a flat world

Journal of Economic LiteratureVol. XLV (March 2007), pp. 83–126

mar07_Article3 3/12/07 5:54 PM Page 83

A Flat World, a Level Playing Field, aSmall World After All, or None of the

Above? A Review of Thomas L.Friedman’s The World is Flat

EDWARD E. LEAMER∗

Geography, flat or not, creates special relationships between buyers and sellers whoreside in the same neighborhoods, but Friedman turns this metaphor inside-out byusing The World is Flat to warn us of the perils of a relationship-free world in whichevery economic transaction is contested globally. In his “flat” world, your wages areset in Shanghai. In fact, most of the footloose relationship-free jobs in apparel andfootwear and consumer electronics departed the United States several decades ago,and few U.S. workers today feel the force of Chinese and Indian competition, notwith-standing the alarming anecdotes about the outsourcing of intellectual services. Ofcourse, standardization, mechanization, and computerization all work to increase thenumber of footloose tasks, but innovation and education work in the opposite direc-tion, creating relationship-based activities—like the writing of this review. It mayonly be personal conceit, but I imagine there is a reason why the Journal of EconomicLiterature asked me to do this review.

1. Prologue

When the Journal of EconomicLiterature asked me to write a review

of The World is Flat: A Brief History of theTwenty-First Century (Farrar, Straus, andGiroux 2005) by Thomas Friedman, Iresponded with enthusiasm, knowing itwouldn’t take much effort on my part. As

∗ Leamer: Anderson Graduate School of Management,Department of Economics and Department of Statistics,UCLA. Thanks for comments from my brother, LaurenceLeamer, my UCLA colleagues at the Political Economylunch (Dan Treisman and Michael Ross), EricRasumusen, John Talbot and my former students,Christopher Thornberg, Peter Schott and BernardoBlum, and to Frank Levy who offered extensive com-ments and several important suggestions.

83

soon as I received a copy of the book, Ishipped it overnight by UPS to India to havethe work done. I was promised a one-dayturn-around for a fee of $100. Here is what Ireceived by e-mail the next day: “This bookis truly marvelous. It will surely change thecourse of human history.” That struck me aspossibly accurate but a bit too short and toogeneric to make the JEL happy, and I decid-ed, with great disappointment, to do thework myself.

2. What Might that “Flat World” MetaphorMean?

Stumbling onto a book titled The World isFlat by Thomas Friedman would leave a

84 Journal of Economic Literature, Vol. XLV (March 2007)


book browser puzzled about its likely con-tent. My first guess would be epistemologyand evolution. The World is Flat must be areference to the pre-Columbian sailor’sworry about falling off the edge of the earth,and the tenacious clinging to that idea bymembers of the Flat-Earth Society in theface of “overwhelming” “scientific” evi-dence. Put that into the current context, thedebate about the intellectual legitimacy of“intelligent design,” and you are led to myconclusion: The World is Flat is probably abook about faith-based decision making andthe teaching of intelligent design in theschools. This book is going to surprise. It willshow that a flat earth is not a straw man atall, and that science is only another kind ofreligion, seeking to burn its heretics at thestake with all the vigor of traditional religion.

Alas, the subtitle A Brief History of theTwenty-First Century unsettles this briefflight of fancy about the content of this book,but it leaves the browser utterly confused.How could The World is Flat and A BriefHistory of the Twenty-First Century haveanything to do with each other? That subti-tle reminds me of the New Yorker cartoonthat hung outside a history professor’s officeat UCLA for many years. It depicted a stu-dent receiving a final exam in a historycourse: “Explain World War II. Use bothsides of the page if necessary.”

Enough with all these diverting thoughts.It’s time to look at the blurb. The blurb pointsin a wholly different direction:

. . . the convergence of technology and eventsthat allowed India, China, and so many othercountries to become part of the global supplychain for services and manufacturing, creatingan explosion of wealth in the middle classes ofthe world’s two biggest nations and givingthem a huge new stake in the success of glob-alization? And with this “flattening” of theglobe, which requires us to run faster in orderto stay in place, has the world gotten too smalland too fast for human beings and their politi-cal systems to adjust in a stable manner?

Huh? That last sentence packs in at leasttwo too many metaphors for me to process:

a flat earth, people running faster but stayingin place, a small world and a fast world. Andthen there is the “explosion” in the previoussentence. What is Friedman getting at withthis mélange of metaphors? I understandthat China has increased it exports of manu-factures to levels far above anyone’s expecta-tions. I understand that call centers andsome back-office functions and some soft-ware coding have moved to India. I under-stand that GDP growth in both China andIndia has been phenomenal and has liftedhundreds of millions out of poverty. Butwhat is the meaning of that flatteningmetaphor? What is the alternative to a flatworld? A smooth sphere? Bumps? That’s apuzzle worth solving.

2.1 Friedman’s “Aha” Flat Moment

Friedman’s “aha” flat moment came on agolfing outing during a Discovery Channelexcursion to Bangalore, India, where, sur-rounded by buildings emblazoned with U.S.tech names, he was told to “Aim at eitherMicrosoft or IBM” (p. 3). Friedmanrecounts that Columbus, sailing in search ofIndia apparently on the premise that theEarth is round, encountered exotic nativeAmericans unlike the Europeans with whichhe was familiar and pronounced themIndians, allowing Columbus to carry thenews back to King Ferdinand and QueenIsabella: the Earth is round. Likening hisDiscovery Channel crew to the sailors onthe Nina, the Pinta, and the Santa Maria,when Friedman found in Bangalore notIndians but Americans in name and speechand business practices, he “shared my dis-covery only with my wife, and only in a whis-per. ‘Honey.’ I confided, ‘I think the world isflat” (p. 5).

Once “flat” was in Friedman’s head, hecouldn’t seem to get it out. When on thatsame trip to Bangalore, Friedman was toldby Nandan Nilekani, CEO of InfosysTechnologies Limited, “Tom, the playingfield is being leveled.” Friedman concludes

85Leamer: A Review of Thomas L Friedman’s The World is Flat


“What Nandan is saying , I thought, is thatthe playing field is being flattened . . . flat-tened? Flattened? My God, he’s telling methe world is flat!” p. 7.

Flattened? I still don’t get it. To digressbriefly into the use of metaphors in eco-nomics I routinely ask my Ph.D. classes ininternational economics what they think Imean when I tell them “Joe’s elevatordoesn’t stop on all floors.” The foreign stu-dents (of which there are many) whosenative language is not English imagine lit-erally an elevator in Joe’s building that isn’tworking right. American students (ofwhich there are only one or two) suspectthat what I mean is “Joe isn’t playing witha full deck.” What I want these students tounderstand is that models in economicsare highly metaphorical and if studentscannot tell the difference between the lit-eral mathematical properties of modelsversus the subtle messages of the models,they are not understanding the language.Decoding metaphors is the hardest part oflearning a new language. When we empha-size only the grammar and the syntax inour economics courses, we leave littleroom for the kind of experience with thelanguage that allows students to learn toconverse in Economics, and we graduatestudents who can parse an Economics sen-tence but cannot carry on an Economicsconversation.

That is my way of saying that our languagereally matters, and metaphors need to bechosen carefully. I am open to a good newmetaphor (economic model), but a metaphorisn’t going to work for me unless I can figureout what message is intended. I know what a“small world” means. I have some ideas whata “level playing field” may mean. But a “FlatWorld” for me is an elusive idea. Worthlearning about.1

1 Jagdish Bhagwati (2005) also has troubles with theFlat World metaphor and combats it with another: “kalei-doscopic comparative advantage.” Sorry, Professor, thatone lacks the necessary punch to do much damage to itscompetitor.

3. Economic Models of Flat Competition

3.1 Flatness is Familiar in EconomicGeography

Flatness is not a new concept in econom-ics: there are Von Thünen ringed cities inthe centers of flat planes of agricultural landand Lösch hexagons that divide a flat planeof demand into identical service areas with asupply point in the middle of each. Andthere are Hotelling competitors that clusterinefficiently together in the center of astraight-line segment along which customersare distributed uniformly (a flat distribu-tion). There is one important message inthese models of economic geography. Bestto reflect on it before we let our brains andspirits be flattened by Friedman’s metaphor.The message is that geography, flat or other-wise, limits competition and creates long-term relationships between buyer and seller.

The German farmer Johann H. VonThünen noticed that farmland closer to thetowns where the produce was sold com-manded a premium price and he is creditedwith being the father of economic geographybecause, in 1826, he postulated a featureless(flat) plane of land with a town in the center.Crops shipped from farm to town had differ-ent ratios of transportation cost to value.Fertilizers and farm implements wereshipped the other way. These assumptionscreate a sequence of “Von Thünen” concen-tric rings of farmland around the town cen-ter, with the land rents highest near thecenter, with “heavy” crops that need fertiliz-ers produced close to market. (A modernversion of this idea is the microeconomicexam question: “Why does the State ofWashington ship its best apples to otherstates, not its worst?”)

Von Thünen’s theory of concentric ringspresumes the existence of towns, but wheredo these towns come from? These might bemarket towns where farmers exchange onecrop for another. They might be industrialcenters. On this question, Alfred W. Weber



(1909) postulates a mathematical program-ming problem of industrial location in whichmaterials can be found at discrete supplypoints on a featureless (flat) plane and man-ufacturing must procure these resources andship finished products to prespecified loca-tions of demand. Should the processing bedone where the materials can be found orwhere the product is sold? It depends onwhether processing adds weight or reducesweight.

