. Since 1880Ian W. McLean

9446 David Cutler Why Have Americans Become More Obese?Edward GlaeserJesse Shapiro

9447 Stephen Redding Distance, Skill Deepening, and Development: Will PeripheralPeter K. Schott Countries Ever Get Rich?

9448 Cameron Howell Legacies in Black and White: The Racial Composition of the Sarah E. Turner Legacy Pool

9449 Steven D. Levitt Testing Theories of Discrimination : Evidence from WeakestLink

9450 David Flath Regulation, Distribution Efficiency, and Retail Density

9451 Hilary Sigman Letting States do the Dirty Work: State Responsibility for Federal Environmental Regulation

9452 Amil Petrin Omitted Product Attributes in Discrete Choice Models Kenneth Train

9453 Harvey S. Rosen Portfolio Choice and Health StatusStephen Wu

9454 Alexander Ljungqvist The Cash Flow, Return, and Risk Characteristics of Private Matthew Richardson Equity

52. NBER Reporter Spring 2003

Paper Author(s) Title

9455 Michael Baker Simulating the Response to Reform of Canada’s Income Jonathan Gruber Security ProgramsKevin Milligan

9456 James M. Poterba Inter-Asset Differences in Effective Estate Tax Burdens Scott Weisbenner

9457 Gerardo della Paolera Gauch Banking ReduxAlan M. Taylor

9458 Michael W. Brandt Time-Consistent No-Arbitrage Models of the Term StructureAmir Yaron

9459 Jean Boivin Has Monetary Policy Become More Effective?Marc Giannoni

9460 Amy Finklelstein Health Policy and Technological Change: Evidence from theVaccine

9461 Bruce N. Lehmann Diversification and the Optimal Construction of Basis David M. Modest Portfolios

9462 Todd Sinai Owner-Occupied Housing as a Hedge Against Rent Risk Nicholas Souleles

9463 Michael R. Darby Universities, Joint Ventures, and Success in the AdvancedLynne G. Zucker Technology ProgramAndrew Wang

9464 Stephen J. Brown Fees on Fees in Funds of FundsWilliam N. GoetzmannBing Liang

9465 William N. Goetmann Rain or Shine: Where is the Weather Effect?Ning Zhu

9466 William N. Goetzmann Efficiency and the Bear: Short Sales and Markets around theNing Zhu WorldArturo Bris

9467 C. Nicholas McKinney The Collapse of a Medical Clearinghouse (and Why Such Muriel Niederle Failures are Rare)Alvin E. Roth

9468 Christopher J. Ruhm Healthy Living in Hard Times

9469 William N. Goetzmann Modeling and Measuring Russian Corporate Governance: TheMatthew Spiegel Case of Russian Preferred and Common Shares

Andrey Ukhov

9470 Stephen J. Brown Investor Sentiment in Japanese and U.S. Daily Mutual FundWilliam N. Goetzmann FlowsTakato HirakiNoriyoshi ShirishiMasahiro Watanabe

9471 Martin Gaynor Competition Among HospitalsWilliam B. Vogt

9472 Jeffrey Grogger Welfare Transitions in the 1990s: The Economy, Welfare Policy,and the EITC

9473 David Card Unionization and Wage Inequality: A Comparative Study of the Thomas Lemieux U.S., the U.K., and CanadaW. Craig Riddell

9474 Lance Lochner Individual Perceptions of the Criminal Justice System

9475 Steven R. Grenadier An Equilibrium Analysis of Real Estate

9476 Amitabh Chandra Is the Convergence in the Racial Wage Gap Illusory?

NBER Reporter Spring 2003 53.

9477 Roger Gordon Do We Now Collect any Revenue from Taxing Capital Income?Laura KalambokidisJoel Slemrod

9478 Raghuram Rajan The Emergence of Strong Property Rights: Speculations fromLuigi Zingales History

9479 Graciela Kaminsky The Center and the Periphery: The Globalization of FinancialCarmen Reinhart Turmoil

9480 Li Gan Individual Subjective Survival CurvesMichael HurdDaniel McFadden

9481 Rodolfo Martell Equity Market Liberalizations as Country IPOsRené M. Stulz

9482 Christopher Avery Do and Should Financial Aid Packages Caroline M. Hoxby Affects Students’ College Choices?

9483 Steven Shavell Economic Analysis of Accident Law

9484 Hiroshi Ono Constraints on the Level and Efficient Use of Labor in Japan Marcus E. Rebick

9485 Edgar Cudmore Border Delays and Trade LiberalizationJohn Whalley

9486 Lars E. O. Svensson Monetary Policy and Real Stabilization

9487 W. Kip Viscusi The Value of a Statistical Life: A Critical Review of Market Joseph E. Aldy Estimates throughout the World

