A Service of

Leibniz-Informationszentrum econstor Wirtschaft Leibniz Information Centre Make Your Publications Visible. zbw for Economics

National Bureau of Economic Research (NBER) (Ed.)

Periodical Part NBER Reporter Online, Volume 2006

NBER Reporter Online

Provided in Cooperation with: National Bureau of Economic Research (NBER), Cambridge, Mass.

Suggested Citation: National Bureau of Economic Research (NBER) (Ed.) (2006) : NBER Reporter Online, Volume 2006, NBER Reporter Online, National Bureau of Economic Research (NBER), Cambridge, MA

This Version is available at: http://hdl.handle.net/10419/61989

Standard-Nutzungsbedingungen: Terms of use:

Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Documents in EconStor may be saved and copied for your Zwecken und zum Privatgebrauch gespeichert und kopiert werden. personal and scholarly purposes.

Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle You are not to copy documents for public or commercial Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich purposes, to exhibit the documents publicly, to make them machen, vertreiben oder anderweitig nutzen. publicly available on the internet, or to distribute or otherwise use the documents in public. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, If the documents have been made available under an Open gelten abweichend von diesen Nutzungsbedingungen die in der dort Content Licence (especially Creative Commons Licences), you genannten Lizenz gewährten Nutzungsrechte. may exercise further usage rights as specified in the indicated licence. www.econstor.eu NBER Reporter NATIONAL BUREAU OF ECONOMIC RESEARCH

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

Program Report

IN THIS ISSUE Program Report: Industrial Organization  Industrial Organization Research Summaries: Europe 8 Consumer Demand for Health Insurance 0 Nancy L. Rose* International…Production and Distribution 3 Historical…U.S. Trade Policy 6

NBER Profile 9 Conferences 20 Bureau News 31 The NBER’s Program on Industrial Organization (IO) celebrates its Current Working Papers 66 fifteenth anniversary this year. Researchers in the IO program explore a wide range of topics within the field. Rather than attempting to skim the full scope of program activity, this report highlights work in three broad areas: regulation and antitrust policy; pricing behavior by firms; and auc- tions markets.1 Discussion of the substantial body of research on technol- ogy and technical change is deferred to reports of the Productivity Program and the NBER Project on Industrial Technology and Productivity. Those interested in learning more about the IO program may visit the NBER website for links to the full set of Industrial Organization Working Papers: http://www.nber.org/programs/io/io.html Regulatory and Antitrust Policy

NBER Service Brings When markets deviate from competitive ideals, assessing the desirabil- You New Data for Free ity of government intervention requires a careful assessment of the costs of market failures relative to the benefits of imperfect regulation. The rec- A new, free NBER email ser- ognition that even imperfect markets may be preferable to regulated out- vice gives you daily email links to comes accompanied a dramatic transformation in the nature and extent all U.S. government data releases, of government intervention across a broad range of markets over the past including unemployment, trade, thirty years. Many industries long subject to price and entry regulation in interest rates, GDP, etc. We keep track of your preferences and the United States — among them airlines, trucking, railroads, and bank- email you the requested links ing — were deregulated. Telecommunications and electric utilities have when they are released. To sign been vertically disintegrated and structurally competitive segments were up for any or all of the govern- opened to market-based outcomes. Privatization of state-owned enter- ment releases, visit www.nber.org/ releases and register your choices. * Rose directs the NBER’s Program on Industrial Organization and is a Professor of Economics at MIT. In this article, the numbers in parentheses refer to NBER Working Papers. NBER Reporter Summer 2006  prises outside the United States has substan- tially increased reliance on market outcomes in NBER Reporter many sectors, although regulators in some cases have replaced government managers in provid- ing oversight. Where government intervention has been maintained, various forms of incen- tive-based regulation increasingly have replaced The National Bureau of Economic Research is a private, nonprofit research orga- state ownership or traditional cost-of-service nization founded in 1920 and devoted to objective quantitative analysis of the rate determination. American economy. Its officers and board of directors are: IO program members are among the lead- President and Chief Executive Officer — Martin Feldstein ing scholars of antitrust and regulatory policy, Vice President for Administration and Budget — Susan Colligan and many have been directly involved in the Controller — Kelly Horak design or implementation of reforms through BOARD OF DIRECTORS their government service, advice to regulatory Chairman — Elizabeth E. Bailey agencies, or consulting to affected firms. In the Vice Chairman — John S. Clarkeson face of continuing policy debates over regula- Treasurer — Robert Mednick tory reform, highlighted more than a decade ago DIRECTORS AT LARGE by Paul Joskow and Roger Noll in the NBER’s Peter Aldrich Jessica P. Einhorn John Lipsky 1994 American Economic Policy in the 1980s, Elizabeth E. Bailey Martin Feldstein Laurence H. Meyer the NBER recently sponsored a research proj- John Herron Biggs Jacob A. Frenkel Michael H. Moskow Andrew Brimmer Judith M. Gueron ect designed to leverage this expertise. Project John S. Clarkeson Robert S. Hamada Rudolph A. Oswald participants were asked to identify key issues Don R. Conlan George Hatsopoulos Robert T. Parry in economic regulation, assess the impact of Kathleen B. Cooper Karen N. Horn Marina v. N. Whitman George Eads Judy Lewent Martin B. Zimmerman regulatory reforms across a variety of indus- tries, and evaluate significant contemporary DIRECTORS BY UNIVERSITY APPOINTMENT concerns about these reforms. Two dozen schol- ars assembled for a September 2005 conference George Akerlof, California, Berkeley Joel Mokyr, Northwestern Jagdish W. Bhagwati, Columbia Andrew Postlewaite, Pennsylvania in Cambridge to discuss the results of this proj- Michael J. Brennan, California, Los Angeles Craig Swan, Minnesota ect, to be assembled in an NBER volume on Glen G. Cain, Wisconsin Uwe Reinhardt, Princeton “Economic Regulation and Its Reform: What Ray C. Fair, Yale Nathan Rosenberg, Stanford Franklin Fisher, MIT David B. Yoffie, Harvard Have We Learned?” This project complements Saul H. Hymans, Michigan Arnold Zellner, Chicago a substantial body of primary research by NBER Marjorie B. McElroy, Duke associates on regulatory and antitrust policy. A DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS selection of research from the conference and from NBER working papers is described below. Richard B. Berner, National Association for Business Economics Gail Fosler, The Conference Board Richard C. Green, American Finance Association Economic Regulation Dr. Arthur Kennickell, American Statistical Association and Its Reform Thea Lee, American Federation of Labor and Congress of Industrial Organizations Robert Mednick, American Institute of Certified Public Accountants Electricity Restructuring: Competition and Angelo Melino, Canadian Economics Association Incentive Regulation Jeffrey M. Perloff, American Agricultural Economics Association John J. Siegfried, American Economic Association William W. Lewis, Committee for Economic Development NBER researchers continue in the van- Gavin Wright, Economic History Association guard of research, market design, and imple- The NBER depends on funding from individuals, corporations, and private foun- mentation of electricity restructuring. Much dations to maintain its independence and its flexibility in choosing its research of the empirical work to date has focused on activities. Inquiries concerning contributions may be addressed to Martin Feldstein, President & CEO, NBER 1050 Massachusetts Avenue, Cambridge, restructured generation markets, in which prices MA 02138-5398. All contributions to the NBER are tax deductible. generally are determined through a competi- tive bidding process. Frank Wolak2 describes The Reporter is issued for informational purposes and has not been reviewed by the Board of Directors of the NBER. It is not copyrighted and can be freely repro- the evolutionary nature of the restructuring duced with appropriate attribution of source. Please provide the NBER’s Public process, emphasizing the tension between an Information Department with copies of anything reproduced. imperfectly competitive market and an imper- Requests for subscriptions, changes of address, and cancellations should be sent fect regulatory process in providing incentives to Reporter, National Bureau of Economic Research, Inc., 1050 Massachusetts for least-cost supply at various stages of the Avenue, Cambridge, MA 02138-5398. Please include the current mailing label. production process. In one of the first empiri-

2 NBER Reporter Summer 2006 cal analyses of restructuring supply-side wholesale electricity prices. Borenstein formation over the past quarter century. benefits (11001), the potential for these (11594) provides insight into contin- Although telecommunications regulators incentives to reduce costs is highlighted: ued resistance to real-time pricing, high- adopted various incentive-based policies Kira Fabrizio, Catherine Wolfram, and I lighting substantial distributional effects early, “forward-looking” cost-based regu- show that restructuring is associated with of real-time prices across heterogeneous lation still plays a prominent role in set- increased productivity, documenting gen- industrial and commercial customers that ting prices for unbundled network ele- erating-plant efficiency gains in the use of may make it difficult to gain political sup- ments (UNEs) that must be leased by labor and materials input from replacing port without some system to compensa- local telephone companies to their com- a regulated monopoly with market com- tion losers. petitors. As noted by Robert Pindyck petition. As Wolak points out, though, For services such as transmission and (10287, 11225), the typical pricing for- the technical characteristics of electricity distribution, which typically remain sub- mulas used to set UNE lease rates induce supply and demand suggest that market ject to regulation even in restructured substantial investment and entry inef- power may be of particular concern, lim- markets, innovations have shifted the ficiencies by failing to account properly iting the benefits of restructuring. Joskow focus from cost-based price setting toward for the substantial sunk costs of tele- (8442) discusses the role of market power incentive mechanisms. Joskow3 provides comm investments.4 Jerry Hausman and and other contributors to the 2000–1 a comprehensive review of the theory and Gregory Sidak5 compare the outcomes California electricity crisis; Ali Hortacsu complexities involved in applying incen- of regulatory approaches in the United and Steven Puller (11123) measure effi- tive-based regulation. He then discusses States, the United Kingdom, and New ciency losses from strategic bidding in the applications of incentive mechanisms to Zealand. They conclude that in both the Texas ERCOT market; and Dae-Wook the regulation of prices and service qual- United States and the United Kingdom, Kim and Chris Knittel (10895) com- ity for “unbundled” electricity transmis- unbundling may have caused an increase pare direct measures of markups to those sion and distribution networks. Further, in competition if one measures com- inferred from oligopoly models of market he assesses the evidence on the perfor- petition by market share of entrants, at power in California generation markets. mance of incentive regulation for electric the cost of adverse investment effects by Wolak also describes market design and distribution and transmission networks both incumbents and new entrants. In regulatory policies that limit the ability and describes challenges for future policy the last section of their paper, they argue of suppliers to exercise unilateral mar- and research. that emerging facilities-based competi- ket power — such as forward contract- Among those challenges are deter- tion should allow the end of telecomm ing, horizontal divestitures, demand-side mining the role of competition in elec- price regulation and the regulatory bur- participation, and local market power tricity retailing and transmission. Joskow den that it creates for both consumers mitigation — and uses examples from and Jean Tirole (9534) analyze the likely and the economy. If the nature of local worldwide wholesale electricity markets performance of competitive merchant exchange competition during the 1990s to illustrate the importance of effectively transmission markets, and conclude that is a guide to the future, then Shane addressing each aspect of the market this model is likely to yield substan- Greenstein and Michael Mazzeo’s (9761) design process to ensure the maximum tial investment inefficiencies. While their research suggests that we may see increas- benefits of electricity restructuring. prognosis for retail electricity competi- ing product differentiation as a result of While early empirical electricity tion is more optimistic (10473), they note local competition. research focused predominantly on gen- a variety of challenges and efficiency limi- Cable Television eration markets, researchers increasingly tations of competitive outcomes. In their have turned their attention to retail mar- work on “Reliability and Competitive Greg Crawford’s 6 analysis of the kets and demand-side policy. Peter Reiss Electricity Markets” (10472), they ana- cable television industry highlights the and Matthew White (8687, 9986) use lyze the complexity involved in inte- impact of economic regulation on prod- data from San Diego households to mea- grating economists’ approach to market uct quality and innovation. Regulation in sure consumer responsiveness to chang- design with engineering system design this industry has varied greatly over time, ing electricity prices and conservation for reliability across the entire electric- as federal legislation has deregulated, re- programs enacted during the California ity network. Finally, they highlight the regulated, and deregulated consumer electricity crisis. They argue that consum- implications for system investment, cable prices. More recently, penetration ers may be more responsive to price fluc- operation, and reliability of interactions by Digital Broadcast Satellites raises tuations than previously thought. Severin among competitive markets, operational questions about the need for regulation Borenstein and Stephen Holland (9922) constraints, and regulatory and adminis- to constrain cable prices (see research by suggest that substantial efficiency gains trative practices. Austan Goolsbee and Amil Petrin on wel- could be obtained from shifting even Telecommunications fare gains from DBS introduction, 8317). modest shares of relatively price-insensi- Crawford analyzes the interplay of price tive customers from fixed retail electricity The telecommunications sector sim- regulation and firm quality choices, with prices to those that reflect time-varying ilarly has undergone a dramatic trans- attention to the implications of satellite

NBER Reporter Summer 2006 3 competition for performance in cable fic control, and security infrastructure, tially reduced the costs of regulatory television markets. His work highlights which have not in general kept pace with distortions, banking efficiency improved ongoing concern over horizontal con- the growth and changes in the industry. following the removal of most price and centration and vertical integration in the Pharmaceutical Regulation entry controls, generating substantial real programming market, and bundling by benefits for the economy as a whole. both cable systems and programmers, the Pharmaceutical regulation has long Patrick Bolton et al. (10571) show that latter being the subject of current policy generated concern over its effect on inno- opening the banking sector to price and debate at the Federal Communications vation incentives and product launch product-offering competition also may Commission. delays. Patricia Danzon and Eric Keuffel8 improve information provision and con- Airline Deregulation tackle these and other issues in their sumer-product matching, given the supe- analysis of pharmaceutical safety, price, rior information that financial services In general, the empirical evidence and marketing regulations on a variety sellers may have about product suitability on deregulation of structurally competi- of industry performance measures. They for buyers of those services. tive industries suggests considerable gains note that regulatory reforms such as the Eric Zitzewitz10 analyzes the impli- from removal of price and entry regula- adoption of user fees, fast track, and pri- cations of such asymmetric information tion, although the transition from reg- ority review may have reduced review- for the regulation of non-banking finan- ulated to competitive markets may be induced delays, especially for priority cial services firms. He argues that agency longer and more costly than academics drugs. For example, Ernst Berndt et al. conflicts created by information asymme- or policymakers originally envisioned. (10822) finds that implementation of tries and consumer behavioral biases may Borenstein and I 7 describe the significant performance goals and user fees for FDA impede market efficiency. For example, consumer benefits from reduced fares drug applications substantially reduced asset management and financial advisor and increased flight frequencies and from approval lags — by an average of roughly firms may have incentives to discriminate nonstop service subsequent to airline six months. They estimate a net savings of according to customer sophistication or deregulation, while acknowledging the more than 125,000 life-years from these search ability, offering low-price, high- industry’s considerable financial volatil- reforms. quality products to sophisticated clients ity. We argue that market power concerns Discouragement of innovation also and high-price, low-quality products to have diminished as growth by low-cost can be a significant hidden cost of price the less sophisticated. Ali Hortacsu and carriers now challenges legacy airlines in regulation. Price controls present in Syverson (9728) provide some evidence virtually all parts of the country. Recent many countries may reduce the price of of this phenomenon in research on prod- research by Goolsbee and Chad Syverson existing pharmaceuticals, but also appear uct differentiation and search costs in the suggests that even the threat of entry by to discourage the development and dif- mutual fund industry. Zitzewitz too dis- carriers such as Southwest may reduce fusion of innovative new treatments cusses the implications of these factors incumbent prices (11072). This surge (see analyses by Danzon, Richard Wang, for regulatory and antitrust policy in this in competition, combined with adverse and Liang Wang (9874); Danzon and sector, with particular attention to recent demand shocks, high fuel prices, and Jonathan Ketcham (10007); and the late interventions by the Attorney high labor costs, has contributed to cur- Jean Lanjouw (11321)). In their work on General and the SEC. rent financial distress among many legacy Medicaid prescription drug purchasing, Antitrust Policy airlines, though. Financial distress and Mark Duggan and Fiona Scott Morton accompanying bankruptcies have been (10930) highlight another indirect effect In regulated industries, firms may costly for shareholders, high-wage work- of price regulation: government rules that be subject to overlapping jurisdiction by ers, and the Pension Benefit Guaranty base Medicaid purchase price on the aver- both regulators and antitrust authori- Corporation, although many costs and age price of that drug across private-sec- ties. Dennis Carlton and Randal Picker11 dislocations may be transitional. For tor purchasers increase equilibrium drug analyze the tension that this produces, example, Borenstein and Rose (9636) prices for non-Medicaid purchasers, and describing the historical origin of antitrust show that schedule disruptions associ- increase a firm’s incentives to introduce and regulation policy and the ongoing ated with airline bankruptcies are largely new versions of a drug at higher prices. struggles to define the appropriate mech- transitory; where they are more perma- Financial Services anism and substantive scope for regulat- nent, they appear to be modest relative ing competition. They note that debates to background fluctuations in flights and Randall Kroszner and Phillip over the role of antitrust and regulation destinations served, and to be isolated to Strahan9 analyze the evolution of bank- continue with particular prominence in medium-sized airports. Overall airline ing regulation of prices (interest rates), today’s network industries, whether tele- investment and consumer benefits con- entry, capital, and investment decisions communications, transportation, or elec- tinue to be substantial. The greater long- from the 1930s to the last part of the tricity. Moreover, core issues such as inter- run challenge may be the performance of twentieth century. They note that while connection and mandatory access have government-controlled airports, air traf- industry adaptations to constraints par- increased in salience as reform-induced

 NBER Reporter Summer 2006 restructuring has led to vertical disinte- to assume that consumers are perfectly com, and conclude that consumers are gration of firms and increased compe- informed about each firm’s single price, quite sensitive to prices, particularly at tition with incumbents in many indus- many markets deviate substantially from Barnes & Noble.com. Glenn Ellison and try segments, while the Supreme Court’s this description. The ability to price dis- Sara Fisher Ellison (10570) show that decision in Trinko leaves open substantial criminate — charging different prices for Internet retailers respond strategically to questions about how these relationships a product either across firms or to differ- the increased pricing pressure imposed will be governed. ent customers of the same firm — may by these price comparison sites, though. For most sectors of the economy, enable firms to extract greater consumer Their analysis of the online computer interactions among firms are governed value from a transaction or to expand components market demonstrates that by court interpretations of antitrust pol- the set of customers they serve. This retailers engage in a variety of practices icy rather than by economic regulatory phenomenon appears ubiquitous in the to mitigate the extreme price-sensitivity agency decisions. NBER researchers have economy. that price comparison sites may induce, explored a variety of aspects of antitrust Consumer search costs may be an allowing firms to mark-up prices at least policy, from theoretical and empirical important source of sustained price vari- enough to cover fixed as well as marginal analyses of merger policy to consider- ability in a market. For example, cash costs for efficient retailers. Glenn Ellison ation of vertical restraints. The appropri- prices for a given prescription drug vary (9721) provides a theoretical model of ate role and application of antitrust policy widely across pharmacies within a par- one such practice, “add-on pricing” — for in innovative sectors has attracted partic- ticular geographic market. The magni- example, advertising low prices for one ular attention, both as a matter of prin- tude of price dispersion is correlated good in the expectation of selling addi- ciple and in the context of high-profile with attributes that appear related to the tional (or higher quality) products to cases such as U.S. v. Microsoft.12 In these costs and benefits of consumer search, consumers at a high price at the point of sectors, the tension between encouraging for example, greater price variation for sale. His work shows that this practice competition through entry and maintain- drugs prescribed for one-time use to treat can sustain softer price competition as a ing profit incentives for dynamic growth an acute condition, relative to those pre- competitive equilibrium. and efficiency is particularly acute. Ilya scribed for ongoing purchase as mainte- A rich body of research by Florian Segal and (11525) nance therapies. Alan Sorensen (8548) Zettelmeyer, Fiona Scott Morton, and focus on this tension in research that ana- uses information on retail price disper- Jorge Silva-Risso uses data on individ- lyzes a number of specific policies, high- sion and prescription attributes to ana- ual consumer automobile purchases to lighting those that benefit both entry and lyze the distribution of consumer search explore the interaction of retail auto pric- innovation. Carlton and Robert Gertner costs for these products. His estimated ing, consumer information, and Internet (8976) argue that dynamic efficiency model of consumer pharmacy choice sug- information and referral services. These requires coordination of antitrust pol- gests that search intensities in this mar- researchers first document significant icy with intellectual property laws in an ket are relatively low — implying that, on reductions in average automobile pur- attempt to resolve tensions created by average, only 5 to 10 percent of prescrip- chase price associated with using an the tendency for network industries to tions are comparison-price shopped. Internet referral service (8667), on the evolve toward closed systems. Michael One way to increase search intensity order of 2.2 percent after controlling for Katz and Howard Shelanski (10710) is to lower its cost. Internet retailing often selection effects in who uses the service. argue that traditional merger analysis, has been cited as intensely price compet- They then show that Internet referrals based on static welfare analyses, may miss itive in large part because of easy con- disproportionately benefit minority buy- important dynamic efficiency implica- sumer search for low-price vendors. Joel ers, offsetting the average 2 percent price tions of mergers in highly innovative sec- Waldfogel and Lu Chen (9942) argue disadvantage these groups incur because tors. Carlton (11645) emphasizes the that this reduces the price advantage of of their personal costs of search or nego- importance of dynamic barrier-to-entry brand-name retailers. They find that con- tiation for purchases made through analysis as one component of this.13 sumer exposure to price comparison sites traditional dealer channels (8668). In such as DealTime.com reduces Amazon research that matches consumer sur- Pricing Behavior in Oligopoly purchase shares for those consumers, and vey and transactions data to explore the Markets that reductions are roughly twice as large mechanism underlying these price effects, for sites that include retailer reliabil- these researchers conclude that increased The behavior of firms in oligopoly ity information (which may substitute transparency of dealer invoice costs com- markets is one of the mainstays of IO for retailer brand reputation) in addi- bine with greater negotiating clout of the research. NBER researchers have made tion to item price. Goolsbee and Judith online referral service to reduce a cus- considerable progress in better under- Chevalier (9085) develop a method of tomer’s price by an average of 1.5 percent standing firms’ pricing decisions, par- estimating price elasticities for online (11515). Consumer information appears ticularly with reference to price disper- booksellers using publicly available to be particularly important in extracting sion. Although competitive models tend data on Amazon and Barnes & Noble. value from auto price negotiations. For

NBER Reporter Summer 2006 5 example, Megan Busse, Zettelmeyer, and and renegotiation costs to procurement is also to understand the performance of Silva-Risso find that purchasers obtain contracts awarded by auction mecha- these markets. Mireia Jofre-Bonet and 80 percent of the value of auto manufac- nisms (12051). Martin Pesendorfer (8626) develop a turer promotions in the form of heavily As Susan Athey and Philip Haile method of estimating a dynamic model of promoted customer rebates, but only 35 point out (12126, p. 1), “auctions have behavior in repeated highway construc- percent of the value of promotions that provided a fruitful area for combining tion procurement auctions with firm- are paid as dealer discounts (10887). This economic theory with econometric anal- level capacity constraints, and then quan- research agenda has provided unparal- ysis … to understand behavior and inform tify efficiency losses that result in this leled insights into pricing determinants policy.” Athey and Haile describe meth- setting. In many repeated auctions set- in a significant consumer market, and odological innovations, many by NBER tings, the potential for collusion among generated important new findings for the researchers, which have facilitated esti- bidders may also be a significant con- role of the Internet in changing outcomes mation of more realistic models and pro- cern. Ken Hendricks, Rob Porter, and in conventional retail channels. vided significant insights into auction Guofu Tan (9836) develop a theory of market operation and performance. A collusion in affiliated private value and Auction Markets significant thrust of this work has been common value auction environments, to allow the data more freedom to drive and use their model to test for bidding While Internet retailing in general results by relaxing parametric and func- rings in federal offshore oil and gas lease has attracted considerable attention and tional form assumptions. For example, auctions. They show that the winner’s interest, the icon of Internet selling may Haile, Han Hong, and Matthew Shum curse in common value settings works well be eBay. Its popularity as a mecha- (10105) develop nonparametric tests of against bid rigging for marginal tracts. nism for matching buyers and sellers has one of the key valuation questions in auc- Bajari and Fox (11671) analyze poten- spawned a rich economics literature as tions: are bidders’ valuations generated tial collusion in FCC spectrum auctions; well as numerous competitors; much of by independent private values for the and Kathryn Graddy this is described in Patrick Bajari and Ali good, in which case bidders need not be (10795) provide a case study of price- Hortacsu’s survey of Internet auctions concerned about the “winner’s curse,” or fixing in auctions using the Sotheby’s/ research (10076). The seeming ubiquity by common values, in which case bidders Christie’s art auctions case, drawing out of auctions, for goods ranging from fine must optimally shade their bids knowing the lessons for auctions and competition art to Beanie Babies, and in settings that that winning means they had an exces- policy from details of this case. range from government procurement to sively optimistic estimate of the good’s pollution permits,14 has prompted sev- true value. They apply this test to differ- Conclusion eral NBER researchers to model the ben- ent types of U.S. Forest Service timber efits of auctions over alternative mar- auctions, and find support for its ability This report of necessity focuses on ket transaction mechanisms. Alexandre to distinguish between settings in which a fraction of the IO research conducted Ziegler and Edward Lazear (9795) ana- common values are likely to be more or by NBER scholars, although I hope it lyze the choice between retail store-based less significant. provides an indication of the breadth and auction markets. They describe the Another approach to allaying con- and depth of contributions made in this relative benefits of each, and character- cerns about constraints imposed by struc- area. Interested readers are encouraged ize the conditions that lead to more effi- tural model estimates of auctions looks to peruse the NBER website to access cient market organization through retail to experimental data. In this spirit, Bajari the entire body of scholarly work in this stores relative to auctions. Eduardo Engel, and Hortacsu (9889) use experimental area. Ronald Fischer, and Alexander Galetovic data to calibrate the quality of struc- (8869) analyze Demsetzian auctions for tural estimation based on four alternative 1 NBER researchers also have been exclusive rights in settings that range theoretical models of bidder behavior. responsible for an extensive body of work from procurement to royalty contracts, Andreas Lange, John List, and Michael on methodological advances in empiri- and conclude that “competition for the Price (10639) develop an innovative cal industrial organization. Ariel Pakes field” through ex ante auctions welfare combination of field data and lab experi- (10154) provides an overview. Areas dominates duopoly competition when- mental data to evaluate the impact of sec- that have attracted considerable atten- ever marginal revenue is decreasing in ondary resale markets for timber on bid- tion include hedonic modeling: see, quantities sold. Bajari, Robert McMillan, ding behavior in timber auctions. for example, papers by Pakes (8715), and Steve Tadelis (9757) highlight lim- Improving the models and meth- , Rosa Matzkin, and itations of auctions relative to negoti- ods available to analyze auction markets Lars Nesheim (9895), Lanier Benkard ations in procurement settings, partic- can yield important economic insights and Patrick Bajari (9980, 10278). ularly those dominated by incomplete into these markets, and can aid partici- Considerable work also has focused on information. With Stephanie Houghton, pants in developing appropriate bidding estimation of dynamic games: for example, Bajari and Tadelis estimate adaptation strategies. But an important policy goal papers by Igal Hendel and Aviv Nevo

 NBER Reporter Summer 2006 (9048, 11307), Jaap Abring and Jeffrey Its Reform: What Have We Learned? 2005 Campbell (9712), Martin Pesendorfer 2005. 11 D. W. Carlton and R. Picker, and Philipp Schmidt-Dengler (9726), 6 G. Crawford, “Cable Television: “Antitrust and Regulation,” NBER Bajari, Benkard, and Jon Levin (10450), Does Cable Need to be Regulated Any Conference on Economic Regulation and Pakes, Michael Ostrovsky, and Steven More?” NBER Conference on Economic Its Reform: What Have We Learned? Berry (10506), Adam Copeland and Regulation and Its Reform: What Have 2005. George Hall (11870), Gabriel Weintraub, We Learned? 2005. 12 David Evans, Albert Nichols, and Benkard, and Ben Van Roy (11900), and 7 S. Borenstein and N. L. Rose, Richard Schmalensee’s analysis of the Guillermo Caruana and Liran Einav “Regulatory Reform in the Airline Microsoft case (11727) argues that the (11958). Industry,” NBER Conference on remedies imposed struck a reasonable bal- 2 F. Wolak, “Regulating Competition Economic Regulation and Its Reform: ance among competing concerns. in Wholesale Electricity Supply,” NBER What Have We Learned? 2005. 13 See also antitrust analysis in more Conference on Economic Regulation and 8 Danzon, P.M. and E. Keuffel, general settings; for example, Charles Its Reform: What Have We Learned? “Regulation of the Pharmaceutical Calomiris and Thanavut Pornrojnangkool 2005. Industry,” NBER Conference on (11351) on anticompetitive effects of bank 3 P. L. Joskow, “Incentive Regulation Economic Regulation and Its Reform: mergers in lending markets, Evans and in Theory and Practice: Electricity What Have We Learned? 2005. Schmalensee (11603) on competition pol- Distribution and Transmission Networks,” 9 R. Kroszner and P. Strahan, icy in markets with two-sided platforms, NBER Conference on Economic “Regulation and Deregulation of the U.S. and Carlton and Michael Waldman Regulation and Its Reform: What Have Banking Industry: Causes, Consequences, (11407) on tying in durable goods mar- We Learned? 2005. and Implications for the Future,” NBER kets. 4 Pindyck also examines the implica- Conference on Economic Regulation and 14 See, for example, Orley Ashenfelter tions of sunk costs for competition policy Its Reform: What Have We Learned? and Kathryn Graddy (8997) on art auc- (11430). 2005. tions, Bajari and Jeremy Fox (11671) on 5 J. Hausman and J.G. Sidak, 10 E. Zitzewitz, “Financial Regulation FCC spectrum auctions, and Luis Cabral "Telecommunications Regulation: Current in the Aftermath of the Bubble,” NBER and Hortacsu (NBER WP 10363) on Approaches with the End in Sight,” NBER Conference on Economic Regulation and reputation mechanisms in e-Bay auctions. Conference on Economic Regulation and Its Reform: What Have We Learned?

*

NBER Reporter Summer 2006 7 Research Summaries

Europe

Alberto Alesina*

Per capita income in Continental United States. United States. Europe was slower to cap- Western Europe (in short, Europe) was There are three reasons why work ture the benefits of the technological rev- catching up with the United States from hours per person are lower in Europe: 1) olution in information technology (IT). the end of the Second World War until fewer people are in the labor force (early Part of the reason was lack of competition the mid-1980s; from 1950 to about 1975, retirements, lower worker participation, in the product and, especially, the ser- we speak of a European miracle. Then, delayed entry into the labor force, all in vice markets. The slow pace of deregula- something changed. The United States various combinations depending on the tion in these sectors created disincentives re-emerged from the difficult decades of country); 2) more vacation time for those to innovation and investment.1 An early the 1970s with a renewed political energy who do work; 3) lower hours in a “nor- deregulation in the1980s, restructuring that led to deregulation, increased compe- mal” work week. In different propor- of companies, and a new generation of tition, reduction of marginal tax rates, and tions for different countries, all three fac- aggressive and powerful CEOs in the restructuring of corporations, which later tors matter. For instance, in Italy the first United States created the fertile ground facilitated the immediate adoption of the factor, lower participation, is the driving on which the IT revolution could gener- innovations from the information revo- force. In France and Germany, all three ate an exceptional period of rapid growth. lution. Europe, instead, seemed “stuck:” factors explain about a third each of the In Europe, instead, incumbent firms con- incapable of gathering sufficient energy difference with the United States. tinued with the old practice of govern- to reform itself. This was especially the Europeans are working less and less ment protection from competition and case for the largest countries: Germany, for three reasons: first, increasing mar- government hidden subsidies. France, Italy, and Spain. ginal tax rates (especially from the 1960s The second major rigidity in Europe to the 1980s); second, a preference for is, of course, in the labor market. Firing What Happened? leisure; and, third, labor regulation and costs interfere with firms’ decisions, mak- union-imposed standards for work time, ing them cautious about hiring. Rather Let’s start with the basics. One of the including retirement regulations. Social than removing these impediments (and most remarkable facts about Europeans is multipliers compounds these effects: introducing well thought out unemploy- that they work much less than Americans. if a family member or friend has more ment insurance schemes), several coun- Europeans worked more than Americans time off, your own benefit from leisure tries, especially France, seem unable to in the 1950s and 1960s, when they were increases, creating more social demand reform. Europeans remain enamored of lowering their heads in the war recon- for leisure. When it becomes the social “job security” which often means “secu- struction efforts, first, and then during norm, because of regulations requiring six rity” for those insiders who have a job a period of boom. But then Europeans weeks of vacation, or retirement at age 60 and no work for those who are not “in”. began to work fewer and fewer hours per because the law imposes it, then it is diffi- Progress in this direction, achieved in capita. While in the early 1970s hours cult to change people’s minds about what Denmark and Sweden, both of which worked per person were about the same is “fair”. have vastly reduced firing costs, has been in Europe and in the United States, today Working less and maintaining reason- immediately followed by a significant the French, German, and Italians work able growth rates is possible if your produc- reduction in unemployment. about 1400 hours per person per year ver- tivity increases at a healthy pace. In fact, In summary, you can work less and sus about 1800 hours per person in the this has been the case for several decades less and not fall behind if you are more in Europe in the 1950s, 1960s, and later and more productive when you work. * Alesina is a Research Associate in the through the 1980s. So, Europeans man- But several countries in Europe may have NBER’s Programs on Public Economics, aged to keep up with the United States pushed this too far, and their policies Monetary Economics, Political Economy, and Economic Fluctuations and Growth. by working less but being more and more are not sufficiently productivity enhanc- He is also a professor of economics at productive. But in the 1990s, European ing. Perhaps Europeans have more or less . productivity growth fell below that of the consciously “chosen” to grow less than

 NBER Reporter Summer 2006 the United States and to fall behind eco- to “plan,” and a favorable view of gov- more “political argument” was put for- nomically and, as a consequence, politi- ernment dictated policies for achieving a ward in favor of the Euro: the common cally, but this does not seem to be the offi- multitude of social goals. The second is currency, by shutting down competitive cial rhetoric emanating from European the resistance of national governments to devaluations as a temporary “drug” for a capitals. truly open their markets, a resistance that national economy, would have facilitated has created obstacles to those European the adoption of structural reforms in the The Role of the EU leaders who indeed understood the ben- labor, goods, and service markets. Thus efits of competition. In European capitals, far it has not quite worked, certainly not Has the European Union been able often one hears a grand pro-European everywhere and with the necessary speed. to increase competition and move the rhetoric, but as soon as non-domestic (but Labor and goods market reforms, with continent to market-friendly growth- European) firms try to acquire domestic the exception of a few northern coun- enhancing reforms? The record is mixed. ones, the European rhetoric is immedi- tries, have been very slow to materialize On the one hand, some progress has been ately forgotten and nationalistic protec- and encounter great political opposition achieved in the area of trade, government tion resurfaces. of two types. One is that of “insiders” subsidies, and goods markets. But, on who fear losing their privileges. Think the other hand, Europe remains full of And the Euro? of incumbent protected firms, unionized impediments to free market competition workers with job security, service provid- across national boundaries, especially in Overall, the first six years of the Euro ers sheltered by international competi- the service sectors and when it comes to have confirmed the pros and cons that tion, banks protected by foreign acquisi- mergers and acquisition. 2 economists had identified regarding the tions, and so on. The second stems from Rather than concentrating on this construction of an average monetary the general European public, which by critical point of a common and truly con- union.5 The Euro has eliminated the pos- and large is “ideologically” averse to free tinental free market, EU institutions have sibility of competitive devaluations that, market thinking. In fact, Europe, espe- vested much energy in areas that are better in the end, cause and has vastly cially France, seems to be embracing more left to national governments. An excessive reduced the risk premise of government and more protectionist policies when it tendency to “coordinate” and “plan” in debt of highly indebted countries, such comes to foreign competition; and the Brussels has led to inconclusive rhetoric as Italy. It also has facilitated financial Anglo-American pro-market approach is and displaced efforts.3 One prime exam- market integration and possibly inter- viewed more and more suspiciously. These ple of this is the glorified “Lisbon process” regional trade and may have increased tendencies are ingrained in European cul- that sets detailed targets on socioeco- cross-national price competition. ture and reflect the history of ideas of this nomic variables, identical for all countries On the other hand, it has imposed continent.6 in Europe, and all to be reached within the same monetary policy on all country Europeans talk about a “European a certain time period. For example, the members. As it turns out, the economic model,” something that Germans refer Lisbon targets specify how many children cycles of countries in “Euro land” have to as a “social market economy,” and they of different age groups should be in public been less correlated than one might have contrast it to the Anglo-American free childcare facilities in 2010 in all European predicted; therefore, the common mon- market model. Unfortunately, much of Union countries, what the women’s par- etary policy does not indeed fit all coun- this thinking about a “European model” ticipation rate should be in the labor tries at all times. The European Central is fuzzy and ends up facilitating politi- force, and so on. Another example is the Bank has been repeatedly accused of cal compromises with privileged insid- “obsession” with the coordination of fis- checking European growth and com- ers. Europe does not have to adopt the cal policy imbedded in the details of the pared negatively to the pro-growth poli- American model; it certainly can have Growth and Stability Pact.4 The need cies of the Bank of Alan something distinct from it, say a system for tight coordination of fiscal policy in Greenspan. However, these criticisms of of efficient competitive markets coupled a monetary union has been vastly over- the European Central Bank are almost with extensive but efficient redistributive emphasized. Part of the problem is that totally misplaced. First, this institution is programs and social protection. Northern certain countries historically incapable of often a convenient scapegoat for govern- European countries are moving in this keeping the fiscal house in order (like Italy ments that do not have the will or courage direction, but the major European coun- and Greece) needed to be “constrained” to implement reforms. Second, these criti- tries are far from it. In these countries, the with some external commitments, but cisms come from the misplaced view that combination of regulation, protectionism, lately even France and Germany have vio- the European sluggish growth problem is high tax rates, and redistributive programs lated these budget rules. demand driven and that monetary policy end up creating unnecessary distortion Why did EU institutions make this could have helped in this regard. and often directing flows of resources not mistake, namely focusing on excessive Above and beyond the economic to the truly needy but to politically pow- coordination? There are two reasons: one arguments in favor or against a momen- erful categories. Measure of effectiveness is, in part, a typical European tendency tary union in Europe, an additional of welfare states in moving in the desired

NBER Reporter Summer 2006 9 direction for income flows from rich to Schucknecht, “What Does the European 6 A. Alesina and M. Angeletos, poor vary dramatically across European Union Do?” NBER Working Paper No. “Fairness and Distribution: U.S. versus countries; northern European countries 8647, December 2001, published in Europe,” NBER Working Paper No. 9502, do relatively well, Mediterranean coun- Public Choice, June 2005, pp. 275–319; February 2003, published in American tries are the worst. Lack of swift reforms A. Alesina and R. Wacziarg, “Is Europe Economic Review, September 2005, in many European countries does not Going Too Far?” NBER Working Paper pp. 913–35; A. Alesina, R. Di Tella, depend only on the inability of their lead- No. 6883, January 1999, published in and R. McCulloch, “Inequality and ers. Europeans themselves remain very Carnegie Rochester Conference Volume, Happiness: Are Europeans and Americans suspicious of market liberalization. An supplement of Journal of Monetary Different?” NBER Working Paper No. interesting case in this respect is Germany. Economics, December 1999, pp. 1–42. 8198, April 2001, published in Journal This country has received recently the 3 A. Alesina and R. Perotti, “The of Public Economics, May 2004, pp. “political shock” due to the assimilation European Union: A Politically Incorrect 837–931; A. Alesina, E. Glaeser, and of former East Germans. Evidence shows View,” NBER Working Paper No. 10342, B. Sacerdote, “Why Doesn’t the U.S. that their Communist experience has March 2004, published in Journal of Have a European-Style Welfare System?” accustomed them to extensive govern- Economic Perspectives, Winter 2004, pp. NBER Working Paper No. 8524, October ment intervention and, as a result, they 26–48. 2001, published in Brookings Paper on have moved the preference of the average 4 A. Alesina, I. Angeloni, and F. Etro, Economic Activity, Fall 2001, pp.178– German in this direction.7 Europe faces “International Unions,” published in 287; A. Alesina, S. Danninger, and M. great challenges in the near future. The American Economic Review, June 2005, Rostagno, “Redistribution through Public need for reforms is clear; the political will pp. 602–15. Employment: The Case of Italy,” NBER is lacking. 5 A. Alesina and R. Barro, “Currency Working Paper No. 7387, October 1999, Unions,” NBER Working Paper No. published in IMF Staff Papers, December 1 A. Alesina, S. Ardagna, G. Nicoletti, 7927, September 2000, published in 2001, pp.447–73. and F. Schiantarelli, “Regulation and Quarterly Journal of Economics, May 7 A. Alesina and N. Fuchs-Schuendeln, Investment,” NBER Working Paper No. 2002, pp. 409–30; A. Alesina, R. Barro, “Good Bye Lenin (or not?) — The Effect 9560, March 2003, published in the and S. Tenreyro, “Optimal Currency of Communism on People’s Preferences,” Journal of the European Economic Areas,” NBER Working Paper No. NBER Working Paper No. 11700, Association, June 2005, pp. 791–825. 9072, July 2002, published in NBER October 2005. 2 See A. Alesina, I. Angeloni, and L. Macroeconomic Annual, 2002.