Next comes August Lösch (1938, 1954)who postulates another mathematical pro-gramming problem with demand uniformlyspread on an infinite plane. The problem isto locate supply points in this plane to min-imize the total distance traveled from cus-tomers to supply points. The most efficientshape that minimizes the average distancebetween customers and supply point is acircle with the supply point in the center.But circles cannot be a solution to Lösch’sproblem because they cannot tile a plane.Of the three regular shapes that can tile aplane—equilateral triangles, squares, andhexagons—it is hexagons that come closestto circles. Thus Lösch’s infinite featureless(flat) plane of demand is cut most efficient-ly into equal-sized hexagons with a supplypoint located at the center of each.2

While Von Thünen describes an econom-ic equilibrium, Weber and Lösch describeonly a mathematical optimization problem

2 I actually wrote my first published paper (my SeniorThesis at Princeton) on the Lösch problem but withdemand limited geographically to squares and circlesrather than infinite planes: “Location Equilibria,” Journalof Regional Science, Vol. 8 (No. 2, 1968), 229–42.

that may or may not be approximated byfirms seeking locations that maximize prof-its. Probably not, since these socially opti-mal solutions leave firms in advantagedpositions relative to their closest customerswho would have to travel far to find anothersupplier. That market power is sure to beexploited with monopoly pricing, which inturn is sure to be contested by locationand/or pricing decisions of other firms.Where might competitive firms locate in aflat world? An economic equilibrium modelof competition on a flat line segment (call ita beach) has been offered by HaroldHotelling (1929) who argues that competi-tion will induce both sellers (hot dog stands)to crowd together at the center, producingan inefficient outcome. The Hotellingmodel is routinely applied to competitionbetween two political parties, each position-ing itself close to the median voter.3

3.1.1. The Message of Models ofCompetition in Flat Planes:Relationships Not Markets

From these references we should receivean important message that needs to beabsorbed as we try to decode Friedman’sflat-world metaphor. Geography, whetherphysical or cultural or informational, limitscompetition since it creates cost-advantagedrelationships between sellers and buyerswho are located “close” to one another. The

3 The traditional models of international comparativeadvantage also embody some geographical features, butvery strange ones. In the traditional Heckscher–Ohlinmodel, with machines and labor as the inputs, countriesare assumed to be close enough that the exchange ofgoods internationally is costless, but so far apart that nei-ther workers nor contracts can find their way from onecountry to another. (If contracts could be formed acrossborders, current account deficits would allow for capitalaccumulation in one country to be financed by another,making the assumption of fixed capital stocks untenable.)Further, inside the countries, industries are so close geo-graphically and technologically, that workers and

machines can move without cost from one job to thenext. Beginning with Paul Krugman (1991), internation-al economists have been working hard to introduce somemore meaningful geography into their models, some-times with more success than other times. Krugman pos-tulates two points at which production can occur.“Peasants” who do the farming are tied to the land ateach of these points, but “workers” who do the manufac-turing can choose either location. Transportation costsare of the “iceberg” type—part of the product melts awayen route. With increasing returns, product differentia-tion and monopolistic competition in manufacturing,what might the equilibrium look like? Is the manufactur-ing all done in one location and manufactured productsshipped one way and agricultural products the other? Ordo both communities grow their own food and exchangemanufactured products? Like Weber’s solution: itdepends.



key word is relationships, with italics addedto emphasize that this is the core propertythat drives my disagreements withFriedman. Flatness doesn’t create a relation-ship-free equilibrium; it merely changes thegeography of relationships. It turns irregularhard-to-define geographic regions of cus-tomers committed to the same supplier intoregular hexagons with the suppliers at thecenters.

Flat featureless planes of competition arethe preferred spaces for the mathematicalmodeling of geographical competition, notbecause flatness closely resembles naturebut only because modeling of competition inreal geographies with rivers and oceans andhills and mountains is beyond the reach ofalgebra. The point of this modeling is not toshow what flatness implies. The point is toshow what transportation costs imply. Thewhat-if question that these economic geog-raphers ask is not: “What if the world wereflat instead of spherical or bumpy?” Theirwhat-if question is: “What if transportationcosts were low instead of high? What dodeclines in transportation costs do to theintensity of competition and the geographyof prosperity?” What if the world weresmall? The answer is clear: the smaller theworld, the more contested are theexchanges, and the weaker are the relation-ships between buyers and sellers. In a smallworld, wages in Los Angeles are set inShanghai.

3.1.2 Some Real Questions for EconomicGeography to Explore

There are countless real “globalization”questions for economic geography to answer.For example: What if Europe were to forman economic union that allowed the fourfreedoms among countries: free movementof goods, people, services, and capital, allsupported by a common currency? Whichkinds of activities would concentrate andwhich would disperse? Do the peripheralcountries, like Sweden, benefit or lose out?What happens to a central country, like

Switzerland, that doesn’t join? On anothercontinent: Can the economic liberalizationsin South America create an export-led pros-perity, or are Argentina and Chile and otherLatin American countries too far from thelarge and wealthy European and NorthAmerican marketplaces for that strategy towork? If closeness matters, how can oneaccount for the historically high per capitaGDPs of New Zealand and Australia, andwhat accounts for the economic successes offar-way countries like Taiwan and Singapore,and now China?

When looking for problems that can bestudied by economic geography, it is wise tokeep in mind that the distance effects neednot come from simple transportation costs.Physical distance may create and reinforcelinguistic and cultural barriers that make itdifficult to exchange thoughts between peo-ple located far from each other in the cultur-al landscape. In that kind of world, therewould be specialized cultural services (e.g.,plays and newspapers and legal services)made by locals for locals. These couldn’t beshipped very far because the messageswould melt away to meaninglessness if thecontent were shipped over great distances tounreceptive cultures.

But suppose that, by an accident of histo-ry, a geographically large area with a largepopulation adopted much the same lan-guage and much the same culture. Thoughlarge physically, this area is small culturally.Call it the United States of America. Withinthis country, how many “Hollywoods” doyou suppose might emerge? If the agglom-erative externalities (economies of scale)were moderate compared with the culturalshipping costs, hundreds of Hollywoodsmight be sprinkled around the UnitedStates each producing specialized productssuited to the local markets. But suppose thebenefits from agglomeration are large andthe competitive outcome allowed only oneHollywood to emerge producing onehomogenous cultural product that is notcustomized for Southern tastes nor Eastern



Tastes nor Midwestern tastes nor Westerntastes. The same movies are displayed intheaters all over the country. Enter into thisdrama a new actor: globalization. Supposethat communications innovations drivedown the “shipping costs” for cultural prod-ucts to points outside the United States.Suppose English is the language of globalcommerce and children all over the worlddrink Coke and wear jeans. Then what hap-pens? Would a Bollywood emerge and com-pete actively for the U.S. market? Or wouldglobal competition allow only a singleHollywood? Does our Hollywood win orlose from this form of cultural globalization?

If your instincts are the same as mine, youare thinking that Hollywood wins big timefrom declining cultural “transportation”costs, and many of the rest of us living in LosAngeles benefit as well. It is the local cultur-al providers in France and Mexico andCanada and India and China that are proba-bly hurt from an extension of the reach ofU.S. cultural services. (As I write these sen-tences, I am resisting the very strong desireto write “The World is Not Flat” to conveymy disagreement with Friedman but at thecost of using a metaphor I do not under-stand. Countries on the receiving end of thiscultural exchange might complain about the“flatness” of a one-cultural world and at thesame time object to the “non-flatness” of thecompetitive landscape that leaves the UnitedStates firmly in a favored position. )

Cultural products are one of many intel-lectual services in which the United States isthe global leader. Like the movie industry,the production of these intellectual servicesis geographically clustered inside the UnitedStates, presumably because of powerful localagglomerative externalities. Finance is donein New York City. Innovations in electronictechnology occur in the Silicon Valley andBoston and Austin. Biotech is clustered inSan Diego. Country music in Nashville.Mundane economics training is centered inCambridge, Massachusetts. Creative eco-nomics education is in Los Angeles. Now

imagine a cultural and telecommunicationsrevolution that allows the reach of these clus-tered intellectual services to be extended,allowing financial innovations made in NewYork to be used in Hong Kong, and softwaredesigned in Seattle to be marketed in India.Does this help or hurt the United States? Iam thinking that the United States is a bigwinner not a loser from the extension of themarket for its intellectual services, provided,of course, that the rest of the world offersintellectual property protection. Then we geta larger market but not much in the way ofnew competitors because of the highly local-ized economies of scale and the very sub-stantial first-mover advantages that historyhas bestowed on the United States.

But the answers to all these questions arethe same as the answer to Weber’s questionof where to locate manufacturing: itdepends. It depends on the power of theagglomeration externalities, the costs ofdelivery, and the distribution of consumersand their preferences across space.

3.1.3 Another Book with Another Title

Clearly there is a book to be written aboutthe impacts of the dramatic decline in trans-portation and communication costs that wehave recently experienced—cargo con-tainerization, air shipment, telecommunica-tions, the Internet, e-mail, voice-mail, andthe cell phone. The title of that book wouldnot be The World is Flat. The title would beIt’s a Small World After All.

I first realized this on a trip to Disneylandin 1975. After a trip through the adventure“It’s a Small World After All,” it dawned onme that, under their outfits, the puppetsdepicting Asians and Europeans andAmericans and Pacific Islanders are all pret-ty much identical. Though clothing andinstruments are different, Walt Disney cre-ated a Small World by having all the puppetssing the same happy song. Disney gave usthe world we should be striving for—a worldwith a rich smorgasbord of cultures but with-out the frictions that cultural differences



usually engender. Not one flat common cul-ture. (I did not whisper my extraordinarydiscovery to my wife. She was alreadyasleep.)