9488 Peter Blair Henry Capital Account Liberalization, The Cost of Capital, and Economic Growth

9489 Magnus Blomstrom The Economics of Foreign Direct Investment Incentives Ari Kokko

9490 Jeffrey D. Sachs Institutions Don’t Rule: Direct Effects of Geography on Per Capita Income

9491 Laurence Ball Monetary Policy for Inattentive EconomiesN. Gregory MankiwRicardo Reis

9492 Shinichi Nishiyama Consumption Taxes and Economic Efficiency in a StochasticKent Smetters OLG Economy

9493 Andrei Shleifer Will the Sovereign Debt Market Survive?

9494 Raphael Desmet Micro-Stimulation of Social Security Reforms in Belgium Alain JoustenSergio PerelmanPierre Pestieau

9495 James Heckman Human Capital PolicyPedro Carneiro

9496 Courtney Coile Retirement Incentives and Couples’ Retirement Decisions

9497 James Heckman Using Matching, Instrumental Variables, and Control FunctionsSalvador Navarro-Lozano

9498 Eric van Wincoop Can Information Heterogeneity Explain the Exchange Rate Philippe Bacchetta Determination Puzzle?

9499 William N. Goetzmann Disposition Matters: Volume, Volatility, and Price Impact of aMassimo Massa Behavioral Bias

9500 Michael W. Klein Capital Account Openness and the Varieties of Growth Experience

Paper Author(s) Title

54. NBER Reporter Spring 2003

9501 Gordon C. Williams Peer Effects in Higher EducationDavid J. Zimmerman

9502 Alberto Alesina Fariness and Redistribution: U.S. versus EuropeGeorge-Marios Angeletos

9503 Roger H. Gordon Taxation of Interest Income

9504 Wolfgang Keller Multinational Enterprises, International Trade, and ProductivityGrowth: Firm-Level Evidence from the United States

9505 Jennifer Ma Education Saving Incentives and Household Saving: Evidence from the 2000 TIAA-CREF Survey of Participant Finances

9506 Eric M. Leeper Fiscal Policy and Inflation: Pondering the Imponderables

9507 Laura Argys Who Shall Live and Who Shall Die? An Analysis of PrisonersNaci Mocan on Death Row in the United States

9508 Melvin Stephens, Jr. Job Loss Expectations, Realizations, and HouseholdConsumption Behavior

9509 John Y. Campbell Bad Beta, Good BetaTuomo Vuolteenaho

9510 Geert Bekaert Market Integration and ContagionCampbell R. HarveyAngela Ng

9511 Margaret Levenstein International Price-Fixing Cartels and Developing Countries:Valerie Suslow A Discussion of Effects and Policy RemediesLynda Oswald

9512 Rajnish Mehra The Equity Premium: Why is it a Puzzle?

9513 Amitabh Chandra Geography and Racial Health DisparitiesJonathan Skinner

9514 Sebastian Edwards A Currency of One’s Own? An Empirical Investigation on Igal Magendzo Dollarization and Independent Currency Unions

9515 Jacob Boudoukh Do Asset Prices Reflect Fundamentals? Freshly SqueezedMatthew Richardson Evidence from the OJ Market Jeffrey (YuQing) ShenRobert F. Whitelaw

9516 Michael J. Rizzo Resident and Nonresident Tuition and Enrollment at Flagship Ronald G. Ehrenberg State Universities

9517 Karolina Ekholm Export-Platform Foreign Direct Investment Rikard ForslidJames Markusen

9518 David Popp Time In Purgatory: Determinants of the Grant Lag for U.S.Ted Juhl Patent ApplicationsDaniel K. N. Johnson

9519 Richard Steckel What Can Be Learned from Skeletons that Might InterestEconomists, Historians, and Other Social Scientists?

9520 Michael Bordo Is Deflation Depressing? Evidence from the Classical Gold Angela Redish Standard

9521 Menzie Chinn Doomed to Deficits? Aggregate U.S. Trade Flows Re-Examined

9522 Hugh Rockoff Deflation, Silent Runs, and Bank Holidays, in the GreatContraction

9523 Sara B. Moeller Do Shareholders of Acquiring Firms Gain Acquisitions? Frederik P. SchlingemannRene M. Stulz

Paper Author(s) Title

NBER Reporter Spring 2003 55.