Consumer Demand for Health Insurance

Thomas C. Buchmueller*

Since the early 1990s, prominent pro- of health plans on the basis of price and Although a “managed competition” posals for health insurance reform have quality. Proponents of these market-ori- model has yet to be adopted as national focused on increasing consumer choice ented plans argue that, in such a system, policy, many large employers organize and competition among integrated health consumers will sort themselves into lower their health benefits programs accord- plans. Under “managed competition” cost, higher quality plans; this pressure by ing to the same basic principles. Research models, consumers choose from a menu consumers will provide strong incentives on the behavior of employees and retir- to health plans and their affiliated provid- ees in these employer programs provides * Buchmueller is a Research Associate in ers to control costs and increase quality a useful laboratory for the role of price the NBER’s Programs in Health Care and Health Economics and a Professor of in order to compete for enrollment. The and quality in consumer health insurance Economic and Public Policy at the Paul Clinton administration’s Health Security decisions. Merage School of Business, University of Act and the “premium support” propos- One distinct problem for market-ori- California, Irvine. His profile appears later als for reforming Medicare are variants ented solutions to health insurance is that, in this issue. of the managed competition approach. when consumers are offered a choice of

10 NBER Reporter Summer 2006 health insurance options, the healthy (less fell by over 25 percent. This was at a time least price-sensitive. risky) consumers may sort themselves into when increased competition among man- The fact that younger, healthier con- certain plans and the more risky consum- aged care health plans was causing premi- sumers are more willing to switch health ers into others. Consequently, some plans ums to decline throughout the country, so plans in response to a change in prices will attract a disproportionate share of less these savings cannot be attributed entirely contributes to adverse selection against costly low risk consumers, while others to the adoption of a fixed dollar contri- plans that are favored by higher-risk con- will attractive older, sicker consumers who bution policy. However, the reduction in sumers. Adverse selection reduces effi- are more costly to insure. This “risk selec- premiums charged to Harvard and UC ciency by distorting the prices of com- tion,” in turn, is influenced in part by the were larger than those observed in the peting plans and, in the extreme, driving rules concerning how insurers are allowed general market, suggesting that the pric- certain options from the market. In both to vary premiums according to subscriber ing reforms enacted by each university did the UC and Harvard examples, this characteristics. State reforms that tight- result in a one-time savings. dynamic caused the most expensive plan ened these rules provide good case stud- In both cases, however, this also came available at each university to experience ies for understanding the relationship at the expense of the number of choices an “adverse selection death spiral” and to between pricing and risk selection. employees had. The most generous indem- be priced out of the market. Cutler and nity insurance — which covered care from Reber estimate that after two years, the The Effect of Premiums the doctor of your choice — was sub- efficiency loss from adverse selection in on Consumer Health Plan ject to an “adverse selection death spi- the Harvard program was roughly 2 per- Choices ral.” Faced with an initial increase in price cent of premiums. for this coverage, the healthiest dropped These and other similar studies3 tell a Two notable experiments in “man- out of indemnity insurance into lower consistent story about the price sensitivity aged competition” took place in the mid- cost plans. Those who remained in the of active employees. Proponents of man- 1990s: the University of California (UC) plan were, therefore, sicker on average. To aged competition Medicare reform pro- and Harvard University both offered a cover their costs, the price of the coverage posals point to these results and the expe- menu of plans that varied in generosity, was raised, which led to more dropouts riences of other large employers to explain but adopted a “fixed dollar contribution” until, after a few years, no one was covered how such reforms would work. However, policy. The plans also varied significantly by the indemnity plan. the extent to which these results general- in cost, so employees had a greater incen- One possibly important impediment ize to elderly adults in the Medicare pro- tive to consider price when selecting a to market based solutions is that consum- gram is unclear. In particular, the switch- health plan. Because out-of-pocket premi- ers often face substantial “switching costs” ing costs that we show to be important ums increased for some employees but not when they try to change their health for active employees are likely to be even for others, these changes provide a natural insurance plan. Under managed care, larger for elderly Medicare beneficiaries. experiment for estimating the impact of switching insurers often means having to In a recent paper, I use administrative price on employee health insurance deci- change providers, and even when that is data from a different employer-sponsored sions. Studies that I have conducted with not necessary, individuals may be reluc- health benefits program to estimate pre- colleagues at the University of California, tant to switch plans for fear of suffering mium elasticities for Medicare-eligible Irvine,1 and by David M. Cutler and an interruption of treatment. “Status quo retirees.4 This employer’s contribution to Sarah J. Reber,2 analyze the effect of these bias” in decisionmaking is another poten- retiree coverage depends on when a per- policy changes on employee plan choices, tial source of persistence. Combining the son retired and her years of service at total spending, and risk selection. UC enrollment files with hospital dis- that date. Therefore, two otherwise sim- The results for UC and Harvard are charge and cancer registry data, Bruce ilar individuals who retired at different strikingly similar. In both cases, employ- Strombom, Paul Feldstein, and I estimate points in time, or at the same time with ees were quite sensitive to price, and were separate premium elasticities for groups different years of service, face very dif- willing to switch plans to save as little as of employees whom we hypothesize to ferent out-of-pocket premiums for the $5 per month in out-of-pocket premiums. face different switching costs. Specifically, same menu of health insurance options. Cutler and Reber estimate a short-run we define 18 distinct groups based on Because this price variation is uncorre- premium elasticity of -2. In addition to age, job tenure, and health risk, where lated with the features of those options, this demand response, participating insur- “high risk” individuals were defined as or other retiree characteristics, it can be ers lowered their premiums in order to those who had recently been hospital- used to obtain unbiased estimates of elas- compete for enrollment. At Harvard, the ized or diagnosed with cancer. Consistent ticity. The results indicate a negative and combined effect of employees shifting to with the switching cost hypothesis, we statistically significant effect of price on lower cost plans and the premium reduc- find that young, low-risk employees who the choice of a health plan, albeit one that tions was a 10 percent reduction in total had recently joined the university were is slightly smaller than the results from spending in one year. Over a three-year the most price-sensitive; older, high-risk the literature on active employees. While period, total spending in the UC program employees with long job tenure were the this is consistent with the hypothesis that

NBER Reporter Summer 2006  older consumers are less price-sensitive is a major issue in the regulation of pri- nearly identical, suggesting that they are than younger ones, the price effects are vate insurance markets, particularly the driven by factors other than the reforms. large enough to suggest that if Medicare small group market (typically defined The fact that New York’s community went from a system of administered pric- as employer-sponsored groups of 50 or rating law did not immediately reduce the ing to competitive bidding, health plans fewer employees) and the individual (or number of people with insurance does would face strong incentives to compete non-group) market. In the early 1990s, not mean it had no effect. It may have on price. nearly every state enacted reforms that influenced the types of plans purchased From a policy perspective, the “near targeted insurers’ underwriting practices by consumers. In fact, consistent with the elderly” adults — that is, those between that were seen as discriminating against simplest Rothschild-Stiglitz model and the ages of 55 and 64 — constitute high-risk groups. The most extreme type with the results from the UC and Harvard another important population. Because of regulation, “pure community rating,” cases, we find that New York’s reforms led many firms have cut back on retiree health mandates that the same premium must be to a shift in enrollment away from tradi- benefits, early retirees in this age group charged for a given plan to all subscrib- tional indemnity plans to health mainte- are especially at risk of being uninsured. ers, regardless of age, gender, or any other nance organizations (HMOs). In a fol- Some recent policy proposals, such as risk characteristic. A main goal of these low-up study, Su Liu and I use national allowing individuals under age 65 to buy new regulations was to increase cover- data to test whether this result from New into Medicare, would address this prob- age, although critics argued that by rais- York generalizes to other states.8 We find lem directly. Other proposals, such as ing premiums for low risk groups the laws that HMO penetration among small tax credits for the purchase of non-group may have reduced coverage, inducing low- employer-sponsored groups increased insurance, do not explicitly target the risk consumers to drop it. more in states that enacted relatively “near elderly” but would be especially rel- Although the Harvard and UC expe- strong small group reforms than in states evant for this group. A key parameter for riences suggest that these “adverse selec- without such reforms. evaluating the cost of such proposals is the tion death spirals” could be important, price elasticity of take-up. the results don’t tell us how adverse selec- Recent Developments Sabina Ohri and I 5 estimate the effect tion affects the overall level of insurance and Future Research of out-of-pocket premiums on the deci- coverage in a market because the univer- sion by early retirees between the ages sities are closed systems. In both cases, Recent developments in public pro- of 55 and 64 to take up insurance cov- employees facing higher premiums simply grams and private health insurance mar- erage offered by their former employer. switched to other, less expensive plans. kets have resulted in significant changes The data are from the same employer- John DiNardo and I test for an effect in the options available to consumers and sponsored program I used to model the of underwriting regulations on insurance suggest fruitful areas for future research. health plan choices of Medicare-eligible coverage, focusing on reforms enacted by January 2006 marked the introduction retirees. We find a statistically significant, New York in 1993.7 New York’s reforms, of Medicare Part D, which provides pre- but small effect of price. The range of our which mandated pure community rat- scription drug coverage. Beneficiaries can elasticity estimates, from -0.10 to -0.16, is ing in both the small group and individ- obtain this coverage through the same consistent with other studies that use dif- ual insurance markets, were the stron- HMOs that were already available in the ferent types of data and different research gest, and hence most controversial, in program or through new stand-alone designs.6 the country. We compare trends in New drug plans offered by private insurers. York to those in two neighboring states: The resulting menu of options is quite Does Limiting Insurers’ Dis- Pennsylvania, which was one of a hand- different from what had been envisioned cretion Help Consumers in ful of states that enacted no reform, and by most managed competition advocates, Insurance Markets? Connecticut, which enacted moderate most notably in the large — some would reforms. We show that, while insurance say bewildering — set of options. Research One factor contributing to adverse coverage did fall in New York after the on the choices made in this new environ- selection in the UC and Harvard cases reforms took effect, coverage also was ment is important not only for evaluating is that, in each system, premium con- falling in Pennsylvania and Connecticut. the prescription drug benefit, but also as tributions faced by employees and pre- One important prediction of the death it may inform policymakers concerning mium payments to plans were “commu- spiral hypothesis is that coverage should future reforms. nity rated” — that is, they did not vary have fallen most among younger con- The legislation that created Medicare with the risk characteristics of those being sumers for whom the reforms caused the Part D also established Health Savings insured. As discussed earlier, one result is greatest premium increases. This should Accounts (HSAs), tax-free savings thus that the most generous plan faced an cause the average age of the insurance accounts that, when used in conjunction adverse selection death spiral. pool to increase. While we find some evi- with a high deductible insurance plan, The relationship between the way dence of such changes in New York, the can be used to fund medical expenses. premiums are rated and risk selection trends for the two comparison states are In the past, consumers have shown lit-

12 NBER Reporter Summer 2006 tle enthusiasm for high deductible poli- and P.J. Feldstein, “Switching Costs, Price Workers: Can Reduced Premiums cies, although certain attractive features Sensitivity and Health Plan Choice,” Achieve Full Coverage?” Health Services of HSAs combined with steady increases Journal of Health Economics, 21(1) Research, 32(4) (1997), pp. 453–70; L. in premiums for other types of health (2002), pp. 89–116. Blumberg, L. Nichols, and J. Banthin, insurance may change this. Some worry 2 D.M. Cutler and S. J. Reber, “Paying “Worker Decisions to Purchase Health that these plans will be most attractive to for Health Insurance: The Tradeoff Insurance,” International Journal of low risk consumers who do not anticipate Between Competition and Adverse Health Care Finance and Economics, a great need for medical care, thus causing Selection,” NBER Working Paper No. 1(3-4) (1997), pp. 305–25. J. Gruber more comprehensive plans to experience 5796, October 1996, and Quarterly and E. Washington, “Subsidies to adverse selection. If this occurs, it may Journal of Economics, 113(2) (1998), pp. Employee Health Insurance Premiums be HMOs, which benefited in the 1990s 433–66. and the Health Insurance Market,” NBER from risk-based sorting in the small group 3 A.B. Royalty and N. Solomon, “Health Working Paper No. 9567, March 2003, market and within employer-sponsored Plan Choice: Price Elasticities in a and Journal of Health Economics, 24(2) programs like those of Harvard and the Managed Competition Setting,” Journal (2005), pp. 253–76. UC, that are adversely affected. How the of Human Resources, 34(1) (1999), pp. 7 T. Buchmueller and J. DiNardo, “Did introduction of HSAs affects consumer 1–41. Community Rating Induce an Adverse health insurance decisions and what these 4 T.C. Buchmueller, “Price and the Selection Death Spiral? Evidence from new products mean for the stability of Health Plan Choice of Retirees,” NBER New York, Pennsylvania and Connecticut,” insurance markets are interesting areas for Working Paper No. 11395, June 2005, NBER Working Paper No. 6782, January future research. and Journal of Health Economics, 25(1) 1999, and American Economic Review, (2006), pp. 81–101. 92(1) (2002), pp. 280–294. 1 T.C. Buchmueller and P.J. Feldstein, 5 T.C. Buchmueller and S. Ohri, 8 T.C. Buchmueller and S. Liu, “Health “The Effect of Price on Switching Among “Health Insurance Take-up by the Near Insurance Reform and HMO Penetration Health Plans,” Journal of Health Elderly,” NBER Working Paper No. in the Small Group Market,” NBER Economics, 16(2) (1997), pp. 231–47; 11951, January 2006, and Health Working Paper No. 11446, June 2005 and T.C. Buchmueller, “Does a Fixed-Dollar Services Research, forthcoming. Inquiry, 42(4) (2005/2006) . Premium Contribution Lower Spending?” 6 M. Chernew, K. Frick, and C. Health Affairs, 17(6) (1998), pp. 228– McLaughlin, “The Demand for Health 35; B.A. Strombom, T.C. Buchmueller, Insurance Coverage by Low-Income

International Organization of Production and Distribution

Elhanan Helpman*

International trade has grown rap- and FDI expanded rapidly, but their of this research in which I was involved. idly since World War II, and in the last nature also has changed as production has In order to understand the new orga- two decades the acquisition of subsidiar- become more fragmented and its indi- nizational forms, it is useful to think ies in foreign countries (that is, foreign vidual stages have been dispersed across about a simple two-dimensional choice direct investment, or FDI) has grown many countries. These trends have been that a business firm has to make concern- even faster. Not only have foreign trade accompanied by growing domestic and ing an intermediate input: it has to decide international outsourcing.1 As a result, we whether to produce it in-house or to out- * Helpman is a Research Associate in the now have a more complex web of inter- source its production to another firm, and NBER’s Program on Political Economy and the Programs on International Trade national trade and FDI than ever before, in either case it has to decide whether to and Investment and International Finance which cannot be explained by traditional make it offshore or not. This yields four and . He is also the Galen trade theory. In response, theorists have possibilities. First, an input can be pro- L. Stone Professor of International Trade at developed new analytical tools for think- duced in-house in the home country of Harvard University. ing about these issues. I will describe some the firm, in which case there is neither

NBER Reporter Summer 2006 3 foreign trade nor FDI. Second, an input Grossman and I explored related downgrading. As a result, countries with can be outsourced in the home country, issues in two additional papers, in which better legal systems gain comparative in which case there is also neither for- the quality of a match is explicitly mod- advantage in sectors with low elastici- eign trade nor FDI. Third, an input can eled and it varies endogenously across ties of substitution, which are particu- be produced in-house in a foreign subsid- countries.4 The first paper focuses on larly sensitive to the incompleteness of iary, in which case there is foreign direct the offshoring decision, the second on contracts. Thus the quality of a coun- investment. If the input is imported back the decision to internalize abroad (that try’s legal system differentially affects its to the home country for further process- is, foreign outsourcing versus FDI). export performance in sectors that vary ing or assembly, there is also intra-firm In both, we introduced variations in by the degree to which they use con- trade. Finally, an input can be outsourced the degree of contract incompleteness tract-sensitive inputs.6 to a foreign supplier, in which case there which allow us to examine how differ- is no FDI, but if the input is imported ences in the quality of legal institutions Sorting into Organizational to the home country for further pro- across countries, or changes in these Forms cessing or assembly there is arm’s-length institutions in one country, affect firm trade. An understanding of what drives structure. To illustrate, consider the out- Scholars have also developed mod- these choices is essential for an under- sourcing decisions of firms in a country els of international trade in which firms standing of the recent trends in the world called North, which can buy inputs in choose which markets to serve and how economy. North or South, where wages are lower to serve them. This work has responded in South and so is the quality of its legal to the accumulated evidence that only a Incomplete Contracts system. In this case, the model finds that small fraction of firms engage in either improvements in North’s legal institu- foreign trade or FDI, that exporting firms Grossman and I started to study tions shift outsourcing from South to are more productive than non-exporters, these issues in the late 1990s, focusing North, as we would expect. Yet improve- multinationals are more productive than first on the internalization decision (that ments in South’s legal institutions shift exporters, and firm productivity disper- is, a firm’s decision to produce in-house outsourcing from North to South only sion varies widely across sectors. Melitz or to outsource). We took an incomplete when the gap in the quality of the legal developed the most useful model.7 In his contracts approach to the theory of the systems is large. The last result shows model, firms within an industry differ by firm. Having in mind dealing with trade how labor and product markets interact productivity and they face fixed costs of and FDI, we first developed an ana- with institutions to produce unexpected exporting. As a result, only the most pro- lytical framework suitable for general outcomes. ductive firms export while the less pro- equilibrium applications.2 In this frame- The impact of legal-system quality ductive firms serve only the domestic work, final goods producers need spe- on trade, via the endogenous formation market. This sorting pattern is consistent cialized intermediate inputs, and they of Ricardian-type comparative advan- with the evidence, and it has important enter an industry as either integrated or tage, is explored in a joint paper with implications for trade structure and the outsourcing enterprises, while suppliers Acemoglu and Antràs.5 We develop a impact of trade liberalization on the reor- of intermediate inputs enter as indepen- simple framework in which final good ganization of industries. dent entities. An outsourcing final goods producers choose a technology from a Melitz, Yeaple, and I extended this producer has to find an input supplier, set that features a tradeoff between costs model to allow firms to serve foreign mar- and a supplier has to find a buyer. An and efficiency, and find that the opti- kets either by exporting or by establishing outsourcing firm pairs up with only one mal choice depends on the degree of subsidiaries in foreign countries that sell supplier, and vice versa. The probabil- contract incompleteness. In our model, directly to the host country (horizontal ity of each side finding a match depends firms want to adopt more efficient tech- FDI).8 In this case, only the most produc- on the number of producers and suppli- nologies, except that their demand for tive firms engage in horizontal FDI, low ers seeking partners. Once a match has better technologies bids up the acqui- productivity firms serve only the domes- formed, the supplier decides on how sition costs of those technologies. As tic market, and firms with intermediate much to invest in the buyer’s special- a result, in countries with better legal productivity export. Moreover, the varia- ized input. This is the point at which institutions, firms upgrade their tech- tion across sectors in the ratio of subsid- the incompleteness of contracts kicks nology only in industries that are rel- iary sales to export sales is positively cor- in. This model implies that trade has atively vulnerable to contract incom- related with the variation across sectors no effect on the organization of indus- pleteness. These happen to be the sectors in the productivity dispersion of firms. tries when matching is subject to con- with relatively low elasticities of sub- The U.S. data support this prediction: stant returns to scale. But when match- stitution across intermediate inputs. In productivity dispersion affects trade and ing leads to increasing returns to scale, sectors with relatively high elasticities FDI. Importantly, the economic size of the model predicts more outsourcing the of substitution, the higher cost of tech- this effect is large; it is of the same order more countries engage in foreign trade.3 nology adoption induces technological of magnitude as the impact of fixed costs

14 NBER Reporter Summer 2006 or freight charges, which are traditional which final good producers provide few 1 The term “outsourcing” has been determinants of the proximity-concen- headquarter services, outsourcing domi- used in more than one way. I use it in the tration tradeoff in the theory of horizon- nates integration. Low-productivity firms traditional sense, as the acquisition of an tal FDI. in these industries outsource at home, input or service from an unaffiliated firm. While the last model focuses atten- while high-productivity firms outsource In this case domestic outsourcing refers to tion on horizontal FDI (that is, FDI to a low-wage foreign country, say South. the acquisition of an input from a domes- designed to serve the host market only) More productivity dispersion in such tic unaffiliated firm while international and the model from the previous section industries raises foreign relative to domes- outsourcing refers to the acquisition of an focuses attention on vertical FDI (that is, tic outsourcing. On the other hand, in input from a foreign unaffiliated firm. FDI designed to reduce manufacturing sectors with a high intensity of headquar- 2 See G.M. Grossman and E. Helpman, costs), this neat distinction between two ter services, all four organizational forms “Incomplete Contracts and Industrial extreme forms of FDI has become less can coexist: the most productive firms Organization,” NBER Working Paper appealing over time, simply because the engage in FDI, the least productive firms No.7303, August 1999, published as data show that multinationals are engaged outsource at home, and in between the “Integration versus Outsourcing in in “complex” integration strategies, which more productive firms outsource to South Industry Equilibrium,” Quarterly Journal are neither purely horizontal nor purely while the less productive firms integrate of Economics 117(1), February 2002, pp. vertical.9 Grossman, Szeidl, and I stud- at home. More productivity dispersion 85–120. ied such complex integration strategies raises offshoring relative to domestic sup- 3 See also J. McLaren, “‘Globalization’ for industries with productivity disper- plying of intermediates, and it raises inte- and Vertical Structure,” American sion across firms.10 In our model, firms gration relative to outsourcing. Higher Economic Review 90(5), December assemble intermediate inputs to manufac- headquarter intensity also makes integra- 2000, pp. 1239–54 on this point. ture final goods, and a firm can locate the tion more prevalent. 4 See G.M. Grossman and E. Helpman, assembly or the production of intermedi- “Outsourcing in a Global Economy,” ates in a combination of countries: home Managerial Incentives NBER Working Paper No.8728, January in North, or foreign in North or South. 2002, published in the Review of The model predicts a strong complemen- Grossman and I have also studied the Economic Studies 72(1), January 2005, tarily between the two forms of FDI. For sorting pattern of heterogeneous firms pp. 135–59, and G.M. Grossman and E. example, a low production cost of compo- when the agency problem arises from Helpman, “Outsourcing versus FDI in nents in South encourages FDI in compo- managerial incentives rather than incom- Industry Equilibrium,” NBER Working nents in South as well as FDI in assembly plete contracts.12 In this model, outsourc- Paper No.9300, November 2002, pub- there. This model produces rich patterns ing provides the supplier with better lished in the Journal of the European of trade and FDI. incentives, but integration gives the final Economic Association 1(2-3), April-May good producer better monitoring oppor- 2003, pp. 317–27. Heterogeneity and Incomplete tunities. As a result, the least and the most 5 See D. Acemoglu, P. Antràs, and E. Contracts productive firms outsource while firms Helpman, “Contracts and the Division of with intermediate productivity integrate. Labor,” NBER Working Paper No. 11356, Combining heterogeneity in the pro- Among those who integrate, the more May 2005. ductivity of firms with incomplete con- productive integrate at home and the less 6 See A.A. Levchenko, “Institutional tracts produces predictions about all four productive engage in FDI. This sorting Quality and International Trade,” IMF organizational forms mentioned at the pattern is quite different from the sorting Working Paper WP/04/231, 2004, beginning of this review: integration at pattern discussed above, where incom- and N. Nunn, “Relationship Specificity, home, outsourcing at home, integration plete contracts were the source of the Incomplete Contracts, and the Pattern in a foreign country, and outsourcing to agency problem. Yet there is evidence for of Trade,” mimeo, University of Toronto, a foreign country. I study this combina- both patterns.13 2005 for empirical evidence. tion in a joint paper with Antràs.11 In I have reviewed a number of stud- 7 See M.J. Melitz, “The Impact of this model, the tradeoff between inte- ies that can be used to explain rich pat- Trade on Intra-Industry Reallocations gration and outsourcing is driven by the terns of trade and FDI, and the relation- and Aggregate Industry Productivity,” tradeoff between agency costs and the ship between them. Much of this theory NBER Working Paper No. 8881, April costs of organization. When integration has been motivated by new evidence, and 2002, published in Econometrica 71(6), has higher fixed costs than outsourcing some new implications of the various November 2003, pp. 1695–725. and offshoring has higher fixed costs than models have been tested. There remains, 8 See E. Helpman, M.J. Melitz, and home sourcing, the model predicts varia- however, much more that needs to be S.R. Yeaple, “Export versus FDI,” NBER tion in the prevalence of the four organi- done, and new data are needed for this Working Paper No. 9439, January zational forms as a function of industry purpose.14 2003, published as “Export versus FDI characteristics. For example, in sectors in with Heterogeneous Firms,” American

NBER Reporter Summer 2006 5 Economic Review 94(1), March 2004, Strategies for the Multinational Firm,” Journal of International Economics pp. 300–16. NBER Working Paper No. 10189, 63(2), July 2004, pp. 237–62. 9 See World Trade Organization, December 2003, forthcoming in the 13 See S.F. Lin and C. Thomas, “When Annual Report 1998 (Geneva: World Journal of International Economics. Do Multinational Firms Outsource? Trade Organization Conference on 11 See P. Antràs and E. Helpman, Evidence From the Hotel Industry,” Trade and Development), 1998, and “Global Sourcing,” NBER Working Paper mimeo, Harvard University, 2005 S.R. Yeaple, “The Complex Integration No. 10082, November 2003, published in 14 See E. Helpman, “Trade, FDI, and Strategies of Multinationals and Cross the Journal of Political Economy 112(3), the Organization of Firms,” mimeo, Country Dependencies in the Structure June 2004, pp. 552–80. Harvard University, February 2006, of Foreign Direct Investment,” Journal of 12 See G.M. Grossman and E. forthcoming in the Journal of Economic International Economics 60(2), August Helpman, “Managerial Incentives Literature, for a detailed review of the lit- 2003, pp. 293–314. and the International Organization of erature on these topics. 10 See G.M. Grossman, E. Helpman, Production,” NBER Working Paper No. and A. Szeidl, “Optimal Integration 9403, December 2002, published in the

Historical Aspects of U.S. Trade Policy

Douglas A. Irwin*

While international trade and trade Early American Trade Policy support of manufacturing through sub- policy continue to be as controver- sidies and import tariffs, but it is - com sial as ever, the United States has been To say much about the stance of a monly believed that the report was never committed for more than half a cen- country’s trade policy requires, at a min- implemented. Although Hamilton’s pro- tury to maintaining an open market. imum, data on the average tariff level. posals for bounties (subsidies) failed to It was not always that way. For most of Unfortunately, standard U.S. trade statis- receive support, my research has shown U.S. history, the United States imposed tics only started calculating average tariff that Congress adopted virtually every tar- fairly substantial barriers to imports in figures from 1821. To fill the gap in the iff recommendation put forward in the an effort to protect domestic producers historical data, I gathered information report by early 1792.3 These tariffs were from foreign competition. from early government documents to cal- not highly protectionist duties, because For the past several years, I have culate the average tariff on total and duti- Hamilton feared discouraging imports, been investigating the historical aspects able imports for the period from 1790 the critical tax base on which he planned of U.S. trade policy as part of the to 1820.2 These figures reveal that tariffs to fund the public debt. Indeed, because NBER’s research on international trade started out at relatively low levels, about his policy toward manufacturing was one and the development of the American 15 percent in the 1790s, but rose there- of limited encouragement and not pro- economy. The purpose of this research after to generate additional revenue and tection, Hamilton was not as much of has been to study the economic effects help finance the War of 1812. Because re- a protectionist as he is often made out of past trade policies on the U.S. econ- exports were a significant component of to be. Hamilton’s moderate tariff poli- omy and understand the political and U.S. foreign trade at this time, my study cies found support among merchants and economic forces that have shaped those suggests that it is important to adjust for traders, the backbone of the Federalist policies.1 drawbacks (rebated tariff revenue on re- Party. But disappointed domestic manu- exported goods) to determine the true facturers shifted their political allegiance level of the tariff. to the Republican Party, led by Thomas * Irwin is a Research Associate in the NBER’s One of the classic, early state- Jefferson and James Madison, both of International Trade and Investment and Development of the American Economy ments on U.S. trade policy is Alexander whom were willing to consider much Programs, and a Professor of Economics at Hamilton’s Report on Manufactures in more draconian trade policies aimed at Dartmouth College. 1791. This report called for government Britain.

16 NBER Reporter Summer 2006 Indeed, as president, Jefferson was permanent effect in reallocating resources ers from foreign competition. The role responsible for one of the most unusual between domestic infant industries (such of the tariff in promoting the expan- policy experiments in the history of U.S. as cotton textiles) and trade-dependent sion of the early American cotton tex- trade policy. At his request, Congress industries (such as shipbuilding). tile industry has been quite controver- imposed a nearly complete embargo on sial, with older scholarship by Frank international commerce from December Antebellum Trade Policy Taussig suggesting that the industry was 1807 to March 1809. The Jeffersonian well established by the 1830s and more trade embargo provides a rare opportu- During the 1820s, the average tariff recent work suggesting that the tariff nity (or natural experiment) to observe on dutiable imports rose sharply, peaking remained critical for some time. In joint the effects of a nearly complete (albeit at over 60 percent in 1830, even higher work, Peter Temin and I concluded that short-lived) elimination of international than under the notorious Hawley-Smoot Taussig was correct in that the cotton trade. Economists usually describe the tariff of 1930. Over the next decade, the textile industry could survive without gains from international trade by com- average tariff fell by half, and stood at the tariff by the early 1830s.7 Using paring welfare at a free-trade equilibrium less than 20 percent by the Civil War. data from 1826 to 1860, we estimated with welfare at an autarky equilibrium. In fact, the period from 1830 to 1860 the sensitivity of domestic production In practice, such a comparison is almost was one of just two in American history to fluctuations in import prices and never feasible because the autarky equi- when tariffs exhibited a secular decline concluded that most of the industry librium is almost never observed, except (the other being from the mid-1930s to could have survived even if the tariff had in unique cases such as this one. By mid- the present). What political economy been completely eliminated. The lack of 1808, the United States was about as factors explain the rise and fall of tariffs responsiveness of domestic production close to being fully shut off from interna- during this period? to changes in import prices was mainly tional commerce as it has ever been dur- A shift in the economic interests due to the specialization of American ing peacetime. of the West (the Midwest of today) and British producers in different types Monthly price data allow us to observe appears to explain this pattern.6 In the of textile products that were imperfect the dramatic impact of the embargo: the highly sectional politics of the day, the substitutes for one another: imports export-weighted average of the prices North favored high tariffs, the South consisted of intricate ginghams, whereas of raw cotton, flour, tobacco, and rice, favored low tariffs, and the West was New England power looms were supply- which accounted for about two-thirds a swing region. In the 1820s, a coali- ing plain weaves, such as sheeting. of U.S. exports in the United States, fell tion in Congress between the North In terms of exports, the United by one third within a month or two of and West raised tariffs by exchanging States produced about 80 percent of the the embargo. The price of imported com- votes on import duties for spending on world’s cotton in the decades prior to the modities rose by about a third as the num- internal improvements, which benefited Civil War, most of which was exported. ber of ships entering U.S. ports fell to a the West. But when President Andrew A long-standing question has been the trickle and imports became increasingly Jackson began vetoing internal improve- degree to which the United States had scarce. According to my calculations, ments bills, he effectively de-linked these market power in cotton and how much the static welfare cost of the embargo issues and destroyed the North-West it could have gained from an optimal was about 5 percent of GDP.4 Thus, the alliance. South Carolina’s refusal to export tax on cotton. To address this embargo inflicted substantial costs on the enforce the existing high tariffs sparked question, I estimated the elasticity of economy during the short period that it the nullification crisis and paved the foreign demand for U.S. cotton exports was in effect. way for the Compromise Tariff of 1833, and used it in a simple partial equilib- The embargo, along with the dra- which phased out tariffs above 20 per- rium model to determine the optimal matic reduction in trade as a result of cent over a nine year period. Although export tax and its probable impact on the War of 1812, is commonly believed Congress could not credibly commit prices, trade, and welfare.8 The results to have spurred early U.S. industrializa- to implementing the staged reductions indicate that the export demand elastic- tion by promoting the growth of nascent or maintaining the lower duties, the ity for U.S. cotton was about -1.7 and domestic manufacturers. Joseph Davis Compromise held because of the grow- that the optimal export tax of about 50 and I used his newly available series on ing economic interest of the West in percent would have raised U.S. welfare U.S. industrial production to investigate exporting grains (due, ironically, to by about $10 million, that is about 0.3 how this protection from foreign compe- transportation improvements as the rail- percent of U.S. GDP or about 1 percent tition affected domestic manufacturing.5 road network expanded) which gave it a of the South’s GDP. (Such a tax was not On balance, the trade disruptions did not stake in maintaining a lower tariff equi- implemented, however, partly because decisively accelerate U.S. industrialization librium in cooperation with the South. the U.S. Constitution forbids the impo- as trend growth in industrial produc- Aside from revenue, the ostensible sition of export taxes.) One implica- tion was little changed over this period. purpose of such import tariffs was to tion is that the welfare gains from an However, the disruptions may have had a protect import-competing manufactur- export tax are not necessarily large, even