3.2 A Level Playing Field?

A “small world” is a clear metaphor, but a“flat world” is for me an elusive new one.The most prevalent metaphor for discussingglobal competition is “a level playing field”which actually is not crystal clear either. A“level playing field” refers to rules of thegame that “unfairly” favor one competitorover another. If the football field is level, thebetter team will emerge victorious, but asloped field confers advantages to the teamdefending the higher goal.

“Fairness” requires a level playing field,but fairness and levelness are in the heartand eye of the beholder. If you are biggerand faster and smarter than I am, justbecause of your genetic draw, that doesn’tseem fair to me. That’s not a level playingfield. Better offer a handicap or a pointspread. That would make it fair. If you wantto push this to the extreme, it is unfair forthere to be a loser at all. Can’t we all be win-ners? (With grade inflation, we are doingthis on campus.)

The literature on the theory of interna-tional trade has many models with slopedplaying fields, including ones with techno-logical differences and with policy differ-ences across countries. The next paragraphhas a model-building exercise that I offer mystudents. It is appealing because it has a geo-graphic aspect to it; because it speaks to thegrowing concern over the U.S. deficit, firstwith Japan and now with China; and becauseit yields a surprising conclusion.

Imagine that there are two countries—Japan and the United States. Japan sits onthe top of a hill and the United States sits atthe bottom. To get U.S. goods to Japan, onehas to hire porters to carry the goods up thehill. But the Japanese can put their prod-ucts in a chute and let gravity do thework—costlessly transporting Japanese

goods down the hill to the U.S. market. Nota level playing field, you should be thinking.Japan is clearly in the advantageous posi-tion. Not so fast, I caution the students.Who pays for lugging the U.S. products upthe hill? Why do you presume it is theUnited States and not the Japanese? Thisshould get them thinking about elasticitiesof supply and demand. If U.S. goods are inshort supply and are desperately desired bythe Japanese, while Japanese goods areabundant and not much desired byAmericans, then it is the United States atthe bottom of the hill that is in the advanta-geous position and it is the Japanese whopay for the lugging of the goods up the hill.If the Japanese build their mountain artifi-cially with trade barriers that make it diffi-cult to ship Washington apples to Japaneseconsumers and if the Japanese consumerswould pay any price for those apples whileAmericans could care less about the latestSony gadget, then it is the Japanese whopay for the barriers, not the Americans. Sobe careful when you put rocks in your har-bor. And be sure to wear the right kind ofglasses when you are viewing the playingfield. What looks tilted one way with yourregular glasses may be tilted the other waywith econ-oculars.

4. More Detective Work: How DoesFriedman Use the “Flat” Word?

Now that we have diverted to a briefreview of flatness in economic thought andwe have put some flesh on the bones of thelevel-playing-field metaphor, it is time toreturn to our first task: trying to decodeFriedman’s flatness metaphor.

One way to uncover what Friedmanmeans by “The World is Flat” is to see howhe uses the f-word. Since it occurs on virtu-ally every page of this 450-page book, thereis plenty of grist for the mill.

. . . what the flattening of the world means isthat we are now connecting all the knowledgecenters on the planet together into a singleglobal network. . . . (p. 8)



from 1492 to around 1800 . . . it shrank theworld from a size large to a size medium. From1800 to 2000 . . . shrank the world from a sizemedium to a size small. . . . around the year2000 we entered a whole new era . . . shrinkingthe world from a size small to a size tiny andflattening the playing field at the same time.(p. 9–10, my italics)to flatten their accents in order to compete ina flatter world. (p. 27)That is why I introduced the idea that theworld has gone from round to flat. Everywhereyou turn, hierarchies are being challengedfrom below or transforming themselves fromtop-down structure into more horizontal andcollaborative ones. (p. 45)common standards create a flatter, more levelplaying field. (p. 52)Just as the national highway system that wasbuilt in the 1950s flattened the United States,broke down regional differences, and made itso much easier to relocate in lower-wageregions, like the South . . . the laying of globalfiber highways flattened the developed world.(p. 69)For the world to get flat, all your internaldepartments—sales, marketing, manufactur-ing, billing, and inventory—had to becomeinteroperable, no matter what machines orsoftware each of them was running. (p. 74)There is no future in vanilla for most compa-nies in a flat world. A lot of vanilla making insoftware and other areas is going to shift toopen-source communities. (p. 91)My bottom line is this: Open-source is animportant flattener because it makes avail-able for free many tools, from software toencyclopedias, that millions of peoplearound the world would have had to buy inorder to use, and because open-source net-work associations—with their open bordersand come-one-come-all approach—can chal-lenge hierarchical structures with a horizontalmodel of innovation . . . . (p 102–03)China will never be truly flat until it gets overthat huge speed bump called “politicalreform.” (p. 126)Insourcing came about because once theworld went flat, the small could act big—smallcompanies could suddenly see around theworld. (p. 143, my italics)Search engines flatten the world by eliminat-ing all the valleys and peaks, all the walls androcks, that people used to hide inside of, atop,

behind or under in order to mask their reputa-tions or parts of their past. (p. 158)The net result of this convergence was the cre-ation of a global, Web-enabled playing fieldthat allows for multiple forms of collabora-tion—the sharing of knowledge and work—inreal time, without regard to geography, dis-tance, or in the near future, evenlanguage. . . . That is what I mean when Isay the world has been flattened. (pp.176–77, Friedman’s italics and boldface)

This tediously long list is meant to helpyou feel my frustration. Friedman’s use ofthe flatness metaphor is virtually all encom-passing. “Flattened” seems to be a synonymfor “changed.” Often, and certainly in thelast quotation, Friedman is really writingabout a small world in which distance, meas-ured physically, linguistically, and culturally,doesn’t isolate your job from competitionfrom far-away workers.4

4.1 Friedman’s Ten Forces that Flattenedthe World

There is one last hope for the decoding ofthe flatness metaphor—a careful look atFriedman’s list of the ten forces that flattenedthe world. Here they are.

1. 11/9/89: The Fall of the Berlin Wall.2. 8/9/95: The Birth of the Internet (When

Netscape Went Public).3. Work Flow Software: Software that co-

ordinates tasks.4. Open-Sourcing: The Self-Organizing

Collaborative Communities that pro-duced Linux and Wikipedia.

5. Outsourcing, Y2K: The fiber-optic bub-ble created the potential to move soft-ware coding to India and Y2K demandfor coding far in excess of U.S. capacitycreated the necessity.

6. Offshoring: The movement of manufac-turing jobs to low-wage developingcountries.

4 Even flatness as a visual metaphor, which is apt, isequivalent to a shrinking globe—in a spherical earth youcannot see around the world and cannot recognize theopportunities very far from where you reside. If the worldwere flat, you could see it all. That’s equivalent to a small-er globe that allows you to “reach” far-away opportunities.


mar07_Article3 3/12/07 5:54 PM 1

7. Supply-Chaining: Wal-Mart supplychain management practices.

8. “Insourcing”: UPS which not only picksup your broken Toshiba laptop butrepairs its as well. Small companies canact large because the fixed-cost logisticsinfrastructure is supplied by a thirdparty.

9. “In-forming”: Google and Internetsearches.

10. The Steroids: Digital, mobile, personaland virtual communications. Videocon-ferences.

Well that’s disappointing. This eclecticlist doesn’t do much to narrow the defini-tion of “flattened.” So it must mean“changed.” Of course, a book titled TheWorld is Different isn’t destined to be a bestseller, no matter what the content may be. Acommercially better title would be TheWorld is Flat.

Each of the items on Friedman’s list of tendeserves comment. The last one, “thesteroids,” is another case of metaphorical mal-function. With the coining of the newwords/new definitions “in-forming” and“insourcing” and “supply-chaining” Friedmanis further working his magic with ourlanguage.

I find myself trying now to resist thedesire to discuss “open-sourcing” which isfourth on Friedman’s list of “flatteners.” Weacademics know more about this than hedoes. We are part of a “Self-OrganizingCollaborative Community” called theresearch universities of the United Statesand increasingly the rest of the world. Unlikecontributors to Wikipedia and Linux, we getpaid for our work, not by those who con-sume the fruits of our labor, but by taxpayersand by donors and by our students, all ofwhom we have convinced are better-off byvirtue of the research that we do. When itgot started fifty years ago, this systemworked great, but it isn’t working as wellanymore. While we are doing plenty ofworthwhile research, we are also doing plen-ty that isn’t worthwhile and the competition

for research talent defined by the fads of themoment is driving up the cost of educationto unaffordable levels. Adam Smith wouldhave understood what’s wrong here. It takessales for the invisible hand to do its magic.Begging in your work clothes when youaren’t working isn’t enough, even though thepastime may be lucrative. On the contrary,the more lucrative is the begging, the morelikely is the conclusion that the work isworthwhile, whether it is or isn’t. But it takesmarket prices to tell us what’s valuable andwhat’s not. Good will and good intentionscan carry a collaborative community produc-tively for a while, but financial rewardsrelentlessly bend the system to their will,slowly perhaps, but inevitably. That’s theinvisible hand at work. Thus, open-sourcinghas the same problems and the same proba-ble longevity as the communes of the1960s—they worked great for a while, butthe participants chose other ways to liveonce they got to know each other.

But, anyway, I don’t see how workflowsoftware, open-sourcing, Wal-Mart supply-chaining, or UPS “insourcing” can be consid-ered revolutionary. They are just additionalsmall steps in the march of progress. Ratherthan ten, my list of revolutionary forces hasthree items:

1. More Unskilled Workers: The economicliberalizations in China and India andRussia and South America and on andon have added to the effective globallabor markets a huge number ofunskilled workers and relatively littlehuman and physical capital.