Paper Author(s) Title

9524 Alan B. Krueger Strikes, Scabs, and Tread Separations: Labor Strife and the Alexandre Mas Production of Defective Bridgestone/Firestone Tires

9525 Rajnish Mehra The Equity Premium in RetrospectEdward C. Prescott

9526 W. Erwin Diewert Measuring Capital

9527 Bruce A. Blonigen CEO Turnover and Foreign Market ParticipationRossitza B. Wooster

9528 Glenn Ellison Knife Edge of Plateau: When Do Market Models Tip? Drew Fudenberg

9529 Michael W. Brandt Price Discovery in the U.S. Treasury Market: The Impact ofKenneth A. Kavajecz Orderflow and Liquidity on the Yield Curve

9530 Martin Feldstein Why is Productivity Growing Faster?

9531 Siaf I. Shah Mohammed Freight Rates and Productivity Gains in British TrampJeffrey G. Williamson Shipping 1869-1950

9532 John J. Donohue III Further Evidence that Legalized Abortion Lowered Crime:Steven D. Levitt A Reply to Joyce

9533 Pierpaolo Battigalli International Agreements on Product Standard: AnGiovanni Maggi Incomplete-Contracting Theory

9534 Paul L. Joskow Merchant Transmission Investment Jean Tirole

9535 Roger Gordon A New Measure of the Effective Tax Rate on Investment Laura KalambokidisJoel Slemrod

9536 Robert Shimer The Cyclical Behavior of Equilibrium Unemployment and Vacancies: Evidence and Theory

9537 Kevin Lang The Effect of the Payroll Tax on Earnings: A Test ofCompeting Models of Wage Determination

9538 Jonathan A. Parker Consumption Risk and Cross-Sectional ReturnsChristian Julliard

9539 David W. Galenson The Life Cycles of Modern Artists: Theory, Measurement,and Implications

9540 Dana P. Goldman The Reallocation of Compensation in Response to HealthArleen A. Leibowitz Insurance Premium IncreasesNeeraj Sood

9541 Julan Du Does Insider Trading Raise Market Volatility?Shang-Jin Wei

9542 Malcolm Baker A Catering Theory of DividendsJeffrey Wurgler

9543 Michael B. Devereaux Endogenous Exchange Rate Pass-through When NominalCharles Engel Prices are Set in AdvancePeter E. Strogaard

9544 Louis K.C. Chan Analysts’ Conflict of Interest and Biases in Earnings ForecastsJason KarceskiJosef Lakonishok

9545 Joshua D. Angrist Does Teacher Testing Raise Teacher Quality? Evidence fromJonathan Guryan State Certification Requirements

56. NBER Reporter Spring 2003

Paper Author(s) Title

9546 Pedro Carneiro Estimating Distributions of Treatment Effects with an Application to the Returns to Schooling and Measurement ofthe Effects of Uncertainty on College Choice

9547 John Y. Campbell Strategic Asset Allocation in a Continuous-Time VAR ModelGeorge ChackoJorge RodriguezLuis M. Viciera

9548 Jonathan A. Parker Consumption Risk and Expected Stock Returns

9549 Werner Troesken Lead Water Pipes and Infant Mortality in Turn-of-the-CenturyMassachusetts

9550 Chulhee Lee Labor Market Status of Older Males in the United States,1880-1940

9551 Rajeev Dehejia Why Does Financial Development Matter? The United StatesAdriana Lleras-Muney from 1900 to 1940

9552 Eric M. Leeper Putting “M” back in Monetary PolicyJennifer E. Roush

9553 Bridget Terry Long The Impact of Federal Tax Credits for Higher EducationExpenses

9554 Jonathan Meer Exploring the Health-Wealth NexusDouglas L. MillerHarvey S. Rosen

9555 Kristin J. Forbes A Decomposition of Global Linkages in Financial MarketsMenzie D. Chinn Over Time

9556 Richard Disney Pension Reform and Economic Performance in Britain in theCarl Emmerson 1980s and 1990sSarah Smith

9557 Guillermo Calvo Inflation Inertia and Credible Disinflation — The Open Oya Celasun Economy CaseMichael Kumhof

9558 Jess Benhabib Backward-Looking Interest-Rate Rules, Interest-RateStephanie Schmitt-Grohe Smoothing, and Macroeconomic InstabilityMartin Uribe