NBER Reporter Summer 2006 7 for a country with considerable market moted the growth of infant industries. during the late nineteenth century? One power. I examined the case that has been her- paper investigates the multiple channels by alded as possibly the best example of infant which tariffs could have promoted growth Post Civil War Industrializa- industry protection: the tinplate industry, during this period.12 I found that 1) late tion and Growth which produces thin sheets of iron or steel nineteenth century growth hinged more that have been coated with tin.11 Although on population expansion and capital accu- After the Civil War, the United States the tinplate industry is an obscure one, it is mulation than on productivity growth; 2) maintained high tariffs to protect domestic unique because, unlike most manufactur- tariffs may have discouraged capital accu- manufacturers from foreign competition. ing industries, it did not receive significant mulation by raising the price of imported Tariff advocates claimed that high import tariff protection after the Civil War, appar- capital goods; and 3) productivity growth duties helped to expand industrial employ- ently because of a mistaken interpretation was most rapid in non-traded sectors (such ment and keep wages high, while also aid- of the tariff code. Left without adequate as utilities and services) whose performance ing farmers by creating a steady demand protection, there was virtually no domestic was not directly related to the tariff.13 in the home market for the food and raw production prior to 1890. The McKinley At the end of the nineteenth cen- materials that they produced. Tariff critics tariff of 1890 substantially raised the duty tury, though, the pattern of U.S. trade charged that those import duties raised the on imported tinplate to encourage the changed dramatically. For most of the cen- cost of living for consumers and harmed entry and growth of domestic producers. tury, the United States had a strong com- agricultural producers by effectively taxing The act also contained an unusual provi- parative advantage in agricultural goods their exports, thus redistributing income sion in which the tariff would be com- and exported mainly raw cotton, grains, from consumers and farmers to big busi- pletely eliminated in six years if, by that and meat products in exchange for imports nesses in the North. time, domestic production did not amount of manufactured goods. But in the mid- One approach to examining the mag- to at least one-third of imports. The tar- 1890s, America’s exports of manufac- nitude of protection given to import-com- iff succeeded in promoting domestic pro- tures began to surge. Manufactured goods peting producers and the costs imposed on duction and output rapidly expanded, and jumped from 20 percent of U.S. exports in export-oriented producers is to focus on by about 1910 the price of U.S. tinplates 1890 to 35 percent by 1900 and nearly 50 changes in the domestic prices of traded fell below those produced in the United percent by 1913. In about two decades, the goods relative to non-traded goods.9 Kingdom. United States reversed a century-old trade Because the tariff also increased the prices The tinplate example has all the ele- pattern and became a large net exporter of of non-traded goods, the degree of protec- ments of an apparently successful appli- manufactured goods. What accounts for tion was much less than indicated by nom- cation of infant industry protection. But this abrupt change in the structure of U.S. inal rates of protection; my results sug- in asking the counterfactual question — exports? gest that the 30 percent average tariff on would the industry have developed any- My research suggests that natural imports yielded just a 15 percent implicit way, and were the tariffs worthwhile? — I resource abundance fueled a dramatic subsidy to import-competing producers answer yes and no. My analysis suggests expansion of iron and steel exports, in while effectively taxing exporters at a rate that tinplate was not an “infant industry” part by enabling a sharp reduction in the of 11 percent. The paper also indicates that floundered because of the lack of pre- price of U.S. exports relative to other com- that the tariff policy redistributed large vious production experience (learning by petitors.14 In this case, the commercial amounts of income (about 9 percent of doing), but rather an industry in which exploitation of the Mesabi iron ore range GDP) across groups, although the impact domestic production was not profitable in Minnesota reduced domestic ore prices on consumers was roughly neutral because because of the high domestic cost of iron by 50 percent in the mid-1890s and was they devoted a sizeable share of their expen- and steel inputs attributable to tariffs. The equivalent to over a decade’s worth of ditures to exportable goods. These findings tinplate industry suffered from negative industry productivity improvement in its may explain why import-competing pro- effective protection due to the existing tar- effect on iron and steel export prices. The ducers pressed for even greater protection iff structure; while a second-best optimal non-tradability of American ore resulted in the face of already high tariffs and why tariff could have corrected that distortion, in its distinctive impact on the pattern of consumers (as voters) did not strongly and improved welfare, such an optimum U.S. trade; whereas raw cotton was trad- oppose the policy. was not imposed. In the absence of the able, and hence the domestic cotton textile Were protectionist policies essential McKinley tariff, the U.S. tinplate industry industry did not reap an advantage from for domestic industries after the Civil War? would have established itself about a decade having local production of cotton, iron ore In the case of pig iron, high import tariffs later as the material input costs of iron and and other minerals were difficult to trade, may have helped those producers, but they steel converged with those in Britain. Over and therefore they were exported in final harmed other manufacturers who needed this time horizon, the McKinley duty fails products, not in raw form. access to cheap iron to produce other prod- to pass a cost-benefit test. ucts, such as machinery and bridges.10 One Were high import tariffs somehow 1 This is an update from my previ- justification for the tariffs is that they pro- related to the strong U.S. economic growth ous report in the Winter 1999 NBER

18 NBER Reporter Summer 2006 Reporter. See D.A. Irwin, “Historical No. 9944, September 2003. Industries? Evidence from the Tinplate Perspectives on U.S. Trade Policy,” NBER 6 D.A. Irwin, “Antebellum Tariff Industry,” NBER Working Paper No. Reporter, Winter 1999. Politics: Coalition Formation and Shifting 6835, December 1998, and Journal of 2 D.A. Irwin, “New Estimates of the Regional Interests,” NBER Working Paper Economic History 60 (June 2000): pp. Average Tariff of the United States, 1790- No. 12161, April 2006. 335–60. 1820,” NBER Working Paper No. 9616, 7 D.A. Irwin and P. Temin, “The 12 D.A. Irwin, “Tariffs and Growth April 2003, and Journal of Economic Antebellum Tariff on Cotton Textiles in Late Nineteenth Century America,” History, 63 (June 2003), pp. 506–13. Revisited,” NBER Working Paper No. NBER Working Paper No. 7639, April 3 D.A. Irwin, “The Aftermath of 7825, August 2000, and Journal of 2000, and The World Economy 24 Hamilton’s Report on Manufactures,” Economic History 61 (September 2001): (January 2001): pp. 15–30. NBER Working Paper No. 9943, pp. 777–98. 13 I am currently studying the role of September 2003, and “The Aftermath 8 D.A. Irwin, “The Optimal Tax on tariffs in promoting structural change dur- of Hamilton’s Report on Manufactures,” Antebellum Cotton Exports,” NBER ing this period. Cross-country analysis is Journal of Economic History, 64 Working Paper No. 8689, December 2001, an alternative way of examining the U.S. (September 2004), pp. 800–21. and Journal of International Economics, experience. D.A. Irwin, “Interpreting the 4 D.A. Irwin, “The Welfare Costs of 60 (August 2003), pp. 275–91. Tariff-Growth Correlation of the Late Autarky: Evidence from the Jeffersonian 9 D.A. Irwin, “Tariff Incidence in Nineteenth Century,” NBER Working Embargo, 1807-1809,” NBER Working America’s Gilded Age,” NBER Working Paper No. 8739, January 2002, and Paper No. 8692, December 2001; Paper No. 12162, April 2006. American Economic Review 91 (May and “The Welfare Costs of Autarky: 10 D.A. Irwin, “Could the U.S. Iron 2002): pp. 165–69. Evidence from the Jeffersonian Embargo, Industry Have Survived Free Trade After 14 D.A. Irwin, “Explaining America’s 1807-1809,” Review of International the Civil War?” NBER Working Paper Surge in Manufactured Exports, 1880- Economics 13 (September 2005): pp. No. 7640, April 2000, and Explorations 1913,” NBER Working Paper No. 7638, 631–45. in Economic History 37 (July 2000): pp. April 2000, and Review of Economics 5 J. H. Davis and D.A. Irwin, “Trade 278–99. and Statistics 85 (May 2003): pp. 364– Disruptions and America’s Early 11 D.A. Irwin, “Did Late Nineteenth 76. Industrialization,” NBER Working Paper Century U.S. Tariffs Promote Infant

NBER Profile: Thomas C. Buchmueller

Thomas C. Buchmueller is a Research economics of health insurance and related Associate in the NBER’s Programs in Health public policy issues. In particular, he is inter- Care and Health Economics. He received his ested in the relationship between health insur- B.A. in Economics from Carleton College in ance and labor market outcomes, the interac- 1985 and his Ph.D. in Economics from the tion between the public sector and private University of Wisconsin, Madison in 1992. insurance markets, and consumer demand Since 1992, he has taught at the University for health insurance. From 2002 to 2006, he of California, Irvine, where he is Professor was a Deputy Editor of Medical Care. He is of Economic and Public Policy at the Paul currently a Deputy Editor of the Journal of Merage School of Business. He has also been Economics and Management Strategy and is a visiting scholar at University of York (UK), on the Editorial Board of Inquiry. INSEAD and CREDES in France, and at Buchmueller lives in Irvine, California the Federal Reserve Bank of San Francisco. with his wife Liz Phillips, a social worker; his Buchmueller will be spending the 2006–7 children, Sophie and Adam; and a 14-year academic year in Australia as a Packer Policy old beagle. He enjoys playing basketball, trav- Fellow. Then, he will be joining the fac- eling, and following the Red Sox from afar. ulty of the University of Michigan’s Stephen Ross School of Business, where he will hold the Waldo O. Hildebrand Chair in Risk Management and Insurance. Buchmueller’s research focuses on the

NBER Reporter Summer 2006 9 Conferences “New Ideas about New Ideas” Considerable work has been done I. The Idiosyncrasy of idiosyncratic connections across distantly on creativity across a wide-range of disci- Innovation related concepts. plines, including business, cognitive neu- John Kounious, a cognitive neurosci- roscience, economics, history, psychology, With innovation viewed as a way for entist at Drexel University, showed that, and sociology, but until recently there the United States to maintain its economic although the final moment of insight is had been little interaction among these position, it is natural to ask what can be sudden, there are substantial changes in researchers. But on March 10 and 11, done to foster it. As Daniel Goroff — a brain activity leading up to insight solu- fifteen experts on innovation and cre- mathematician and the Dean of Faculty tions to problems, such as a quieting of ativity from these disciplines — both at Harvey Mudd College — discussed, the sensory areas of the brain in the sec- established senior scholars and emerg- colleges and universities increasingly onds before the solution reaches con- ing younger researchers — met at the are prioritizing creativity, and the Mills sciousness. He interpreted these results NBER in Cambridge for a “New Ideas Commission is recommending systematic as the unconscious brain searching for About New Ideas Conference” designed testing of higher education outcomes. solutions, but having to quiet down exter- to foster cross-disciplinary dialogue on Our working definition of creativ- nal inputs to bring a candidate solution creativity and innovation. The confer- ity was “the production of novel and into consciousness. Moreover, patterns ence, supported by the Sloan Foundation, useful ideas or artifacts.” The discussion of brain activity before people even see a was organized by NBER Research touched on the arts, industry, and sci- problem predict whether they will solve Associate Richard Freeman of Harvard ences. Perhaps the single point on which that problem with insight, or more ana- University and Faculty Research Fellow there was the widest agreement was that, lytically. Finally, he also reports that mod- Bruce Weinberg of Ohio State University. while there are recognizable patterns in erate rates of arousal are best for prob- Weinberg also prepared this summary creativity, the motivations of creators and lem solving, with the optimal amount of article for the NBER Reporter. the processes by which creative ideas arise arousal being lower as problems become Given the varied group and the nature are frequently specific to the individual, more difficult. of scientific papers presented, one might or idiosyncratic. The idiosyncratic nature Sohee Park, a cognitive neuroscien- have been concerned about the ability to of innovation showed up in brain images, tist at Vanderbilt University, discussed the communicate across disciplinary lines, let in problem-solving experiments, and in link between psychosis and creativity. She alone to find common interest, but soon analyses of historical and contemporary showed that while psychosis itself may cognitive neuroscientists were discussing innovations and innovators. Participants interfere with creativity, the more idiosyn- history, psychologists were talking about were optimistic about our ability to foster cratic associations among people who are economics, and everyone was poring over creativity; however, we agreed that, to be psychosis-prone (but clinically normal) images of the brain. And, after close to successful, we must attempt to confront enhance creativity. She interpreted these 20 hours of discussions, some during an the idiosyncratic nature of creativity. results by arguing that psychosis increases informal stroll along the Charles River, a The Idiosyncratic Nature of the Creative the novelty of associations, but interferes set of themes emerged quite clearly along Brain with memory and other processes that are with policy implications and directions essential for creativity. Healthy but psy- for future research. At the finest level, the cognitive neu- chosis-prone people also show increased Indeed, the timing was also fortu- roscientists at the conference showed the use of their right frontal lobe when they itous from a policy perspective. As the idiosyncratic nature of creativity in brain are generating novel ideas. most recent State of the Union Address functioning. Mark Jung-Beeman, a cog- Teresa Amabile, a psychologist at indicated, the United States increasingly nitive neuroscientist at Northwestern Harvard Business School, discussed how sees its economic position challenged, University, showed that distinct brain emotions can influence the creative pro- and creativity and innovation are viewed areas contribute when people solve prob- cess. Although research on psychopa- as the most promising directions for us to lems with insight, that is, when solu- thology has found a connection between take in maintaining our position. On the tions are accompanied by “Aha” moments. depression and creativity among artists other hand, this increased interest in cre- The patterns of brain activity suggest an and writers, the emotion-creativity con- ativity and innovation may conflict with increase in “top-down” processing, and nection has been virtually unexamined demographics — the workforce has been increased contributions from the brain’s among adults working in business orga- aging, and creativity and innovation tra- right hemisphere. Beeman attributed the nizations. Amabile discussed the results ditionally have been regarded as linked latter effect to the more diffuse links in of an extensive study that found a consis- to youth. the right brain, which allow for novel or tent, positive relationship between posi-

20 NBER Reporter Summer 2006 tive emotion and creativity. Exploiting to the development of quantum mechan- being designed by God shaped the ques- the longitudinal nature of the data, she ics arose accidentally, but later contri- tions he asked and the answers he gave. showed that positive affect preceded butions were self-conscious attempts to Thus, thema are individual influences that creativity in the coming days. She also explain earlier results. shape a person’s creative work. found that getting a new idea or having Consistent with these findings, David Kaiser, a physicist and his- an insight can evoke immediate (though Teresa Amabile’s research has revealed an torian at MIT, presented a cautionary short-lived) feelings of elation — even intrinsic motivation principle of creativ- tale from the rapid, post-war expansion when the idea is a relatively minor one. ity: people will be most creative when of physics. He showed how the expan- Thus, her study revealed bi-directional they are motivated primarily by the inter- sion led teachers to emphasize the most causality in the connection between posi- est, enjoyment, satisfaction, meaningful- mechanical sides of quantum mechan- tive emotion and creativity at work, and ness, and personal challenge of the work ics, which were easier to teach, while shy- it also suggested an incubation effect itself, rather than by extrinsic induce- ing away from more qualitative questions whereby positive emotion on one day can ments or constraints. This principle is of interpretation — the “What does it stimulate new cognitive associations that best understood within a social-psycho- all mean?” musings that had so exercised bear fruit in the coming days. Given the logical view of creativity. Although peo- the discipline’s leaders before the war. different domains studied, these results ples’ production of creative (novel and In this way, concrete pedagogical pres- are consistent with those of Park and oth- appropriate) work certainly depends on sures helped to change how modern phys- ers who study the link between psychosis both their domain expertise and their cre- ics was handled in the classroom, and, and creativity. ative thinking skills, it also depends on indeed, what would count as “creative” Perhaps the most extreme statement their level of intrinsic motivation for the among the younger generation. of the view of innovation as idiosyncratic work, which can be strongly influenced Thus, cognitive neuroscientists, econ- is the chance permutation model of Dean by the inducements and constraints in omists, historians, and psychologists all Keith Simonton, a psychologist at the their social environment. Experimental see creativity as being idiosyncratic in University of California, Davis. In his and non-experimental research has terms of the processes through which model, important contributions are the revealed several aspects of work environ- it develops and the motivations of cre- result of purely random combinations of ments, such as a primary focus on tangi- ators. Given this view, there was great ideas. He discussed a broad range of evi- ble rewards or critical evaluation, which concern about efforts to test outcomes in dence supporting his approach. can undermine intrinsic motivation and higher education. Similarly, the group was The Idiosyncratic Nature of Creative creativity. Also, several aspects of work concerned about the ability to identify Motivations environments — such as autonomy and areas for innovation and target support optimal challenge in the work — can sup- to them, as opposed to supporting inno- The motivations of innovators also port intrinsic motivation and creativity. vation more broadly. Nevertheless, the are idiosyncratic, and this is particularly For example, using their diary database, Unites States must strive for excellence in true in the initial development of an Amabile and her colleagues have discov- education and support scientific research innovation. Josh Lerner, an economist ered a number of specific leader behaviors that will provide the basis for future eco- at Harvard Business School, showed that in the everyday work environment that nomic growth in a flexible way. many of the early developers of open- have positive or negative effects on daily source code were hackers who contrib- perceptions of leader support for creativ- II. The Geography uted to open source code for the stimu- ity and, thus, on creativity itself. of Creativity lation of programming, the recognition Richard Freeman sought to under- of their peers, or to oppose commercial stand gender differences in involvement Technological centers, such as Silicon software manufacturers. Companies only in the sciences in terms of gender differ- Valley and the Route 128 Corridor outside contribute to open source code once a ences in the response to incentives. He of Boston, have been attributed to knowl- large body of code has been developed noted that incentives for innovation often edge spillovers that arise among innova- and the commercial benefits are clear. take the form of tournaments, where the tors. In other words, the presence of many David Galenson and Bruce first person to succeed receives most or all others working on related problems is Weinberg, economists at the University of the returns. The evidence indicates that assumed to lead to informal interactions of Chicago and Ohio State University, women shy away from these situations, that foster creativity. This phenomenon find that early phases of revolutions in providing an explanation for the under- can operate at the level of cities and even the arts, industry, and science are more representation of women in the sciences. nations, and is an important motivation likely to arise from individuals pursuing Gerald Holton, a physicist and his- for public investment in research. their aesthetic goals or from serendipity. torian of science, discussed the role of Perhaps the finest grained evidence As revolutions develop, market factors thema: unquestioned principles held by here comes from the work of David become more important. In physics for individuals that guide their creativity. For Kaiser, who has traced the flow of ideas instance, many of the discoveries that led instance, Newton’s view of the universe as among physicists, looking at the develop-

NBER Reporter Summer 2006 21 ment, mutations, and spread of Feynman While there seems to be a presump- the advantage of being able to see further diagrams. Using these diagrams, which tion that creativity is associated with by standing on the shoulders of giants, illustrate interactions between particles, youth, there was a consensus that older they suffer from having longer climbs. He he shows how interacting communities individuals can, and frequently are highly showed that the age at which innovators modify and define techniques. creative. David Galenson outlined a dis- in the sciences and industry do important Weinberg has shown that geography tinction between experimental and con- work has been increasing over the twen- affects the probability of contributing to a ceptual innovators. Conceptual innova- tieth century. While his view is pessimis- scientific revolution. People who went to tors work deductively and frequently make tic at some level — it implies that inno- graduate school at a place where they were their most important contributions early vators will spend more of their careers exposed to the people who pioneered in their careers. Experimental innovators getting to the knowledge frontier and the new paradigm were more likely to work inductively, accumulating knowl- less time innovating — now we may have make contributions to that paradigm than edge from trial-and-error experiments, the advantage of having a population that people who attended other schools. The and tending to do their most important has reached an age where they are largely nature of their work also was affected. work later in their careers. These experi- done climbing. Lynne Zucker and Michael Darby, mental innovators may be entering their Weinberg’s work has shown that peo- a sociologist and economist at the peak years of creativity. ple who make contributions to new scien- University of California, Los Angeles, Dean Keith Simonton discussed a tific paradigms tend to have been exposed have studied the flow of knowledge from different approach, one that focuses on to them in their formative professional academia to industry. For a variety of lead- disciplines rather than the styles of indi- years. While this result would suggest that ing technologies — including semicon- vidual innovators. In his view, creativity younger individuals are more involved ductors, biotechnology, and nanotech- varies across disciplines depending on the with important innovations, he has also nology — startup firms are more likely to rate at which ideas can be developed and found that older individuals often make develop in cities where there are more star elaborated. He argued that in many fields, the contributions that set off innovative academic researchers. This work provides creativity increases for much of life; and, revolutions. Thus, older individuals have an important, direct link between aca- even in fields where creativity is great- a crucial role to play in the innovative demic research and industrial innovation. est at early ages, older individuals make process. While there was considerable opti- important contributions at the same rate mism about the future of the United as younger individuals after one controls IV. Future Work States as a leader in creativity and inno- for their lower rate of publication. vation, there was a general view that the Regardless of the approach — and Many directions for future work distance between the United States and there was an active, scholarly discussion emerged from the meeting. There was other countries likely would shrink. As of the relative merits of the two approach- considerable interest in using the emerg- other countries develop their research es — it is clear that older individuals are ing tools of cognitive neuroscience to test capabilities, scientific breakthroughs and often highly creative. Both approaches the foundations of other theories. Among their commercial applications likely will imply that there will be differences across the possibilities would be to probe the shift overseas, at least to some extent. For fields in the age at which people are effect of age on creativity and receptivity us to maintain a strong position, we will most creative: in Galenson’s approach, to new ideas by looking at changes in cog- have to invest in our scientific and indus- the shares of conceptual innovators, who nition over the life cycle. Another area in trial innovative communities. tend to be most creative when young, which links could be made was the rela- and experimental innovators, who tend tionship between affect and psychosis and III. Innovation and an Aging to be most creative later in their careers, creativity across domains. There was also Workforce vary across fields. One wonders whether interest in linking observational data on the development of information technol- the creative output of scientists or indus- Work on the effect of age on creativity ogy was due at least in part to the relative trial innovators to information about cog- dates back at least to Harvey Lehmann’s youth of the workforce and if technologi- nitive functioning. Similarly, it would be Age and Achievement, published in 1953. cal progress may shift to other areas as the valuable to study cognition under various The relationship between age and creativ- workforce ages. incentives and other aspects of the social ity is particularly important today with Ben Jones, an economist at environment. Work is also necessary to new technologies developing rapidly and Northwestern University’s Kellogg School reconcile differences in views of creativ- the workforce aging, driven by the large of Management, and Bruce Weinberg ity that have emerged across the various baby boom generation. Will our ability took another view of the relationship disciplines. to innovate and take advantage of innova- between age and creativity. Jones argued tions be affected by the aging workforce? that the accumulation of knowledge over References How can companies that need to inno- time generates a burden of knowledge. In vate adapt to an aging workforce? his words, while later generations have Amabile, T. M. (1996), Creativity in con-

22 NBER Reporter Summer 2006 text: Update to the social psychology of from Impressionism to Cubism,” American Simonton, Dean Keith, “Creativity: creativity, Boulder, CO: Westview Press. Economic Review 91 (no. 4, September Cognitive, Personal, Developmental, and 2001): pp. 1063–71. Social Aspects,” American Psychologist, Amabile, T.M., Schatzel, E.A., Moneta, January 2000. G.B., and Kramer, S.J. (2004), Leader Jones, Benjamin F. “Age and Great behaviors and the work environment for Invention,” Northwestern University Weinberg, Bruce A., “Which Labor creativity: Perceived leader support, The Working Paper, 2005. Economists Invested in Human Capital? Leadership Quarterly, 15:1, pp. 5–32. Geography, Vintage, and Participation Jung-Beeman, Mark, Edward M. in Scientific Revolutions,” Ohio State Amabile, T.M., Barsade, S.G., Mueller, Bowden, Jason Haberman, Jennifer L. University Working Paper, February 2006. J.S., and Staw, B.M (2005), Affect and Frymiare, Stella Arambel-Liu, Richard Creativity at Work, Administrative Greenblatt, Paul J. Reber, and John Zucker, Lynne G., Michael R. Darby, Science Quarterly, 50:3, pp. 367–403. Kounios, “Neural Activity When People and Marilynn B. Brewer, “Intellectual Solve Verbal Problems with Insight,” PloS Human Capital and the Birth of U.S. Folley, Bradley S., and Sohee Park, Biology, vol. 2, issue 4, April 2004. Biotechnology Enterprises,” American “Verbal creativity and schizotypal person- Economic Review, vol. 88, no. 1, March ality in relation to prefrontal hemispheric Kaiser, David, Drawing Theories Apart: 1998. laterality: A behavioral and near-infrared The Dispersion of Feynman Diagrams in optical imaging study,” Schizophrenia Postwar Physics (Chicago: University of Research, 80, 2005. Chicago Press), 2005.

Galenson, David W. “A Portrait of Kaiser, David. “Training Quantum the Artist as a Very Young or Very Old Mechanics: Enrollments and Epistemology Innovator,” NBER Working Paper No. in Modern Physics.” 10515, May 2004. Lerner, Joshua, and Jean Tirole, “The Galenson, David W. and Bruce A. Dynamics of Technology Sharing: Open Weinberg, “Creating Modern Art: The Source and Beyond,” NBER Working Changing Careers of Painters in France Paper No. 10956, December 2004.

*

NBER Reporter Summer 2006 23 Twenty-first Annual Conference on Macroeconomics

The NBER’s twenty-first Annual Dispersion in Business Growth Rates: Discussants: Jordi Gali, MIT and Conference on Macroeconomics, orga- Publicly Traded versus Privately Held NBER, and Christopher A. Sims, nized by NBER Research Associates Firms” Princeton University and NBER Daron Acemoglu, MIT, Kenneth Discussants: Chris Foote, Federal Rogoff, Harvard University, and Michael Reserve Bank of Boston, and Eva Mikhail Golosov, MIT and NBER; Woodford, Columbia University, took Nagypal, Northwestern University Aleh Tsyvinski, Harvard University place in Cambridge on April 7 and 8. and NBER; and Ivan Werning, MIT The program was: Lars Ljungqvist, Stockholm School and NBER, “New Dynamic Public of Economics, and Thomas J. Finance: a User’s Guide” Lawrence J. Christiano and Martin Sargent, New York University and Discussants: Peter A. Diamond, MIT Eichenbaum, Northwestern University NBER, “Indivisible Labor, Human and NBER, and Kenneth L. Judd, and NBER, and Robert Vigfusson, Capital, Lotteries and Personal and NBER Federal Reserve Board, “Assessing Savings: Do Taxes Explain European Structural VARs” Unemployment?” Monika Piazzesi, University of Discussants: Patrick Kehoe, Federal Discussants: Olivier J. Blanchard, MIT Chicago and NBER; and Martin Reserve Bank of Minneapolis, and and NBER, and Edward C. Prescott, Schneider, Federal Reserve Bank Mark W. Watson, Princeton University Arizona State University and NBER of Minneapolis, “Equilibrium Yield and NBER Curves” Troy Davig, Federal Reserve Bank Discussants: Pierpaolo Benigno, New Steven J. Davis, University of Chicago of Kansas City, and Eric M. Leeper, York University and NBER, and John and NBER; John C. Haltiwanger, Indiana University and NBER, Y. Campbell, Harvard University and University of Maryland and NBER; “Fluctuating Macro Policies and the NBER and Ron Jarmin and Javier Miranda, Fiscal Theory”(NBER Working Paper U.S. Census Bureau, “Volatility and No. 11212)

Christiano, Eichenbaum, and run restrictions. Confidence intervals are rise in volatility among publicly traded Vigfusson analyze the quality of VAR- wider in the latter case. Even so, long run firms using the LBD, but show that its based procedures for estimating the identified VARs can be useful for dis- impact is overwhelmed by declining vola- response of the economy to a shock. criminating between competing economic tility among privately held firms. The ris- They focus on two key questions. First, models. ing activity share, higher volatility, and do VAR-based confidence intervals accu- Davis, Haltiwanger, Jarmin, and increasingly volatile character of newly rately reflect the actual degree of sam- Miranda study the distribution of growth listed firms after 1979 explains much of pling uncertainty associated with impulse rates among establishments and firms in the trend increase in volatility among pub- response functions? Second, what is the the U.S. private sector from 1976 onwards. licly traded firms. They also show that size of bias relative to confidence inter- To carry out their study, they exploit the business volatility and dispersion declined vals, and how do coverage rates of confi- recently developed Longitudinal Business much more rapidly in Retail Trade and dence intervals compare to their nominal Database (LBD), which contains annual Services than in Manufacturing. size? They address these questions using observations on employment and pay- To appreciate the role of a “not-so-well- data generated from a series of estimated roll for all business establishments and known aggregation theory” that underlies dynamic, stochastic general equilibrium firms. Their main finding is a large secular Prescott’s (2002) conclusion that higher models. They organize most of their anal- decline in the cross-sectional dispersion of taxes on labor have depressed Europe rela- ysis around a particular question that has firm growth rates and in the average mag- tive to the United States, Ljungqvist and attracted a great deal of attention in the nitude of firm level volatility. Measured in Sargent compare aggregate outcomes for literature: how do hours worked respond the same way as in other recent research, economies with two alternative arrange- to an identified shock? In all of their the employment-weighted mean volatil- ments for coping with indivisible labor: examples, as long as the variance in hours ity of firm growth rates in the private sec- employment lotteries plus complete con- worked attributable to a given shock is tor has declined by more than 40 percent sumption insurance; and individual con- above the remarkably low number of 1 since 1982. This result stands in sharp sumption smoothing via borrowing and percent, structural VARs perform well. contrast to previous findings of rising vol- lending at a risk-free . Under This is true regardless of whether iden- atility for publicly traded firms based on idealized conditions, the two arrange- tification is based on short-run or long COMPUSTAT data. They confirm the ments support equivalent outcomes when

24 NBER Reporter Summer 2006 human capital is not present; when it and produce a stationary rational expecta- aggregate shocks. is present, outcomes are naturally differ- tions equilibrium in which (lump-sum) Piazzesi and Schneider consider how ent. Households’ reliance on personal sav- tax shocks always affect output and infla- the role of inflation as a leading business- ings in the incomplete markets model tion. Tax non-neutralities in the model cycle indicator affects the pricing of nomi- constrains the “career choices” that are arise solely through the mechanism artic- nal bonds. They examine a representative- implicit in their human capital acquisi- ulated by the fiscal theory of the price agent asset-pricing model with recursive tion plans relative to those that can be level. This paper quantifies that mecha- utility preferences and exogenous con- supported by lotteries and consumption nism and finds it to be important in U.S. sumption growth and inflation. They insurance in the complete markets model. data, reconciling a popular class of mon- solve for yields under various assump- Lumpy career choices make the incom- etary models with the evidence that tax tions on the evolution of investor beliefs. plete markets model better at coping with shocks have substantial impacts. Because If inflation is bad news for consumption a generous system of government funded long-run policy behavior determines exis- growth, the nominal yield curve slopes up. compensation to people who withdraw tence and uniqueness of equilibrium, in a Moreover, the level of nominal interest from work. Adding generous government regime-switching environment more accu- rates and term spreads are high in times supplied benefits to Prescott’s model with rate qualitative inferences can be gleaned when inflation news are harder to inter- employment lotteries and consumption from full-sample information than by con- pret. This is relevant for periods such as insurance causes employment to implode ditioning on policy regime. the early 1980s, when the joint dynam- and prevents the model from matching Golosov, Tsyvinski, and Werning ics of inflation and growth was not well outcomes observed in Europe. present a simple dynamic Mirrleesian understood. Davig and Leeper estimate regime- model. There are two main goals for this These papers will appear in an annual switching rules for monetary policy and paper: to review some recent results and volume published by the MIT Press. tax policy over the post-war period in the contrast the Mirrlees approach with the Its availability will be announced in a United States and impose the estimated Ramsey framework in a dynamic setting; future issue of the Reporter. They can also policy process on a model with nominal and to present new numerical results for be found at “Books in Progress” on the rigidities. Decision rules are locally unique a flexible two-period economy featuring NBER’s website.

Innovation Policy and the Economy

The NBER’s seventh annual and NBER, “Is the Pharmaceutical Conference on Innovation Policy and the Industry in a Productivity Crisis?” Erik Brynjolfsson and Xiaoquan Economy took place in Washington on (Michael) Zhang, MIT, “Innovation April 19. The conference was organized Fiona Murray, MIT, and Scott Stern, Incentives for Information Goods” by NBER Research Associates Adam Northwestern University and NBER, B. Jaffe of Brandeis University, Joshua “When Ideas Are Not Free: The Impact Daniel Diermeier, Wallace J. Hopp, Lerner of Harvard University, and Scott of Patents on Scientific Research” and Seyed Iravani, Northwestern Stern of Northwestern University. The University, “Innovating under following papers were discussed: Paula Stephan, Georgia State Pressure — Towards a Science of Crisis University, “Wrapping it up in a Person: Management” Iain M. Cockburn, Boston University The Mobility Patterns of New PhDs”

Rising R and D expenditures and fall- while trends in R and D expenditure are actions costs and other inefficiencies in ing counts of new drug approvals since seriously overstated by failing to account the market for basic research, and by late- 1996 have led many observers to con- for inflation in R and D input costs. Some stage abandonment of drug development clude that there has been a sharp decline of the increase in R and D investment is projects on purely economic grounds. in research productivity in the pharma- a necessary, indeed welcome, response Policies that make “small” markets more ceutical industry over the past decade. to new technological opportunities and attractive, build capacity in translational But a close look at the underlying data, can be expected to deliver a handsome medicine, reduce the cost, time, and Cockburn suggests, shows that these return of innovative drugs in future years. uncertainty of regulatory review, maxi- trends are greatly exaggerated: properly The rising cost per new drug approved is mize access to basic research, and encour- measured, research output is unlikely to nonetheless a serious cause for concern, age greater cooperation and collaborative have fallen as much as these figures imply, particularly where this is driven by trans- research within the industry can all con-

NBER Reporter Summer 2006 25 tribute to greater R and D efficiency. of patents may be indirect; rather than vations. In contrast, they propose and Murray and Stern describe the impact directly affecting behavior through patent analyze a profit-maximizing mechanism of formal intellectual property rights enforcement, scientific conduct may be for bundles of digital goods which is (IPR) on the production and diffusion of influenced by related mechanisms, such more efficient and more accurately pro- “dual knowledge” — ideas that are simul- as material transfer agreements. Not sim- vides innovation incentives for informa- taneously of value as a scientific discovery ply a legal document within a seamless tion goods. Their “statistical couponing and as a useful, inventive construct. They web of cooperation, nor a bludgeon to mechanism” does not rely on the univer- argue that a great deal of knowledge gen- stop scientific progress in its tracks, pat- sal excludability of information goods, erated in academia, particularly in the life ents seem to be changing the “rules of the which creates substantial deadweight loss, sciences, falls into this category (some- game” for scientific exchange, coopera- but instead estimates social value created times referred to as Pasteur’s Quadrant). tion, and credit. from new goods and innovations by offer- The production and diffusion of dual Stephan analyzes data concerning the ing coupons to a relatively small sample of purpose knowledge challenges the prem- placements of new PhDs who had definite representative consumers. They find that ise of most science policy analysis, implic- plans to go to work in industry for the the statistical couponing mechanism can itly based on a clear separation between period 1997-2002. Her data come from operate with less than 0.1 percent of the basic scientific knowledge and applied the Survey of Earned Doctorates over- deadweight loss of the traditional price- knowledge useful in the development of seen by the National Science Foundation. based system, while more accurately align- new technology. Instead, dual knowledge She finds knowledge sources to be heav- ing incentives with social value. simultaneously makes both a basic and an ily concentrated in certain regions and Diermeier, Hopp, and Iravani pro- applied contribution. The authors review states. Moreover, the geographic distribu- pose a rigorous modeling framework qualitative and quantitative evidence tion of knowledge sources, as measured by for characterizing the structural ability relating to the policy challenges raised where PhDs going to work in industry are of organizations to respond quickly and by the production and dissemination of trained, is different than other measures effectively to unanticipated events. As dual knowledge, highlighting three broad of knowledge sources, such as university such, the authors seek to provide a theo- findings. First, rather than facing a funda- R and D-expenditure data, would sug- retical basis for improved crisis manage- mental tradeoff between applied research gest. A major headline is the strong role ment strategies. Their framework con- and more fundamental scientific knowl- played by Midwestern universities, which ceptualizes organizations as adaptive, edge, research agencies can and do invest educate over 26.5 percent of all PhDs responsive networks. Most of the exist- in dual purpose knowledge. Indeed, the going to industry but are responsible for ing models of complex social networks to dual purpose knowledge framework sug- only 21.1 percent of university R and date, however, have not explicitly mod- gests a distinct rationale for public sector D. Stephan finds that only 37 percent of eled human capacity constraints or system involvement in the funding and conduct PhDs trained in science and engineering congestion. As a result, no viable frame- of research: the social impact of a given stay in their state of training. Particularly works exist for investigating the respon- piece of knowledge may be enhanced among certain Midwestern states, many siveness of various organizational struc- when knowledge is produced and dis- students leave for employment on the tures under crisis conditions. The authors closed in accordance with the norms of Coasts. The firms most likely to hire new propose to integrate the social-network the scientific research community (par- PhDs are found in computer and electri- approach to modeling communication ticularly compared to secrecy). Second, cal products, followed by firms in publish- and collaboration with the flow-network within Pasteur’s Quadrant, the increased ing and professional, scientific, and tech- approach from production-systems mod- use of formal IPR seems to be significantly nical services. Almost one out of ten new eling as a way of representing task pro- shaping the structure, conduct, and per- PhDs going to work for industry heads cessing and flow under crisis conditions. formance of both university and indus- to San Jose; 58 percent go to work in one By providing an analytic structure for try researchers. On the one hand, from of twenty cities. The placement data also decisionmaking environments currently the perspective of individual research- suggest that small firms play a larger role viewed as not amenable to formal meth- ers, patenting does not seem to come at in innovation than R and D expenditure ods, this research may improve the per- the expense of scientific publication, and data would suggest. formance of various organizations in both both respond to the process of scientific Innovations can often be targeted the private and public sectors. discovery. However, there is some evi- to be more valuable for some consumers These papers will appear in an annual dence that patent grants may reduce the than others. This is especially true for dig- volume published by the MIT Press. extent of use of knowledge: the citation ital information goods. Brynjolffson and Its availability will be announced in a rate to a scientific article describing a dual- Zhang show that the traditional price sys- future issue of the Reporter. They can also purpose discovery experiences a modest tem not only results in significant dead- be found at “Books in Progress” on the decline after patent rights are granted weight loss, but also provides incorrect NBER’s website. over that knowledge. Finally, the impact incentives to the creators of these inno-

26 NBER Reporter Summer 2006 Conference on Asset Prices and Monetary Policy

The NBER held a conference on Macroeconomic Dynamics, and the Bad News About Inflation Good News Asset Prices and Monetary Policy on Term Structure of Interest Rates” for the Exchange Rate?” May 5 and 6. Organized by NBER Discussant: Jordi Gali, Universitat Discussant: Charles M. Engel, Research Associate John Y. Campbell Pompeu Fabra and NBER University of Wisconsin and NBER of Harvard University, the conference brought together academic research- Jeffrey A. Frankel, Harvard University Roberto Rigobon, MIT and NBER ers with economists from the Federal and NBER, “Commodity Prices, and Brian Sack, Macroeconomic Reserve System, as well as individuals Monetary Policy, and Currency Advisers, LLC, “Noisy Macroeconomic with experience both in academe and in Regimes” Announcements, Monetary Policy, and the policy sphere. The program was: Discussant: Lars E. O. Svensson, Asset Prices” Princeton University and NBER Discussant: Leonardo Bartolini, Federal Simon Gilchrist, Boston University Reserve Bank of New York and NBER, and Masashi Saito, Bank Stephen G. Cecchetti, Brandeis of Japan, “Expectations, Asset Prices, University and NBER, “GDP at Risk: Tommaso Monacelli, Bocconi and Monetary Policy: The Role of A Framework for Monetary Policy Universit, “Optimal Monetary Policy Learning” Responses to Asset Price Movements” with Collateralized Household Debt Discussant: Michael Woodford, Discussant: Andrew Levin, Federal and Borrowing Constraints” Columbia University and NBER Reserve Board Discussant: Hanno Lustig, University of California, Los Angeles and NBER Glenn D. Rudebusch and John C. Panel Discussion: Chair, Martin Williams, Federal Reserve Bank of Feldstein, Harvard University and Monika Piazzesi, University of San Francisco, “Revealing the Secrets NBER Chicago and NBER, and Martin of the Temple: The Value of Publishing , Federal Reserve Board Schneider, New York University, Interest Rate Projections” Laurence Meyer, Macroeconomic “Inflation Illusion, Credit, and Asset Discussant: Marvin Goodfriend, Advisers, LLC Prices” Carnegie Mellon University William Dudley, Goldman Sachs Discussant: Markus Brunnermeier, Princeton University and NBER Hans Dewachter, Katholiek Richard H. Clarida, Columbia Universiteit Leuven, and Marco Lyrio, University and NBER, and Daniel University of Warwick, “Learning, Waldman, Columbia University, “Is

Gilchrist and Saito study the impli- technology growth makes the calculation central bank’s expectations about the cations of financial market imperfections of potential imprecise and thus reduces future path of the policy rate. They show represented by a countercyclical external the benefit of responding to the asset that, in an economy where private agents finance premiums and gradual recogni- price gap. Asset price targeting which have imperfect information about the tion of changes in the drift of technology does not take into account changes in determination of monetary policy, cen- growth for the design of an interest rate potential tends to be welfare reducing. tral bank communication of interest rate rule. Asset price movements induced by The modern view of monetary policy projections can help shape financial mar- changes in trend growth influence balance stresses its role in shaping the entire yield ket expectations and improve macroeco- sheet conditions which in turn determine curve of interest rates in order to achieve nomic performance. the premium on external funds. The pres- various macroeconomic objectives. A Dewachter and Lyrio present a mac- ence of financial market frictions pro- crucial element of this process involves roeconomic model in which agents learn vides a motivation for responding to the guiding financial market expectations of about the central bank’s inflation tar- gap between the observed asset price and future central bank actions. Recently, a get and the output-neutral real interest the potential asset price in addition to few central banks have started to explic- rate. They use this framework to explain responding strongly to inflation. This is itly signal their future policy intentions the joint dynamics of the macroeconomy, because the asset price gap represents dis- to the public, and two of these banks and the term structures of interest rates tortions in the resource allocation induced have even begun publishing their inter- and inflation expectations. Introducing by financial market frictions more -dis nal interest rate projections. Rudebusch learning into the macro model generates tinctly than inflation. Policymakers’ and Williams examine the macroeco- endogenous stochastic endpoints which imperfect information about the drift of nomic effects of direct revelation of a act as level factors for the yield curve.