2. New Equipment for KnowledgeWorkers: The Internet and the PersonalComputer have fundamentally changedthe nature of knowledge work, raisingproductivity, emphasizing talent andreducing the need for “helpers.”

3. Communications Innovations: The cell-phone and the beeper and e-mail andvoice-mail keep us all wired and con-nected 24/7, thus eliminating the bor-derline between time at work and time



at leisure. These same communicationtools, together with the Internet and vir-tually costless telecommunications haveextended the geographic reach of sup-pliers, and have increased the intensityof competition for mundane work andstandardized products.

The first two of these are not “flattening”forces at all. The increase in supply ofunskilled workers is bad news for those whocompete in that market segment but goodnews for those who do not. That doesn’tsound like flattening to me. The new equip-ment for knowledge workers greatly increas-es the productivity of those with naturaltalent but that leaves the less talented withless to do and with lower pay. Today, ThomasEdison would be surprised to see the PC andthe Internet doing most of the perspiring,and shocked to discover that genius is 99percent inspiration and 1 percent perspira-tion. Since most of us are better at perspira-tion than inspiration, that doesn’t sound very“flat” to me.

It is the third of these revolutionarychanges that may be a force for “flattening,”or more accurately, a smaller world. I canbuy an electric drill from my local hardwarestore or I can use the Internet to buy thedrill from a supplier in Dallas or Newark andhave it shipped by next day air. That meansthat my local hardware store is competingover a much larger geographic area and therents that come from location are reduced.This can occur in the labor markets as well,with far-away workers bidding to do mytasks.

5. Other Models of Global Competition

In addition to economic geography, thereare countless models of global competitionthat are worth mentioning in a review of TheWorld is Flat, but with too many words inthis review already, some choices have to bemade. The simple Ricardian model dis-cussed below can be used to make an impor-tant point about the offshoring of intellectual

work that so unsettled Friedman on his tripto Bangalore. I will also use this theory sec-tion to express the opinion that the marketmodel we economists are so fond of appliesto very few exchanges, most of which takeplace in the context of long-term relation-ships. Relationships greatly limit the con-testability of our labor exchanges by makingfar-away low-skilled workers an unlikelyalternative to nearby folks whom we havecome to know and trust.

5.1 Did David Ricardo UnderstandOutsourcing?

As we wander through the thicket of theo-ry that relates to Friedman’s Flat World, weneed to revisit David Ricardo’s model ofcomparative advantage, which has been thebedrock argument for the gains from trade,but even this one is the target of attack.

The public debate about the benefits andcosts of “outsourcing” has been a heavy-weight slugfest. Are there gains from tradeor are there not? In the corner on the right,we have Professor Greg Mankiw, chairmanof the Council of Economic Advisors, wear-ing the Crimson trunks and representingtraditional thinking about the benefits offree trade. In the corner on the left, we havethe newly formed tag-team of SenatorCharles Schumer and Dr. Paul CraigRoberts, loudly and publicly promising afirst-round knock-out of Mankiw’s tradition-al way of thinking. (Yes, that is the samePaul Craig Roberts who has always foughtfrom the corner on the right.) In the frontrow, behind the Schumer/Roberts corner, isthe former Democratic Presidential nomi-nee, Senator John Kerry, pointing a longthin finger at the “Benedict Arnold” busi-nessmen whose company logos he imaginesemblazon Chairman Makiw’s trunks. To thesurprise of some, sitting to the left of JohnKerry is none other than Paul Samuelson,holding aloft a clove of garlic to ward offsimple-minded thinkers. Thomas Friedman,perhaps not realizing the company he is



keeping, has chosen to sit with Mankiw andthe other free traders.5

Here is how the match is going:Commenting on the offshoring of intellec-

tual services, Chairman Mankiw, in theEconomic Report of the President, 2004,advises:

When a good or service is produced morecheaply abroad, it makes more sense to importit than to make or provide it domestically(p. 229).

But, based apparently on some seriouslate-night library work, Schumer andRoberts reply in the New York Times6

When Ricardo said that free trade wouldproduce shared gains for all nations, heassumed that the resources used to producegoods—what he called the ‘factors ofproduction’—would not be easily moved overinternational borders. Comparative advantageis undermined if the factors of production canrelocate to wherever they are most productive:in today’s case, to a relatively few countries withabundant cheap labor. In this situation, thereare no longer shared gains—some countrieswin and others lose.

And one thing is certain: real and effectivesolutions will emerge only when economistsand policymakers end the confusion betweenthe free flow of goods and the free flow offactors of production.

My first reaction to Schumer and Robertswas: You need to write on the blackboard100 times: “There are gains from exchange.”Ricardo’s Principles of Political Economyand Taxation (1817) was a good start, buttake a look at the book that got Ricardothinking about the issues: Adam Smith’s AnInquiry into the Nature and Causes of the

5 “The current debate about off-shoring is dangerouslyhot. But the debate about work going to India, China andMexico is actually no different from the debate once heldabout . . . shoe work leaving Massachusetts or textile workleaving North Carolina. Work gets done where it can bedone most effectively and most efficiently. . . . Every per-son, just as every corporation, must tend to his or her owneconomic destiny, just as our parents and grandparentsin the mills, shoe shops and factories did.” (Friedman,p. 20–21).

6 New York Times Op-ed Piece, January 6, 2004.

Wealth of Nations (1776). It was AdamSmith who emphasized the gains fromexchange and the division of labor: “Thepropensity to truck, barter and exchange onething for another is common to all men, andto be found in no other race of animals.” It isthe opportunity to exchange that allows thedivision of labor, from which flow the incred-ible efficiencies characteristic of moderneconomies.

The very fact that there is an exchangeconfirms that there are gains to the partiesinvolved. When ownership is freely passedfrom seller to buyer, the product or serviceincreases in value, passing from one whovalues it less to one who values it more.The gains from exchange occur whenbuyer and seller reside in the same country,and occur likewise when buyer and sellerreside in different countries. Locations ofresidence of buyer and seller are utterlyirrelevant.

What about factor mobility? Does this viti-ate the gains-from-exchange argument asSchumer and Roberts suggest? Not in theslightest. It is irrelevant to the argumentwhether or not the buyer or seller chooses tomove in order to effect the exchange. Whena New Yorker hears of a better job inCalifornia and decides to make the move,there are gains from the exchange thatexceed the costs of the move. If the NewYorker can deliver the services over theInternet, all the better, since the costs of themove are not subtracted from the gains ofexchange. Substituting “New Delhi” for“New York” in this factor mobility argumentmatters not in the slightest.

“Nonsense, nonsense. Leamer, you are justanother John Doe,” Professor Samuelson(2004) would disdainfully observe:

Prominent and competent mainstream econo-mists enter into the debate to educate and cor-rect warm-hearted protestors who are againstglobalization. Here is a fair paraphrase of theargumentation that has been used recently byAlan Greenspan, Bhagwati, Gregory Mankiw,Douglas Irwin and economists John or JaneDoe spread widely throughout academia . . . .



Correct economic law recognizes that someAmerican groups can be hurt by dynamic freetrade. But correct economic law vindicates theword “creative” destruction by its proof [sic]that the gains of the American winners are bigenough to more than compensate the losers.The present paper provides explication of thepopular polemical untruth (p. 135).

If you find this an elliptical way of express-ing disagreement with Mankiw et al., try dig-ging the deeper explanation from the body ofthe text, Samuelson (2004). Fortunately I(Leamer 2004) had some class notes on thisissue that stood ready on the shelf that I canshare with you. It’s a terms of trade effect thatSamuelson is talking about. Disturbances toequilibria can change the terms of trade andalter the distribution of the gains. (Even Johnand Jane Doe know that.) A flow of migrants isone such disturbance. That doesn’t mean thatthere are no gains from trade. It only meansthat the gains are distributed differently.

I am going to call it “immiserizing out-sourcing” in honor of Bhagwati, who standsjust above John and Jane Doe in Samuelson’shierarchy. The model of outsourcing pre-sented here illustrates what might happen ifthe United States loses geographical controlover its knowledge assets. This puts U.S.prosperity at risk. It’s a possibility, not veryclose to reality, I think.

Before we get into that material I have tomake a confession. The argument that thereare gains from exchange presumes that thereare no “externalities” caused by the transac-tion, like the pollution caused when youexchange your hard-earned dollars for thatgas-guzzling foul-exhausting SUV. Here is aquip that makes the point in our context:International trade is an exchange thatmakes both parties worse off. We get theirwages and they get our culture.

5.1.1 Immiserizing Outsourcing: AdverseTerms of Trade Effect on U.S.Intellectual Property

Bhagwati (1958) raised the possibility of“immiserizing growth” almost fifty years agoin a classic paper. Bhagwati warned that a

country that grows more rapidly than itstrading partners inevitably floods the mar-kets with its export goods, which can lowerthe country’s terms of trade and make thegrowing country actually worse off.7

Factor mobility can also affect the terms-of-trade, as explored by Murray C. Kemp(1966) and Ronald W. Jones (1967), whoraised the possibility of immiserizing capitalflows in a two-sector model with technologi-cal differences. Reverse the labels K and Land you have a long list of theorems on theeffect of labor mobility on welfare. For a sur-vey of the articles on the effects of tariffs inthis kind of model, see Roy J. Ruffin (1984).This literature directly addresses the implic-it policy question lying behind the com-ments by Schumer and Roberts: Should theUnited States attempt to halt outsourcingwith some form of tax policy?