9559 Chiaki Moriguchi Implicit Contracts, the Great Depression, and Institutional Change: A Comparative Analysis of U.S. and JapaneseEmployment Relations, 1920-1940

9560 Alberto Alesina Regulation and InvestmentSilvia ArdagnaGiuseppe NicolettiFabio Schiantarelli

9561 John Cawley Lighting Up and Slimming Down: The Effects of Body WeightSara Markowitz and Cigarette Prices on Adolescent Smoking InitiationJohn Tauras

9562 William J. Collins The Housing Market Impact of State-Level Anti-Discrimination Laws, 1960-1970

9563 Gordon H. Hanson What Has Happened to Wages in Mexico since NAFTA?

9564 John Pencavel The Surprising Retreat of Union Britain

9565 David Weil Individual Rights and Collective Agents: The Role of Old andNew Workplace Institution in the Regulation of Labor Markets

NBER Reporter Spring 2003 57.

Paper Author(s) Title

9566 Alexei Onatski Modeling Model UncertaintyNoah Williams

9567 Jonathan Gruber Subsidies to Employee Health Insurance Premiums and the Ebonya Washington Health Insurance Market

9568 Douglas Laxton Monetary Rules for Small, Open Emerging Economies Paolo Pesenti

9569 Sandra E. Black How Workers Fare When Employers Innovate Lisa LynchAnya Krivelyova

9570 Ricardo Caballero On the International Financial Architecture: Insuring EmergingMarkets

9571 Mila Getmansky An Economic Model of Serial Correlation and Illiquidity inAndrew W. Lo Hedge Fund ReturnsIgor Makarov

9572 James E. Anderson Traders, Cops, and RobbersOriana Bandiera

9573 Rachel Griffith Characteristics of Foreign-Owned Firms in BritishHelen Simpson Manufacturing

9574 Chen Chien-Hsun Initial Public Offering and Corporate Governance in China’sShih Hui-Tzu Transitional Economy

9575 Youngjae Lim Sources of Corporate Financing and Economic Crisis in Korea:Micro-evidence

9576 John Van Reenen Active Labor Market Policies and the British New Deal for the Young Unemployed in Context

9577 Laurence Ball Does Inflation Targeting Matter?Niamh Sheridan

9578 Robert E. Baldwin Openness and Growth: What’s the Empirical Relationship?

9579 Roland G. Fryer Categorical Cognition: A Psychological Model of Categories Matthew O. Jackson and Identification in Decisionmaking

9580 Alan M. Taylor Foreign Capital in Latin America in the Nineteenth andTwentieth Centuries

9581 Baruch Lev The Measurement of Firm-Specific Organization Capital Suresh Radhakrishnan

9582 Raymond Fisman Financial Dependence and Growth RevisitedInessa Love

9583 Raymond Fisman Financial Development and the Composition of Industrial Inessa Love Growth

9584 Kevin Milligan Does Education Improve Citizenship? Evidene from the U.S.Enrico Moretti and the U.K.Philip Oreopoulos

9585 John Kennan The Effect of Expected Income on Individual MigrationJames R. Walker Decisions

9586 Michael Hurd The Retirement-Consumption Puzzle: Anticipated and ActualSusann Rohwedder Decline in Spending at Retirement

9587 Andrew Leigh What Do Financial Markets Think of War in Iraq?Justin WolfersEric Zitzewitz

58. NBER Reporter Spring 2003

Paper Author(s) Title

9588 Thomas S. Dee Are There Civic Returns to Education?

9589 Takatoshi Ito Market Evaluations of Banking Fragility in Japan: JapanKimie Harada Premium, Stock Prices, and Credit Derivatives

9590 Eileen Appelbaum Union Participation in Strategic Decisions of CorporationsLarry W. Hunter

9591 Joni Hersch A Workers’ Lobby to Provide Portable Benefits

9592 Helen Owens Rail Reform Strategies: The Australian Experience

9593 Dora L. Costa Race and Pregnancy Outcomes in the Twentieth Century:A Long-Term Comparison

9594 Sukkoo Kim Historical Perspectives on U.S. Economic Geography Robert A. Margo

9595 Luigi Zingales Banks and Markets: The Changing Character of EuropeanRaghuram Rajan Finance

9596 Louis Kaplow Transition Policy: A Conceptual Framework

9597 William A. Brock The Kindergarten Rule of Sustainable GrowthM. Scott Taylor

9598 Tomas Philipson Intellectual Property and External Consumption Effects:Stephane Mechoulan Generalizations from Pharmaceutical Markets