NBER Reporter Summer 2006 27 These endpoints are sufficiently volatile to thing about how monetary policy is con- number of important insights about the account for most of the variation in long- ducted. Specifically, they show that “bad effects of macroeconomic data releases on term yields and inflation expectations. As news” about inflation – that it is higher monetary policy expectations and asset such, this paper complements the cur- than expected – can be “good news” for prices. However, one puzzling aspect rent macro-finance literature in explain- the nominal exchange rate – which appre- of that literature is that the estimated ing long-term movements in the term ciates on this news – if the central bank responses are quite small. Indeed, these structure without reference to additional has an inflation target that it implements studies typically find that the major eco- latent factors. with a Taylor Rule. The model is one of nomic releases, taken together, account Commodity prices are back. Frankel inflation – not price level – targeting, so for only a small amount of the varia- looks at connections between monetary that in the model a shock to inflation has a tion in asset prices – even those closely policy and agricultural and mineral com- permanent effect on the price level. Since tied to near-term policy expectations. modities. He begins with the monetary PPP holds in the long run of the model, Rigobon and Sack argue that this appar- influences on commodity prices, first for the nominal exchange rate depreciates in ent detachment arises in part from the dif- a large country such as the United States, the long run to an inflation shock, even ficulties associated with measuring -mac then for smaller countries. The claim is though, on impact, it can appreciate in roeconomic news. They propose two new that low real interest rates lead to high real response to this shock. The empirical econometric approaches that allow them commodity prices. The theory is an anal- work in this paper examines point sam- to account for the noise in measured data ogy with Dornbusch overshooting. The pled data on inflation announcements surprises. Using these estimators, they relationship between real interest rates and the reaction of nominal exchange find that asset prices and monetary policy and real commodity prices also is sup- rates in 10-minute windows around these expectations are much more responsive to ported empirically. One channel through announcements for ten countries and sev- incoming news than previously believed. which this effect is accomplished is a neg- eral different inflation measures for the Their results also clarify the set of facts ative effect of interest rates on the desire period July 2001 through March 2005. that should be captured by any model to carry commodity inventories. Frankel Eight of the countries in the study are attempting to understand the interactions concludes with a consideration of the inflation targeters, and two are not. In between economic data, monetary policy, reverse causality: the possible influence the data, the authors indeed find that bad and asset prices. of commodity prices on monetary policy, news about inflation is good news for Monacelli studies optimal monetary under alternative currency regimes. The the nominal exchange rate, and that the policy in an economy with nominal pri- new proposal for PEPI — Peg the Export results are statistically significant. They vate debt, borrowing constraints and price Price Index — is compared (favorably) also find significant differences comparing rigidity. Private debt reflects equilibrium with the popular regime of CPI targeting the inflation targeting countries and the trade between an impatient borrower, by the criterion of robustness with respect two non-inflation targeting countries. For who faces an endogenous collateral con- to changes in the terms of trade such as oil the non-IT countries, there is no signifi- straint, and a patient saver, who engages in price shocks. cant impact of inflation announcements consumption smoothing. Since inflation Cecchetti uses data from a broad on the nominal exchange rate, although can positively affect borrower’s net worth, cross-section of countries to examine the estimated sign is indeed in line with monetary policy optimally balances the GDP at risk and price-level at risk aris- the story here. For each of the IT coun- incentive to offset the price stickiness dis- ing from booms and crashes in equity and tries, the sign is as predicted by the the- tortion with the one of marginally affect- property markets. He shows that the dis- ory and quite significant. Finally, Clarida ing the borrower’s collateral constraint. tribution of GDP and price-level devia- and Waldman study two countries, the He finds that the optimal volatility of tions from their trends both have fat tails, United Kingdom and Norway, in which inflation is increasing in three key param- so the probability of extreme events is there was a clear regime change during eters: 1) the borrower’s weight in the plan- higher than implied by a normal distri- a period for which they have data. They ner’s objective function; 2) the borrower’s bution. Specifically, housing booms cre- study the granting of independence to the impatience rate; 3) the degree of price ate outsized risks of output declines. This Bank of England in 1997 and the shift to flexibility. In general, however, deviations means that policymakers who are intent formal inflation targeting by Norway in from price stability are small for a small on averting catastrophes should react. The 2001. For both countries, the correlation degree of price stickiness. In a two-sector question is: How? between the exchange rate and the infla- version of the model, in which durable Clarida and Waldman make a theo- tion surprise before the regime change price movements can directly affect the retical point and provide some empirical reveal that “bad news about inflation was ability of borrowing, the optimal volatil- support for it: they show in a simple, but bad news about the exchange rate.” After ity of (non-durable) inflation is more size- robust, theoretical monetary exchange the regime change, they find, “bad news able. In the context used here, and relative rate model that the sign of the covari- about inflation is good news about the to simple Taylor rules, the Ramsey-opti- ance between an inflation surprise and the exchange rate.” mal allocation entails a partial smoothing nominal exchange rate can tell us some- The current literature has provided a of real durable goods prices.

28 NBER Reporter Summer 2006 Piazzesi and Schneider consider find in cross-country data. The key mech- These papers will be published by the asset pricing in a general equilibrium anism is that illusionary and smart inves- University of Chicago Press in an NBER model in which some, but not all, agents tors disagree about the level of real inter- conference volume. Its availability will be suffer from inflation illusion. The model est rates, especially when inflation is far announced in a future issue of the NBER predicts that housing booms occur both from its historical average. This disagree- Reporter. They are also available at “Books when inflation is unusually high and ment stimulates borrowing and lending in Progress” on the NBER’s website. when it is unusually low, which they also and drives up the price of collateral.

Firms and the Evolving Structure of Global Economic Activity The NBER held an NBER- Josh Ederington, University of Nick Bloom, Stanford University, and Universities Research Conference on Kentucky, and Phillip McCalman, Raffaella Sadun and John Van Reenan, “Firms and the Evolving Structure of University of California, Santa Cruz, London School of Economics, “It Ain’t Global Economic Activity” in Cambridge “Shaking All Over? International Trade What You Do It’s the Way That You on May 19 and 20. Organizers Andrew and Industrial Dynamics” Do I.T. Investigating the Productivity Bernard of NBER and Dartmouth Discussant: Andres Rodriguez-Clare, Miracle Using Multinationals” College and Marc Melitz of NBER and Pennsylvania State University and Discussant: Stephen Yeaple, University Harvard University chose these papers NBER of Pennsylvania and NBER for discussion: Patrick Conway, University of North Marc-Andreas Muendler, University Jens Arnold, Beata S. Javorcik, and Carolina at Chapel Hill, “Import Price of California, San Diego, and Sascha Aaditya Mattoo, The World Bank, Pressure on Firm Productivity and O. Becker, University of Munich, “The Productivity Effects of Services Employment: The Case of U.S. Textiles” “Margins of Multinational Labor Liberalization Evidence from the Czech Discussant: James Harrigan, Federal Substitution” Republic” Reserve Bank of New York and NBER Discussant: , Discussant: Bruce Blonigen, University Princeton University of Oregon and NBER Albert Park and Xinzheng Shi, University of Michigan; Dean Yang, Huiya Chen, University of California, Rebecca Hellerstein, Federal Reserve University of Michigan and NBER; Davis, and Deborah Swenson, Bank of New York, and Sofia Villas- and Yuan Jiang, National Bureau of University of California, Davis and Boas, University of California, Statistics, China, “Exporting and Firm NBER, “Multinational Firms and New Berkeley, “Arm’s Length Transactions Performance: Chinese Exporters and Chinese Export Transactions” as a Source of Incomplete Cross-Border the Asian Financial Crisis” Discussant: Juan Carlos Hallak, Transmission: The Case of Autos” Discussant: Roberto Rigobon, MIT University of Michigan and NBER Discussant: Gita Gopinath, Harvard and NBER University and NBER

While there is considerable empiri- sectors. When several aspects of services occurs through intra-firm transactions: cal evidence on the impact of liberaliz- liberalization are considered – namely those between domestic and foreign sub- ing trade in goods, the effects of services the presence of foreign providers, priva- sidiaries of a multinational firm. The diffi- liberalization have not been established tization, and the level of competition – culties associated with writing and enforc- empirically. Arnold, Javorcik, and they find that allowing foreign entry into ing a vertical contract are compounded Mattoo examine the link between ser- service industries may be the key chan- when a product must cross a national bor- vice sector reforms and the productiv- nel through which services liberalization der, and this may explain the high rate of ity of manufacturing industries that rely contributes to improved performance of multinational trade across such borders. on service inputs. Their results, based on downstream manufacturing sectors. As Hellerstein and Villas-Boas show that firm-level data from the Czech Republic most barriers to foreign investment today this common cross-border organization for the period 1998-2003, show a posi- are not in goods but in services sectors, of the firm may have implications for the tive relationship between service sector these findings may strengthen the argu- well-documented incomplete transmis- reform and the performance of domes- ment for reform in this area. sion of shocks across borders. They pres- tic firms in downstream manufacturing A growing share of international trade ent new evidence of a positive relationship

NBER Reporter Summer 2006 29 between an industry’s share of multina- increased exit of marginal firms. Conway the differential effect of IT appears to tional trade and its rate of exchange rate uses detailed plant-level information avail- account for almost all of the difference pass-through to prices. They then develop able in the U.S. Census of Manufacturers in total factor productivity between U.S.- a structural econometric model with both and the Annual Survey of Manufacturers owned and all other establishments. This manufacturers and retailers to quantify for the period 1983-2000 to decompose finding holds in the cross section, when how firms’ organization of their activities these effects. He derives the relative con- fixed effects are included, and even when across borders affects their pass-through tribution of technology and import com- a sample of establishments taken over by of a foreign cost shock. They apply the petition to the increase in productivity U.S. multinationals (relative to takeovers model to data from the auto market. and to the decline in employment in by other multinationals and by domestic Counterfactual experiments show why textiles production in the United States firms) is examined. The authors find that cross-border transmission may be much in recent years. He then simulates the the U.S. multinational effect on IT is par- higher for a multinational transaction impact of removal of quota protection on ticularly strong in the sectors that inten- than for an arm’s-length transaction. In the scale of operation of the average plant sively use information technologies (such the structural model, firms’ pass-through and on the incentive for plant closure. as retail and wholesale), the very same of foreign cost shocks is on average 29 Park and his coauthors analyze firm industries that accounted for the U.S.- percentage points lower in arm’s-length panel data to examine how export demand European productivity growth differen- transactions than in multinational trans- shocks associated with the 1997 Asian tial since the mid-1990s. actions because the higher markups from financial crisis affected Chinese export- Multinational labor demand responds a double optimization along the distribu- ers. They construct firm-specific exchange to wage differentials at the extensive mar- tion chain create more opportunity for rate shocks based on the pre-crisis des- gin, when a multinational enterprise markup adjustment following a shock. tinations of firms’ exports. Because the (MNE) expands into foreign locations, Since arm’s-length transactions account shocks were unanticipated and large, they and at the intensive margin, when an for about 60 percent of U.S. imports, this are an ideal instrument for identifying the MNE operates existing affiliates across difference may explain up to 20 percent impact of exporting on firm productiv- locations. Muendler and Becker derive of the incomplete transmission of for- ity and on other aspects of firm perfor- conditions for parametric and nonpara- eign-cost shocks to the United States in mance. The authors find that firms whose metric identication of an MNE model the aggregate. export destinations experience greater cur- to infer elasticities of labor substitution Ederington and McCalman develop rency depreciation have slower growth in at both margins, controlling for location a model of international trade and indus- exports; export growth also increases firm selectivity. Prior studies have rarely found trial evolution. Evolution is driven by productivity, as well as other measures foreign wages or operations to affect the endogenous technology choices of of firm performance. Consistent with employment. The strategy here detects firms, which generate a rich industrial the “learning-by-exporting” hypothesis, salient adjustments at the extensive margin environment that includes the possibility greater exports increase the productivity for German MNEs. With every percent- of a dramatic shakeout. The likelihood, of firms exporting to developed countries age increase in German wages, German magnitude, and timing of this shakeout but not of firms exporting via Hong Kong MNEs allocate 2,000 manufacturing jobs depends not only on the size of the inno- or directly to poorer destinations. to Eastern Europe at the extensive margin vation but also on the structure of pro- Productivity growth in sectors that and 4,000 jobs overall. duction costs. In this setting, trade lib- intensively use information technologies Chen and Swenson study Chinese eralization reduces the likelihood of a (IT) appears to have accelerated much trade between 1997 and 2003 to see shakeout, resulting in a more stable indus- faster in the United States than in Europe how the presence of multinational firms trial structure. However, when shakeouts since 1995, leading to the U.S. “produc- affected the quality, frequency, and - sur arise in global markets, the distribution of tivity miracle.” If this was partly attribut- vival of new export transactions by pri- firm exits can vary widely across countries. able to the superior management/organi- vate Chinese traders. By exploiting the Furthermore, conditions exist so that a zation of U.S. firms (rather than simply richness of the data that come from the shakeout occurs in a closed economy but the U.S. geographical or regulatory envi- fine geographical and product detail, they not in an open economy. The empirical ronment), then we would expect to see show how own-industry multinational evidence presented here is consistent with a stronger association of productivity firm presence helped to stimulate new the prediction that the more internation- with IT for U.S. multinationals located in trade, and to elevate the quality of those ally integrated sectors are less likely to Europe than for other firms. Bloom and trades. In contrast, they find that greater experience a shakeout. his coauthors examine a large panel of concentrations of other industry multina- Theoretical research has predicted U.K. establishments and show that U.S.- tional activity were associated with less- three different effects of increased import owned establishments have a significantly favorable outcomes, as one would predict competition on plant-level behavior: higher productivity of IT capital than if multinational presence brought with reduced domestic production and sales; either non-U.S. multinationals or domes- it local factor market congestion, or ele- improved average efficiency of plants; and tically owned establishments do. Indeed, vated levels of competition.

30 NBER Reporter Summer 2006 Bureau News

National Security Working Group The NBER’s Working Group on and NBER; David Fisman, Princeton James Dertouzos, RAND National Security met in Cambridge on University; and Rakesh Khurana Corporation, “Recruiter Missioning, February 24 and 25. NBER President and Julia Galef, Harvard University, Market Quality, Recruiter Effort, and and Research Associate Martin Feldstein, “Estimating the Value of Connections Enlistment” who directs the group, organized the to Vice-President Cheney” meeting, at which the following topics David Loughran, RAND Corporation, were discussed: Alexander Gelber, Harvard University, “Earnings Loss of Activated Reservists” “Military Enlistments: A Study of Eli Berman, University of California, Compensating Differentials and Labor James Hosek, RAND Corporation, San Diego and NBER, and David D. Supply” “Analysis of Reserve Retirement Laitin, Stanford University, “Hard Reform” Targets: Theory and Evidence on Edward Miguel, University of Suicide Attacks”(NBER Working Paper California, Berkeley and NBER, and John T. Warner and Curtis J. Simon, No. 11740) John Bellows, University of California, Clemson University, “Uncertainty Berkeley, “War and Institutions in about Job Match Quality and Youth Joel Slemrod, University of Michigan Sierra Leone” Turnover: Evidence From U.S. Military and NBER, and Naomi Feldman, Attrition” Ben-Gurion University, “War, Social Neil F. Johnson, Oxford University; Identity, and Taxation: Capitalizing and Michael Spagat, University of Claude Berrebi, RAND Corporation, Patriotism Through Voluntary Tax London; Jorge Restrepo, Universidad and Esteban F. Klor, Hebrew Compliance” Javeriana; Oscar Becerra and Nicolas University of Jerusalem, “The Impact Suárez, Conflict Analysis Resource of Terrorism Across Industries: An Eugene N. White, Rutgers University Center, Bogotá, Colombia; Juan Empirical Study” and NBER; Kim Oosterlinck, Free Camilo Bohórquez, Elvira Maria University of Brussels; and Filippo Restrepo, and Roberto Zarama, Darius Lakdawalla, RAND Occhino, Rutgers University, “How Universidad de los Andes, Bogotá, Corporation and NBER, and Eric Occupied France Financed Its Own Colombia; “Universal Patterns Talley, RAND Corporation, “Optimal Exploitation in World War II” Underlying Ongoing Wars and Liability for Terrorism” Terrorism” Raymond Fisman, Harvard University

Berman and Laitin model the choice as a tactic, predicting that suicide will be ceived military threat to a country, and the of tactics by rebels, bearing in mind that used when targets are well protected and willingness to pay in turn affects the mar- a successful suicide attack imposes the when damage is great. Those predictions ginal efficiency cost of raising resources, ultimate cost on the attacker and the are consistent with the patterns described both via taxes and conscription. A model organization. They first ask what a - sui above. The model has testable implica- of the interactions generates predictions cide attacker would have to believe to be tions for tactic choice of terrorists and for about the effect of the external threat on deemed rational. They then embed the damage achieved by different types of ter- military spending, non-military spend- attacker and other operatives in a club- rorists, which the authors find to be con- ing, and the share of military resources good model that emphasizes the func- sistent with the data. raised via conscription, as well as predic- tion of voluntary religious organizations Feldman and Slemrod explore the tions concerning the effect of wars and as providers of benign local public goods. relationship among war, government external threats on the willingness to vol- The sacrifices that these groups demand financing, and citizens’ willingness to vol- untarily pay taxes. The authors test these make clubs well suited for organizing sui- untarily comply with tax and other obli- predictions empirically using cross-coun- cide attacks, a tactic in which defection by gations because of social identity. Their try data from 1970 to the present on operatives (including the attacker) endan- motivating idea is that the willingness to government finances, the Correlates of gers the entire organization. The model voluntarily comply with obligations to the War Militarized Interstate Disputes data- also analyzes the choice of suicide attacks government may be a function of the per- set, and data on attitudes toward tax eva-

NBER Reporter Summer 2006 31 sion and military service from the World when a profession is particularly pleas- Becerra, Bohórquez, Johnson, Values Survey. ant, the wage tends to be low but the sup- Restrepo, Spagat, Suárez, and Zarama Most studies of war finance have ply of labor high. In studying the supply report a remarkable universality in the focused on how belligerent powers funded of recruits to the military, Gelber solves frequency of violence arising in two high- hostilities with their own resources. The this problem by instrumenting for the profile ongoing wars, and in global terror- collapse of the Third Republic in1 940 left endogenous military wage — he uses a ism. Their results suggest that these quite Berlin in control of a nearly equally pow- statutory formula that usually governs different conflict arenas currently feature erful industrial economy. The resources increases in the wage. Using Department a common type of enemy, that is the var- extracted from France by the Nazis rep- of Defense administrative data on all 3.5 ious insurgent forces are beginning to resent perhaps the largest international million enlistment contracts signed by operate in a similar way regardless of their transfer. Occhino, Oosterlinck, and recruits over 16 recent years, he esti- underlying ideologies, motivations, and White assess the welfare costs of the poli- mates that elasticities of labor supply with the terrain they operate in. The authors cies that the French chose to fund pay- respect to wages are quite high. Ordinary provide a theory to explain their main ments to Germany and alternative plans least squares regressions sometimes show observations which treats the insurgent with a neoclassical growth model that a negative or insignificant impact of the forces as a generic, self-organizing net- incorporates essential features of the occu- military wage on enlistments, but instru- work, dynamically evolving through the pied economy and the postwar stabiliza- mental variables regressions show a posi- continual coalescence and fragmentation tion. Although the mix of taxes, bonds, tive and significant effect. The high elas- of attack units. and seigniorage employed by Vichy, ticities imply that enlarging the military Dertouzas documents research meth- resembles methods chosen by belliger- would be substantially less costly than ods, findings, and policy conclusions from ents, the French economy sharply con- previous estimates have suggested. a project analyzing human resource man- tracted. Vichy’s postwar debt overhang Bellows and Miguel study the after- agement options for improving recruiting would have required substantial budget math of the brutal 1991–2002 Sierra production. He details research designed surpluses; but inflation, which erupted Leone civil war. One notable aspect of to develop new insights to help guide after Liberation, reduced the debt well their project is the availability of extensive future recruiter-management policies. below its steady state level and redistrib- household data on conflict experiences The research involves econometric anal- uted the adjustment costs. The Marshall and local institutions (broadly defined) yses of three large and rich datasets. The Plan played only a minor direct role, and for Sierra Leone. They first confirm that first analysis compares the career paths of international credits helped to substan- there are no lingering effects of war vio- enlisted personnel, including recruiters. tially lower the nation’s burden. lence on local socioeconomic conditions, The second analyzes individual recruiter Fisman, Fisman, Galef, and Khurana a mere three years after the end of the characteristics and links those character- estimate the value of personal ties to civil war, in line with the existing war istics with their productivity, control- Richard Cheney through three distinct impact studies. They find that measures ling for a variety of independent factors. approaches that have been used recently of local community mobilization and col- Finally, the research focuses on station- to measure the value of political connec- lective action — including the number level recruiting outcomes, paying close tions. Their proxies for personal ties are of village meetings and the voter regis- attention to the management options that based on corporate board linkages that tration rate — are significantly higher in can affect recruiter production and effort. are prevalent in the network sociology areas that experienced more war violence, These empirical analyses demonstrate that literature. They measure the value of these conditional on extensive prewar and geo- various types of human resource man- ties using three event studies: 1) market graphic controls. In other words, if any- agement policies can be very helpful in reaction of connected companies to news thing, areas where there was greater vio- meeting the Army’s ambitious recruit- of Cheney’s heart attacks; 2) correlation lence against civilians during the recent ing requirements. For example, the find- of the value of connected companies with war have arguably better local outcomes. ings have implications for human resource probability of Bush victory in 2000; and These findings speak to the remarkable policies in the areas of selecting soldiers 3) correlation of the value of connected resilience of ordinary Sierra Leoneans. for recruiting duty, assigning recruiters companies with the probability of war The authors view these results as comple- to stations, missioning to promote equity in Iraq. In all cases, the value of ties to mentary to the other recent studies of across recruiters, missioning to increase Cheney is precisely estimated as zero. The war, none of which examines local insti- recruiter productivity, using promotions authors interpret this as evidence that tutional or political economy impacts. to motivate and reward recruiters, and U.S. institutions are effective in control- These findings echo the claims of other screening out recruiters who are under- ling rent seeking through personal ties observers of Sierra Leone (including Keen producing. Although the gains from any with high-level government officials. 2005 and Ferme 2002) who also argue individual policy appear to be modest, the A general problem facing estimates that the war increased political aware- cumulative benefits of implementing mul- of the elasticity of labor supply to a pro- ness and mobilization and generated far- tiple policies could save the Army over fession is that the wage is “endogenous”: reaching institutional changes. $50 million in recruiting resources on an

32 NBER Reporter Summer 2006 annual basis. This work will interest those the congressional proposals was found to reduce attrition longer term. involved in the day-to-day management be cost effective. Berrebi and Klor use scoring match- of recruiting resources as well as research- Simon and Warner model first-term ing techniques and event study analysis to ers and analysts engaged in analyses of enlisted attrition as the outcome of a elucidate the impact of terrorism across military enlistment behavior. process of learning about true tastes for different economic sectors. Using the Loughran describes research using a service. Attrition occurs when recruits Israeli-Palestinian conflict as a case study, sample of Army and Air Force reservists learn that their true tastes for service are they differentiate between Israeli com- activated in 2001 and 2002 for the Global sufficiently lower than their forecasted panies that belong to the defense, secu- War on Terrorism. It combines informa- tastes as to render their gain to staying rity, or anti-terrorism related industries tion on their civilian earnings from Social negative. Preference shocks might arise and other companies. Their findings show Security Administration (SSA) data for from different sources, but in this model that, whereas terrorism has a significant 2001 with information on military earn- they arise when youth are ill informed negative impact on non-defense-related ings from Department of Defense (DoD) about the actual on-the-job effort require- companies, the overall effect of terrorism administrative files to estimate the effect ment and (optimistically) understate this on defense and security-related companies of activation on their earnings. This mea- requirement prior to entry. Larger mis- is significantly positive. Similarly, using sure of military earnings includes pays, takes in forecasting the effort require- panel data on countries’ defense expen- allowances, and an approximation to ment lead to higher early attrition, but ditures and imports from Israel, they find the value of the federal tax preference a steeper decline in attrition relative to that terror fatalities in Israel have a posi- accorded military allowances and mili- the better-informed groups. The authors’ tive effect on Israeli exports of defense tary pay received while serving in a com- empirical analysis provides evidence sup- products. These results suggest that the bat zone. The results on earnings and porting this view of the attrition process. expectation of future high levels of ter- activation reported in this document are More educated groups, males, and non- rorism has important implications for early and subject to a number of impor- whites are estimated to have lower, flat- resource allocation across industries. tant caveats, but the estimates do imply ter attrition profiles, a result consistent Lakdawalla and Talley analyze the less prevalent and severe earnings losses with the model. The model also explains normative role for civil liability in align- among activated reservists than do esti- the empirical finding of lower and - flat ing terrorism pre-caution incentives, mates derived from DoD survey data. ter attrition profiles for individuals who when the perpetrators of terrorism are Congress has put forward several entered and remained longer in Delayed unreachable by courts or regulators. The proposals to increase the generosity of Entry Program (DEP). This last result authors consider the strategic interaction the retirement benefits payable to reserv- has important implications for current among targets, subsidiary victims, and ter- ists. The proposals have the potential military manpower policy. The length rorists within a sequential, game-theoretic to affect reserve retention behavior, yet and lethality of the second Iraq war has model. Their model reveals that, while an also could create cross effects on reten- strained the existing force, the Army in “optimal” liability regime indeed exists, its tion in the active-duty force. Hosek and particular, which along with the Marine features appear at odds with conventional his colleagues, Beth Asch and Daniel Corps has borne the brunt of the con- legal templates. For example, it frequently Clendenning, developed a dynamic pro- flict. As public support for the mission in prescribes damages payments from seem- gramming model of active and reserve Iraq has declined, the Army has missed ingly unlikely defendants, directing them retention, estimated it on actual data, its recruiting targets in recent months. to seemingly unlikely plaintiffs. The chal- and used it to simulate the effects of the In response, the Army has reduced the lenge of introducing such a regime using proposals. The most generous proposal, time that newly signed recruits spend in existing tort law doctrines, therefore, which starts retirement benefits as soon the DEP in order to place them in ser- is likely to be prohibitive. Instead, the as the individual leaves the reserves after vice more quickly. In addition to reducing authors argue, efficient precaution incen- 20 or more years of active and reserve ser- the pipeline of future manpower supply, tives may be best provided by alternative vice, increased mid-career retention in the the empirical results here suggest that the policy mechanisms, such as a mutual pub- actives but increased the outflow from the result will also entail higher attrition in lic insurance pool for potential targets of actives to the reserves. Reserve retention service. The Army recognizes the problem terrorism, coupled with direct compensa- increased prior to 20 years but decreased and has adjusted basic training to reduce tion to victims of terrorist attacks. afterwards, by so much that expected attrition. It remains to be seen whether years of service declined on net. None of this adjustment in training policies will

NBER Reporter Summer 2006 33 Entrepreneurship Working Group The NBER’s Entrepreneurship Of Internet-Related IPOs” 11 and the Effect on Firm Outcomes?” Working Group met in Cambridge on Discussant: Baruch Lev, New York Discussant: Karin Thorburn, March 10. NBER Research Associate University Dartmouth College Josh Lerner of the Harvard Business School, who directs this group, orga- David G. Blanchflower, Dartmouth Robert W. Fairlie, University of nized the following program: College and NBER, and Jon California, Santa Cruz, and Harry A. Wainwright, NERA Economic Krashinsky, University of Toronto, Entrepreneurship in the Service Sector Consulting, “An Analysis of the Impact “Liquidity Constraints, Household of Affirmative Action Programs on Wealth, and Entrepreneurship Francine Lafontaine, University of Self-Employment in the Construction Revisited” Michigan, and Renata Kosová, George Industry”(NBER Working Paper No. Discussant: Annamaria Lusardi, Washington University, “Firm Survival 11793) Dartmouth College and NBER and Growth in Retail and Service Discussant: Scott Wallsten, AEI- Industries: Evidence from Franchised Brookings Joint Center for Regulatory Boyan Jovanovic, New York University Chains” Studies and NBER, and Balazs Szentes, Discussant: Eric Van Den Steen, MIT University of Chicago, “An Estimated Entrepreneurial Finance Model of the Market for Venture Iain M. Cockburn, Boston University Capital” and NBER, and Stefan Wagner, Antoinette Schoar, MIT and NBER, Discussant: Lucy White, Harvard INNO-tech, “Patents and the Survival “ Judge Specific Differences in Chapter University

Kosová and Lafontaine analyze the size no longer affect exit. However, they acquisition compared to exit via delisting survival and growth of franchised chains find that chain size continues to have a from the exchange because of business fail- using an unbalanced panel data set that negative effect on chain growth, a result ure. Firms that applied for more patents covers about 1000 franchised chains each that implies that chains converge in size were less likely to be acquired, although year from 1980 to 2001. The empirical to chain-specific levels. if they obtained unusually highly cited literature on firm survival and growth has Cockburn and Wagner examine the patents, they meght be a more attractive focused almost exclusively on manufac- effect of patenting on the survival pros- acquisition target. These findings do not turing. This analysis allows the authors pects of 356 internet-related firms that hold true for business method patents, to explore whether chain age and size made an initial public offering on the which do not appear to confer a survival have the same effect on the survival and NASDAQ at the height of the stock mar- advantage. growth of retail and service chains as firm ket bubble of the late 1990s. By March Blanchflower and Wainwright and establishment age and size have been 2005, almost two thirds of these firms find that despite the existence of vari- found to have on survival and growth in had delisted from the exchange. Changes ous affirmative action programs designed manufacturing. In addition, while the in the legal environment in the United to improve the position of women and researchers focus on the effect of age and States in the 1990s made it much easier minorities in public construction, little size as the prior literature has done, their to obtain patents on software, and ulti- has changed in the last 25 years. They large and long panel data set allows them mately, on business methods, although show that where race-conscious affirma- to control for the first time for chain-spe- less than half of the firms in the sample tive action programs exist, they appear to cific effects as well as for other chain char- obtained, or attempted to obtain, patents. generate significant improvements: when acteristics that might affect chain survival For those that did, the authors hypoth- these programs are removed or replaced and growth. They find that controlling esize that patents conferred competitive with race-neutral programs, the utiliza- for chain-level unobserved heterogeneity advantages that translated into higher tion of minorities and women in public is statistically warranted, and affects the probability of survival, although they may construction declines rapidly. They also conclusions they reach on the effect of also simply have been a signal of firm qual- show that the programs have not helped chain age and size in our regressions. They ity. Controlling for other determinants of minorities to become self-employed or also find that other chain characteristics firm survival, such as age, venture-capi- to raise their earnings over the period affect the survival and growth of individ- tal backing, financial characteristics, and 1979–2004, using data from the Current ual chains. Finally, their long panel allows stock market conditions, patenting is pos- Population Survey and the Census, but them to examine a subsample of mature itively associated with survival. Quite dif- have improved the position of white chains, for which they find that age and ferent processes appear to govern exit via females. There has been a growth in incor-

34 NBER Reporter Summer 2006 porated self-employment rates of white shows that the use of cash collateral and and Jovanovic (1989) reveals that the two women in construction such that cur- the extension of the exclusivity period groups face different incentives, and thus rently their rate is significantly higher increase a firm’s likelihood of re-filing for have different solutions to the entrepre- than that of white men. The data are sug- bankruptcy. Finally, based on the judge neurial decision. Second, they use micro- gestive of the possibility that some of fixed effects, she also creates an aggregate data from matched Current Population these companies are “fronts” which are index to measure the pro-debtor (pro- Surveys (1993–2004) to demonstrate that actually run by their white male spouses creditor) friendliness of the judges. She unanticipated housing appreciation mea- or sons to take advantage of the affirma- provides suggestive evidence that a pro- sured at the MSA-level is a significantly tive action programs. management bias leads to increased rates positive determinant of entry into self- Schoar uses information on Chapter of re-filing and lower post-bankruptcy employment. In addition, they perform a 11 filings for almost 5000 private- com credit ratings. duration analysis to demonstrate that pre- panies across five district courts in the Hurst and Lusardi (2004) recently entry assets are an important determinant United States between 1989 and 2003. challenged the long-standing belief that of entrepreneurial longevity. For each case, she codes the entire docket, liquidity constraints are important causal Jovanovic and Szentes model the in particular all of the decisions that the determinants of entry into self-employ- market for venture capital. VCs have the judge made during a Chapter 11 pro- ment. They demonstrated that the oft- expertise to assess the profitability of proj- cess. She first establishes that while there cited positive relationship between entry ects, and have liquidity to finance them. are some significant differences across rates and assets is actually unchanging as The scarcity of VCs enables them to inter- districts in the types of firms that file assets increase from the first to the 95th nalize their social value, so that the com- for Chapter 11, within districts, cases percentile of the asset distribution, but petitive equilibrium is socially optimal. appear to be assigned randomly to judges. rise drastically after this point. They also This optimality obtains on an open set of She then estimates judge-specific fixed applied a new instrument, unanticipated parameter values. The scarcity of VCs also effects to analyze whether judges differ changes in house prices, for wealth in the leads to an equilibrium return on venture systematically in their Chapter 11 rulings. entry equation, and showed that instru- capital higher than the market rate, but She finds very strong and economically mented wealth is not a significant deter- the preliminary estimates here show this significant differences across judges in minant of entry. Fairlie and Krashinsky excess return to be negligible. The abil- their propensity to grant or deny specific reinterpret these findings: first, they dem- ity to earn higher returns makes VCs less motions. Some judges appear to rule per- onstrate that bifurcating the sample into patient when waiting for a project to suc- sistently more favorably towards allow- workers who enter self-employment after ceed; this explains why companies backed ing the use of cash collateral, lifting the job loss and those who do not reveals by venture capitalists reach IPOs earlier automatic stay, or conversion of cases into steadily increasing entry rates as assets than other start-ups and why they are other chapters, such as 7. Next, she uses increase in both subsamples. They argue worth more at IPO. the estimated judge fixed effects to instru- that these two groups merit a separate ment for the exogenous variation in the analysis, because a careful examination of propensity to grant a specific motion. She the entrepreneurial choice model of Evans

NBER Director John Lipsky to the IMF NBER Director-at-Large John Lipsky, succeeds Anne O. Krueger, an NBER from 1974-84. In 1984 he joined Salomon who was elected to the Board in 1998 Research Associate and international econ- Brothers, where he spent 13 years. He then and became a member of the Executive omist, in that position. moved to Chase Manhattan Bank, serving Committee in 2002, will become first dep- Lipsky is currently vice chairman as Chief Economist for three years. He was uty managing director of the International of J.P.Morgan Chase & Company, but appointed chief economist of J.P. Morgan Monetary Fund on September 1. Lipsky worked at the IMF earlier in his career, Chase when the two banks merged.