We can capture this immiseration idea in asimple Ricardian model. Figure 1 illustratesthe production possibilities of typical U.S.and Indian workers.8 They are equally pro-ductive in making apparel, but the UnitedStates has a technological superiority in writ-ing software. This superiority may come onlyfrom geography and history: Designing andwriting software code requires close com-munications among the participants, and theIndian programmers are too far from theUnited States to benefit from the agglomer-ation externalities that come from the needfor clear communications. You cannot be aprogrammer unless you are “here.”

7 The traditional limit to growth comes from thediminishing marginal product of capital as capital accu-mulates. A startling implication of the small-countryHeckscher–Ohlin model is that even though technologiesat the sectoral level exhibit diminishing marginal prod-ucts, the GDP function does not, since capital accumula-tion is accompanied by a shift from the labor-intensivesectors to the capital-intensive sectors, leaving the capitalintensities at the sectoral level fixed. Immizerising growthrecreates diminishing marginal product of capital for theGDP function through terms-of-trade changes.

8 For expositional purposes, I am illustrating the pro-duction possibilities of a typical worker, not the economyoverall. For this reason, the diagram says nothing aboutthe trade balance.



Apparel

Software

India

U.S.

Figure 1. Production Possibilities: U.S. and Indian Workers

Apparel

Software

Consumption Options

Production

Consumption

U.S.

India

India

U.S.

Figure 2. Specialization and Consumption: U.S. and Indian Workers

Figure 2 illustrates the usual Ricardianequilibrium under these conditions: In addi-tion to the production possibilities solidlines, this figure has two diamonds repre-senting the production choices for India andthe United States, two dotted consumption

possibility lines, two “indifference” curvesthat determine consumer choice and twosolid circles representing the consumerchoices in the United States and in India. Inthis figure, the United States specializes insoftware and India specializes in apparel.



Apparel

Software

India Initial

U.S. and Indian production possibilities with offshoring

U.S. initial consumption

U.S. and Indian consumption with offshoring

Figure 3. Common Consumption Point, After Offshoring of Software Coding

Workers in both countries are made betteroff by this exchange. In this equilibrium, U.S.workers have higher living standards than theIndians because of the U.S. technologicalsuperiority in software.

Figure 3 illustrates the effect of a changein communication technology that allowsIndian programmers to participate in theconversations that are essential to the designand writing of computer code. The straightlines represent the initial production possi-bilities and the open circles the initial con-sumption points. The effect of deliveringIndians to the U.S. workforce over wires is toflood the global marketplace with cheapsoftware, which causes a terms-of-tradedeterioration for the United States. Thisterms of trade deterioration for the UnitedStates and improvement for India musteliminate the income differences in the twocountries to stop the flow of Indians to theUnited States through wires. In the finalequilibrium, the technological differencebetween U.S. workers and Indian workers iscompletely eliminated. The U.S. workers aremade worse off by this offshoring since theylose completely the gains from trade that

came from their monopoly position (marketpower) in the software industry.

There is an important message here. Inthis model, the United States enjoys monop-oly rents from the intellectual property thatit creates, but those rents may dissipate if thefranchise for creating intellectual property isextended to our customers. While this is areal issue, I doubt that this Ricardian modelcaptures the production of intellectual prop-erty very accurately. A critical implicitassumption of this Ricardian model is thatsoftware programs are as alike as plain whitet-shirts. If there are more t-shirts produced,with downward sloping demand, the price oft-shirts must fall. That is what causes theterms-of-trade deterioration in thisRicardian immisserizing outsourcing modelas more coding is done in India. But intel-lectual properties are not stacks of identicalplain-white t-shirts. While some softwareprograms compete with each other, many donot, and the value of one can be enhanced bythe existence of others. For this reason, Idon’t think we need to worry about terms-of-trade deterioration for our intellectual prop-erties—movies, drugs, financial derivatives,



equipment, and software. We do need toworry about intellectual property protection,without which the export value of our intel-lectual property may be seriously impaired.

We also need to worry about the newcompetition for mundane coding tasks. Inthose same notes (Edward E. Leamer2004), I have a Heckscher–Olhin modelwith production done by masters andhelpers. The masters all live in the UnitedStates. Helpers reside in the United Statesand in India. It takes masters and helpers tomake software. It takes only helpers to makeapparel or nontraded services. In the initialequilibrium, externalities and transportationcosts keep the masters and their helpersclustered together in the United States, andthe United States exchanges software forapparel with India. A change in the commu-nications technology that allows U.S. soft-ware masters to hire India helpers caneliminate all the U.S. helpers jobs in thesoftware sector. These U.S. helpers move tothe U.S. nontraded service sector. Who winsand who loses from this change? U.S. mas-ters are sure to gain since they can hirecheaper Indian helpers. U.S. helpers mighthave a different outcome. A flood of dis-placed helpers in the United States seemslikely to drive down their wages, but keep inmind that there are global efficiency andincome gains some of which will raise thedemand for U.S. helpers. Might the incomegains that accrue to U.S. masters translateinto increased demand for U.S. helpers inthe service sector? Call that trickle-downoffshoring.

5.1.2 Production and Innovation areComplementary

A terms-of-trade deterioration afflictingU.S. intellectual products is far down the listof things to worry about. Higher up the listof concerns is the likely collateral but unin-tended transfer overseas of the innovativeprocess in manufacturing along with the pro-duction work. Historically, the research labshave not been far from the manufacturing

plants since figuring out how to make “it”economically feasible was just as importantto the innovative process as discovering “it.”In addition, ongoing cost-reducing processinnovations can only be made by those whoare actually producing the product. As wemove more manufacturing to China, we arecertainly also moving process innovationsthere as well. It remains to be seen if prod-uct innovation can remain in the UnitedStates when production is done elsewhere.

In case you hadn’t noticed how much thestructure of U.S. production is changing, fig-ure 4 illustrates the fraction of value addedoriginating in the three sectors that haveexperienced the greatest declines and figure5 illustrates the same for the four sectorsthat have experienced the greatest increases.It is the “manly man” work in manufactur-ing9 and farming and transportation that ison the decline, falling from 40 percent ofGDP in 1950 to 15 percent today.Meanwhile, it is the intellectual “geek work”in finance, professional services, and infor-mation that is on the rise, as is health care.These high growth sectors have grown from18 percent in 1950 to 44 percent today. Ithink we need to be concerned about theinnovative process in an economy with somuch finance and so little manufacturing.Dude, what is the latest financial derivativereally worth? How much is it going tochange your life?

So what? Does it matter if we transfersome innovative functions out of the UnitedStates? To find an answer, I suggest anotherquestion: Why does the federal governmentof the United States subsidize research anddevelopment, but the City of Los Angelesdoes not? The answer is that the City of LosAngeles would not benefit much from theinnovations the subsidies might support.Most of the value of that subsidizedresearch would leak out of the city, and asubsidy would be a waste of the taxpayer’s

9 If it is the kind of thing you like to do, please substitute“It is the ‘womanly woman’ work in womanufacturing.”


Value Added Fractions of GDP,,,,


Value Added Fractions of GDPGrowth in Geek Work

Information

Educational services, health care, etc.

business services

Finance, insurance, real estate, rental, and

0

5

10

15

20

25

30

35

40

45

50

1947

1950

1953

1956

1959

1962

1965

1968

1971

1974

1977

1980

1983

1986

1989

1992

1995

1998

2001

2004

lea sing

Professional a nd Per c

ent

Figure 5. The Growth of Geek Jobs

Manufacturing

Agriculture, forestry, fishing, and hunting

Transportation and warehousing

0

5

10

15

20

25

30

35

40

45

50

1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002

Decline in the Manly SectorsPe

rcen

t

Figure 4. The Disappearance of Manly Work

money. For the United States, it has beendifferent. Much of the research and devel-opment done here has raised the productiv-ity of the U.S. workforce relative tocompetitors. However, “globalization” ismaking the United States overall prettyleaky, and we taxpayers need to insist thatour taxes subsidize the creation of intellectu-al assets that are sufficiently immobile that

the local return is high enough to justify thesubsidy. For the mobile intellectual assets, itdoesn’t matter to me whether they are cre-ated in Seattle or downtown Los Angeles orin Beijing. Friedman’s call10 for more R&D

10 “At a time when we need to be doubling our invest-ments in basic research to overcome the ambition and edu-cation gaps, we are actually cutting that funding” p. 268.



support by the U.S. government thus seemsmistaken policy in the “flat” world he imag-ines. Better to concentrate on the immo-bile assets—the researchers (not theresearch), and infrastructure, including theparks and public spaces that these highlypaid knowledge workers enjoy.

5.2 Markets or Relationships: Contestableor Negotiable Exchanges?11

The worry that your job is going to betaken away by a cheaper worker in India orChina is based on the supposition that thereis a “market” for the tasks that you perform.Though we economists talk as if mostexchanges were meditated by markets towhich our simple supply and demand modelapplies, in fact most exchanges require long-term relationships between buyer and seller.The “capital” invested in these long-termrelationships creates a situation of bilateralbargaining that responds to competitivepressures in much softer ways than would amarket. It’s the difference between con-testable versus negotiable exchanges. A mar-ket has contestable exchanges that cannot beconsummated if they deviate in the slightestfrom the “market.” If the global marketoffers to do your work for one cent lesstoday, then you either accept a wage reduc-tion tomorrow morning, or you lose your job.That isn’t your situation, is it, o ye of tenurestatus? Your university job is not contested,is it? Your job is negotiated. The supply anddemand balance for academic economistscan change sharply but it will take a verylong time for you to feel that change. Younegotiate with your employer (theUniversity) and your employer negotiateswith the paying customers (state legislatorsand voters and students and donors).