9599 Ricardo Caballero Inflation Targeting and Sudden StopsArvind Krishnamurthy

9600 Philip L.Williams Antitrust Merger Policy: Lessons from the AustralianGraeme Woodbridge Experience

9601 Stephen Coate Power-hungry Candidates, Policy Favors, and Pareto ImprovingCampaign Finance Policy

9602 Alma Cohen The Effect of Automobile Insurance and Accident Liability Rajeev Dehejia Laws in Traffic Fatalities

9603 Jeffrey A. Groen In-State versus Out-of-State Students: The Divergence ofMichelle J. White Interest between Public Universities and State Governments

9604 Rodolfo E. Manuelli Frictionless Technology Diffusion: The Case of TractorsAnanth Seshadri

9605 Martin Lettau Expected Returns and Expected Dividend GrowthSydney C. Ludvigson

9606 Assaf Razin A Brazilian-Type Debt Crisis: Simple AnalyticsEfraim Sadka

9607 Emmanuel Saez The Evolution of High Incomes in Canada, 1920-2000 Michael R. Veall

9608 Simeon Djankov The New Comparative EconomicsEdward L. GlaeserRafael La PortaFlorencio Lopez-de-SilanesAndrei Shleifer

9609 Roberto Rigobon The Effects of War Risk on U.S. Financial MarketsBrian Sack

9610 Nancy E. Reichman Effects of Child Health on Parents’ Relationship Status Hope CormanKelly Noonan

NBER Reporter Spring 2003 59.

Paper Author(s) Title

9611 Yacine Ait-Sahalia How Often to Sample a Continuous-Time Process in the Per A. Mykland Presence of Market Microstructure Noise

9612 Alan L. Olmstead Hog Round Marketing, Seed Quality, and Government Policy:Paul W. Rhode Institutional Change in U.S. Cotton Production, 1920-1960

9613 Bengt Holmstrom The State of U.S. Corporate Governance: What’s Right andWhat’s Wrong?

9614 Ross Levine Migration, Spillovers, and Trade Diversion: The Impact ofSergio L. Schmukler Internalization on Stock Market Liquidity

9615 Patricia M. Danzon Productivity in Biotechnology R&D: The Role of ExperienceSean Nicholson and AlliancesNuno Sousa Pereira

9616 Douglas A. Irwin New Estimates of the Average Tariff of the United States,1790-1820

9617 Haijime Katayama Why Plant-Level Productivity Studies are Often Misleading,Shihua Lu and an Alternative Approach to InterferenceJames Tybout

9618 Joshua Aizenman Military Expenditure, Threats, and GrowthReuven Glick

9619 Justin Wolfers Is Business Cycle Volatility Costly? Evidence from Surveys ofSubjective Wellbeing

9620 Thorsten Beck Bank Supervision and Corporate FinanceAsli Demirguc-KuntRoss Levine

9621 Tetsushi Sonobe Productivity Effects of TVE Privatization: The Case Study ofKeijiro Otsuka Garment and Metal Casting Enterprises in the Greater

Yangtze River Region

9622 Louis Kaplow Fairness Versus Welfare: Notes on the Pareto Principle,Steven Shavell Preferences, and Distributive Justice

TECHNICAL WORKING PAPERS

288 William A. Brock Multinomial Choice with Social InteractionsSteven N. Durlauf

289 Christopher R. Bollinger Iatrogenic Specification Error: A Cautionary Tale of CleaningAmitabh Chandra Data

290 Kent Smetters The (Interesting) Dynamic Properties of the Neoclassical Growth Model with CES Production

291 Viliam Druska Generalized Moments Estimation for Panel DataWilliam C. Horrace

292 Nelson C. Mark Dynamic Seemingly Unrelated Cointegrating Regression Masao OgakiDonggyu Sul

NA

TIO

NA

L B

UR

EA

U O

F E

CO

NO

MIC

RE

SE

AR

CH

NB

ER

Reporter

1050 M

assa

ch

use

tts Ave

nu

e

Cam

brid

ge,

Massa

ch

use

tts 0213

8-5

398

(617

) 868-3

900

Ch

an

ge S

erv

ice R

eq

ueste

d

Nonprofit O

rg.U

.S. P

ostageP

AID

National B

ureau ofE

conomic R

esearch