NBER Researcher to Head Philadelphia Fed NBER Research Associate Charles Administration, University of Rochester, a member of the NBER’s Program on I. Plosser, a professor of econom- has been named president of the Federal Economic Fluctuations and Growth. ics and former dean of the William E. Reserve Bank of Philadelphia. He takes Simon Graduate School of Business office on August 1. Plosser had been

NBER Reporter Summer 2006 35 Labor Studies

The NBER’s Program on Labor Sandler, University of Maryland, Labor Market Outcomes and Social Studies met in Cambridge on March 17. “The Changing Impact of Fathers on Behavior”(NBER Working Paper No. NBER Research Associates Lawrence Women’s Occupational Choices” 12006) F. Katz and Richard B. Freeman, both of Harvard University, organized this James J. Heckman, University of Hanh Ahee, Stanford University, program: Chicago and NBER, and Jora Stixrud and Ulrike Malmendier, Stanford and Sergio Urzua, University of University and NBER, “Biases in the Judith K. Hellerstein, University of Chicago, “The Effects of Cognitive Market: the Case of Overbidding in Maryland and NBER, and Melinda and Non-Cognitive Abilities on Auctions”

Over the past century, the labor force daughters. They show that this is a phe- that explains schooling choices, wages, participation rate of women has increased nomenon unique to women, as it should employment, work experience, and choice dramatically. Hellerstein and Sandler be if it is a response to rising female labor of occupation explains these behavioral examine one potential ramification of force participation rates. The magnitude outcomes. this, namely whether the transmission of of the shift in women working in their Ahee and Malmendier argue that occupation-specific skills between fathers fathers’ occupations that results from individual biases inducing overpayment and daughters has increased. They develop increased transmission is large — about are exacerbated in auctions. If consumers a model of intergenerational human capi- 20 percent of the total increase in the are heterogeneous in their ability to iden- tal investment in which increased labor probability a woman enters her father’s tify the lowest-price item of a given quality, force participation by women gives fathers occupation over our sample period — and then the auction mechanism will system- more incentives to invest in daughters’ this is an estimate that they argue is likely atically select as winners those consum- skills that are specific to the fathers’ occu- a lower bound. ers whose estimate is most biased upward. pations. As a result, daughters are more Heckman, Stixrud, and Urzua estab- Using a novel dataset on eBay auctions of likely to enter the labor market and to lish that a low-dimensional vector of cog- a popular board game, the authors find take up their fathers’ occupations. Testing nitive and noncognitive skills explains a that buyers neglect lower prices once they whether the transmission of occupa- variety of labor market and behavioral have started bidding. In 51 percent of all tion-specific skills between fathers and outcomes. For many dimensions of social auctions, the price is higher than the”buy- daughters has increased is confounded by performance, cognitive and noncognitive it-now” price at which the same good is the fact that occupational upgrading of skills are equally important. Their analy- available for immediate purchase from women alone will generate an increased sis addresses the problems of measure- the same website. However, only 12 per- probability over time that women work ment error, imperfect proxies, and reverse cent of bidders systematically overbid. in their fathers’ occupations. The authors causality that plague conventional stud- The authors also find that prices are more show that, under basic assumptions of ies of cognitive and noncognitive skills likely to be above the buy-it-now price in assortative mating, a comparison of the that regress earnings (and other out- longer auctions, auctions with more bids, rates of change over time in the probabil- comes) on proxies for skills. Noncognitive and if the seller’s item description explic- ity that a woman enters her father’s occu- skills strongly influence schooling deci- itly mentions the (higher) retail price of pation relative to her father-in-law’s occu- sions, and also affect wages given school- the manufacturer. Experience does not pation can be used to test whether there ing decisions. Schooling, employment, diminish the suboptimal bidding behav- has been increased transmission of occu- work experience, and choice of occupa- ior. Instead, high experience is correlated pation-specific human capital. Using data tion are affected by latent noncognitive with more distortion, such as higher bid- for the birth cohorts of 1909–77 contain- and cognitive skills. These authors study ding in auctions where the manufacturer’s ing information on womens’ occupations a variety of correlated risky behaviors, price is mentioned. The latter result sug- and the occupations of their fathers and such as teenage pregnancy and marriage, gests that overbidding reflects individual fathers-in-law, Hellerstein and Sandler smoking, marijuana use, and participa- biases rather than search cost or other demonstrate an increase in occupation- tion in illegal activities. They find that the standard explanations for suboptimal pur- specific transmission between fathers and same low-dimensional vector of abilities chase decisions.

36 NBER Reporter Summer 2006 Productivity Program Meeting The NBER’s Program on Produc- Evidence from U.S. States” Evidence from Colombian Plants” tivity met in Cambridge on March 17. Discussant: Chad Syverson, University Program Director Ernst Berndt of MIT Carol A. Corrado, Federal Reserve of Chicago and NBER organized the meeting, where the fol- Board; Charles R. Hulten, University lowing papers were discussed: of Maryland and NBER; and Daniel Johannes Van Biesebroeck, University E. Sichel, Federal Reserve Board, of Toronto and NBER, “Wages Equal Bruce A. Weinberg, Ohio State “Intangible Capital and Economic Productivity: Fact or Fiction” University and NBER, “Which Growth”(NBER Working Paper No. Discussant: Wayne Gray, Clark Labor Economists Invested in Human 11948) University and NBER Capital? Geography, Vintage, and Participation in Scientific Revolutions” Marcela Eslava, Universidad de Los Boyan Jovanovic, New York University Andes; John Haltiwanger, University and NBER, and Chung-Yi Tse, David H. Autor, MIT and NBER; of Maryland and NBER; Adriana “Creative Destruction in Industries” William R. Kerr, Harvard University; Kugler, University of Houston Discussant: Shane Greenstein, and Adriana D. Kugler, University of and NBER; and Maurice Kugler, Northwestern University and NBER Houston and NBER, “Do Employment University of Southhampton, “Factor Protections Reduce Productivity? Adjustments After Deregulation: Panel

Weinberg studies how proximity and they find that wrongful-discharge pro- intangible assets makes a significant differ- vintage are related to innovation, using tections significantly reduce employ- ence in the observed patterns of U.S. eco- evidence from the human capital revolu- ment flows. Moreover, analysis of nomic growth. The rate of change of out- tion in labor economics. He finds a strong plant-level data provides evidence of put per worker increases more rapidly effect of geography on the probability of capital deepening and a decline in when intangibles are counted as capital, making a contribution and on the nature total factor productivity following the and capital deepening becomes the unam- of the contribution. Contributors to the introduction of wrongful-discharge biguously dominant source of growth in human capital paradigm are significantly protections. This last result is poten- labor productivity. The role of multifactor more likely to have studied at the University tially quite important, suggesting that productivity is correspondingly dimin- of Chicago or Columbia University and mandated employment protections ished, and labor’s income share is found to have been in graduate school in the reduce productive efficiency, as theory to have decreased significantly over the early years of the human capital revolu- would suggest. However, the analysis last 50 years. tion, earning their doctorates during the also presents some puzzles including, Eslava, Haltiwanger, Kugler, and mid-1960s. These results also indicate that most significantly, evidence of strong Kugler analyze employment and capi- a small number of contributors played a employment growth following adop- tal adjustments using a panel of plants large role in the development of human tion of dismissal protections. In light from Colombia. They allow for nonlin- capital, especially at the beginning. of these puzzles, the authors read their ear adjustment of employment to reflect Theory predicts that mandated findings as suggestive but tentative. not only adjustment costs of labor but employment protections may reduce Published macroeconomic data also adjustment costs of capital, and productivity by distorting production traditionally exclude most intangi- vice versa. Using data from the Annual choices. Firms facing (non-Coasean) ble investment from measured GDP. Manufacturing Survey, which include worker dismissal costs will curtail hiring This situation is beginning to change, plant-level prices, they generate mea- below efficient levels and retain unpro- but the estimates here suggest that as sures of plant-level productivity, demand ductive workers, both of which should much as $800 billion is still excluded shocks, and cost shocks, and use them to affect productivity. Autor, Kerr, and from U.S. published data (as of 2003), measure desired factor levels. They then Kugler use the adoption of wrong- and that this leads to the exclusion estimate adjustment functions for capi- ful-discharge protections by U.S. state of more than $3 trillion of business tal and labor as a function of the gap courts over the last three decades to intangible capital stock. To assess between desired and actual factor lev- evaluate the link between dismissal the importance of this omission, co- els. As in other countries, they find non- costs and productivity. Drawing on authors Corrado, Hulten, and Sichel add linear adjustments in employment and establishment-level data from the intangible capital to the standard sources- capital in response to market fundamen- Annual Survey of Manufacturers and of-growth framework used by the BLS, tals. In addition, they find that employ- the Longitudinal Business Database, and find that the inclusion of our list of ment and capital adjustments reinforce

NBER Reporter Summer 2006 37 each other, in that capital shortages ated with them. He finds that equality productivity effects. Localized labor reduce hiring and labor shortages reduce holds strongly in Zimbabwe (the most markets and imperfect substitutabil- investment. Moreover, they find that the developed country in the sample), but ity of different worker-types provide a market-oriented reforms introduced in not at all for Tanzania (the least devel- partial explanation. Colombia after 1990 increased employ- oped country). The results for Kenya are Most industries go through a “shake- ment adjustments, especially on the job intermediate. Differences between wage out” phase during which the number destruction margin, while reducing capi- and productivity premiums are most pro- of producers in the industry declines. tal adjustments. Finally, they find that nounced for characteristics that are clearly Industry output generally continues to while completely eliminating frictions related to human capital, such as school- rise, though, which implies a reallocation from factor adjustments would yield a ing, training, experience, and tenure. of capacity from exiting firms to incum- dramatic increase in aggregate productiv- Moreover, where the wage premium bents and new entrants. Thus, shakeouts ity through improved allocative efficiency, differs from the productivity benefit, seem to be classic creative destruction epi- the reforms introduced in Colombia gen- general human capital tends to receive sodes. Shakeouts of firms tend to occur erated relatively modest improvements. a wage return that exceeds the produc- sooner in industries where technological Using a matched employer-employee tivity return, and the reverse holds for progress is more rapid. Existing models data set of manufacturing plants in three more specific human capital invest- do not explain this. In fact, the relation- sub-Saharan countries, Van Biesebroeck ments. Schooling tends to be over- ship emerges in a vintage-capital model compares the marginal productivity of rewarded, even though most of the in which shakeouts of firms accompany different categories of workers with the productivity benefit comes from job the replacement of capital, and in which wages they earn. In each country, he training. Wages tend to rise with expe- a shakeout is the first replacement echo observes approximately 135 firms and rience, even though productivity is of the capital created when the industry an average of 5.5 employees per firm. mostly increasing in tenure. Sampling is born. Jovanovic and Chung-Yi fit the Under certain conditions, the wage pre- errors, nonlinear effects, and non- model, with some success, to the Gort- miums for worker characteristics should wage benefits are rejected as explana- Klepper data. equal the productivity benefits associ- tions for the gap between wage and

Cohort Studies The NBER’s Working Group on “Long-Term Care of the Disabled Pittsburgh and NBER, and Karen Cohort Studies, directed by Dora Costa Elderly: Do Children Increase Clay, Carnegie Mellon University, of MIT, met in Cambridge on March Caregiving by Spouses?” “Deprivation and Disease in Early 24. These papers were discussed: Twentieth-Century America” Hoyt Bleakley, University of Robert A. Pollak, Washington Chicago, “Malaria in the Americas: A Chulhee Lee, Seoul National University and NBER; Liliana E. Retrospective Analysis of Childhood University, “Socioeconomic Differences Pezzin, Medical College of Wisconsin; Exposure” in Wartime Morbidity and Mortality of and Barbara S. Schone, Agency for Black Union Army Soldiers” Healthcare Research and Quality, Werner Troesken, University of

Do adult children influence the care may respond by not providing future care ication campaigns in the United States that elderly parents provide for each other? for the nondisabled spouse when neces- (circa 1920), and in Brazil, Colombia, Pezzin, Pollak, and Schone develop two sary. Joint children act as a commitment and Mexico (circa 1955), with a specific models in which the anticipated behavior mechanism, increasing the probability goal of measuring how much childhood of adult children provides incentives for that elderly spouses will provide care; exposure to malaria depresses labor pro- elderly parents to increase care for their stepchildren may provide weaker incen- ductivity. These eradication campaigns disabled spouses. The “demonstration tives for spousal care. Using data from the happened because of advances in medi- effect” assumes that children learn from Health and Retirement Study, the authors cal and public-health knowledge, which a parent’s example that family caregiving find some evidence that spouses provide mitigates concerns about reverse causal- is appropriate behavior. For the “punish- more care when they have children with ity of the timing of eradication efforts. ment effect,” if the nondisabled spouse strong parental attachment. Bleakley collects data from regional fails to provide spousal care, then children Bleakley considers the malaria-erad- malaria eradication programs and col-

38 NBER Reporter Summer 2006 lates them with publicly available cen- heart and kidney disease later in life. In immunity against diseases and nutritional sus data. Malarious areas saw large drops the typical city, eradicating typhoid fever status prior to enlistment are responsi- in their malaria incidence following the in 1900 would have reduced the death ble for the observed mortality differen- campaign. In both absolute terms and rel- rate from heart disease by 21 percent, and tials. The patterns of wartime mortality ative to those in non-malarious areas, the the death rate from kidney disease by 23 of black and white soldiers are generally cohorts born after eradication had higher percent. These results are obtained after similar, but the relative effects of the two income as adults than the preceding gen- the authors include controls for the con- factors were somewhat different by race. eration. Similar increases in literacy and temporaneous disease environment and It appears that the health of white recruits the returns to schooling also occur. The lagged values of the dependent variable was more strongly influenced by the dis- results for years of schooling are mixed, and the overall disease environment. ease environment they were exposed to though. Lee investigates the patterns of socio- prior to enlistment. For black soldiers, on Troesken and Clay confirm that economic differences in wartime morbid- the other hand, socioeconomic status, a deprivation early in life can have lingering ity and mortality of black Union Army proxy for nutritional status and general physiological effects. In particular, their soldiers, and compares them with white economic wellbeing, was a perhaps more results suggest that crowded housing con- recruits. Light-skinned soldiers, former powerful determinant of health. Lee sug- ditions in early life facilitate the spread slaves who had been engaged in non-field gests that the larger occupational differ- of tuberculosis, which in turn, increases occupations, men from large plantations, ences in wartime mortality among blacks the risk of cancer and stroke later in life. and enlistees from urban areas were less could reflect the differences in health and In the typical city, eradicating tuberculo- likely to contract diseases and/or to die living conditions of blacks and whites sis in 1900 would have reduced the death from disease while in service than, respec- prior to enlistment. The stronger health rates from cancer and stroke in 1915 by tively, dark-skinned soldiers, field hands, effect of prior residence in urban areas 32 percent. Similarly, drinking impure men from small farms, and enlistees from among whites could be explained by the water in early life raises the likelihood rural areas. Patterns of disease-specific differences in prior exposure to disease that one will be infected with typhoid mortality and timing of death suggests between blacks and whites. fever, which in turn, increases the risk of that the differences in development of

International Finance and Macroeconomics

The NBER’s Program on Inter- Mark M. Spiegel, Federal Reserve Bank Reserve Board national Finance and Macroeconomics of San Francisco, “Offshore Financial met in Cambridge on March 24. NBER Centers: Parasites or Symbionts? Michael Kumhof and Stijn Van Research Associates Menzie D. Chinn, (NBER Working Paper No. 12044) Nieuwerburgh , IMF, “Monetary University of Wisconsin, and Lars E.O. Discussant: Sebnem Kalemli-Ozcan, Policy in an Equilibrium Portfolio Svensson, Princeton University, orga- University of Houston and NBER Balance Model” nized this program: Discussant: Michael Devereux, Romain Ranciere, IMF; Aaron University of British Columbia Ricardo J. Caballero, MIT and Tornell, University of California, NBER; Emmanuel Farhi, MIT; and Los Angeles and NBER; and Frank Eduardo A. Cavallo, Harvard University, Pierre-Olivier Gourinchas, University Westermann, University of Munich, and Jeffrey A. Frankel, Harvard of California, Berkeley and NBER, “Systemic Crises and Growth” University and NBER, “Does Openness “An Equilibrium Model of ‘Global Discussant: Roberto Chang, Rutgers to Trade Make Countries More Vulnerable Imbalances’ and Low Interest Rates” University and NBER to Sudden Stops, or Less? Using Gravity to Discussant: Paolo A. Pesenti, Federal Establish Causality” Reserve Bank of New York Julian Di Giovanni and Andrei A. Discussant: Graciela L. Kaminsky, George Levchenko, IMF, “Openness, Volatility, Washington University and NBER Andrew K. Rose, University of and the Risk Content of Exports” California, Berkeley and NBER, and Discussant: Sylvain Leduc, Federal

Three of the most important recent deficit, the stubborn decline in long-run as anomalies from the perspective of con- facts in global macroeconomics — the real rates, and the rise in the share of ventional wisdom and models. Caballero, sustained rise in the U.S. current account U.S. assets in global portfolio — appear Farhi, and Gourinchas provide a model

NBER Reporter Summer 2006 39 that rationalizes these facts as an equilib- effect on GDP growth. This link coexists Standard theory shows that sterilized rium outcome of two observed forces: 1) with the negative link between variance foreign exchange interventions do not potential growth differentials among dif- and growth typically found in the litera- affect equilibrium prices and quantities, ferent regions of the world and, 2) hetero- ture. To explain the link between crises and that domestic and foreign currency- geneity in these regions’ capacity to gener- and growth, the authors present a model denominated bonds are perfect substi- ate financial assets from real investments. in which contract enforce-ability prob- tutes. Kumhof and Van Nieuwerburgh In extensions of the basic model, they also lems generate financial constraints and show that when fiscal policy is not suf- generate exchange rate and gross flows low growth. Systemic risk-taking relaxes ficiently flexible in response to spend- patterns that are broadly consistent with borrowing constraints and increases ing shocks, exchange rates must adjust the recent trends observed in these vari- investment. This leads to higher long-run to restore budget balance. This exchange ables. Unlike the conventional wisdom, growth, but also to a greater incidence rate adjustment generates a capital gain in the absence of a large change in the two of crises. The authors find that the nega- or loss for holders of domestic currency forces, the model does not augur any cata- tive link between skewness and growth denominated bonds and causes perfect strophic event. More generally, the frame- emerges under similar restrictions in the substitutability to break down. Because work is flexible enough to shed light on a model and in the data. of imperfect asset substitutability, uncov- range of scenarios in a global equilibrium It has been observed that more ered interest rate parity no longer holds. environment. open countries experience higher out- Government balance sheet operations can Rose and Spiegel analyze the causes put growth volatility. DiGiovanni and be used as an independent policy instru- and consequences of offshore financial Levchenko use an industry-level panel ment to target interest rates. Sterilized centers (OFCs). Since OFCs are likely dataset of manufacturing production and foreign exchange interventions should be to be tax havens and money launderers, trade to analyze the mechanisms through most effective in developing countries, they encourage bad behavior in source which trade can affect the volatility of where fiscal volatility is large and where countries. Nevertheless, OFCs may also production. They find that sectors with the fraction of domestic currency denom- have unintended positive consequences higher trade are more volatile and that inated government liabilities is small. for their neighbors, since they act as a trade leads to increased specialization. Openness to trade is one factor competitive fringe for the domestic bank- These two forces act to increase over- that has been identified as determining ing sector. The authors derive and simu- all volatility. They also find that sectors whether a country is prone to sudden late a model of a home country monopoly that are more open to trade are less cor- stops in capital inflow, currency crashes, or bank facing a representative competitive related with the rest of the economy, an severe recessions. Some believe that open- OFC that offers tax advantages attained effect that acts to reduce aggregate vola- ness raises vulnerability to foreign shocks, by moving assets offshore at a cost that is tility. The point estimates indicate that while others believe that it makes adjust- increasing in distance between the OFC each of the three effects has an apprecia- ment to crises less painful. Several authors and the source. The model predicts that ble impact on aggregate volatility. Added have offered empirical evidence that hav- proximity to an OFC is likely to have pro- together they imply that a single stan- ing a large tradable sector reduces the con- competitive implications for the domestic dard deviation change in trade openness traction necessary to adjust to a given cut- banking sector, although the overall effect is associated with an increase in aggregate off in funding. This would help explain on welfare is ambiguous. Rose and Spiegel volatility of about 15 percent of the mean lower vulnerability to crises in Asia than test and confirm the predictions empiri- volatility observed in the data. The results in Latin America. Such studies may, how- cally. OFC proximity is associated with a are also used to provide estimates of the ever, be subject to the problem that trade more competitive domestic banking sys- welfare cost of increased volatility under is endogenous. Cavallo and Frankel use tem and greater overall financial depth. several sets of assumptions. The authors the gravity instrument for trade openness, Ranciere, Tornell, and Westermann then propose a summary measure of the which is constructed from geographical document the fact that countries that riskiness of a country’s pattern of export determinants of bilateral trade. They find have experienced occasional financial cri- specialization, and analyze its features that openness indeed makes countries less ses have, on average, grown faster than across countries and over time. There is vulnerable, both to severe sudden stops countries with stable financial conditions. a great deal of variation in countries’ risk and currency crashes, and that the rela- The authors measure the incidence of cri- content of exports, but it does not have a tionship is even stronger when correcting sis using the skewness of credit growth, simple relationship to the level of income for the endogeneity of trade. and find that it has a robust negative or other country characteristics.

40 NBER Reporter Summer 2006 Asset Pricing The NBER’s Program on Asset the Long Run and the Real Exchange Discussant: Tano Santos, Columbia Pricing met in Chicago on March 31. Rate” University and NBER Program Director John H. Cochrane Discussant: Adrien Verdelhan, Boston and Research Associate Lars P. Hansen, University Torben G. Andersen, Northwestern both of the University of Chicago, orga- University and NBER, and Luca nized this agenda: Ravi Jagannathan, Northwestern Benzoni, “Can Bonds Hedge Volatility University and NBER; Alexey Risk in the U.S. Treasury Market? Luca Benzoni, University of Malakhov, University of North A Specification Test for Affine Term Minnesota; Robert S. Goldstein, Carolina; and Dmitry Novikov, Structure Models” University of Minnesota and NBER; Goldman Sachs, “Do Hot Hands Discussant: Jun Pan, MIT and NBER and Pierre Collin-Dufresne, Persist Among Hedge Fund Managers? University of California, Berkeley and An Empirical Evaluation” Brad Barber and Ning Zhu, University NBER, “Can Standard Preferences Discussant: David Hsieh, Duke of California, Davis, and Terrance Explain the Prices of Out-of-the-Money University Odean, University of California, S&P 500 Put Options?” Berkeley, “Do Noise Traders Move Discussant: George Constantinides, Stavros Panageas and Jiangeng Markets?” University of Chicago and NBER Yu, University of Pennsylvania, Discussant: Sheridan Titman, “Technological Growth, Asset Pricing, University of Texas and NBER Riccardo Colacito and Mariano M. and Consumption Risk Over Long Croce, New York University, “Risk for Horizons”

Before the stock market crash of market crash. To this end, the authors tility of the depreciation rate in the order 1987, the Black-Scholes model implied extend their model to a Bayesian setting of 12 percent and the cross-country cor- that volatilities of S&P 500 index options in which the agents update their beliefs relation of consumption growth around were relatively constant. Since the crash, about the average jump size in the event 30 percent. though, deep out-of-the money S&P of a jump. Such beliefs only update at Jagannathan, Malakhov, and 500 put options have become “expen- crash dates, and hence can explain why Novikov empirically demonstrate that sive” relative to the Black-Scholes bench- the volatility smirk has not diminished both hot and cold hands among hedge mark. Many researchers have argued over the last 18 years. The authors find fund managers tend to persist. To measure that such prices cannot be justified in a that the model can capture the shape of performance, they use statistical model- general equilibrium setting if the rep- the implied volatility curve both pre- and selection methods for identifying style resentative agent has “standard prefer- post-crash while maintaining reasonable benchmarks for a given hedge fund, and ences.” However, Benzoni, Goldstein, estimates for expected returns, price-divi- they allow for the possibility that hedge and Collin-Dufresne demonstrate that dend ratios, and risk-free rates. fund net asset values may be based on the “volatility smirk” can be rationalized Brandt, Cochrane, and Santa-Clara stale prices for illiquid assets. They are if the agent is endowed with Epstein-Zin (2004) pointed out that the implicit sto- able to eliminate the backfill bias by delet- preferences and if the aggregate dividend chastic discount factors computed using ing all of the backfill observations in and consumption processes are driven by prices, on the one hand, and consump- their dataset. They also take into account a persistent stochastic growth variable tion growth, on the other hand, have the self-selection bias introduced by the that can jump. They identify a realistic very different implications for their cross- fact that both successful and unsuccessful calibration of the model that simultane- country correlation. They leave this as an hedge funds stop reporting information ously matches the empirical properties of unresolved puzzle. Colacito and Croce to the database provider. The former stop dividends, the equity premium, the prices explain it by combining Epstein and Zin accepting new money and the latter get of both at-the-money and deep out-of- (1989) preferences with a model of pre- liquidated. The authors find statistically the-money puts, and the level of the risk- dictable returns and by positing a very as well as economically significant persis- free rate. A more challenging question correlated long-run component. They tence in the performance of funds rela- (that apparently has not been previously also assume that the intertemporal elas- tive to their style benchmarks. It appears investigated) is whether one can explain ticity of substitution is larger than one. that half of the superior or inferior per- within a standard preference framework This setup brings the stochastic discount formance during a three-year interval will the stark regime change in the volatil- factors computed using prices and quan- spill over into the following three-year ity smirk that has existed since the 1987 tities close together, by keeping the vola- interval.

NBER Reporter Summer 2006 41 Panageas and Jianfeng develop a yield curve fails to span yield volatility, week (month) that they did the previous theoretical model in order to understand as the systematic volatility factors appear week (month). Third, when measured comovements between asset returns and largely unrelated to the cross-section of over one year, the imbalance between consumption over longer horizons. They yields. They conclude that a broad class purchases and sales of each stock by indi- develop an intertemporal general equilib- of affine diffusive, quadratic diffusive, and vidual investors forecasts cross-sectional rium model featuring two types of shocks: affine jump-diffusive models is incapa- stock returns the next year. Stocks heavily “small,” frequent, and disembodied shocks ble of accommodating the observed yield bought by individuals one year underper- to productivity and “large” technological volatility dynamics at daily, weekly, and form stocks heavily sold by 4.4 percentage innovations, which are embodied in new monthly horizons. Hence, yield volatil- points in the following year. The spread in vintages of the capital stock. The lat- ity risk per se cannot be hedged by tak- returns of stocks bought and stocks sold ter affect the economy with significant ing positions in the Treasury bond mar- is greater for small stocks and stocks heav- lags, because firms need to make irre- ket. The authors also advocate using these ily traded by individual investors. Among versible investments in the new types of empirical yield volatility measures more stocks heavily traded by individual inves- capital and there is an option value to broadly as a basis for specification testing tors, the spread in returns between stocks waiting. The model produces endogenous and (parametric) model selection within bought and stocks sold is 13.5 percentage cycles, countercyclical variation in risk the term structure literature. points the following year.Over shorter premia, and only a very modest degree of Barber, Ning, and Odean study periods, such as a week or a month, a predictability in consumption and divi- the trading behavior of individual inves- different pattern emerges. Stocks heav- dend growth as observed in the data. The tors using the Trade and Quotes (TAQ) ily bought by individual investors one authors then use their model as a labora- and Institute for the Study of Security week earn strong returns in the subse- tory to show that, in their simulated data, Markets (ISSM) transaction data for the quent week, while stocks heavily sold one the unconditional consumption Capital period 1983 to 2001. They document week earn poor returns in the subsequent Asset Pricing Model performs badly, three results: First, order imbalance based week. This pattern persists for a total of while its “long-horizon” version performs on buyer- and seller-initiated small trades three to four weeks and then reverses for significantly better. from the TAQ/ISSM data correlates well the subsequent several weeks. In addition Andersen and Benzoni investigate with the order imbalance based on trades to examining the ability of small trades to whether bonds can hedge volatility risk of individual investors from brokerage forecast returns, the authors look at the in the U.S. Treasury market, as predicted firm data. This indicates that trade size predictive value of large trades. In striking by most “affine” term structure models. is a reasonable proxy for the trading of contrast to their small trade results, they To this end, they construct powerful and individual investors. Second, order imbal- find that stocks heavily purchased with model-free empirical measures of the qua- ance based on TAQ/ISSM data indicates large trades one week earn poor returns in dratic yield variation for a cross-section of strong herding by individual investors. the subsequent week, while stocks heavily fixed-maturity zero-coupon bonds (“real- Individual investors predominantly con- sold one week earn strong returns in the ized yield volatility”) over daily, weekly, temporaneously buy (sell) the same stocks subsequent week. and monthly maturities through the use as each other. Furthermore, they predom- of high-frequency data. They find that the inantly buy (sell) the same stocks in one

42 NBER Reporter Summer 2006 Corporate Finance The NBER’s Program on Corporate University of Colorado at Boulder, and Firm Investment” Finance met in Chicago on March 31. “Back to the Beginning: Persistence and The meeting was organized by Kose the Cross-Section of Corporate Capital Steven N. Kaplan, University of John, New York University, and Ivo Structure” Chicago and NBER; Berk A. Sensoy, Welch, NBER and Brown University. Discussant: Ivo Welch University of Chicago; and Per The program was: Stromberg, SIFR, “What Are Firms? Sreedhar T. Bharath and Paolo Evolution From Early Business Plans to SESSION 1: Geography and Corporate Pasquariello, University of Michigan; Public Companies” Finance and Guojun Wu, University of Houston, “Does Asymmetric SESSION 4: Self-dealing and Augustin Landier, New York Information Drive Capital Structure Dividends University, and Vinay B. Nair and Decisions?” Julie Wulf, University of Pennsylvania, Discussant: Stewart C. Myers, MIT and Simeon Djankov, The World Bank; “Tradeoffs In Staying Close: Corporate NBER Rafael La Porta, Dartmouth College Decisionmaking and Geographic and NBER; Florencio Lopez-de- Dispersion” SESSION 3: Regulation, Contracts, Silanes, University of Amsterdam; and and Firms Andrei Shleifer, Harvard University Simi Kedia, Rutgers University; and NBER, “The Law and Economics Venkatesh Panchapagesan, Goldman Efraim Benmelech, Harvard of Self-Dealing” Sachs; and Vahap B. Uysal, University University; and Tobias J. Moskowitz, Discussant: Paul Mahoney, University of Oklahoma, “Geography and University of Chicago and NBER, of Virginia Acquirer Returns” “The Political Economy of Financial Discussant for both papers: Joshua Regulation: Evidence from U.S. State Gerard Hoberg and Nagpurnanand Coval, Harvard University and NBER Usury Laws in the 18th and 19th R. Prabhala, University of Maryland, Century” “Dividend Policy, Risk, and Catering” SESSION 2: Capital Structure Discussant: Malcolm Baker, Harvard Mark J. Garmaise, University of University and NBER Michael L. Lemmon, University of California, Los Angeles, “Ties that Utah; Michael R. Roberts, University Truly Bind: Non-Competition of Pennsylvania; and Jaime F. Zender, Agreements, Executive Compensation,

Landier, Nair, and Wulf docu- proximate employees is stronger in such imity on the acquisition decisions of ment the role of geographic dispersion soft-information industries. Second, they U.S. public firms over the period 1990- on corporate decisionmaking. They find investigate how headquarter proximity to 2003. Transactions in which the acquirer that geographically dispersed firms are employees affects managerial alignment and target firms are located within 100 less employee-friendly. Also, using divi- with shareholder objectives. They docu- kilometers of each other are classified sion-level data, they find that employee ment that the protection of proximate as local transactions. The authors find dismissals are less common in divisions employees only holds when the headquar- that acquirer returns in local transactions located close to corporate headquar- ters are located in less-populated counties, are more than twice those in non-local ters. Finally, it turns out that firms are suggesting concern for such employees. transactions. The higher returns to local reluctant to divest in-state divisions. To Moreover, stock markets respond favor- acquirers are, at least partially, attribut- explain these findings, the authors con- ably to divestitures of close divisions, able to information advantages arising sider two mechanisms. First, they inves- especially for these smaller-county firms. from geographical proximity. These infor- tigate whether headquarter proximity to These findings suggest that social fac- mation advantages facilitate acquisition divisions is related to internal informa- tors work alongside informational con- of targets that, on average, create higher tion flows. They find that firms are geo- siderations in making geographic disper- overall return. However, bidders use their graphically concentrated when informa- sion an important factor in corporate information advantages to earn a higher tion is more difficult to transfer over decision-making. share of the surplus created. long distances (soft information indus- Kedia, Panchapagesan, and Uysal Lemmon, Roberts, and Zender tries). Additionally, the protection of examine the impact of geographical prox- examine the evolution of the cross-sec-

NBER Reporter Summer 2006 43 tional distribution of capital structure and adverse selection. Overall, this evidence ers, competitors, alliances, top manage- find it to be remarkably stable over time: explains why the pecking order theory ment, ownership structure, and the board firms with high (low) leverage remain is only partially successful in explaining of directors. Their analysis focuses on the relatively high (low) levered for over 20 all of firms’ capital structure decisions. It nature and stability of those firm attri- years. Additionally, this relative ranking also suggests that the theory finds support butes. Firm business lines remain remark- is observed for both public and private when its basic assumptions hold in the ably stable from business plan through firms, and is largely unaffected by the pro- data, as should reasonably be expected of public company. Within those business cess of going public. These persistent dif- any theory. lines, non-human capital aspects of the ferences in leverage across firms are associ- Benmelech and Moskowitz study businesses appear more stable than human ated with the presence of an unobserved the political economy of financial reg- capital aspects. In the cross-section, firms firm-specific effect that is responsible for ulation by examining the determinants with more alienable assets have substan- the majority of variation in capital struc- and effects of U.S. state usury laws during tially more human capital turnover. ture. Over 90 percent of the explained the eighteenth and nineteenth centuries. Djankov, La Porta, Lopez-de- variation in leverage is captured by firm They argue that regulation is the outcome Silanes, and Shleifer present a new fixed effects, whereas previously identi- of private interests using the coercive measure of legal protection of minor- fied determinants (for example, size, mar- power of the state to extract rents from ity shareholders against expropriation by ket-to-book, industry) are responsible for other groups. They find that the strictness corporate insiders: the anti-self-dealing less than 10 percent. These findings show of usury coexists with other exclusion- index. Assembled with the help of Lex that firms use net security issuances to ary policies, such as suffrage laws and lack Mundi law firms, the index is calculated maintain their leverage ratios in relatively of general incorporation, or free banking for 72 countries based on legal rules pre- confined regions around their long-run laws, which also respond less to competi- vailing in 2003, and focuses on private means, consistent with a dynamic rebal- tive pressures for repeal. Furthermore, enforcement mechanisms, such as disclo- ancing of capital structure. Importantly, the same determinants of financial reg- sure, approval, and litigation, governing a the results imply that the primary deter- ulation that favor one group and limit specific self-dealing transaction. This the- minants of cross-sectional variation in access to others, are associated with lower oretically-grounded index predicts a vari- corporate capital structures are largely future economic growth rates, highlight- ety of stock market outcomes, and gener- time invariant, which significantly reduces ing the endogeneity of financial develop- ally works better than the commonly used the set of candidate explanations to those ment and growth. index of anti-director rights. based on factors that remain relatively sta- Garmaise studies the effects of non- Fama and French (2001a) show that ble over long periods of time. competition agreements by analyzing the propensity to pay dividends declines Using an information asymmetry time-series and cross-sectional variation significantly in the 1990s, the disappear- index based on measures of adverse selec- in the enforceability of these contracts ing dividends puzzle. Baker and Wurgler tion developed by the market microstruc- across U.S. states. He finds that increased (2004a, 2004b) suggest that these appear- ture literature, Bharath, Pasquariello, enforceability reduces executive compen- ing and disappearing dividends are an out- and Guojun test whether information sation and shifts its form towards greater come of firms “catering” to transient fads asymmetry is the sole determinant of use of salary. He also shows that tougher for dividend paying stocks. Hoberg and capital structure decisions, as suggested non-competition enforcement reduces Prabhala empirically examine disappear- by the pecking order theory. Their tests research and development spending and ing dividends and its catering explana- rely exclusively on measures of the mar- capital expenditures per employee. Non- tion through the lens of risk. They report ket’s assessment of adverse selection risk, competition agreements promote execu- two main findings: 1) risk is a significant rather than on ex-ante firm characteris- tive stability and board participation, but determinant of the propensity to pay divi- tics. They find that information asymme- higher quality managers apparently shun dends and explains up to 40 percent of the try does affect capital structure decisions firms in high-enforcement jurisdictions. disappearing dividends puzzle; 2) cater- of U.S. firms over the period 1973-2002, These results have implications for theo- ing is insignificant once they account for especially when firms’ financing needs ries of executive compensation and firm risk. Risk is also related to payout poli- are low and when firms are financially organization. cies in general: it explains the decision constrained. They also find a significant Kaplan, Sensoy, and Strömberg to increase dividends and to repurchase degree of intertemporal variability in study how firm characteristics evolve from shares. These findings affirm theories and firms’ degree of information asymmetry, early business plan to initial public offer- field evidence on the role of risk in divi- as well as in its impact on firms’ debt issu- ing to public company for 49 venture cap- dend policy and suggest that the 1990s ance decisions. These findings, based on ital financed companies. The average time increase in volatility noted by Campbell, the information asymmetry index, are elapsed is almost six years. They describe Lettau, Malkiel, and Xu (2001) has cor- robust to sorting firms based on size and the financial performance, business idea, porate finance implications. firm insider trading activity, two popu- point(s) of differentiation, non-human lar alternative proxies for the severity of capital assets, growth strategy, custom-

44 NBER Reporter Summer 2006 International Trade and Investment The NBER’s Program on NBER; Luis Garicano, University Bank; and Irene Brambilla, Yale International Trade and Investment met of Chicago; and Esteban Rossi- University and NBER, “Export Crops, in Cambridge on March 31. Giovanni Hansberg, Princeton University and Marketing Costs, and Poverty” Maggi, NBER and Princeton University, NBER, “Organizing Offshoring: and Andres Rodriguez-Clare, NBER Middle Managers and Communication Christian Broda, University of and Pennsylvania State University, orga- Cost” Chicago; Joshua Greenfield, Columbia nized the meeting. These papers were University; and David Weinstein, discussed: Doireann Fitzgerald, University of Columbia University and NBER; California, Santa Cruz, “Trade Costs, “From Groundnuts to Globalization: Julian Di Giovanni and Andrei Limited Enforcement, and Risk A Structural Estimate of Trade and Levchenko, IMF, “Openness, Volatility, Sharing: A Joint Test” Growth” and the Risk Content of Exports” Pol Antras, Harvard University and Jorge Balat and Guido Porto, World

It has been observed that more open ticularly attractive locations to offshoring? trade costs and asset market imperfections countries experience higher output growth Antràs, Garicano, and Rossi-Hansberg are necessary in order to explain the failure volatility. DiGiovanni and Levchenko address these questions with a theory of of perfect consumption risk sharing. The use an industry-level panel dataset of man- international production hierarchies in rejection of complete markets is weaker ufacturing production and trade to ana- which teams arise endogenously to make for developed than developing countries. lyze the mechanisms through which trade efficient use of agents’ knowledge. Their At the same time, financial autarky is also can affect the volatility of production. theory highlights the role of host-coun- rejected, indicating that some risk sharing They find that sectors with higher trade try management skills (middle manage- is possible through asset markets. are more volatile and that trade leads to ment) in bringing about the emergence of Balat, Brambilla, and Porto advance increased specialization. These two forces international offshoring. By shielding top a hypothesis to explain the small estimated act to increase overall volatility. They also management in the source country from impacts of trade barriers on poverty, espe- find that sectors that are more open to routine problems faced by host country cially in rural Africa. They study the case trade are less correlated with the rest of workers, the presence of middle managers of Uganda and claim that high marketing the economy, an effect that acts to reduce improves the efficiency of the transmission costs prevent the realization of the gains aggregate volatility. The point estimates of knowledge across countries. The model from trade. Their basic hypothesis is that indicate that each of the three effects has further predicts that the positive effect of the availability of markets for agricultural an appreciable impact on aggregate vola- middle management skills on offshoring export crops leads to a higher participa- tility. Added together, they imply that a is weaker, the more advanced are commu- tion into export cropping and that this, in single standard deviation change in trade nication technologies in the host country. turn, leads to lower poverty. They use data openness is associated with an increase The authors provide evidence consistent from the Uganda National Household in aggregate volatility of about 15 per- with this prediction. Survey to test it. They first establish that cent of the mean volatility observed in the Fitzgerald addresses the question of farmers living in villages with fewer outlets data. The authors also use these results to whether both goods and asset market fric- for sales of agricultural exports are likely provide estimates of the welfare cost of tions are necessary to explain the failure to be poorer than farmers residing in mar- increased volatility under several sets of of consumption risk sharing across coun- ket-endowed villages. Further, they show assumptions. They then propose a sum- tries. She presents a multi-country model that market availability leads to increased mary measure of the riskiness of a coun- with Armington specialization. There household participation in export crop- try’s pattern of export specialization, and are iceberg costs of shipping goods across ping (coffee, tea, cotton, fruits) and that analyze its features across countries and countries. In asset markets, contracts are households engaged in export cropping over time. There is a great deal of varia- imperfectly enforceable. Both frictions are less likely to be poorer than subsistence- tion in countries’ risk content of exports, separately limit the extent to which coun- based households. They conclude that the but it does not have a simple relationship tries can pool risk. The model suggests a presence of marketing costs affects the to the level of income or other country test for the presence of each of the two way that trade lowers poverty by hinder- characteristics. types of friction that exploits data on ing farmers from engaging in export crop- Why do firms decide to offshore cer- bilateral imports. Fitzgerald implements ping. In addition, these effects are non-lin- tain parts of their production process? this test using a sample of developed and ear: the poverty impacts of higher market What qualifies certain countries as par- developing countries. She finds that both availability are much stronger in low mar-

NBER Reporter Summer 2006 45 ket density villages than in medium or Second, increases in the number of variet- typical country in the world, new high market villages. This uncovers the ies drive down the cost of innovation and imported varieties contribute 0.13 per- role of market access and price competi- result in ever more variety creation. Using centage points per year to total factor pro- tion among buyers and intermediaries as highly disaggregated trade data — for ductivity or 12 percent of their productiv- key building blocks in the link between example Gabon’s imports of Gambian ity growth. Individual country experiences export opportunities and the poor. raw, unshelled, groundnuts — Broda, vary substantially, with trade explaining 5 Starting with Romer (1987) and Greenfield, and Weinstein structurally percent of the productivity growth in the Rivera-Batiz-Romer (1991), economists estimate the impact that new imports typical developed country but about a have been able to model how trade have had in approximately 4000 markets quarter of productivity growth in the typ- enhances growth through the creation and per country. They then move from ground- ical developing country. They also find import of new varieties. In this frame- nuts to globalization by building an exact that the creation of new varieties is corre- work, international trade increases eco- total factor productivity index that aggre- lated with R and D activities across coun- nomic output through two channels. First, gates these micro gains to obtain an esti- tries in ways consistent to semi-endoge- trade raises productivity because produc- mate of trade on productivity growth nous growth models proposed by Jones ers gain access to new imported varieties. around the world. They find that in the (1995).