That isn’t like a market at all. At theeconomist’s hypothetical “market,” count-less faceless buyers meet countless faceless

11 “Markets and hierarchies” is Oliver E. Williamson’s(1975) dichotomy. I prefer “markets or relationships” and“contestable or negotiable exchanges.”

sellers, and carry out exactly the sametransaction—exchanging at one “instant” oftime x units of a good or service A whosevalue is transparent to both buyer and sell-er for y units of good or services B, alsowith a transparent value.12 The market isthick with buyers and sellers but there areno relationships. These market participantsdon’t even know each other.

Buyers and sellers in this hypotheticalmarket are all brought together into thesame “space” so that they can overhear theterms of the exchanges being offered by oth-ers and, from that information, cut a betterdeal for themselves. With all that overhear-ing, the participants haggle their way to acollective rate of exchange, the “marketprice” at which total sales exactly match totalpurchases. Having found this market rate,the exchanges take place, and the partici-pants go on their merry separate ways, neverto see each other again.

In this hypothetical market, the identities ofbuyer and seller are totally irrelevant. Itdoesn’t matter where the buyers or sellersreside, what their nationalities are, who theywork for, who their parents are, where theywent to school, how big are their bankaccounts, how attractive they are, what theirbeliefs are, whether they are men or women,old or young, tall or short, convicts or ex-con-victs, sinners or saints, con men or paragons.The exchange is completely impersonal.

A market is infinitesimally close geographi-cally, but infinitely remote personally. “Arm’slength” is the way we describe it. Closeenough to overhear, but too far to shake handsand way too far to hug. (Gestures of trust.)

In fact, there are very few exchanges thatare mediated by “markets.” There are veryfew “commodities” whose value is transparentenough to allow the formation of a market.There are very few exchanges that take placeat a frequency high enough and transparent

12 Often, one of the items exchanged is “money,”standing for a basket of goods or services.



enough to other potential participants thatmarket prices can emerge. Most exchangestake place within the context of long-termrelationships that create the language need-ed for buyer and seller to communicate,that establish the trust needed to carry outthe exchange, that allow ongoing servicingof implicit or explicit guarantees, that mon-itor the truthfulness of both parties, andthat punish those who mislead. Manyexchanges occur between colleagues whowork for the same firm. Indeed, about 40percent of U.S. imports are carried outinternal to multinational enterprises.

The fear that seems to underlie much ofFriedman’s flat earth metaphor is that workis becoming commoditized and sold in glob-al markets. What got him to his “flat world”conclusion was his observation that softwarecoding in Seattle seems contestable fromBangalore.

But: Is Friedman’s job commoditized? IsFriedman’s job contested?

I think I am getting close again to under-standing the Flat World Metaphor. It ismostly about the commoditization of work,and the extension geographically of the con-test for mundane work in manufacturing andservices like sewing apparel and readingmanuals at call centers over the phone andalso not-so-mundane intellectual activities.

The central issue is whether Americans aregoing to sell their products and services in aglobal “market” that completely determinesthe wages and working conditions, or areAmerican jobs going to be protected by rela-tionships and by geography. Will export andimport-competing jobs be confined to spe-cialized negotiable “craftwork” in which theforces of global competition are greatlyattenuated, while the contestable jobs arelimited to nontraded sectors with a contestthat does not extend beyond borders? Or, asFriedman’s flat earth title suggests, are thedramatic changes in communications tech-nology, including the Internet and also essen-tially free telecommunications extending thecontest for nontraded contestable work across

borders and also turning secure negotiablejobs into insecure contestable jobs?

5.2.1 The Luddites Preferred the OldRelationship-Based Way of SellingCloth

An important determinant of contestabili-ty is product maturity. Product innovationsare the waves that bring new high-payingskilled “craft” jobs to innovating communi-ties, but standardization of the product andmechanization of the process inevitablycause the waves to recede as routinized jobsat low wages replace skilled jobs at highwages. Attitudes toward technology andtrade naturally depend on whether the waveis advancing or receding. As we currently feelthe undertow of the jobs receding from theUnited States, it is wise to understand thatwaves have been advancing and recedingsince the Industrial revolution began to stan-dardize and mechanize the weaving of clothat the beginning of the nineteenth century.

Prior to 1780, British cotton textiles wereproduced in small artisan shops and on fami-ly farms using spinning wheels to make thethread and hand-driven looms to weave thecloth. The Industrial Revolution broughtnumerous new machines to help do the work,including John Kay’s flying shuttle patented in1733, and James Hargreaves’s spinning jennypatented in 1767 and Richard Arkwright’swater-powered spinning frame in 1769. Theavailability of this machinery moved the jobsout of homes into small mills near streamsthat could supply the waterpower. Thensteam power allowed the mills to leave thestreams of rural England in search of cheaperurban labor. It was here that workers, caughtin the undertow of mechanization and stan-dardization, responded with the Ludditedestruction of textile machinery.

The Luddites were upset not just with themachines but also with the way that a marketsystem prices standardized products. Whenartisans made customized goods, the price wasnegotiated and skilled workers’ interests couldbe pursued by wise and/or benevolent master



craftsmen. That was relationship-basedexchange. But, once the product is standard-ized and the process of production is mecha-nized, the price is set by the “market” andlaborers are forced in effect to bid againsteach other to do the work. Turned over com-pletely to market forces, workers lose controlof their workplace and their financial securityand even their pride. The Luddite song,“General Ludd’s Triumph” points to these twoconcerns: machines and pricing:

The guilty may fear, but no vengeance he aimsAt the honest man’s life or EstateHis wrath is entirely confined to wide framesAnd to those that old prices abate

Here “wide frames” is a reference to themachines and “old prices” is a reference tothe negotiated/administered prices that werea characteristic of artisan shops. Smash themachines and give us back our old negotiatedprices.

5.2.2 Which Activities are Contestable:Mundane Work or Creative Work?

The Luddites correctly perceived thatmechanization and standardization of textileproduction was turning their jobs into taskscontestable by workers near and far.

The key word here is “contestable.”Friedman’s flat world hypothesis seems to bethat there are or will be many U.S. jobs thatare contested by Chinese and Indians. Thisstrikes me as rather far from reality. It is onlythe mundane codifiable tasks in tradables for

which there are global markets.13 You’d besurprised how few of those remain in theUnited States. Table 1 reports imports andexports divided by U.S. value added fromthe 1998 and 2004 Input–Output Tablesprepared by the BEA. Also included in thetable is the share of GDP that originates inthe sector, the cumulative thereof, and, inthe last column, the change in the GDPshare from 1998 to 2004.

Sectors are sorted by the 2004 importpenetration ratio—the ratio of imports tovalue-added. At the top is apparel andleather products, with a 2004 import pene-tration ratio of 341 percent, up from 154percent in 1998. While imports in that sec-tor were on the rise, exports were stagnant,and value added as a share of GDP wasdeclining. That’s a contested sector. Anothersector that lost a lot of value added wascomputer and electronic products, whichexperienced a rise in the import penetrationratio from 99 percent to 147 percent. At thebottom of this list is motion picture andsound recording industries, which has highlevels of exports and low levels of imports,though both were on the rise in this period.With imports in 2004 only 4 percent ofvalue added, while exports are 29 percent itseems likely that the jobs in this movie sectorare only minimally contested globally.

I have labeled this table “hints of con-testability” because some serious econo-metric work would be needed to translate

13 These ideas were developed in Leamer andMichael Storper (2001). Levy and Murnane (2004) offera highly interesting and closely related framework forthinking about cognitive tasks. What I am calling mun-dane, codifiable tasks, Levy and Murnane call rule-based deductive tasks. For Levy and Murnane, thenonroutine, noncodifiable tasks are pure pattern recog-nition which cannot be articulated. Though it’s a greatinsight, I am not so sure that pattern recognition versusdeduction is the key divider between codifiable andnoncodifiable tasks, since simple pattern recognition iscarried out well by computer algorithms of variousforms, including the statistical tool that is wisely avoid-ed by most economists: stepwise regression. (We don’trely on stepwise regression or any other automated sta-tistical pattern recognition to pull understanding from

our data sets because there is currently no way of pro-viding the critical contextual inputs into those algo-rithms and because an understanding of the context isabsolutely critical to making sense of our noisy nonex-perimental data. The last person you want to analyze aneconomics data set is a statistician, which is what you getwhen you run stepwise regression.) Furthermore,though Levy and Murnane’s classification is intended toapply only to intellectual work, it does not extend natu-rally to physical activities, like sports or ballet, wheresome part of the tasks are codifiable and learnabale in aclassroom, but most require extensive learning-by-doing. But regardless, Levy and Murnane and I are incomplete agreement on the critical point: the footloosecontestable tasks are the routine codifiable tasks, bothin manufacturing and intellectual services.