Behavioral Finance

The NBER’s Working Group on Princeton University and NBER, Predictable Returns” Behavioral Finance met in Chicago on “Advisors and Asset Prices: A Model of Discussant: Josef Lakonishok, April 1. Directors Robert J. Shiller of the Origins of Bubbles” University of Illinois and NBER and Richard H. Thaler Discussant: Pietro Veronesi, University of the University of Chicago organized of Chicago Amil Dasgupta, Andrea Prat, and this program: Michela Verardo, London School of Malcolm Baker, Harvard University Economics “The Price of Conformism” Ulrike Malmendier, Stanford and NBER; Stefan Nagel, Stanford Discussant: Markus Brunnermeier, University and NBER, and Enrico University and and NBER; and Jeffrey Princeton University Moretti, University of California, Wurgler, New York University and Berkeley and NBER, “Winning by NBER, “The Effects of Dividends on Nicholas Barberis, Yale University Losing: Evidence on Overbidding in Consumption” and NBER, and Wei Xiong, “What Mergers” Discussant: Erik Hurst, University of Drives the Disposition and Momentum Discussant: Yiming Qian, University of Chicago and NBER Effects? An Analysis of a Recent Iowa Preference-Based Explanation” Lauren Cohen, Yale University, Discussant: Bing Han, Ohio State Harrison Hong, Princeton University, and Andrea Frazzini, University University and Jose Scheinkman and Wei Xiong, of Chicago, “Economic Links and

Do acquiring companies profit from bidding contests to address this iden- before the merger fight, but diverge signif- acquisitions, or do acquiring CEOs tification issue. They construct a novel icantly after one bidder has completed the destroy shareholder value? Answering dataset on all mergers with overlapping merger. Winners significantly underper- this question empirically is difficult since bids of at least two potential acquirors form losers over a five-year horizon. the hypothetical counterfactual is hard between 1983 and 2004. They then com- Many asset price bubbles occur dur- to determine. While negative stock reac- pare adjusted abnormal returns of all can- ing periods of excitement about new tech- tions to the announcement of merg- didates both before and after a merger nologies. Hong, Scheinkman, and Xiong ers are consistent with value-destroying fight. The key identifying assumption is focus on the role of advisors and the mergers, they are also consistent with that the returns and other corporate out- communication process with investors in overvaluation of the acquiror at the time comes of losing bidders are a valid coun- explaining this stylized fact. Advisors are of the announcement. Similarly, stud- terfactual for the winner, after employing well-intentioned and want to maximize ies of long-term returns to acquirors are the usual controls and matching crite- the welfare of their advisees (like a parent affected by slowly declining overvalu- ria. The authors find that stock returns and child). But only some of them under- ation. Malmendier and Moretti study of bidders are not significantly different stand the new technology (the tech-sav-

46 NBER Reporter Summer 2006 vys); others do not and can only make a also increase with mutual fund and special bought (sold) by institutions in the past 3 downward-biased recommendation (the dividends. to 5 quarters underperform (overperform) old-fogies). While smart investors recog- Cohen and Frazzini find evidence the rest of the market in the next 12 to 30 nize the heterogeneity in advisors, naive of return predictability across economi- months. These results are of similar mag- ones mistakenly take whatever is said at cally linked firms. They test the hypoth- nitude to, but distinct from, other known face value. Tech-savvys inflate their fore- esis that, in the presence of investors sub- asset pricing anomalies. The findings chal- casts to signal that they are not old-fogies ject to attention constraints, stock prices lenge the mainstream view of the roles because more accurate information about do not promptly incorporate news about played by individuals and institutions in their type improves the welfare of inves- economically related firms, generating generating asset pricing anomalies. tors in the future. A bubble arises for a return predictability across assets. They One of the most striking portfolio wide range of parameters and its size is use a dataset of firms’ principal customers puzzles is the “disposition effect”: the maximized when there is a mix of smart to identify a set of economically related tendency of individuals to sell stocks in and naive investors in the economy. firms, and show that stock prices do not their portfolios that have risen in value, Classical models predict that the divi- incorporate news involving related firms, rather than fallen in value, since pur- sion of stock returns into dividends and generating predictable subsequent price chase. Perhaps the most prominent expla- capital appreciation does not affect inves- moves. A long/short equity strategy based nation for this puzzle is based on pros- tor consumption patterns, while men- on this effect yields monthly alphas of pect theory. Despite its prominence, this tal accounting and other economic fric- over 150 basis points, or over 18 percent hypothesis has received little formal scru- tions predict that investors are more likely per year. tiny. Barberis and Xiong take up this to consume from stock returns in the As previous agency models have task, and analyze the trading behavior form of dividends. Using two microdata shown, fund managers with career con- of investors with prospect theory prefer- sets, Baker, Nagel, and Wurgler find cerns have an incentive to imitate the ences. Surprisingly, they find that, in its that investors are indeed far more likely recent trading strategy of other managers. simplest implementation, prospect theory to consume from dividends than capi- Dasgupta, Prat, and Verardo embed this often predicts the opposite of the disposi- tal gains. In the Consumer Expenditure rational conformist tendency in a styl- tion effect. They provide intuition for this Survey, household consumption increases ized financial market with limited arbi- result, and identify the conditions under with dividend income, after controlling trage. Equilibrium prices incorporate a which the disposition effect holds or fails. for total wealth, total portfolio returns, reputational premium or discount, which They also discuss the implications of their and other sources of income. In a sample is a monotonic function of past trade results for other disposition-type effects of household investment accounts data between career-driven traders and the rest that have been documented in settings from a brokerage, net withdrawals from of the market. Their prediction is tested such as the housing market, futures trad- the accounts increase one-for-one with with quarterly data on U.S. institutional ing, and executive stock options. ordinary dividends of moderate size, after holdings from 1983 to 2004. They find controlling for total portfolio returns, and that stocks that have been persistently

International Trade and Organization

The NBER’s Working Group on California, Davis and NBER, and Hierarchies and International Trade: International Trade and Organization Barbara J. Spencer, University Theory and Evidence” met in Cambridge on April 1. Director of British Columbia and NBER, Gordon H. Hanson of the University “Contractual versus Generic Volker Nocke, University of of California, San Diego, organized the Outsourcing: The Role of Proximity” Pennsylvania, and Stephen Yeaple, meeting, at which these papers were University of Pennsylvania and NBER, discussed: Dalia Marin, University of Munich, “Endogenizing Firm Scope: Trade and Thierry Verdier, Centre for Liberalization and the Size Distribution Robert C. Feenstra, University of Economic Policy Research, “Corporate of Multiproduct Firm”

Feenstra and Spencer explore the “contractual” — in which suppliers under- through contracts abroad rather than at relationship between proximity of buy- take specific investments — or involve home, but does not change the range of ers and sellers and the organizational “generic” market transactions. Proximity generic imports. A higher-quality foreign form of outsourcing. Outsourcing can be expands the variety of products sourced workforce raises the variety of contrac-

NBER Reporter Summer 2006 47 tual trade, but at the expense of gener- stylized facts and link these features to Nocke and Yeaple develop a theory ics. The authors confirm these predictions the trade environment that countries of multiproduct firms and endogenous using data for ordinary versus processing and firms face. They introduce heterog- firm scope that can explain a well-known exports from Chinese provinces to desti- enous firms with internal hierarchies into empirical puzzle: larger firms appear to be nation markets and the predictions of an a Krugman (1980) model of trade. The less efficient in that they have lower val- extended model that allows for multina- model simultaneously determines firms’ ues of Tobin’s Q. The authors extend this tional production. organizational choices and heterogene- theory to study the effects of trade liber- Corporate organization varies within ity across firms in size and productivity. alization and market integration on the a country and across countries with coun- They show that international trade and size distribution of firms. They show that try size. Larger countries have larger firms the toughness of competition in interna- a symmetric bilateral trade liberalization with flatter more decentralized corporate tional markets induce a power struggle in leads to a less skewed size distribution. hierarchies than smaller countries. Firms firms, eventually leading to decentralized The opposite result obtains in the case of a in larger countries change their corporate corporate hierarchies. They show further unilateral trade liberalization in the liber- organization more slowly than firms in that trade triggers inter-firm reallocations alizing country. In this model, trade liber- smaller countries. Furthermore, corporate towards more productive firms in which alization affects not only the distribution diversity within a country is correlated CEOs have power. Based on unique data of observed productivities but also pro- with the pattern of heterogeneity among from 660 Austrian and German corpo- ductivity at the firm level. In the empiri- firms in size and productivity. Marin and rations, they offer econometric evidence cal section, the authors show that the key Verdier develop a theory to explain these consistent with the model’s predictions. predictions are consistent with the data.

Public Economics Program Meeting

The NBER’s Program on Public of California, San Diego, “$2.00 Medicare” Economics met in Cambridge on April 6– Gas! Studying the Effect of Gas Tax Discussant: Jonathan Skinner, Dartmouth 7. Program Director James M. Poterba of Moratorium” College and NBER MIT organized the meeting at which these Discussant: Andrew Samwick, Dartmouth papers were discussed: College and NBER Raj Chetty, University of California, Berkeley and NBER, “Why Do Seth H. Giertz, Congressional Budget Douglas A. Shackelford, University of Unemployment Benefits Raise Office, “The Elasticity of Taxable Income North Carolina and NBER; Zhonglan Unemployment Durations? Moral Hazard During the 1990s: A Sensitivity Analysis” Dai and Harold H. Zhang, University of vs. Liquidity” Texas; and Edward Maydew, University Discussant: Mark Duggan, University of Monica Singhal, Harvard University of North Carolina, “Capital Gains Taxes Maryland and NBER and NBER, and Adam Looney, Federal and Asset Prices: Capitalization or Reserve Board, “The Effect of Anticipated Lock-in?” Stephen Coate, Cornell University and Tax Changes on Intertemporal Labor Discussant: Scott Weisbenner, University NBER, and Marco Battaglini, Princeton Supply and the Realization of Taxable of Illinois and NBER University, “A Dynamic Theory of Public Income” Spending, Taxation, and Debt” Robert Moffitt, John Hopkins University Discussant: Aleh Tsyvinski, Harvard Bruce D. Meyer, University of Chicago and NBER, “Welfare Work Requirements University and NBER and NBER, and Bradley T. Heim, Duke with Paternalistic Government University, “Identification and Estimation Preferences” Patrick Bayer, Yale University and of Structural Models of Labor Supply and Discussant: Emmanuel Saez, University of NBER; Nathaniel Keohane, Yale Program Participation” California, Berkeley and NBER University; and Chris Timmins, Duke (No discussants for these three papers.) University, “Migration and Hedonic Amy Finkelstein, MIT and NBER, “The Valuation : The Case of Air Quality” Joseph J. Doyle, Jr., MIT and NBER, Aggregate Effects of Health Insurance: Discussant: Kenneth Chay, University of and Krislert Samphantharak, University Evidence from the Introduction of California, Berkeley and NBER

Giertz examines alternative meth- 1990 and 1993 federal tax increases. The by Gruber and Saez (2002), Carroll odologies for measuring responses to the methodologies build on those employed (1998), Auten and Carroll (1999), and

48 NBER Reporter Summer 2006 Feldstein (1995). Internal Revenue tage of this approach is that the changes year among those working increased in Service tax return data for the project are are anticipated and therefore should not response to the EITC expansions. This from the Statistics of Income, which heav- cause re-evaluations of lifetime income. change occurred even though both the ily oversamples high-income filers. Special Consequently, the estimates of substitu- income effect of the larger credits and the attention is paid to the importance of tion effects should not be confounded substitution effect of the higher phase- sample income restrictions and meth- by life-cycle income effects. The empiri- out rates implied a decline in hours. They odology. Estimates are broken down by cal design also allows for comparison of address these surprising EITC results as income group to measure how responses similar families and can be used to esti- well as the effects of welfare budget set to tax changes vary by income. In general, mate elasticities across the income dis- changes using a joint model of labor sup- estimates are quite sensitive to a number tribution. In particular, the authors pro- ply and program participation. of different factors. Using an approach vide estimates for low and middle income Despite the considerable attention similar to Carroll’s yields elasticity of tax- families. Using data from the Survey paid to the theory of tax incidence, there able income (ETI) estimates as high as of Income and Program Participation are surprisingly few estimates of the pass- 0.54 and as low as 0.03, depending on the (SIPP), they estimate an intertemporal through rate of sales taxes on retail prices. income threshold for inclusion into the elasticity of family labor earnings close to Doyle and Samphantharak estimate the sample. Gruber and Saez’s preferred spec- one for families earning between $30,000 effect of a suspension and subsequent ification yields estimates for the 1990s and $75,000.The estimates for families in reinstatement of the gasoline sales tax of between 0.20 and 0.30. Yet another the EITC phase-out range are lower but in Illinois and Indiana on retail prices. approach compares behavior in a year still substantial. Estimates from the IRS- Earlier laws set the timing of the rein- before a tax change to behavior in a year NBER individual tax panel are consistent statements, providing plausibly exoge- after the tax change. That approach yields with the SIPP estimates. Tests using alter- nous changes in the tax rates. Using a estimated ETIs ranging from 0 to 0.71. nate control groups and simulated ““pla- unique dataset of daily gasoline prices The results suggest tremendous variation cebo” tax schedules support the identify- at the station level, the authors find that across income groups, with people at the ing assumptions. The high-end estimates retail gas prices drop by 3 percent follow- top of the income distribution showing suggest substantial efficiency costs of ing the elimination of the 5 percent sales the greatest responsiveness. In fact, the taxation. tax, and increase by 4 percent following estimates suggest that the ETI could be Heim and Meyer estimate a struc- the reinstatements, compared to neigh- as high as 1.2 for those at the very top of tural model of employment, hours, and boring states. Some evidence also suggests the income distribution. The major con- program participation choices of single that the tax reinstatements are associated clusion, however, is that isolating the true women over the 1984–96 period. During with higher prices up to an hour into taxable income responses to tax changes the 1980s and 1990s, tax and welfare pol- neighboring states, which provides some is extremely complicated by a myriad of icy dramatically altered the labor supply evidence on the size of the geographic other factors and thus little confidence and program participation incentives of market for gasoline. should be placed on any single estimate. single mothers. The authors use this set- Shackelford and his co-authors Additionally, focusing on particular com- ting to explore identification in struc- examine the impact on asset prices of a ponents of taxable income might yield tural labor supply models. Through the reduction in the long-term capital gains more insight. judicious use of special samples (specific tax rate using an equilibrium approach Looney and Singhal use antici- states, years, women with certain num- that considers both buyers’ and sellers’ pated changes in tax rates associated with bers of children), control variables, and responses. They demonstrate that the changes in family composition to estimate separate coefficients for different types of equilibrium impact of capital gains taxes intertemporal labor supply elasticities and income, they isolate the different sources reflects both the capitalization effect (cap- elasticities of taxable income with respect of variation in the aftertax reward to ital gains taxes decrease demand) and the to the net-of-tax wage rate. A number of work. They explore the role of the inten- lock-in effect (capital gains taxes decrease provisions of the tax code are tied explic- sive hours choice versus the extensive supply). Depending on time periods and itly to child age and dependent status. work/nonwork decision, examine the stock characteristics, either effect may Changes in the ages of children can thus role of the point-in-time shape of the dominate. Using the Taxpayer Relief Act affect marginal tax rates through phase-in tax schedule for a given demographic of 1997 as their event, they find evi- or phase-out provisions of tax credits or group versus changes over time, the tax dence supporting a dominant capitaliza- by shifting individuals across tax brackets. treatment of children, and the role of tion effect in the week following news The authors identify the response of labor functional form. They also provide sub- that sharply increased the probability of a and taxable income to these tax changes stantive results on effects of the Earned reduction in the capital gains tax rate and by comparing families who experienced a Income Tax Credit (EITC) and welfare a dominant lock-in effect in the week after tax-rate change to families who had a sim- programs. Studies analyzing the effects the rate reduction became effective. Non- ilar change in dependents but no result- of the EITC using difference-in-differ- dividend paying stocks (whose sharehold- ing tax-rate change. A primary advan- ence methods have found that hours per ers only face capital gains taxes) experi-

NBER Reporter Summer 2006 49 ence higher average returns during the evidence of these types of effects. A back “business-as-usual” in which legislators week the capitalization effect dominates of the envelope calculation based on the bargain over the allocation of pork, and and stocks with large embedded capi- estimated impact of Medicare suggests “responsible-policymaking” in which tal gains and high individual ownership that the overall spread of health insurance policies maximize the collective good. exhibit lower average returns during the between 1950 and 1990 may be able to Transitions between the two regimes are week the lock-in effect dominates. They explain about half of the increase in real brought about by shocks in the value of also find that the tax cut increases the per capita health spending over this time the public good. In the long run, equilib- trading volume in non-dividend paying period. rium tax rates are too high and too vola- stocks during the dominant capitalization It is well known that unemployment tile, public good provision is too low and week and in stocks with large embedded benefits raise unemployment durations. debt levels are too high. In some environ- capital gains and high individual owner- This result has traditionally been inter- ments, a balanced budget requirement ship during the dominant lock-in week. preted as a substitution effect caused by can improve citizen welfare. Work requirements in means-tested a distortion in the price of leisure relative Conventional hedonic techniques for transfer programs have grown in impor- to consumption, leading to moral hazard. estimating the value of local amenities rely tance in the United States and in some Chetty questions this interpretation by on the assumption that households move other countries. The theoretical litera- showing that unemployment benefits can freely among locations. Bayer, Keohane, ture that considers their possible optimal- also affect durations through an income and Timmins show that when moving ity generally operates within a traditional effect for agents with limited liquidity. is costly, the variation in housing prices welfarist framework where some function The empirical relevance of liquidity con- and wages across locations may no lon- of the utility of the poor is maximized. straints and income effects is evaluated ger reflect the value of differences in local Here Moffitt argues that society instead in two ways. First, he divides households amenities. They develop an alternative has preferences over the actual work allo- into groups that are likely to be con- discrete-choice approach that considers cations of welfare recipients and that the strained and unconstrained based on prox- the household location decision directly, resulting paternalistic social welfare func- ies such as asset holdings. He finds that and apply it to the case of air quality tion is more consistent with the historical increases in unemployment benefits have in U.S. metro areas in 1990 and 2000. evidence. With this social welfare func- small effects on durations in the uncon- Because air pollution is likely to be cor- tion, optimality of work requirements is strained groups but large effects in the related with unobservable local charac- possible but depends on the accuracy of constrained groups. Second, he finds that teristics such as economic activity, they the screening mechanism which assigns lump-sum severance payments granted at instrument for air quality using the con- work requirements to some benefit recip- the time of job loss significantly increase tribution of distant sources to local pol- ients and not others. Numerical simula- durations among constrained households. lution — excluding emissions from local tions show that the accuracy must be high These results suggest that unemployment sources, which are most likely to be corre- for such optimality to occur. The simula- benefits raise durations primarily because lated with local conditions. Their model tions also show that earnings subsidies of an income effect induced by liquidity yields an estimated elasticity of willing- can be justified with the type of social wel- constraints rather than moral hazard from ness to pay with respect to air quality fare function used here. distorted incentives. of 0.34 to 0.42. These estimates imply Finkelstein investigates the effects of Battaglini and Coate present a that the median household would pay market-wide changes in health insurance dynamic political economy theory of pub- $149 to $185 (in constant 1982-4 dol- by examining the single largest change in lic spending, taxation, and debt. Policy lars) for a one-unit reduction in aver- health insurance coverage in American choices are made by a legislature con- age ambient concentrations of particulate history: the introduction of Medicare in sisting of representatives elected by geo- matter. These estimates are three times 1965. She estimates that the impact of graphically-defined districts. The legisla- greater than the marginal willingness to Medicare on hospital spending is over ture can raise revenues via a distortionary pay estimated by a conventional hedonic six times larger than what the evidence income tax and by borrowing. These rev- model using the same data. The results from individual-level changes in health enues can be used to finance a national are robust to a range of covariates, instru- insurance would have predicted. This dis- public good and district-specific trans- menting strategies, and functional form proportionately larger effect may arise if fers (interpreted as pork-barrel spending). assumptions. The findings also confirm market-wide changes in demand alter the The value of the public good is stochastic, the importance of instrumenting for local incentives of hospitals to incur the fixed reflecting shocks such as wars or - natu air pollution. costs of entering the market or of adopt- ral disasters. In equilibrium, policymak- ing new practice styles. She presents some ing cycles between two distinct regimes:

50 NBER Reporter Summer 2006 Health Economics

The NBER’s Program on Health “Consumer Welfare Implications of the Schooling Reduce Mortality Risk Economics met in Cambridge on April Nursing Home Ownership Mix” Among the Elderly?” 7. Program Director Michael Grossman and NBER Research Associate Ted Joyce Carlos Dobkin, University of William N. Evans, University of organized the program, at which these California, Santa Cruz and NBER, and Maryland and NBER, and Heng Wei, papers were discussed: Steven Puller, Texas A&M University, University of Maryland, “Postpartum “The Effects of Government Transfers Hospital Stay and the Outcomes of Christopher Carpenter, University of on Monthly Cycles of Drug Abuse, Mothers and Newborns” California, Irvine, “How Do Workplace Crime, and Mortality” Smoking Laws Work?” Jens Ludwig, Georgetown University Douglas Almond, Columbia and NBER; Dave Marcotte, University John Vernon, University of University and NBER, and Bhashkar of Maryland; and Karen Norberg, Connecticut and NBER, and Rexford Mazumder, Federal Reserve Bank Washington University and NBER, Santerre, University of Connecticut, of Chicago, “How Did Compulsory “Anti-Depressants and Suicide”

A large literature shows that state and findings confirm that workplace smoking significantly increase their consumption local laws requiring smoke-free workplaces bans do reduce smoking; they document of drugs when their checks arrive at the are associated with improved worker out- the underlying mechanisms through which beginning of the month and, as a result, comes (lower secondhand smoke exposure local smoking by-laws improve health out- experience adverse events including arrest, and own smoking rates.) Carpenter pro- comes; and they show that the effects of hospitalization, and death. Using data vides new quasi-experimental evidence on these laws are strongly heterogeneous with from California, they find that the over- the effects of workplace smoking laws by respect to occupation. all rate of drug related hospital admissions using the differential timing of adoption of Vernon and Santerre compare the increases abruptly at the beginning of the over 100 local smoking by-laws in Ontario, likely consumer benefits of higher qual- month, with admissions increasing 25 per- Canada over the period 1997-2004. He is ity with the potentially higher production cent during the first five days of the month. able to control for demographic character- costs that result from increased not-for- This cycle is driven largely by recipients of istics, year fixed effects, and county fixed profit activity in a nursing home services Supplemental Security Income (SSI). SSI effects. Because he observes the respon- market area. They compare consumer ben- recipients also experience an abrupt 22 dent’s report of the smoking policy at her efits and costs by observing empirically percent increase in within-hospital mortal- worksite, he can test directly for compli- how an increased market penetration of ity after receiving their checks on the first ance. Although the results indicate that not-for-profit facilities affects the utili- of the month. The authors also document local by-laws increase workplace bans in zation of private-pay nursing home care. pronounced monthly cycles in drug related the aggregate, Carpenter finds that the Increased (decreased) utilization of nurs- crimes. On the first of the month, arrests effects are driven entirely by blue collar ing home care reflects that the consumer for drug possession and sale increase by workers. Among blue collar workers, local benefits associated with additional not- 20 percent and prostitution arrests decline by-laws significantly reduced the fraction for-profit nursing homes are greater (less) by 16 percent. These findings suggest that of worksites without any smoking restric- than consumer costs. The empirical results “full wallets” adversely affect some aid tions (that is, where smoking is allowed indicate that, from a consumer’s perspec- recipients and that policymakers should anywhere at work), by over half. These tive, too few not-for-profit nursing homes explore alternate disbursement regimes, local policies also improved health out- exist in the typical market area of the such as a staggered disbursement schedule comes: adoption of a local by-law signifi- United States. The policy implication is or in-kind support, that have the potential cantly reduced second hand smoke expo- that more quality of care per dollar might to reduce the rate of adverse events. sure among blue collar workers, by 25-30 be obtained by attracting a greater percent- It is widely believed that education percent, and workplace smoking laws age of not-for-profit nursing homes into improves health. And, empirical evidence did reduce smoking. For all of the out- most market areas. substantiating that the relationship is causal comes, Carpenter finds plausibly smaller Dobkin and Puller analyze the has progressed in recent years. A pinnacle in and insignificant estimates for white collar monthly patterns of adverse outcomes this progression is arguably Lleras-Muney’s and sales/service workers, the vast major- attributable to the consumption of illegal 2005 analysis of state compulsory school ity of whom worked in places with pri- drugs by recipients of government transfer law changes in the United States, which vately initiated smoking bans well before payments. They find evidence that certain were found to improve educational attain- local by-laws were adopted. Overall, these subpopulations on government cash aid ment and consequently to reduce mortal-

NBER Reporter Summer 2006 51 ity. Almond and Mazumder revisit these Evans and Heng use a restricted-use data- social costs associated with severe depres- results, noting that they are not robust to set of all births in California over a six-year sion and suicide; by plausible clinical and state time trends, even when the Census period to examine the effect of these early behavioral arguments, SSRIs could have sample is tripled and a coding error recti- discharge laws.They demonstrate that early either positive or negative effects on suicide fied. They use a new dataset with greater discharge laws considerably reduced the mortality. Randomized clinical trials on detail on health outcomes and statistical fraction of newborns and mothers who this topic have not been very informative power, yielding two primary findings: 1) were discharged early. They also find that because of small samples and other prob- they replicate Lleras-Muney’s results for an additional day in the hospital reduced lems. Ludwig, Marcotte, and Norberg use aggregate measures of health; and 2) the the probability of readmission by about data from 27 countries for up to twenty pattern of effects for specific health con- one percentage point for vaginal deliver- years to estimate the association between ditions appears to depart from theoretical ies with complications and for c-sections SSRI sales and suicide mortality using only predictions of how education should affect of all types. The former result is statisti- the variation across countries in the timing health. They also find that state differences cally significant at conventional levels but of when SSRIs were first sold that can be in vaccination rates against smallpox dur- the latter result is only significant at a p- explained by differences in the speed with ing the period of compulsory school law value of around 0.10. There was no statis- which countries approve new drugs for sale reform may account for the improvement tically significant change in 28-day new- more generally. This source of variation is in health and its association with educa- born readmission rates for babies whose plausibly unrelated to unmeasured mental tional attainment. mothers had uncomplicated vaginal deliv- health conditions or other factors that may During the 1980s and 1990s, the eries. Finally, although the statutes did not influence both SSRI sales and suicide out- lengths of postpartum hospital stays cover Medicaid patients and patients with comes. The authors find that an increase in declined for both vaginal and cesarean no insurance, their postpartum length of SSRI sales of 1 pill per capita (about a 13 births. Health professionals and policy- stay was affected by the changes in the law percent increase over 1999 sales levels) is makers expressed concern that shorter hos- as well. associated with a decline in suicide mortal- pital stays might jeopardize the health of Does drug treatment for depression ity of around 3-4 percent. These estimates both mothers and infants. The federal gov- with selective serotonin reuptake inhibi- imply a cost per statistical life saved of ernment and states responded by passing tors (SSRIs) increase or decrease the risk of around $66,000, far below most other gov- laws requiring that insurance carriers pro- completed suicide? The question is impor- ernment interventions to improve health vide coverage for longer postpartum stays. tant in part because of the substantial outcomes.

Environmental Economics

The NBER’s Working Group on Pfaff, Alexander van Geen, Joseph Discussant: Larry Karp, University of Environmental Economics met in Graziano, Iftikhar Hussein, Hasina California, Berkeley Cambridge on April 7-8. Group Director Momotaj, Roksana Sylvi, and Don Fullerton of the University of Texas Habibul Ahsan, Columbia University, William A. Pizer, Resources for organized this program: “Can Information Alone Change the Future, and Richard G. Newell, Behavior? Arsenic Contamination of “Indexed Regulation” Meredith Fowlie, University of Groundwater in Bangladesh” Discussant: Ian Sue Wing, Boston California, Berkeley, “Emissions Discussant: David Bloom, Harvard University Trading, Electricity Industry University and NBER Restructuring, and Investment in Hilary Sigman, Rutgers University and Pollution Abatement” Stephen Polasky, University of NBER, “Environmental Liability and Discussant: Erin Mansur, Yale Minnesota, and Nori Tarui, Columbia Redevelopment of Old Industrial Land” University University, “Environmental Regulation Discussant: Kathleen Segerson, in a Dynamic Model with Uncertainty University of Connecticut Malgosia Madajewicz, Alexander and Investment”

Policymakers increasingly rely on emissions trading program face differ- mize the total cost of achieving pollution emissions trading programs to address ent economic regulations and investment reductions. Fowlie analyzes an emissions the environmental problems caused by incentives in their respective industries, trading program that was introduced to air pollution. If polluting firms in an then emissions markets may fail to mini- reduce smog-causing pollution from large

52 NBER Reporter Summer 2006 stationary sources (primarily electricity environmental regulation with endoge- cies, where quantities are proportional to generators) in 19 eastern states. She devel- nous choice of emissions abatement tech- an index, such as output. They find that ops and estimates a model of a plant’s envi- nology by regulated firms and exogenous policy preferences hinge on additional ronmental compliance decision. Using learning about environmental damages parameters describing the first and sec- variation in state-level electricity-industry from emissions by a regulator. Invest- ond moments of the index and the ex post restructuring activity, she identifies the ments in abatement technology by one optimal quantity level. When the ratio of effect of economic regulation on pollu- firm that lower abatement costs for the these variables’ coefficients of variation tion permit market outcomes. She finds firm may also lower abatement costs of divided by their correlation is less than 2, first that plants in states that have restruc- other firms (technology spillovers). There indexed quantities are preferred to fixed tured electricity markets are less likely are two issues facing environmental regu- quantities. A slightly more complex con- to adopt more capital intensive compli- lators: setting regulation to achieve opti- dition determines when indexed quanti- ance options. Second, this economic reg- mal abatement given available informa- ties are preferred to prices. Applied to the ulation effect, together with a failure of tion; and setting regulations to achieve case of climate change, the authors find the permit market to account for spatial optimal investment given possible strate- that quantities indexed to GDP are pre- variation in marginal damages from pol- gic investment and technology spillovers. ferred to fixed quantities for about half lution, has resulted in increased health The authors compare taxes, standards, of the 19 largest emitters, including the damages. Had permits been defined in and marketable permits under flexibil- United States and China, while prices terms of units of damages instead of units ity, in which the policy is updated upon dominate for all other countries. of emissions, more of the mandated emis- learning new information, versus under Many communities are concerned sions reductions would have occurred in commitment, in which the policy is not about the reuse of old industrial land and restructured electricity markets, thereby updated. Flexible policy allows regula- believe that environmental liability is a hin- avoiding on the order of hundreds of pre- tion to reflect the most up-to-date infor- drance to redevelopment. However, with mature deaths per year. mation. However, under flexible policy, land price adjustments, liability might not Pfaff and his co-authors study how firms can invest strategically to influence impede redevelopment. Existing literature effectively information alone induces future regulation. The authors find that has found price reductions in response to people to incur the cost of avoiding a an optimal solution for both investment liability, but few studies have looked for health risk. Arsenic contamination of and abatement decisions can be achieved an effect on redevelopment.Sigman stud- the groundwater in Bangladesh provides under a flexible marketable permit scheme ies variations in state liability rules — spe- an unfortunate natural experiment. The in which the permit allocation to a firm is cifically, strict liability and joint and sev- authors find that the response to specific increasing in the firm’s investment. No eral liability — that affect the level and information about the safety of one’s well other policy scheme, taxes, auctioned per- distribution of expected private cleanup is large and rapid; having an unsafe well mits, or standards, under either flexibility costs. She explores the effects of this varia- raises the probability by 0.5 that the indi- or commitment, will guarantee achiev- tion on industrial land prices and vacancy vidual changes to another well within one ing an optimal solution. These results run rates in a panel of cities across the United year. The estimate of the impact of infor- counter to prior literature that finds that States from 1989 through 2000. In the mation is unbiased, because arsenic lev- price-based mechanisms are superior to estimated equations, joint and several lia- els are uncorrelated with individual char- quantity-based mechanisms, or that such bility reduces land prices and increases acteristics. The evidence suggests that a comparisons depend on conditions. vacancy rates in central cities. Neither a media campaign communicates general Weitzman (1974) revealed that prices price nor quantity effect is estimated from information about arsenic as effectively are preferred to quantities when mar- strict liability. The results suggest that as a more expensive door-to-door effort ginal benefits are relatively flat as - com liability is at least partly capitalized, but does. pared to marginal costs. Newell and Pizer does still deter redevelopment. Tarui and Polasky study dynamic extend this comparison to indexed poli-

NBER Reporter Summer 2006 53 Monetary Economics Program Meeting

The NBER’s Monetary Economics in Economies with Complementarities Discussant: John Williams, Federal Program met in Cambridge on April and Asymmetric Information” Reserve Bank of San Francisco 21. Jean Boivin, NBER and Columbia Discussant: Hyun Shin, Princeton Business School, and John Leahy, NBER University Jordi Gali, MIT and NBER, and and New York University, organized this Tommaso Monacelli, IGIER, program: Lars E.O. Svensson and Noah “Optimal Monetary and Fiscal Policy in Williams, Princeton University and a Currency Union” Bartosz Mackowiak and Mirko NBER, “Monetary Policy with Model Discussant: Paolo Pesenti, Federal Wiederholt, Humboldt University Uncertainty: Distribution Forecast Reserve Bank of New York Berlin, “Optimal Sticky Prices under Targeting” Rational Inattention” Discussant: Alexei Onatski, Columbia Virgiliu Midrigan, Ohio State Discussant: Laura Veldkamp, New York University University, “Menu Costs, Multi- University Product Firms, and Aggregate Frederic S. Mishkin, Columbia Fluctuations” George-Marios Angeletos, MIT University and NBER, and Niklas Discussant: Ariel Burstein, University and NBER, and Alessandro Pavan, Westelius, Hunter College, “Inflation of California, Los Angeles Northwestern University, “Efficient Use Band Targeting and Optimal Inflation of Information and Welfare Analysis Contracts”

In standard sticky price models, fre- characterize the efficient use of infor- provide an algorithm for finding the opti- quent and large price changes imply that mation, which allows them to address mal policy, as well as solutions for arbi- the aggregate price level responds quickly whether such heightened sensitivity is trary policy functions. This allows them to nominal shocks. Mackowiak and socially undesirable. They show how the to compute and plot consistent distri- Wiederholt present a model in which efficient use of information trades off bution forecasts — fan charts — of target price setting firms optimally decide what volatility for dispersion, and how this variables and instruments. Their methods to pay attention to, subject to a constraint relates to the socially optimal degree of hence extend certainty equivalence and on information flow. When idiosyncratic coordination. Finally, they show how the “mean forecast targeting” to more general conditions are more variable or more relation between equilibrium and effi- certainty non-equivalence and “distribu- important than aggregate conditions, cient use of information is instrumental in tion forecast targeting.” then firms pay more attention to idiosyn- understanding the social value of informa- Inflation band targeting is a simpler cratic conditions than to aggregate con- tion. They conclude with a few applica- form of an inflation contract that is widely ditions. When the authors calibrate the tions, including production externalities, used in practice and is more politically model to match the large average absolute Keynesian frictions, inefficient fluctua- tenable than alternative strategies, such size of price changes observed in the data, tions, efficient market competition, and as appointment of a conservative central prices react fast and by large amounts to large Cournot and Bertrand games. banker, or an optimal, pecuniary, inflation idiosyncratic shocks, but prices react only Svensson and Williams examine contract. Mishkin and Westelius provide slowly and by small amounts to nominal optimal and other monetary policies in the first theoretical treatment (that they shocks. Nominal shocks have persistent a linear-quadratic setup with a relatively know of ) of how inflation target bands real effects. The authors use their model general form of model uncertainty: so- can be designed to mitigate the time- to investigate how the optimal allocation called Markov jump-linear-quadratic sys- inconsistency problem. Their paper ana- of attention and the dynamics of prices tems extended to include forward-looking lyzes inflation target ranges in the context depend on the firms’ environment. variables. The form of model uncertainty of a Barro-Gordon (1983) type model, Angeletos and Pavan analyze equi- that their framework encompasses but has a more realistic setting, in that librium and welfare for a tractable class includes: simple i.i.d. model deviations; the time-inconsistency problem stems not of economies with externalities, strate- serially correlated model deviations; esti- from the preferences of the central bank, gic complementarity or substitutabil- mable regime-switching models; more as in Barro-Gordon, but instead from ity, and incomplete information. They complex structural uncertainty about very political pressures from the government. first characterize the equilibrium use of different models, for instance, backward- They demonstrate that inflation target information and show how strategic pay- and forward-looking models; time-vary- bands, or a range, can achieve many of the off effects can heighten the sensitivity of ing central-bank judgment about the state benefits of these other strategies, provid- equilibrium actions to noise. Then they of model uncertainty; and so forth. They ing a possible reason why this strategy has

54 NBER Reporter Summer 2006 been used by so many central banks. Their the optimal monetary policy requires that ner price data collected in retail stores theoretical model also enables them to inflation be stabilized at the union level. to document that: 1) although the aver- outline how an inflation targeting range On the other hand, the relinquishment of age magnitude of price changes is large, should be designed optimally and how it an independent monetary policy, coupled a substantial number of price changes should change when there are changes in with nominal price rigidities, generates are small in absolute value; 2) the distri- the nature of shocks to the economy. a stabilization role for fiscal policy, one bution of non-zero price changes has fat Gali and Monacelli lay out a tracta- beyond the efficient provision of public tails; and 3) stores tend to adjust prices ble model for fiscal and monetary policy goods. Interestingly, the stabilizing role of goods in narrow product categories analysis in a currency union, and analyze for fiscal policy is shown to be desirable, simultaneously. He extends the standard its implications for the optimal design not only from the viewpoint of each indi- menu costs model to a multi-product of such policies. Monetary policy is con- vidual country, but also from that of the setting in which firms face economies of ducted by a common central bank, which union as a whole. In addition, this paper scale in the technology of adjusting prices. sets the interest rate for the union as a offers some insights on two aspects of pol- The model, because of its ability to repli- whole. Fiscal policy is implemented at the icy design in currency unions: the condi- cate this additional set of microeconomic country level, through the choice of gov- tions for equilibrium determinacy and the facts, can generate aggregate fluctuations ernment spending level. The model incor- effects of exogenous government spend- much larger than those in standard menu porates country-specific shocks and nom- ing variations. costs economies. inal rigidities. Under these assumptions, Midrigan uses a large set of scan-

Education Program Meeting

The NBER’s Program on Education of Boston, and Tim R. Sass, Florida Education: Evidence from the UK” met on April 27 in Cambridge. Program State University, “Classroom Peer Director Caroline M. Hoxby of Harvard Effects and Student Achievement” Andrew Leigh, Australian National University organized the meeting at University, “ Teacher Pay and Teacher which these papers were discussed: Michael Anderson, MIT, “Uncovering Aptitude” Gender Differences in the Effects of Caroline Hoxby, and Gretchen Early Intervention: A Reevaluation of Harry Krashinsky, University of Weingarth, Harvard University, the Abecedarian, Perry Preschool, and Toronto, “ How Would One Extra “Taking Race Out of the Equation: Early Training Projects” Year of High School Affect Academic School Reassignment and the Structure Performance in University? Evidence of Peer Effects” Erich Battistin and Barbara from a Unique Policy Change” Sianesi, Institute for Fiscal Studies, Mary A. Burke, Federal Reserve Bank “Misreported Schooling and Returns to

In the last and current decade, the ally appears to have been random, condi- ways) and peers who create an unfocused Wake County school district has reas- tional on a student’s fixed characteristics (bimodal or “schizophrenic”) classroom. signed numerous students to schools, such as race and income. Using panel data These results also indicate that, when we moving up to 5 percent of the student on students before and after they expe- properly account for the effects of peers’ population in any given year. Before 2000, rience policy-induced changes in peers, achievement and peers’ race, the student’s the explicit goal was balancing schools’ Hoxby and Weingarth explore which ethnicity, income, and parental education racial composition; after 2000, it was models of peer effects explain the data. have no, or at most very slight, effects. balancing schools’ income composition. Their results reject the models in which a Burke and Sass analyze a unique Throughout, finding space for the area’s peer has a homogeneous effect that does micro-level panel dataset encompassing rapidly expanding student population was not depend on the student’s own char- all public school students in grades 3- the most important concern. The reas- acteristics. They find support for models 10 in the state of Florida for each of the signments generate a very large number in which a student benefits from peers years 1999/2000 to 2003/4. The authors of natural experiments in which students who are somewhat higher achieving than are able to directly link each student and experience new peers in the classroom. himself but not very different. A student teacher to a specific classroom and thus As a matter of policy, exposure to an benefits least from peers who are very can identify each member of a student’s “experiment” should have been and actu- different (in either positive or negative classroom peer group. The ability to track

NBER Reporter Summer 2006 55 individual students through multiple the domain of total years of education. gle percent rise in the salary of a starting classrooms over time, and multiple classes However, there were no significant long- teacher boosts the average aptitude of stu- for each teacher, enables the authors to term benefits for boys. These conclusions dents entering teacher education courses control for many sources of spurious peer change little when allowance is made for by 0.6 percentile ranks, with the effect effects including fixed individual student attrition and possible violations of ran- being strongest for those at the median. characteristics and fixed teacher inputs, dom assignment. This result is robust to instrumenting for and allows them to compare the strength Battistin and Sianesi study the teacher pay using uniform salary sched- of peer effects across different groupings impact of misreported treatment status ules for public schools. Leigh also finds of peers and across grade levels. They on the estimation of causal treatment some evidence that pay dispersion in the are also able to compare the effects of effects. Although the bias of matching- non-teaching sector affects the aptitude of fixed versus time-varying peer character- type estimators computed from misclas- potential teachers. istics. The authors find mixed results on sified data cannot, in general, be signed, Krashinsky uses a unique policy the importance of peers in the linear-in- the authors show that the bias is most change in Canada’s most populous prov- means model, and resolve some of these likely to be downward if misclassification ince, Ontario, to provide direct evidence apparent conflicts by considering non- does not depend on variables entering the on the effect of reducing the length of linear specifications of peer effects. Their selection-on-observables assumption, or if high school on student performance in results suggest that some grouping by abil- it only depends on such variables via the university. In 1999, the Ontario govern- ity may create Pareto improvements over propensity score index. They extend the ment eliminated the fifth year of - edu uniformly mixed classrooms. In general, framework to multiple treatments and cation from its high schools, and man- they find that contemporaneous behavior then provide results to bound the returns dated a new four-year program. This wields stronger influence than peers’ fixed to a number of educational qualifications policy change created two cohorts of stu- characteristics. in the United Kingdom, semi-parametri- dents who graduated from high school The view that the returns to invest- cally. By using the unique nature of their together and entered university with dif- ments in public education are highest for data, they assess the plausibility for the ferent amounts of high school education, early childhood interventions primarily two biases – from measurement error and thus making it possible to indentify the stems from several influential random- from omitted variables – to cancel each effect of one extra year of high school ized trials — Abecedarian, Perry, and the other out. education on university academic perfor- Early Training Project — that point to Can changes in teacher pay encour- mance. Using several different economet- super-normal returns to preschool inter- age more able individuals to enter the ric approaches on original survey data, ventions. Anderson implements a uni- teaching profession? So far, studies of the the results demonstrate that students who fied statistical framework to present a de impact of pay on the aptitude distribution receive one less year of high school edu- novo analysis of these experiments, focus- of teachers have provided mixed evidence cation perform significantly worse than ing on core issues that have received little on the extent to which altering teacher sal- their counterparts in all subjects, even attention in previous analyses: treatment aries represents a feasible solution to the after accounting for the age difference effect heterogeneity by gender, over-rejec- teacher quality problem. Using a unique between cohorts. Overall, both in terms tion of the null hypothesis due to multi- dataset of test scores for every individual of individual courses and grade point ple inference, and robustness of the find- admitted into an Australian university average, four-year graduates perform 5 ings to attrition and deviations from the between 1989 and 2003, Leigh explores percentage points, or approximately one- experimental protocol. The primary find- how changes in average pay or pay disper- half of a letter grade, lower than under- ing of this reanalysis is that girls garnered sion affect the decision to enter teacher graduates with one more year of high substantial short- and long-term benefits education courses in the eight states and school education. from the interventions, particularly in territories that make up Australia. A sin-

56 NBER Reporter Summer 2006 Program Meeting on Children

The NBER’s Program on Children Todd Elder and Darren Lubotsky, Elizabeth O. Ananat, MIT, and Daniel met in Cambridge on April 28. Program University of Illinois, Urbana- M. Hungerman, University of Notre Director Jonathan Gruber of MIT orga- Champaign, “Kindergarten Entrance Dame, “The Pill and Reproductive nized the meeting. These papers were Age and Children’s Achievement: Behavior of Young Women” discussed: The Impact of State Policies, Family Background, and Peers” Margherita Fort, University of Padova, Samuel Berlinski, University “Education and the Timing of Births: College, London; Sebastian Galiani, Gustavo J. Bobonis, University Evidence from a Natural Experiment in Universidad de San Andres; and of Toronto, and Frederico Finan, Italy” Paul Gertler, University of University of California, Berkeley, California, Berkeley and NBER, “Endogenous Peer Effects in School Emily Oster, University of Chicago, “The Effect of Pre-primary Education Participation” “Does a Rising Tide Lift All Boats on Primary School Performance” Evenly? Health Investments and Gender Inequality in India”

Although the theoretical case for uni- countries, they find that children who The use of experimental designs has versal pre-primary education is strong, the are older when they enter kindergarten enabled researchers to identify social empirical foundation is weak. Berlinski, have higher reading and math test scores. interactions or neighborhood effects on Galliani, and Gertler contribute to the Older children are also less likely to be individual behavior. However, a remain- empirical case by investigating the effect held back later and less likely to be diag- ing obstacle in the literature has been of a large expansion of universal pre-pri- nosed with a learning disability. Next the the inability to distinguish between peer mary education on subsequent primary authors ascertain why entrance age affects effects that are determined by a person’s school performance in Argentina. They achievement. They find evidence of sub- reference group behavior (endogenous estimate that one year of pre-primary stantial heterogeneity in the impact of peer effects) and effects that are gener- school increases average third grade test entrance age, with richer children benefit- ated as a result of specific background scores by 8 percent of a mean, or by 23 ing considerably more than poorer chil- characteristics of the groups themselves percent of the standard deviation, of the dren. This is consistent with the idea that (contextual peer effects). Bobonis and distribution of test scores. They also find wealthy parents are more willing or more Finan identify and estimate endogenous that pre-primary school attendance posi- able to develop their children’s human peer effects on children’s school partici- tively affects student’s self-control in the capital prior to the start of kindergar- pation decisions using evidence from the third grade as measured by behaviors such ten. Although older children in kinder- Progresa program. Under Progresa, pay- as attention, effort, class participation, garten tend to be taller, the authors find ments were provided to poor mothers and discipline. no evidence of heterogeneity in this cor- conditional upon school enrollment of The average age at which children relation. They conclude that older chil- their children. Because program eligibil- enter kindergarten has climbed steadily dren tend to have more cognitive skills ity was randomly assigned, the authors in the past 25 years. Using data from the and better preparation prior to entering use this exogenous variation in school Early Childhood Longitudinal Study and kindergarten, and that physical maturity participation to identify peer effects on the National Educational Longitudinal is not the primary cause of the entrance the school enrollment of ineligible chil- Study of 1988, Elder and Lubotsky age effect. Finally, school entrance cutoffs dren residing in the same communities. examine the effect of age at entrance to also influence the average age of classes, They find that peers have considerable kindergarten on achievement test perfor- and the authors are able to separately influence on the enrollment decision of mance, grade retention, and diagnoses of identify the influence of an individual’s program-ineligible children, and these learning disabilities. State kindergarten own age at entry from the influence of effects are nonlinear and concentrated cutoff dates generate two sources of plau- his or her classmates’ age. They find some among children from relatively poorer sibly exogenous variation in the age at evidence for both positive and negative households. These findings imply that which children enter kindergarten, which peer effects: conditional on a child’s own educational policies aimed at encouraging these authors use to estimate instrumental age, older classmates increase test scores, enrollment can produce large social mul- variables models of the effect of entrance but also increase the probability of being tiplier effects. age on outcomes. Consistent with recent retained in grade or being diagnosed with A growing literature documents how work in the United States and other a learning disability. the availability of oral contraception

NBER Reporter Summer 2006 57 affected the outcomes of young women effects while controlling for self-selection first part of her paper explores the prox- in the 1960s and 1970s, but the effects of of women into education. Identification imate causes of the gender imbalance in oral contraception on women’s fertility relies on exogenous variation in schooling mortality in India. She finds that a large remains disputed. In this paper, Ananat induced by a mandatory school reform share of the gender imbalance (about and Hungerman examine whether rolled out nationwide in Italy in the early 30 percent) can be explained by differ- increased access to the pill affected wom- 1960s. Findings based on Census data ential access to vaccination. The second en’s fertility and whether the pill sub- (Italy, 1981) suggest that a large fraction part of the paper estimates the effect of stituted for other fertility technologies. of the women affected by the reform post- changes in access to vaccination on gender Using census data, the authors document pone the time of the first birth but catch inequality. Oster argues that the direction that access to the pill led to falling short- up with this fertility delay before turning of these effects is not obvious. A simple term fertility rates for young women. 26. There is some indication that the fer- model of (gender-biased) parental invest- They then use two different datasets to tility behavior of these women is different ments, and empirical work using varia- examine the impact of legal oral contra- from that of the average women in the tion in access to vaccination, both suggest ception on unwanted pregnancies, and in population. that initial increases in vaccination avail- particular on abortions, in order to iden- Gender inequality in South Asia is ability from low levels will increase gen- tify whether these fertility technologies an important policy issue; gender imbal- der inequality; further increases will then were viewed to some extent as substitut- ances in mortality have been of partic- decrease inequality. This non-monotonic able by young women. In both datasets, ular concern. Policymakers often argue pattern is also reflected in differences in the authors find that access to oral contra- that increasing the level of development mortality. This result may shed light on ception lowers the number of abortions and access to health care are crucial to the contrast between the cross-sectional for young women. addressing this inequality. Oster analyzes and time-series evidence on gender and Fort assesses the causal effects of the relationship between access to child development. education on the timing of first order health investments and gender inequal- births, allowing for heterogeneity in the ity in those investments in India. The

Higher Education The NBER’s Working Group on on Academic Achievement “ University and NBER Higher Education met in Cambridge Discussant: William Becker, Indiana on April 28. Group Director Charles T. University John Bound, University of Michigan Clotfelter of Duke University organized and NBER; Michael Lovenheim, this program: James D. Adams, Rensselaer University of Michigan; and Sarah Polytechnic Institute and NBER, Turner, University of Virginia and Jill M. Constantine and Neil S. and J. Roger Clemmons, University NBER, “Understanding the Increased Seftor, Mathematica Policy Research, of Florida, “The Growing Allocative Time to the Baccalaureate Degree” “A Study of the Effect of Talent Search Inefficiency of the U.S. Higher Discussant: Thomas Kane, Harvard on Secondary and Postsecondary Education Sector” University and NBER Outcomes in Florida, Indiana, and Discussant: Richard Jensen, University Texas” of Notre Dame Peter Arcidiacono, Duke University; Discussant: Scott Carrell, Dartmouth Shakeeb Khan, University of College David Figlio, University of Florida Rochester; and Jacob L. Vigdor, and NBER; Colleen Donovan, Duke University and NBER, “Race- Florian Hoffman, University of University of California, Berkeley; Conscious Admissions and Inter-Racial Toronto, and Philip Oreopoulos, and Mark Rush, University of Florida, Contact: Does Homophily Matter?” University of Toronto and NBER, “Cramming: The Effects of School Discussant: Jesse Rothstein, Princeton “Do Better Professors Produce Better Accountability on College Study University and NBER Students? The Effects of Objective and Habits and Performance” Subjective Measures of Teacher Quality Discussant: Christopher Avery, Harvard

Low-income students and students typically are less likely than middle- and school and attend college, and are thus whose parents have not attended college upper-income students to complete high less likely to reap the benefits of attend-

58 NBER Reporter Summer 2006 ing college. By providing information on first-year students taking same courses hint at growing financial pressures on U.S. the types of high school courses students but with different instructors is small, public universities. should take to prepare for college and on but not trivial: a two standard deviation School accountability -- the practice the financial aid available to pay for col- switch in instructor quality would be of evaluating schools on the basis of the lege, the Talent Search program seeks to expected to lower the likelihood of drop- observed performance of students and address substantial informational hurdles. ping the course by 1.5 percentage points, rewarding and punishing schools accord- Using a large amount of administrative and increase the number of course sub- ing to these evaluations -- is ubiquitous data compiled in Florida, Indiana, and jects enrolled in the following year by less in the world today, with nations on every Texas for one complete cohort of stu- than 0.05 of its standard deviation. They continent experimenting with such poli- dents, Constantine and Seftor were able also find little evidence that student eval- cies. While there has been considerable to use complex propensity score match- uations of teacher quality significantly research attention paid to the effects of ing models to identify nonparticipating relate to student achievement. school accountability plans on the stan- students who were most similar to Talent Adams and Clemmons present new dardized test scores of average students or Search participants. They find that Talent evidence on research and teaching pro- low-performing students, as well as evi- Search participants were more likely than ductivity in universities. Their findings dence concerning the incentives embed- comparison students to apply for fed- are based on a panel that covers 1981- ded within school accountability plans, eral financial aid and enroll in public 99 and includes 102 top U.S. schools. there has been no published research to postsecondary institutions in all three Faculty size grows at 0.6 percent per year date investigating the effects of these plans states. These findings suggest that assist- compared with growth of 4.9 percent on the high end of the academic distri- ing low-income students who have col- in the industrial science and engineering bution: those students who would surely lege aspirations to overcome information workforce. Measured by papers and cita- have attained proficiency in the absence barriers—an important objective of the tions per researcher, productivity grows of school accountability plans. Figlio, Talent Search program—may be effective at 1.4-6.7 percent per year and produc- Donovan, and Rush seek to fill this void; in helping these students achieve their tivity and its rate of growth are higher in they exploit data from a state that changed aspirations. private than public universities. Measured the basis of its accountability system in Teaching is the central task that col- by baccalaureate and graduate degrees 1999. This change directly influenced a leges and universities perform for stu- per teacher, teaching productivity grows large number of schools that immediately dents. Differences in teaching quality, at 0.8-1.1 percent per year and growth either transitioned from being threatened across instructors and institutions, thus is faster in public than private universi- with sanctions to not being threatened at may play a key role in determining stu- ties. A decomposition analysis shows that all, or vice versa. Using this identification dents’ academic experiences, interests, growth in research productivity within strategy, they can measure the impact on and successful transition into the labor universities exceeds overall growth. This students of the school they attend either force. Yet research about the importance is because research shares grow more rap- becoming threatened or becoming less of teacher quality focuses almost exclu- idly in universities whose productivity threatened. In order to implement this sively on the primary or secondary level. grows less rapidly. Likewise, the research identification strategy, the authors use a Hoffmann and Oreopoulos use adminis- share of public universities increases even remarkable dataset from a large selective trative data from a large Canadian univer- though productivity grows less rapidly in public university in the state in question. sity between 1996 and 2005, matched to public universities. Together, these find- They observe that school accountabil- course instructors and instructors’ teach- ings imply that allocative efficiency of U.S. ity plans have the potential to substan- ing evaluations. Their main approach is to higher education may have declined dur- tially affect high-achieving students’ per- use the fact that many entering first-year ing the late twentieth century. Regression formance and study habits in college. students end up with different instructors analysis of individual universities finds They observe that accountability systems, because of scheduling conflicts or year-to- that R and D stock, endowment, and whether they are based on a low-level year replacements. They estimate the over- post-doctoral students increase research test of basic skills or based on a higher- all influence of postsecondary instruc- productivity, that the effect of nonfederal level standards-based assessment, tend to tors on course dropout, enrollment, and R and D stock is less, and that research lead to increased cramming behavior and grade outcomes by estimating the vari- is subject to decreasing returns. Since poorer study habits in college. However, ance of the value added that instruc- the nonfederal R and D share grows and the two types of accountability systems tors contribute to various outcome mea- is much higher in public universities, apparently lead to very different outcomes sures of academic achievement. They also this could account for some of the ris- in college, as measured by course grades. examine more direct effects by estimat- ing allocative inefficiency. The evidence An accountability system based on a low- ing the consequences of entering a class for decreasing returns in research also level test of basic skills is associated with with an instructor who ranks high or low suggests limits on scale that restrict the unambiguously worse performance by in subjective teaching evaluations. Their ability of more efficient institutions to students in college. On the other hand, an main finding is that the variance among expand. Besides all this, the data strongly accountability system based on a rigorous

NBER Reporter Summer 2006 59 standards-based assessment apparently diation, which lengthens the course of employed and greater propensity to trans- results in significantly improved mathe- study; inability to finance full-time atten- fer among institutions. matics performance in college, as well as dance, requiring part-time enrollment and Two recent U.S. Supreme Court rul- improved performance in other technical employment; or simply a desire to extend ings have armed the constitutionality of mathematics-based courses such as chem- the consumption experience of collegiate certain types of racial preferences in col- istry, economics, engineering and phys- life. The consequences of extended time- lege admissions. The main premise behind ics that were not directly covered by the to-degree may include individual loss of the Supreme Courts decisions was that accountability system, with no ill effects earnings, as well as the social cost of affirmative action benefits not only the on performance in less technical courses. potentially reduced economic growth as minority students targeted by the policy, These results indicate that the design of the supply of college-educated workers is but majority students as well. The pur- accountability systems is critically impor- limited by tradeoffs between school and ported benefit to majority students is tant in determining the degree to which work. Bound, Lovenheim, and Turner rooted in an increased likelihood of inter- high-performing students obtain skills find that the increase in time to degree is racial contact associated with increased that help them succeed in college. localized among graduates of non-selec- minority representation on campus. Time to completion of the baccalaure- tive public colleges and universities. They However, affirmative action itself may not ate degree has increased markedly among find no evidence that changes in student foster these relationships. Arcidiacono, college graduates in the United States over characteristics, including pre-collegiate Khan, and Vigdor show that inter-racial the last two decades. Between the cohorts achievement or parental characteristics, relationships are more likely to form graduating from high school in 1972 and explain the observed increase in time to among students with comparable test 1992, the percent of graduates receiving a degree. The increase in time to degree has scores. Their simulations suggest that less degree within 4 years dropped from 57.6 been the largest within states that have the aggressive admissions policies would actu- percent to 44 percent. Among the rea- largest increases in cohort size, consistent ally increase inter-racial contact by reduc- sons that students may extend the col- with dilution in resources per student at ing the disparities between majority and legiate experiences beyond the four-year public colleges. As proximate causes, the minority student characteristics. norm are: the need for academic reme- authors find evidence of increased hours

Health Care Program Meeting The NBER’s Program on Health University; and Jonathan Ketcham, Market for Corporate Care met in Cambridge on May 5. Arizona State University, “Specialization Control” NBER Research Associate Douglas O. and Sorting in the Obstetrics Market” Staiger of Dartmouth College organized Avi Dor, Case Western Reserve this program: Tomas J. Philipson, University of University and NBER, and William Chicago and NBER, and Anupam B. Encinosa, AHRQ, “Does Cost-Sharing Amitabh Chandra, Harvard University Jena, University of Chicago, “Surplus Affect Compliance? Insurance and the and NBER; Jonathan Gruber, MIT Appropriation from R&D and Market for Prescription Drugs” and NBER; and Robin McKnight, Health Care Technology Assessment University of Oregon and NBER, Procedures” David M. Cutler, Harvard University “Medical Price Sensitivity and Optimal and NBER, and Angus S. Deaton Health Insurance for the Elderly” Leemore Dafny, Northwestern and Adriana Lleras-Muney, University and NBER, and David Princeton University and NBER, “The Sean Nicholson, Cornell University Dranove, Northwestern University, Determinants of Mortality”(NBER and NBER; Andrew Epstein, Yale “Regulatory Exploitation and the Working Paper No. 11963)

As private insurers and the govern- question of the sensitivity of medical con- California Public Employees Retirement ment attempt to constrain elderly medical sumption to its price, but RAND did not System (CalPERS) enacted a series of sub- spending in the coming years, a first-order include the elderly in its HIE. The purpose stantial copayment increases for both active consideration is the price sensitivity of the of Gruber, Chandra, and McKnight’s employees and retirees, first for the state’s medical consumption of the elderly. For paper is to remedy this deficiency by study- PPO plans, and then for its HMO plans. the non-elderly, the famous RAND Health ing a major set of copayment changes in a The result was a staggered set of copay- Insurance Experiment (HIE) addressed the modern, managed care environment. The ment changes that allow these authors to

60 NBER Reporter Summer 2006 carefully evaluate the impact on the medi- tively low observed risk, and there is more tem, presenting hospitals with an opportu- cal care utilization of the elderly. To evalu- than twice as much variation across physi- nity to increase operating margins by 5 or ate these policy changes, the authors have cians within a practice in patients’ observed more percentage points simply by “upcod- compiled a comprehensive database of all health than one would expect if weekday ing” patients to more lucrative codes. The medical claims for those enrolled contin- patients were randomly assigned. authors find that “room to upcode” is a uously in several of the CalPERS plans. Given the rapid growth in health care statistically and economically significant They find that both physician office visits spending that is often attributed to tech- predictor of for-profit but not of not-for- and prescription drug utilization are price nological change, many private and public profit acquisitions in the period immedi- sensitive among the elderly, although the institutions are grappling with how to best ately following this policy change. They elasticities are modest, as with the RAND assess and adopt new health care technolo- also find that hospitals acquired by for- HIE. Unlike the HIE, however, this paper gies. The leading technology adoption cri- profit systems subsequently upcoded more finds significant “offset” effects in terms of teria proposed in theory and used in prac- than a sample of similar hospitals that were increased hospital utilization in response tice involve so called “cost-effectiveness” not acquired, as identified by propensity to the combination of higher copayments measures. However, little is known about scores. These results suggest that firms that for physicians and prescription drugs. The the dynamic efficiency implications of do not fully exploit regulatory loopholes most chronically ill individuals are equally such criteria, in particular how they influ- are vulnerable to takeover. responsive to copay increases in terms of ence the R and D investments that make Insurance for prescription drugs is their reduced use of physician care or pre- technologies available in the first place. characterized by two regimes: flat copay- scription drugs, but there are much larger Philipson and Jena argue that such criteria ments and variable coinsurance. Dor and offset effects for these populations, so that implicitly concern maximizing consumer Encinosa develop a simple model to show there is little net gain from higher copay- surplus, which many times is consistent that patient compliance is lower under ments for that group. This suggests that with maximizing static efficiency after an coinsurance because of uncertainty in cost copayment increases targeted to health innovation has been developed. Dynamic sharing. Empirically, the authors derive would be part of an optimal health insur- efficiency, however, concerns aligning the comparable models for compliance behav- ance arrangement for the elderly. social costs and benefits of R and D and ior in the two regimes. Using claims data The welfare implications of variations is therefore determined by how much of from nine large firms, they focus on diabe- in how physicians treat patients depend the social surplus from the new technol- tes, a common chronic condition that leads on whether patients have different opti- ogy is appropriated as producer surplus. The to severe complications when inappropri- mal treatments and are treated by physi- authors analyze the relationship between ately treated. In the coinsurance model, an cians who are likely to provide those opti- cost-effectiveness measures and the degree increase in the coinsurance rate from 20 to mal treatments. Epstein, Ketcham, and of surplus appropriation by innovators 75 percents results in the share of persons Nicholson examine the extent to which driving dynamic efficiency. They illustrate who never comply to increase by nearly 10 expecting mothers direct themselves, or are how to estimate the two for the new HIV/ percent and reduces the share of fully com- directed to, particular physicians based on AIDS therapies that entered the market pliant persons by almost 25 percent. In their preferences for physicians’ treatment after the late 1980s and find that only 5 the co-payment model, an increase in the styles and the patients’ health conditions. percent of the social surplus is appropri- copayment from $6 to $10 results in a 6.2 The authors capitalize on the largely ran- ated by innovators. They show how this percent increase in the share of never-com- dom assignment to weekdays of weekend finding can be generalized to other exist- pliers, and a concomitant 9 percent reduc- patients because of physicians’ call sched- ing cost-effectiveness estimates by deriv- tion in the share of full compliers. Similar ules and the concentration of induced ing how those estimates identify innovator results hold when the level of cost-shar- deliveries and scheduled c-sections, both appropriation for a set of studies of over ing is held constant across regimes. While opportunities for a patient to choose her 200 drugs. They find that these studies non-compliance reduces expenditures on physician. Using Florida and New York implicitly support a low degree of appro- prescription drugs, it may also lead to discharge data from 1999 to 2004 linked priation as well. Despite the high annual increases in indirect medical costs attrib- to information on physician practices, the cost of drugs to patients, very low shares of utable to avertable complications. Using authors find that one-third of the varia- social surplus may go to innovators, which available aggregate estimates of the cost tion in treatment styles across physicians may imply that cost-effectiveness is too of diabetic complications, the authors cal- is attributable to patient-physician match- high in a dynamic efficiency sense. culate that the $6-$10 increase in copay- ing on unobserved characteristics, which Dafny and Dranove evaluate the pos- ment would have the direct effect of reduc- implies that a considerable part of the sibility that a failure to exploit regulatory ing national drug spending for diabetes variation in medical treatment rates may loopholes could motivate corporate take- by $125 million. However, the increase enhance welfare. In one-third of the group overs. They use the U.S. hospital industry in non-compliance rates is expected to practices, certain physicians specialize in in 1985-96 as a case study. A 1988 change increase the rate of diabetic complications, treating weekday patients with relatively in Medicare rules widened a pre-existing resulting in an additional $360 million in high observed risk and others with rela- loophole in the Medicare payment sys- treatment costs. These results suggest that

NBER Reporter Summer 2006 61 both private payers and public payers may poor countries too of about 30 years. This countries, and across groups within coun- be able to reduce overall medical costs by difference persists despite the remarkable tries. While there is no consensus about the switching from coinsurance to copayments progress in health improvement in the last causal mechanisms, they tentatively iden- in prescription drug plans. half century, at least until the HIV/AIDS tify the application of scientific advance Mortality rates have fallen dramati- pandemic. In both the time-series and the and technical progress (some of which cally over time, starting in a few countries cross-section data, there is a strong correla- is induced by income and facilitated by in the eighteenth century, and continuing tion between income per capita and mor- education) as the ultimate determinant to fall today. In just the past century, life tality rates, a correlation that also exists of health. Such an explanation allows a expectancy has increased by over 30 years. within countries, where richer, better-edu- consistent interpretation of the histori- At the same time, mortality rates remain cated people live longer. Cutler, Deaton, cal, cross-country, and within-country evi- much higher in poor countries, with a dif- and Lleras-Muney review the determi- dence. They downplay direct causal mecha- ference in life expectancy between rich and nants of these patterns: over history, over nisms running from income to health.