TABLE 1HINTS OF CONTESTABILITY

Value Added Shares andTrade Penetration Trade to Value Added Ratios GDP Share

1998 2004 1998 2004 ChangeImports Exports Imports Exports Share Cum Share Cum

Apparel and leather and alliedproducts 154% 17% 341% 18% 0.62% 0.6% 0.27% 0.3% –0.35%Fabricated metal products 130% 97% 340% 161% 0.19% 0.8% 0.09% 0.4% –0.11%Electrical equipment, appliances,and components 97% 57% 205% 80% 0.37% 1.2% 0.20% 0.6% –0.17%Plastics and rubber products 115% 93% 194% 108% 0.16% 1.3% 0.12% 0.7% –0.04%Paper products 131% 105% 174% 104% 0.14% 1.5% 0.11% 0.8% –0.04%Computer and electronic products 99% 68% 147% 71% 1.97% 3.5% 1.28% 2.1% –0.69%Textile mills and textile productmills 69% 37% 114% 47% 0.23% 3.7% 0.16% 2.2% –0.07%Miscellaneous manufacturing 73% 23% 90% 27% 0.70% 4.4% 0.64% 2.9% –0.06%Chemical products 67% 56% 80% 54% 1.24% 5.6% 1.36% 4.2% 0.12%Motor vehicles, bodies and trailers,and parts 55% 24% 77% 25% 2.79% 8.4% 2.26% 6.5% –0.54%Printing and related supportactivities 35% 43% 77% 59% 0.04% 8.5% 0.02% 6.5% –0.02%Machinery 42% 40% 57% 47% 1.90% 10.4% 1.35% 7.8% –0.55%Furniture and related products 22% 5% 50% 5% 0.60% 11.0% 0.42% 8.3% –0.18%Petroleum and coal products 24% 19% 41% 14% 0.47% 11.4% 0.86% 9.1% 0.39%Farms 39% 55% 33% 49% 0.43% 11.9% 0.41% 9.5% –0.02%Air transportation 33% 37% 32% 38% 0.68% 12.5% 0.61% 10.1% –0.07%Other transportation equipment 24% 61% 28% 53% 1.24% 13.8% 0.89% 11.0% –0.35%Mining, except oil and gas 5% 91% 18% 64% 0.07% 13.9% 0.05% 11.1% –0.02%Food and beverage and tobaccoproducts 9% 10% 13% 7% 3.76% 17.6% 3.35% 14.4% –0.41%Insurance carriers and relatedactivities 5% 2% 13% 3% 1.97% 19.6% 1.96% 16.4% –0.01%Miscellaneous professional,scientific and technical services 8% 17% 8% 13% 0.79% 20.4% 0.79% 17.2% 0.00%Motion picture and soundrecording industries 2% 24% 4% 29% 0.30% 20.7% 0.29% 17.5% 0.00%

these numbers into legitimate measures ofcontestability, recognizing among otherthings that exports as well as imports cancreate global contests. It does seem likelythat the industries at the bottom of this listwith low ratios of imports to value addedare minimally contested and the ones notshown but further down the list are evenless so. Referring now to the cumulativeGDP share, the sectors in this table com-prised only 21 percent of GDP in 1998 and17.5 percent of GDP in 2004. The other 80

percent are mostly nontraded sectorsincluding, for example, 10 percent of valueadded in real estate, 9 percent in state andlocal government, 9 percent in construc-tion, 8 percent in retail trade, 6 percent inthe federal government, 6 percent in hospi-tals, 5 percent in ambulatory care and so on.But, I hasten to add, wages in Los Angeleshospitals can be set in Shanghai by a chainof contests. If apparel producers in LA arecompeting in the same product markets asapparel made in China, and if workers in



LA can choose between hospital work andsewing garments, then there is a chain ofcontests linking U.S. hospitals to garmentfactories in China. To summarize, deter-mining the degree of global contestabilityfor U.S. work is an important task that hashardly begun.

Though the evidence is imperfect, thedata in table 1 do suggest that the contest forU.S. jobs is not as great as newspapers sug-gest. You should not be alarmed by each lit-tle bit of additional news on the subject. Agreat example of the hype of the offshoringthreat is in the area of radiology recentlyscrutinized by Levy and Goelman (2005)who start with a quotation from the NewYork Times:

It turns out that even American radiologists,with their years of training and annual salariesof $250,000 or more, worry about their jobsmoving to countries with lower wages, in muchthe same way that garment knitters, blast-fur-nace operators and data-entry clerks do.Radiology may just be the start of patient careperformed overseas. (The New York Times“Who is Reading Your X-Ray?” Pollack,November 16, 2003)

But, Levy and Goelman argue, issues ofreliability and liability are sure to keep mostof the complex x-ray reading right here athome, close enough so you or your generalpractitioner can shake your radiologist’shand, and well within the reach of America’strial attorneys.

5.3 What I Mean by Mundane

The words “mundane” and “codifiable”may not be altogether clear and, to help out,here are some tasks, ordered by their mun-danity:

• Type this page.• Edit this page.• Write an article for an Economics

journal.• Write a good joke.

I am told that Ph.D. students atNorthwestern when faced with the task of

transcribing printed pages of numbers intomachine-readable files scan the documentsand e-mail them to India for overnight typ-ing. I wonder if very many students sendtheir dissertations to India for editing? Myexperience with U.S.-based editors rangesfrom annoyance to outright outrage. If youfind an editor who makes you happy, you’dbetter cling to him for dear life. Create arelationship and for heaven’s sake don’tthreaten to send the job to India if he doesn’tcut his fee. He will dump you in an instant,and you will have to do the editing yourself,or not.

In the case of typing a page, both partiesto the transaction understand what it meansto carry out the task and both can easily andaccurately determine if the task was donewell. But when I first ask you to edit mymanuscript, neither you nor I know whatthat means. Once you have made anattempt, I can tell you if I like what you havedone or not. If we pass manuscripts back andforth enough, you and I are going to learnwhat I mean when I ask, “Please edit thismanuscript.” That’s an investment in lan-guage that is specific to you and me. Thatlinguistic investment creates a strong bondbetween us—a relationship. That’s whythere are no markets for editorial tasks—ittakes a relationship.

Writing a publishable Economics article isa more difficult craft that can only belearned in a slow and sometimes painfultrial-and-error process. Part of the learningprocess involves face-to-face conversationswith the other members of the guild, espe-cially the master craftsmen. Absent frequentattendance at seminars and conferences, it isvirtually impossible to learn this craft.Neither the Internet nor free telecommuni-cations has had much impact on that geo-graphical reality. For the creation of newideas—it takes a neighborhood.

Friedman seems to know this on somepages of this book and ignore it otherwise.According to Jaithruth Rao, MphasiS, (“oneof the first people I met in Bangalore”)



What we have done is taken the grunt work.You know what is needed to prepare a taxreturn? Very little creative work. This is whatwill move overseas. . . . The accountant whowants to stay in business in America will be theone who focuses on designing creative complexstrategies. . . . (p. 13).

And a few pages later, Tom Glocer, CEOof Reuters

We think we can off-load commoditized report-ing and get that done efficiently somewhere elsein the world (p. 18).

6. The Mobility of Ideas

Most discussions of globalization and flatearths emphasize the markets for goods andsome services, but the phenomenon isbroader than that.

Globalization is the increased internation-al mobility of goods, people, contracts(including financial claims) and thoughts(facts, ideas, and beliefs).

There is a difference between mobilityand movement. Think about two reservoirsof water at different heights that are keptfrom seeking a common level by a separat-ing barrier. Thus neither movement normobility. Next tear down the barrier andmake the water completely mobile. Onepossibility is that that there is a movementof water from the high side to the low side.Thus mobility and movement. Another pos-sibility is that in anticipation of the removalof the barriers, the folks on the high enddrain their reservoir to exactly the sameheight as the low reservoir. Now tear downthe barrier, and there is mobility but notmovement. Don’t think “silly” when youread this. U.S. wages can move down to for-eign levels from the threat of competition,with no flows of goods or workers across theborders.

Of these four mobilities, it is the last thatis probably the most important—the mobili-ty of ideas. Evidence for this comes from thetwo great tests of the power of globalization.The first test was created by Nature and thesecond by Man. Jared Diamond, in his

remarkable book Guns, Germs and Steel,poses the question: why did Europe conquerthe Americas, rather than the other wayaround? One of the answers is globalization.The north–south orientation of the Americasdid not support the transfer of technologybecause innovative crops and livestock thatwould flourish in one latitude would not sur-vive in another. In contrast, the middle lati-tudes stretched all the way from Europe toChina, allowing grains discovered in theMiddle East to be grown in Spain or inChina, and gunpowder discovered in Chinato be deployed in European arms.

Though those east–west trade routesbetween Europe and China were establishedto exchange goods, it was the stowaway ideasthat traveled with the goods that really mat-tered. Europe conquered the Americasbecause of it superior mobility of thoughts.

The second test of globalization came inthe aftermath of World War II when abouttwo thirds of humanity lived in inward-look-ing isolationist countries, and the other onethird (Western Europe, the United States,Canada, Japan, Korea,Taiwan, and a few oth-ers) created a great trading network withinwhich stowaway ideas traveled freely alongwith the goods. The isolationist policies ofthe governments of the two thirds impededthe flow of ideas across borders, but worsestill, in an effort to maintain political supportfor those external barriers, these countriestypically controlled and thus greatly slowedthe exchange of ideas internally, throughcontrol of the media and spying on domesticcitizens. The technological progress createdin the one third group of trading nations leftthe two thirds far behind, and eventuallycaused those inward-looking isolationist gov-ernments to pursue new policies.

6.1 The Speed at Which Ideas areExchanged Determines the Pace ofProgress

It is the speed at which ideas are passedbetween people that determines the rate of



14 http: / /en.wikipedia .org/wiki /A%26E%27s_Biography_of_the_Millennium.