Political Economy

The NBER’s Political Economy Discussant: Luigi Zingales, Harvard and NBER Program met in Cambridge on May 6. University and NBER Program Director Alberto Alesina of Edward L. Glaeser and Andrei Shleifer, Harvard University organized the meet- Alberto Alesina; William Easterly, Harvard University and NBER, and ing at which these papers were discussed: New York University; and Janina Giacomo Ponzetto, Harvard University, Matuszeski, Harvard University, “Why Does Democracy Need Enrico Spolaore, Tufts University, “Artificial States” Education?” and Romain Wacziarg, Stanford Discussant: Ernesto Dalbo, University of Discussant: Simon Johnson, MIT and University and NBER, “The Diffusion California, Berkeley NBER of Development” Discussant: Francesco Caselli, London Efraim Benmelech, Harvard University, Raghuram G. Rajan, IMF and NBER, School of Economics and NBER and Tobias J. Moskowitz, University and Luigi Zingales, “The Persistence of Chicago and NBER, “The Political of Underdevelopment: Institutions, Paola Giuliano and Antonio Economy of Financial Regulation: Human Capital, or Constituencies?” Spilimbergo, IMF, and Giovanni Evidence from U.S. State Usury Laws in Discussant: Pedro Dalbo, Brown Tonon, Dana-Farber Cancer Institute, the 18th and 19th Century” University “Genetic Map for Economist” Discussant: Antoinette Schoar, MIT

Spolaore and Wacziarg study the determinants of income differences, as a valid instrument for disentangling barriers to the diffusion of develop- particularly for differences in human the causal relationship between culture ment across countries over the very long capital and institutions. This paper dis- and economic outcomes? Guiliano, run. They find that genetic distance, a cusses the economic mechanisms that Spilimbergo, and Tonon examine how measure associated with the amount are consistent with these facts. It pres- economists may interpret the corre- of time elapsed since two populations’ ents a framework in which differences lation between genetic distance and last common ancestors, bears a statisti- in human characteristics transmitted cultural and economic variables. They cally and economically significant cor- across generations — including cultur- argue that currently used measures of relation with pairwise income differ- ally transmitted characteristics — can genetic distance are a poor proxy for ences, even after controlling for various affect income differences by creating cultural differences. Rather, genetic dis- measures of geographical isolation, and barriers to the diffusion of innovations, tance, being determined among other other cultural, climatic, and historical even when they have no direct effect on things by geographical barriers, reflects differences. These results hold not only productivity. The empirical evidence transport costs between countries. To for contemporary income differences over time and space is consistent with demonstrate this point, the authors but also for income differences mea- this “barriers” interpretation. construct a new measure of geographic sured since 1500, and for income differ- What does genetic distance distance within Europe that takes into ences within Europe. Similar patterns between populations measure? And, account the existence of major geo- of coefficients exist for the proximate is it a good proxy for culture as well graphical barriers. They show that this

62 NBER Reporter Summer 2006 measure explains both genetic distance associated with lower future economic explain why poor economic institu- and trade between European countries. growth rates, highlighting the endo- tions persist even in fairly democratic Artificial states are those in which geneity of financial development and but poor societies, and the second political borders do not coincide with growth. because it does not explain why poor a division of nationalities desired by Across countries, education and education is so persistent. Rajan and the people on the ground. Alesina, democracy are highly correlated. Zingales try to reconcile these two Easterly, and Matuszeski propose and Glaeser, Shleifer, and Ponzetto views by arguing that the underlying compute for all countries in the world empirically motivate and then model cause of underdevelopment is the ini- two new measures of the degree to a causal mechanism that explains this tial distribution of factor endowments. which states are artificial. One is based correlation. In their model, schooling Under certain circumstances, this leads on measuring how borders split ethnic teaches people to interact with others to self-interested constituencies that, in groups into two separate adjacent coun- and raises the benefits of civic partici- equilibrium, perpetuate the status quo. tries. The other measures how straight pation, including voting and organiz- In other words, poor education pol- land borders are, under the assumption ing. In the battle between democracy icy might well be the proximate cause the straight land borders are more likely and dictatorship, democracy has a wide of underdevelopment, but the deeper to be artificial. The authors then show potential base of support but offers (and more long lasting) cause is the ini- that these two measures seem to be only weak incentives to its defenders. tial conditions (like the distribution highly correlated with several measures Dictatorship provides stronger incen- of education) that determine political of political and economic success. tives, but to a narrower base. As edu- constituencies, their power, and their Benmelech and Moskowitz study cation raises the benefits of civic par- incentives. Although the initial condi- the political economy of financial reg- ticipation, it also raises the support for tions may well be a legacy of the colo- ulation by examining the determinants more democratic regimes relative to nial past, and may well create a perverse and effects of U.S. state usury laws dur- dictatorships. This increases the likeli- political equilibrium of stagnation, per- ing the 18th and 19th centuries. They hood of democratic revolutions against sistence does not require the presence argue that regulation is the outcome dictatorships, and reduces that of suc- of coercive political institutions. The of private interests using the coercive cessful anti-democratic coups. authors present some suggestive empir- power of the state to extract rents from Why is underdevelopment so per- ical evidence. On the one hand, such other groups. They find that strict- sistent? One explanation is that poor an analysis offers hope that the des- ness of usury coexists with other exclu- countries do not have institutions that tiny of societies is not preordained by sionary policies, such as suffrage laws can support growth. Because institu- the institutions they inherited through and lack of general incorporation or tions (both good and bad) are persis- historical accident. On the other hand, free banking laws, which also respond tent, underdevelopment is persistent. it suggests that we need to understand less to competitive pressures for repeal. An alternative view is that underde- better how to alter factor endowments Furthermore, the same determinants velopment comes from poor educa- when societies may not have the inter- of financial regulation that favor one tion. Neither explanation is fully sat- nal will to do so. group and limit access to others, are isfactory, the first because it does not

NBER Reporter Summer 2006 63 Market Microstructure The NBER’s Working Group Rochester “Liquidity in the Pricing of Syndicated on Market Microstructure met Laura Frieder and Rodolfo Martell, Loans” in Cambridge on May 12. NBER Purdue University, “On Capital Discussant: Amir Sufi, University of Research Associate Bruce Lehmann Structure and the Liquidity of a Firm’s Chicago of the University of California, San Stock” Diego, Duane Seppi of Carnegie Discussant: Naveen Khanna, Michigan Terrence Hendershott and Mark S. Mellon’s Tepper School of Business, State University Seasholes, University of California, and Avanidhar Subrahmanyam of the Berkeley, “Market Maker Inventories University of California, Los Angeles’s Ronald L. Goettler and Christine A. and Stock Prices” Anderson School of Management, orga- Parlour, Carnegie Mellon University, Discussant: Marc Lipson, University of nized the meeting. These papers were and Uday Rajan, University of Virginia discussed: Michigan, “Information Acquisition in a Limit Order Market” Amrut Nashikkar and Marti Michael J. Fleming, Federal Reserve Discussant: Dmitry Livdan, Texas Subrahmanyam, New York University, Bank of New York, and Monika A&M University “Latent Liquidity and Corporate Bond Piazzesi, University of Chicago Yield Spreads” and NBER, “Monetary Policy Anurag Gupta and Ajai Singh, Case Discussant: Michael Piwowar, U.S. Tick-by-Tick” Western Reserve University, and Allan Securities and Exchange Commission Discussant: Clara Vega, University of Zebedee, San Diego State University,

Analysis of high-frequency data determined. Consistent with the idea the market, after choosing whether to shows that yields on Treasury notes are that debt forces managers to make bet- purchase information about the com- highly volatile around FOMC ter investment decisions, they find that mon value. They may either post prices announcements, even though the aver- as leverage increases, equity bid-ask or accept posted prices. If a trader’s age effects of fed funds target rate sur- spreads decrease. Using the fitted val- order has not executed, then he ran- prises on such yields are fairly modest. ues from their first-stage regression, the domly re-enters the market, and may Fleming and Piazzesi partially resolve results from the second-stage regres- change his previous order. The model this puzzle by showing that yield sion further imply that as liquidity is thus a dynamic stochastic game with changes seem to depend not only on decreases, leverage increases. This is asymmetric information. The authors the surprises themselves, but also on consistent with the notion that manag- numerically solve for the equilibrium the shape of the yield curve at the time ers rely on debt financing when the cost of the trading game, and characterize of announcement. They also show that of equity financing increases. While equilibria with endogenous informa- the reaction of yields to FOMC controlling for the endogenous rela- tion acquisition. Agents with the low- announcements is sluggish, but that tionship between spreads and leverage est intrinsic benefit from trade have the much of this sluggishness can be attrib- greatly reduces the impact of spreads on highest value for information and also uted to the few inter-meeting moves. leverage, the results here suggest that a tend to supply liquidity. As a result, Market liquidity around FOMC single standard deviation increase in market observables, such as bid and announcements behaves in a manner spreads results in a 3 percent increase in ask quotes, in addition to transaction generally consistent with that found leverage. Not only do these results add prices, are informative about the com- for other announcements, although the to the understanding of the complex mon value of the asset. Asymmetric richness of FOMC announcement relationship between capital structure information creates a volatility multi- release practices induces differences in and liquidity, but they also shed light plier (prices are more volatile than the the market-adjustment process. on the determinants of leverage and fundamental value of the asset) that is Earlier literature on the capital bid-ask spreads. especially severe when the fundamental structure has only touched on a causal Goettler, Parlour, and Rajan volatility is high. In the latter case, the relation between liquidity and lever- model endogenous information acqui- time to execution of each type of agent age (that is, liquidity affects leverage). sition in a limit-order market for a sin- increases, and there is a change in the Frieder and Martell use a two-stage gle financial asset. The asset has a com- composition of trader types in the mar- least squares analysis to explore the mon value and each trader has a private ket at any given time. notion that these variables are jointly value for it. Traders randomly arrive at Gupta, Singh, and Zebedee exam-

64 NBER Reporter Summer 2006 ine whether banks price expected positively correlated with subsequent folio holdings database of the largest liquidity in syndicated loan spreads. changes. A portfolio that is long in custodian in the market, Nashikkar Using extensive data on U.S. term loans, stocks with the highest inventory posi- and Subrahmanyam relate the liquid- they show that banks have the ability to tions and short in stocks with the low- ity of corporate bonds, as measured by discern the expected liquidity of a loan est inventory positions has returns of their ease of market access, to the non- at the time of origination. More impor- 0.10 percent and 0.33 percent over default component of their corporate tantly, they show that loans with higher the next one and five days, respec- bond yields. They estimate the ease of expected liquidity have significantly tively. These findings empirically vali- access of a bond using a recently devel- lower spreads at origination, after con- date the causal mechanism — liquid- oped measure called latent liquidity, trolling for other determinants of loans ity supplier inventory — that underlies which weights the turnover of funds spreads such as borrower, loan, syn- models linking liquidity provision and holding the bond by their proportional dicate, and macroeconomic variables. asset prices. Inventories complement holdings of the bond. They use the Therefore, they identify a new factor past returns when predicting return credit default swap (CDS) prices of the (expected liquidity) being priced in reversals. A portfolio long on high- bond issuer to control for the credit syndicated term loans, which, in the inventory/low-return stocks and short risk of a bond. At an aggregate level, aggregate, results in an annual saving on low-inventory/high-return stocks they find a contemporaneous relation- of $1.5 billion to the borrowing firms yields 1.05 percent over the follow- ship between aggregate latent liquidity in their sample. Thus, for the first time ing five days. Order imbalances calcu- and the average non-default component in the literature, they document a link lated from signing trades relative to in corporate bond prices. Additionally, between the secondary market liquid- quotes also predict reversals and are for individual bonds, they find that ity of an asset and its pricing in the pri- complementary to inventories and past bonds with higher latent liquidity have mary market. returns. Finally, specialist inventories a lower non-default component of their Using eleven years of NYSE special- can be used to predict return continua- yield spread. Bonds that are held by ist data, Hendershott and Seasholes tions over a one-day horizon. funds exhibiting greater buying activ- examine daily inventory/asset price Recent research has shown that ity command lower spreads (are more dynamics. The unique length and default risk explains only part of the expensive), while the opposite is true breadth of their sample enables the total yield spread on risky corporate for those that exhibit greater selling first longer-horizon testing of market bonds relative to their riskless bench- activity. Also, the liquidity in the CDS making inventory models. They con- marks. One candidate for the unex- market has an impact on bond pricing, firm such models’ predictions: that spe- plained portion of the spread is a over and above bond-specific liquidity cialists’ positions are negatively cor- premium for the illiquidity in the cor- effects. related with past price changes and porate bond market. Using the port-

NBER Reporter Summer 2006 65 Current Working Papers

NBER Working Papers On-Line

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

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

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

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

NBER Working Papers Paper Author(s) Title 12111 Karen Clay Deprivation and Disease in Early Twentieth-Century Werner Troesken America 12112 Amy Finkelstein Testing for Adverse Selection with “Unused Observables” James Poterba 12113 Robert Kaestner Effects of Title IX and Sports Participation on Girls’ Xin Xu Physical Activity and Weight 12114 John H. Tyler Prison-Based Education and Re-Entry into the Mainstream Jeffrey R. Kling Labor Market 12115 Catherine Weinberger The Narrowing of the U.S. Gender Earnings Gap, 1959– Peter Kuhn 1999: A Cohort-Based Analysis 12116 Richard B. Freeman Learning from Other Economies: The Unique Institutional and Policy Experiments Down Under 12117 Philippe Aghion Exchange Rate Volatility and Productivity Growth: The Philippe Bacchetta Role of Financial Development Romain Ranciere 12118 Benjamin M. Friedman The Greenspan Era: Discretion, Rather Than Rules

66 NBER Reporter Summer 2006 Paper Author(s) Title

12119 David Popp Exploring Links Between Innovation and Diffusion: Adoption of NOx Control Technologies at U.S. Coal-Fired Power Plants 12120 Frank R. Lichtenberg The Impact of New Laboratory Procedures and Other Medical Innovations on the Health of Americans, 1990-2003: Evidence from Longitudinal, Disease-Level Data 12121 Linda S. Goldberg Distribution Margins, Imported Inputs, and the Insensitivity José Manuel Campa of the CPI to Exchange Rates 12122 Robert W. Fogel Why China is Likely to Achieve its Growth Objectives 12123 Dhaval Dave The Effects of Retirement on Physical and Mental Health Inas Rashad Outcomes Jasmina Spasojevic 12124 David M. Cutler Why Do Europeans Smoke More Than Americans? Edward L. Glaeser 12125 Francesco Caselli On the Theory of Ethnic Conflict Wilbur John Coleman II 12126 Susan Athey Empirical Models of Auctions Philip A. Haile 12127 Galina Hale Institutional Weakness and Stock Price Volatility Assaf Razin Hui Tong 12128 Edward L. Glaeser Why Does Democracy Need Education? Giacomo Ponzetto Andrei Shleifer 12129 Ricardo J. Caballero Zombie Lending and Depressed Restructuring in Japan Takeo Hoshi Anil K. Kashyap 12130 Ellen R. McGrattan Does Neoclassical Theory Account for the Effects of Big Lee E. Ohanian Fiscal Shocks? Evidence From World War II 12131 Johannes Van Biesebroeck Complementarities in Automobile Production 12132 Tomas J. Philipson Antitrust in the Not-For-Profit Sector Richard A. Posner 12133 James A. Robinson A Political Theory of the Soft Budget Constraint Ragnar Torvik 12134 Li Gan The Thick Market Effect on Housing Markets Transactions Qinghua Zhang 12135 John G. Shoven Political Risk versus Market Risk in Social Security Sita N. Slavov 12136 Ricardo J. Caballero Flight to Quality and Collective Risk Management Arvind Krishnamurthy 12137 Filippo Occhino How Occupied France Financed Its Own Exploitation in Kim Oosterlinck World War II Eugene N. White 12138 Eugene N. White Bubbles and Busts: The 1990s in the Mirror of the 1920s 12139 Claudia Goldin The Homecoming of American College Women: The Lawrence F. Katz Reversal of the College Gender Gap Ilyana Kuziemko 12140 Heinrich Hock The Dynamics of Age Structure, Dependency, and David N. Weil Consumption 12141 Gordon H. Hanson Illegal Migration from Mexico to the United States NBER Reporter Summer 2006 67 Paper Author(s) Title

12142 Charles Yuji Horioka The Causes of Japan’s “Lost Decade”: The Role of Household Consumption 12143 Jacob L. Vigdor The New Promised Land: Black-White Convergence in The American South, 1960–2000 12144 Jules H. van Binsbergen Optimal Decentralized Investment Management Michael W. Brandt Ralph S.J. Koijen 12145 Justine S. Hastings Preferences and Heterogeneous Treatment Effects in a Thomas J. Kane Public School Choice Lottery Douglas O. Staiger 12146 Clemens Sialm Investment Taxes and Equity Returns 12147 David N. Weil Population Aging 12148 Stephen M. Maurer Open Source Software: The New Intellectual Property Suzanne Scotchmer Paradigm 12149 John Y. Campbell Household Finance 12150 Elizabeth Oltmans Ananat Abortion and Selection Jonathan Gruber Phillip B. Levine Douglas O. Staiger 12151 Ravi Jagannathan Why Do IPO Auctions Fail? Ann E. Sherman 12152 Steven D. Levitt An Economist Sells Bagels: A Case Study in Profit Maximization 12153 Enrico Spolaore The Diffusion of Development Romain Wacziarg 12154 Sebastian Edwards The U.S. Current Account Deficit: Gradual Correction or Abrupt Adjustment? 12155 Thomas J. Kane What Does Certification Tell Us About Teacher Jonah E. Rockoff Effectiveness? Evidence from Douglas O. Staiger 12156 Mark Coppejans Consumer Demand Under Price Uncertainty: Empirical Donna Gilleskie Evidence from the Market for Cigarettes Holger Sieg Koleman Strumpf 12157 Paul Oyer The Macro-Foundations of Microeconomics: Initial Labor Market Conditions and Long-Term Outcomes for Economists 12158 Pierpaolo Benigno Optimal Inflation Targeting Under Alternative Fiscal Michael Woodford Regimes 12159 Phil Oreopoulos The Short- and Long-Term Career Effects of Graduating Till von Wachter in a Recession: Hysteresis and Heterogeneity in the Market Andrew Heisz for College Graduates 12160 Douglas A. Irwin The Impact of Federation on Australia’s Trade Flows 12161 Douglas A. Irwin Antebellum Tariff Politics: Coalition Formation and Shifting Regional Interests 12162 Douglas A. Irwin Tariff Incidence in America’s Gilded Age 12163 Sebastian Edwards The Relationship Between Exchange Rates and Inflation Targeting Revisited 12164 Nathan Nunn Putting the Lid on Lobbying: Tariff Structure and Long- Daniel Trefler Term Growth When Protection is for Sale

68 NBER Reporter Summer 2006 Paper Author(s) Title

12165 Robert F. Engle Execution Risk Robert Ferstenberg 12166 Gadi Barlevy Identification of Search Models with Initial Condition H.N. Nagaraja Problems 12167 Steven J. Davis The Flow Approach to Labor Markets: New Data Sources R. Jason Faberman and Micro-Macro Links John Haltiwanger 12168 Martin Feldstein The 2006 Economic Report of the President: Comment on Chapter One (The Year in Review) and Chapter Six (The Capital Account Surplus) 12169 Stefano Della Vigna The Fox News Effect: Media Bias and Voting Ethan Kaplan 12170 Edward J. Kane Inadequacy of Nation-Based and VaR-Based Safety Nets in the European Union 12171 Janet Currie Does Child Abuse Cause Crime? Erdal Tekin 12172 Lynne G. Zucker Movement of Star Scientists and Engineers and High-Tech Michael R. Darby Firm Entry 12173 Peter K. Schott The Relative Sophistication of Chinese Exports 12174 Roland G. Fryer, Jr. Belief Flipping in a Dynamic Model of Statistical Discrimination 12175 Torsten Persson Democratic Capital: The Nexus of Political and Economic Guido Tabellini Change 12176 Alan Manning Comprehensive versus Selective Schooling in England in Jörn-Steffen Pischke Wales: What Do We Know? 12177 Florian Zettelmeyer Scarcity Rents in Car Retailing: Evidence from Inventory Fiona Scott Morton Fluctuations at Dealerships Jorge Silva-Risso 12178 John Whalley The Post-MFA Performance of Developing Asia 12179 Elizabeth Cascio Public Preschool and Maternal Labor Supply: Evidence From the Introduction of Kindergartens into American Public Schools 12180 Andreas Ammermueller Peer Effects in European Primary Schools: Evidence from Jörn-Steffen Pischke PIRLS 12181 Sebnem Kalemli-Ozcan AIDS, Reversal of the Demographic Transition, and Economic Development: Evidence from Africa 12182 Shane Greenstein The Diffusion of the Internet and the Geography of the Jeff Prince Digital Divide in the United States 12183 Long Chen The Expected Value Premium Ralitsa Petkova Lu Zhang 12184 David M. Cutler Intensive Medical Care and Cardiovascular Disease Mary Beth Landrum Disability Reductions Kate A. Stewart 12185 David W. Galenson Analyzing Artistic Innovation: The Greatest Breakthroughs Of the Twentieth Century 12186 Anusha Chari Firm-Specific Information and the Efficiency of Investment Peter Blair Henry 12187 Serkan Arslanalp Debt Relief Peter Blair Henry NBER Reporter Summer 2006 69 Paper Author(s) Title

12188 Marianne Baxter What Determines Bilateral Trade Flows? Michael A. Kouparitsas 12189 Michael Bordo One World Money, Then and Now Harold James 12190 Allan Drazen Pork Barrel Cycles Marcela Eslava 12191 Andrew K. Rose Size Really Doesn’t Matter: In Search of a National Scale Effect 12192 Richard E. Baldwin Trade Liberalization with Heterogenous Firms Rikard Forslid 12193 Richard H. Clarida Are There Thresholds of Current Account Adjustment in Manuela Goretti the G7? Mark P. Taylor 12194 Richard H. Clarida G7 Current Account Imbalances: Sustainability and Adjustment 12195 Bronwyn H. Hall The Private Value of Software Patents Megan MacGarvie 12196 Pol Antràs Organizing Offshoring: Middle Managers and Luis Garicano Communication Costs Esteban Rossi-Hansberg 12197 Hope Corman The Effects of Health on Health Insurance Status in Fragile Anne Carroll Families Kelly Noonan Nancy E. Reichman 12198 Eiji Ogawa Stabilization of Effective Exchange Rates Under Common Junko Shimizu Currency Basket Systems 12199 Thomas Dee Do High School Exit Exams Influence Educational Brian A. Jacob Attainment or Labor Market Performance? 12200 Justin Wolfers Interpreting Prediction Market Prices as Probabilities Eric Zitzewitz 12201 Martin Feldstein The Effect of Taxes on Efficiency and Growth 12202 Alexandre Mas Pay, Reference Points, and Police Performance 12203 Andrew Ang Is IPO Underperformance a Peso Problem? Li Gu Yael V. Hochberg 12204 John Y. Campbell Intergenerational Risksharing and Equilibrium Asset Prices Yves Nosbusch 12205 Amy Finkelstein Redistribution by Insurance Market Regulation: Analyzing James Poterba a Ban on Gender-Based Retirement Casey Rothschild 12206 Daron Acemoglu Technology, Information, and the Decentralization of the Philippe Aghion Firm Claire Lelarge John Van Reenen Fabrizio Zilibotti 12207 Charles T. Clotfelter The Academic Achievement Gap in Grades 3 to 8 Helen F. Ladd Jacob L. Vigdor 12208 Lara Gardner The Effects of State Medicaid Policies on the Dynamic Donna Gilleskie Savings Patterns of the Elderly

70 NBER Reporter Summer 2006 Paper Author(s) Title

12209 Pablo T. Spiller Buy, Lobby, or Sue: Interest Groups’ Participation in Policy Sanny Liao Making — A Selective Survey 12210 Francis A. Longstaff An Empirical Analysis of the Pricing of Collateralized Arvind Rajan Debt Obligations 12211 Patrick Bayer Identifying Individual and Group Effects in the Presence of Stephen L. Ross of Sorting: A Neighborhood Effects Application 12212 Stefania Albanesi Home Production, Market Production, and the Gender Claudia Olivetti Wage Gap: Incentives and Expectations 12213 Michael B. Devereux Expectations and Exchange Rate Policy Charles Engel 12214 Charles Engel Portfolio Choice in a Monetary Open-Economy DSGE Akito Matsumoto Model 12215 Michael B. Devereux Expenditure Switching vs. Real Exchange Rate Charles Engel Stabilization: Competing Objectives for Exchange Rate Policy 12216 Nick Bloom Measuring and Explaining Management Practices Across John Van Reenen Firms and Countries 12217 Irene Brambilla Multinationals, Technology, and the Introduction of Varieties of Goods 12218 William D. Nordhaus Baumol’s Diseases: A Macroeconomic Perspective 12219 Pierpaolo Benigno Are Valuation Effects Desirable From a Global Perspective? 12220 Fang Cai International Diversification at Home and Abroad Francis E. Warnock 12221 Susumu Imai Crime and Young Men: The Role of Arrest, Criminal Hajime Katayama Experience, and Heterogeneity Kala Krishna 12222 Christian Leuz Do Foreigners Invest Less in Poorly Governed Firms? Karl V. Lins Francis E. Warnock 12223 Mark Grinblatt Sensation Seeking, Overconfidence, and Trading Activity Matti Keloharju 12224 Daron Acemoglu Markets versus Governments: Political Economy of Michael Golosov Mechanisms Aleh Tsyvinski 12225 Joel Slemrod Tax Competition with Parasitic Tax Havens John D. Wilson 12226 Joshua Aizenman Prizes for Basic Research – Human Capital, Economic Ilan Noy Might, and the Shadow of History 12227 Susan M. Dynarski The Cost of Complexity in Federal Student Aid: Lessons Judith E. Scott-Clayton From Optimal Tax Theory and Behavioral Economics 12228 Leonard Caruana An Elephant in the Garden: The Allies, Spain, and Oil in Hugh Rockoff World War II 12229 Sebastian Edwards Monetary Unions, External Shocks, and Economic Performance: A Latin American Perspective 12230 Robert Shimer On the Optimal Timing of Benefits with Heterogenous Iván Werning Workers and Human Capital Depreciation 12231 Chad Syverson Prices, Spatial Competition, and Heterogenous Producers: An Empirical Test

NBER Reporter Summer 2006 71 Paper Author(s) Title

12232 Alison Cuellar Medicaid Policy Changes in Mental Health Care and their Sara Markowitz Effect on Mental Health Outcomes 12233 Viviana Fernandez The International CAPM and a Wavelet-Based Decomposition of Value at Risk 12234 Evan Gatev Managing Bank Liquidity Risk: How Deposit-Loan Til Schuermann Synergies Vary with Market Conditions Philip E. Strahan 12235 Barry Eichengreen Sudden Stops and IMF-Supported Programs Poonam Gupta Ashoka Mody 12236 John Mullahy Econometric Risk Adjustment, Endogeneity, and Extrapolation Bias 12237 Rena M. Conti Early Retirement and Public Disability Insurance Ernst R. Berndt Applications: Exploring the Impact of Depression Richard G. Frank 12238 Rachel M. McCleary U.S.-Based Private Voluntary Organizations: Religious Robert J. Barro and Secular PVOs Engaged in International Relief & Development 12239 Louis Kaplow Discounting Dollars, Discounting Lives: Intergenerational Distributive Justice and Efficiency 12240 Jagadeesh Sivadasan Tax Law Changes, Income Shifting, and Measured Wage Joel Slemrod Inequality: Evidence from Data 12241 Karen Eggleston Hospital Ownership and Quality of Care: What Explains Yu-Chu Shen the Different Results? Joseph Lau Christopher H. Schmid Jia Chan 12242 Glenn Ellison Internet Retail Demand: Taxes, Geography, and Online- Sara Fisher Ellison Offline Competition 12243 Giorgio E. Primiceri Intertemporal Disturbances Ernst Schaumburg Andrea Tambalotti 12244 Robert Town The Welfare Consequences of Hospital Mergers Douglas Wholey Roger Feldman Lawton R. Burns 12245 E. Han Kim Are Elite Universities Losing Their Competitive Edge? Adair Morse Luigi Zingales 12246 Sukkoo Kim Division of Labor and the Rise of Cities: Evidence from U.S. Industrialization, 1850-1880 12247 Geert Bekaert Stock and Bond Returns with Moody Investors Eric Engstrom Steven R. Grenadier 12248 Geert Bekaert Risk, Uncertainty, and Asset Prices Eric Engstrom Yuhang Xing 12249 John Whalley China’s FDI and Non-FDI Economies and the Xian Xin Sustainability of Future High Chinese Growth 12250 Kevin H. O’Rourke Democracy and Protectionism Alan M. Taylor

72 NBER Reporter Summer 2006 Paper Author(s) Title

12251 Zeynep Hansen The Impact of Group Diversity on Performance and Hideo Owan Knowledge Spillover – An Experiment in a College Classroom Jie Pan 12252 Ana Carla A. Costa Judicial Risk and Credit Market Performance: Micro Joao M.P. De Mello Evidence from Brazilian Payroll Loans 12253 Leigh Linden There Goes the Neighborhood? Estimates of the Impact of Jonah E. Rockoff Crime Risk on Property Values From Megan’s Laws 12254 Daron Acemoglu Input and Technology Choices in Regulated Industries: Amy Finkelstein Evidence From the Health Care Sector 12255 Carolyn Kousky Private Investment and Government Protection Erzo F.P. Luttmer Richard J. Zeckhauser 12256 James Dow Noise Traders Gary Gorton 12257 Kevin Lang Education and Labor-Market Discrimination Michael Manove 12258 Susumu Imai Protection for Sale or Surge Protection? Hajime Katayama Kala Krishna 12259 Louis Kaplow Capital Levies and Transition to a Consumption Tax 12260 Martín Uribe Individual Versus Aggregate Collateral Constraints and the Overborrowing Syndrome 12261 James J. Choi Why Does the Law of One Price Fail? An Experiment on David Laibson Index Mutual Funds Brigitte C. Madrian 12262 Dongguyn Shin New Evidence on Real Wage Cyclicality Gary Solon 12263 Ken Hendricks Information Spillovers in the Market for Recorded Music Alan Sorensen 12264 Valerie A. Ramey A Century of Work and Leisure Neville Francis 12265 Young Jun Chun Population Aging, Fiscal Policies, and National Saving: Predictions for Korean Economy 12266 Joseph J. Doyle, Jr. $2.00 Gas! Studying the Effects of a Gas Tax Moratorium Krislert Samphantharak 12267 Anne Case Socioeconomic Status and Health in Childhood: A Christina Paxson Comment on Chen, Martin, and Matthews (2006) Tom Vogl 12268 Yuriy Gorodnichenko Monetary Policy When Potential Output is Uncertain: Matthew D. Shapiro Understanding the Growth Gamble of the 1990s 12269 Daron Acemoglu Disease and Development: The Effect of Life Expectancy Simon Johnson on Economic Growth 12270 Alessandro Beber Resolving Macroeconomic Uncertainty in Stock and Bond Michael W. Brandt Markets 12271 Robert E. Lipsey Measuring International Trade in Services 12272 Gilbert E. Metcalf Energy Conservation in the United States: Understanding Its Role in Climate Policy 12273 Kanika Kapur Love or Money? Health Insurance and Retirement Among Jeannette Rogowski Married Couples

NBER Reporter Summer 2006 73 Paper Author(s) Title

12274 Marianne Bertrand Does Corruption Produce Unsafe Drivers? Simeon Djankov Rema Hanna Sendhil Mullainathan 12275 Philippe Aghion When Does Domestic Saving Matter for Economic Diego Comin Growth? Peter Howitt 12276 Mihir A. Desai Capital Structure with Risky Foreign Investment C. Fritz Foley James R. Hines Jr. 12277 Vitor Gaspar Stability First: Reflections Inspired by Otmar Issing’s Anil K. Kashyap Success as the ECB’s Chief Economist 12278 Emre Ozdenoren Willpower and the Optimal Control of Visceral Urges Stephen Salant Dan Silverman 12279 Martin Feldstein Balancing the Goals of Health Care Provision 12280 Pierre Mohnen Innovativity: A Comparison Across Seven European Jacques Mairesse Countries Marcel Dagenais 12281 Geoffrey Heal Supermodality and Tipping Howard Kunreuther 12282 Erica Field Educational Debt Burden and Career Choice: Evidence From a Financial Aid Experiment at NYU Law School 12283 Bernardo S. de M. Carvalho Ineffective Controls on Capital Inflows Under Márcio G.P. Garcia Sophisticated Financial Markets: Brazil in the Nineties 12284 Louis Kaplow Optimal Income Transfers 12285 Charles Clotfelter Would Higher Salaries Keep Teachers in High-Poverty Elizabeth Glennie Schools? Evidence From a Policy Intervention in North Helen Ladd Carolina Jacob Vigdor 12286 Julie Berry Cullen Tinkering Toward Accolades: School Gaming Under a Randall Reback Performance Accountability System 12287 Reuben Gronau Home Production and the Macro Economy – Some Lessons From Pollak and Wachter and from Transition Russia 12288 Malcolm Baker The Effect of Dividends on Consumption Stefan Nagel Jeffrey Wurgel 12289 Hanming Fang Sources of Advantageous Selection: Evidence From the Michael P. Keane Medigap Insurance Market Dan Silverman 12290 Andrew B. Abel Equity Premia with Benchmark Levels of Consumption: Closed-Form Results 12291 John Cawley Beyond BMI: The Value of More Accurate Measures of Richard V. Burkhauser Fatness and Obesity in Social Science Research 12292 Roger Gordon Dividends and Taxes Martin Dietz 12293 Andrew B. Bernard Multi-Product Firms and Product Switching Stephen J. Redding Peter K. Schott 12294 W. Kip Viscusi Rational Discounting for Regulatory Analysis

74 NBER Reporter Summer 2006 Paper Author(s) Title

12295 Michelle Lowry The Variability of IPO Initial Returns Micah S. Officer G. William Schwert 12296 Hui Huang Baumol-Tobin and the Welfare Costs of National Security John Whalley Border Delays 12297 Vojislav Maksimovic The Industry Life Cycle and Acquisitions and Investment: Gordon Phillips Does Firm Organization Matter? 12298 Eric I. Knudsen Economic, Neurobiological, and Behavioral Perspectives James J. Heckman on Building America’s Future Workforce Judy L. Cameron Jack P. Shonkoff 12299 Michael D. Bordo Currency Mismatches, Default Risk, and Exchange Rate Christopher M. Meissner Depreciation: Evidence From the End of Bimetallism Marc D. Weidenmier 12300 Sara Markowitz Even For Teenagers, Money Does Not Grow on Trees: John Tauras Teenage Substance Use and Budget Constraints 12301 Martin Gaynor What Do We Know About Competition and Quality in Health Care Markets? 12302 Wilson Au-Yeung Australian Government Balance Sheet Management Jason McDonald Amanda Sayegh 12303 Axel Börsch-Supan Early Retirement, Social Security, and Well-Being in Hendrik Jürges Germany 12304 Weili Ding The Impact of Poor Health on Education: New Evidence Steven F. Lehrer Using Genetic Markers J. Niels Rosenquist Janet Audrain-McGovern 12305 Weili Ding Do Peers Affect Student Achievement in China’s Steven F. Lehrer Secondary Schools? 12306 Richard B. Freeman Searching for the EU Social Dialogue Model 12307 Alan J. Auerbach The Choice Between Income and Consumption Taxes: A Primer 12308 Urban Jermann Financial Innovations and Macroeconomic Volatility Vincenzo Quadrini 12309 Paul Willen Collateralized Borrowing and Life-Cycle Portfolio Choice Felix Kubler 12310 Alex Bryson Worker Needs and Voice in the US and the UK Richard B. Freeman 12311 Robert J. Gordon The Boskin Commission Report: A Retrospective One Decade Later 12312 Raymond Fisman Cultures of Corruption: Evidence From Diplomatic Edward Miguel Parking Tickets 12313 Stephen Coate Socially Optimal Districting: An Empirical Investigation Brian Knight 12314 Diego Comin An Exploration of Technology Diffusion Bart Hobijn 12315 Richard B. Freeman People Flows in Globalization 12316 Rafael Dobado González Globalization, De-Industrialization, and Mexican Aurora Gómez Galvarriato Exceptionalism, 1750–1879 Jeffrey G. Williamson

NBER Reporter Summer 2006 75 Paper Author(s) Title

12317 Felix Oberholzer-Gee Media Markets and Localism: Does Local News en Joel Waldfogel Español Boost Hispanic Voter Turnout? 12318 James Heckman Contributions of Zvi Griliches 12319 Matthias Doepke Inflation as a Redistribution Shock: Effects on Aggregates Martin Schneider and Welfare 12320 Bronwyn H. Hall Empirical Studies on Innovation in the Knowledge Driven Jacques Mairesse Economy 12321 Francine D. Blau New Evidence on Gender Difference in Promotion Rates: Jed DeVaro An Empirical Analysis of a Sample of New Hires 12322 Volker Nocke Globalization and Endogenous Firm Scope Stephen Yeaple 12323 Mark Duggan Federal Policy and the Rise in Disability Enrollment: Robert Rosenheck Evidence for the VA’s Disability Program Perry Singleton 12324 James H. Stock Why Has U.S. Inflation Become Harder to Forecast? Mark W. Watson 12325 Dean Yang International Migration, Remittances, and Household Investment: Evidence from Philippine Migrants’ Exchange Rate Shocks 12326 Richard E. Baldwin Trade and Growth with Heterogenous Firms Frédéric Robert-Nicoud 12327 Abdurrahman Aydemir A Comparative Analysis of the Labor Market Impact of George J. Borjas International Migration: Canada, Mexico, and the United States 12328 Alberto Alesina Artificial States William Easterly Janina Matuszeski 12329 Justin McCrary The Effect of Female Education on Fertility and Infant Heather Royer Health: Evidence From School Entry Policies Using Exact Date of Birth 12330 Miki Kohara Do Borrowing Constraints Matter? An Analysis of Why Charles Yuji Horioka the Permanent Income Hypothesis Does Not Apply in Japan 12331 Janna L. Matlack Do Rising Tides Lift All Prices? Income Inequality and Jacob L. Vigdor Housing Affordability 12332 Gene M. Grossman Separation of Powers and the Budget Process Elhanan Helpman 12333 Elias Papaioannou Optimal Currency Shares in International Reserves: The Richard Portes Impact of the Euro and the Prospects for the Dollar Gregorios Siourounis 12334 Muriel Niederle Competitive Wages in a Match With Ordered Contracts 12335 José J. Escarce Hospital Competition, Managed Care, and Mortality After Arvind K. Jain Hospitalization for Medical Conditions: Evidence From Three Jeannette Rogowski States 12336 Ruediger Bachmann Lumpy Investment in Dynamic General Equilibrium Ricardo J. Caballero Eduardo M.R.A. Engel 12337 Anders B. Trolle A General Stochastic Volatility Model for the Pricing and Eduardo S. Schwartz Forecasting of Interest Rate Derivatives 12338 Dean Karlan Does Price Matter in Charitable Giving? Evidence From a John A. List Large-Scale Natural Field Experiment

76 NBER Reporter Summer 2006 Paper Author(s) Title

12339 Louis Kaplow Optimal Control of Externalities in the Presence of Income Taxation 12340 Marco Arena The Lending Channel in Emerging Economies: Are Carmen Reinhart Foreign Banks Different? Francisco Vázquez 12341 Gary D. Hansen Consumption Over the Life Cycle: The Role of Annuities Selahattin Imrohoroglu 12342 Zhonglan Dai Capital Gains Taxes and Asset Prices: Capitalization or Edward Maydew Lock-In? Douglas A. Shackelford Harold H. Zhang 12343 William O. Brown, Jr. Competing With the NYSE J. Harold Mulherin Marc D. Weidenmier 12344 Laurence J. Kotlikoff On The General Relativity of Fiscal Language Jerry Green 12345 Eduardo Levy Yeyati Liquidity Insurance in a Financially Dollarized Economy 12346 Charles P. Thomas The Performance of International Equity Portfolios Francis E. Warnock Jon Wongswan 12347 Ann Harrison Globalization and Poverty 12348 W. Kip Viscusi The Catastrophic Effects of Natural Disasters on Insurance Patricia Born Markets 12349 N. Gregory Mankiw The Macroeconomist as Scientist and Engineer 12350 Daniel S. Hamermesh Time Zones as Cues for Coordination: Latitude, Longitude, Caitlin Knowles Myers and Letterman Mark L. Pocock 12351 Charles Yuji Horioka The Dissaving of the Ages Revisited: The Case of Japan 12352 David M. Cutler Education and Health: Evaluating Theories and Evidence Adriana Lleras-Muney 12353 Lawrence J. Christiano Assessing Structural VARs Martin Eichenbaum Robert Vigfusson 12354 Steven J. Davis Volatility and Dispersion in Business Growth Rates: John Haltiwanger Publicly Traded versus Privately Held Firms Ron Jarmin Javier Miranda

TECHNICAL PAPERS 323 James H. Stock Heteroskedasticity-Robust Standard Errors for Fixed Effects Mark W. Watson Panel Data Regression 324 Richard K. Crump Nonparametric Tests for Treatment Effect Heterogeneity V. Joseph Hotz Guido W. Imbens Oscar A. Mitnik 325 Alberto Abadie On the Failure of the Bootstrap for Matching Estimates Guido W. Imbens

NBER Reporter Summer 2006 77 Nonprofit Org. U.S. Postage NBERReporter PAID National Bureau of NATIONAL BUREAU OF ECONOMIC RESEARCH Economic Research

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

Change Service Requested