15 “Block printing, whereby individual sheets of paperwere pressed into wooden blocks with the text and illus-trations carved in, was first recorded in Chinese history,and was in use in East Asia long before Gutenberg. By thetwelfth and thirteenth century many Arabic and Chineselibraries contained tens of thousands of printed books.The Koreans and Chinese knew about movable metaltypes at the time, but arising from the complexity of theChinese writing system, movable type printing wasn’t aswidely used as that of Renaissance Europe.”http://en.wikipedia.org/wiki/Johann_Gutenberg.

progress. As long as the only way to transferideas was via person-to-person conversa-tions, progress was measured in tens of cen-turies starting with the evolution of homosapiens in 200,000 BC. Early written lan-guage such as Egyption hieroglyphs in 4000BC was an important step forward since itallowed an innovator to express her ideas in“writing” and then have those thoughts readby others for as many years and viewings asthe parchment and tablets could withstand.While this increased the mobility of ideas,hand-transcription was an expensive anderror-ridden process, restricting access tothe texts to a very few. Enter JohannGutenberg, the most influential person of themillennium per the Arts and EntertainmentNetwork.14 Gutenberg’s movable type andprinting press allowed the mass printing ofbooks, beginning with the Gutenberg Biblein 1455.15 Without printed books or anequivalently efficient system for passingideas between people, it is hard to imaginethat there could have been an IndustrialRevolution in the eighteenth and nineteenthcenturies—no Renaissance, no ScientificRevolution, no steam engine, no electricmotor, no internal combustion engine, notelegraph, no telephone . . . . At least notso soon.

But like tablets and parchment, printedbooks and physical documents have impor-tant limitations. There are still substantialfixed costs of going from author’s text toprinted document, and there are substantialdistribution and storage costs to create the

libraries of paper documents and books, andthere are very substantial search costs to findthe right pages of the right book at the righttime. The Internet has changed all this andhas done so virtually overnight. That con-trasts greatly with the printing press. It tookseveral centuries after Gutenberg for theneeded infrastructure investments in press-es, literacy and libraries to provide access tothoughts to significant fractions of the globalpopulation.

The Internet is like the printing press interms of its potential impact on the mobilityof thoughts. The Internet gives access todayto the thoughts you had just yesterday toevery web-enabled person all over the globe.You just have to post your thoughts on theweb and hope that the search engines willpoint the relevant readers to your materials.The “publication” costs on the Internet areminimal, not even any pesky editors and pub-lishers standing between you and your audi-ence. The distribution costs are virtually zeroonce the Internet and literacy infrastructureare in place. And digitalized electronic searchis ever so much more efficient than the “ana-log” visual browsing that you and I used toconduct in the bowels of university libraries.Absent efficient search, the massive amountsof textual, numerical, and visual informationposted on the Web is a mountain of noisethat completely covers the faint signal ofvaluable data. To turn that mountain of datafirst into information and later into wisdomand insight, we need a filtering system thatseparates the signal from the noise, that sortsthe important from the relevant and the rel-evant from the irrelevant. Thank you Google.

How can one be anything but optimisticabout the impact that the Internet will haveon progress? If you think of humans onEarth as a single thinking organism, then,until the 1980s, we have been using onlyabout one third of our global brain becausetwo thirds of humanity were shut in closetswhere they couldn’t communicate with therest of us. The liberalizations that have sweptthe globe have thus tripled the number of



useable neurons in our collective brain.This has to be a good thing for all of us.Unlike the production of standard washingmachines and toaster ovens, the innovativethoughts produced by the newly reintegrat-ed two thirds do not necessarily competewith the innovative thoughts produced byour one third. It’s not a zero sum game. It’sa cumulative process with the innovators oftoday standing on the shoulders of thegiants of yesterday. And add to that thepower of the Internet that increases thespeed of our global brain by many multi-plies. We are a heck of a lot smarter nowthan we were a couple of decades ago. Thetwenty first century is sure to have a paceof innovation that is unrivaled in humanhistory.

7. Facts:

That’s the theory. What about the facts?Thomas Friedman and most journalists

use what Dan Suits at the University ofMichigan called “man-who” statistics, as in “Iknow a man who . . . ” This book starts withan interview of Nandam Nailekani, CIA ofInfosys Technologies, and then DavidNeeleman, CEO of JetBlue AirwaysCorporation, and then General RichardMeyers, chairman of the Joint Chiefs of Staff,and then Colin Powell, and then StevenHolmes, UPS spokesman, and then DavidGlass, CEO of Walmart, and then and thenand then an obscure economist or two(Robert Lawrence and Paul Romer) (anotherjoke, of course.)

Economists create knowledge, or thinkthey do, using an entirely differentmethodology that keeps them always veryfar from CEOs and any of the other actorsin the drama they study. Some economistsnoodle away at the blackboard or on padsof paper. We call them theorists. Othereconomists stare at computer screensdownloading data from various websitesand organizing those data with econometricsoftware packages. I am one of these. We

look not at man-who statistics but ratherpopulation means, or the multivariateequivalent, estimated models.

There is something we can learn fromFriedman’s methodology. He does anextraordinarily good job creating ideas, espe-cially when you consider that he is not acard-carrying member of our guild. But, Ithink, he could benefit from some theoryand some evidence, put together the way wedo. If he had some of these, he would havebeen led to an alarmist view of another sort.

From the countless facts that are materialto whether or not The World is Flat, I haveselected just a few.

• The income distribution is not becom-ing flatter, neither between countriesnor within the United States.

• Trade is a neighborhood phenomenon,close to home geographically and orga-nizationally.

• Trade contributes to the decline in man-ufacturing jobs but doesn’t seem to bethe primary driver.

• Outsourcing of intellectual work is asmall drop in a very large bucket.

• The United States is extremely wellpositioned to compete in the Internet-based segment of the economy.

7.1 The Tale of the Three-Legged Dog

Before we begin to look at some facts, Ihave to tell you a story about the use of the-ory and evidence to guide decision making.

When asked to determine what wouldhappen if a dog lost one of it’s front legs,one group of researchers went to Toys R USand purchased a toy model of a dog thatbounced around and barked when turnedon. After many hours of play with thismechanical marvel, these researchersdeveloped an affection for their toy modelthat was as deep and intense as for the realthing. They were deeply saddened to dis-cover, when they removed the front leg,their dog pitifully lay on the ground wig-gling in a circle, though still barking as ithad before.



0

20 ,00 0

30 ,00 0

0% 20% 40% 60 % 80 % 100 %

USA

China

Average per capita GDP = $5079

1980 Gl obal Income Distribution

India

1 0,0 00

Cumulative Population Fraction

Figure 6. The 1980 Global Labor Pool

Per

Cap

ita I

ncom

e, $

1995

Another group of researchers searchedfor real examples of three-legged dogs andfound, to their amazement, that those realthree-legged dogs could run and jump,and seemed hardly different from realfour-legged dogs.

I am the proud owner of a three-leggeddog that has been teaching me about life andeconomics.

7.2 No “Flattening” of Global and LocalInequality

7.2.1 The Global Labor Pool: No FlatteningHere

Figure 6 illustrates the global income dis-tribution in 1980. This figure ignoresinequality within countries and merelyassigns to each individual the per capitaincome of his or her country. The verticalaxis is real GDP per capita in $1995 usingPPP exchange rates and the horizontal axis isthe fraction of the global population. Eachcountry in this figure is represented by a rec-tangle with height equal to per capitaincome, with width equal to the populationfraction and with area representing totalGDP (population times GDP per capita).Three rectangles are shaded: the USA on the

left with a high per capita income and a largepopulation fraction, and India and China onthe right with low per capita incomes andvery large population fractions.

Countries in this figure are sorted by percapita incomes, left to right, with the wealthycountries on the left and the poor countrieson the right. Call this the global labor “pool,”a very strange pool indeed, with the liquidpiled up high on one end and hardly presentat the other end. What could possibly beholding up the high end? A prevalent answerin 1980 was that the two thirds of humanityin the low-income part of the global laborpool lived in countries with governmentsthat adopted inward-looking isolationistpolicies that prevented businesses frompursuing the obvious arbitrage opportunityof buying labor where it is cheap and sellingthe products where labor is dear. The liber-alizations that swept the globe promised tounleash these arbitrage opportunities and tocreate a global labor market—a flat worldper Friedman. Call it the Great Equalizationin which wages and per capita incomes inthe poorest parts of the globe would rise dra-matically, but only by “draining” the high-income end of the pool. If the liquid in thispool were to find a common level, it would



16 Maddison data, http://english.historia.se/.

leave every country with a per capita incomeequal to $5079, a little less than one fourth ofthe 1980 per capita income in the UnitedStates. That’s turning the clock back to 1910,when the United States last had a per capitaincome that low.16

Don’t be alarmed. Equalization of percapita incomes isn’t going to happen. Onereason why it can’t happen is that globalarbitrage works to equalize only the earn-ings of identical workers performing identi-cal tasks. Much of the difference in GDPper capita among countries comes from thegreater amounts of physical and human cap-ital in the West, which advantages aren’tgoing to go away any time soon. A secondview about the potential effect of these lib-eralizations focuses on the huge unskilledlabor forces in China and India andIndonesia and the Soviet bloc and Africaand on and on. These liberalizationsincreased the effective global supply ofunskilled workers enormously with no com-mensurate increase in human capital orphysical capital or natural resources.Inevitably, this would mean a decline in theglobal wage rates of unskilled workers andtheir substitutes, but an increase in theglobal rate of compensation for comple-mentary factors. Some serious econometricwork is needed to determine which factorsof production are complements and whichare substitutes for unskilled labor, but youand I can guess. Both physical capital andsemi-skilled workers are substitutes forunskilled workers in some activities. Thinkabout construction. You can get that oldconcrete removed by hiring one skilledworker operating a jackhammer or fiveunskilled workers using sledgehammers.Agriculture is the same thing. Workers onU.S. farms operate extremely expensiveequipment for planting, cultivating and har-vesting. Elsewhere in the globe, much ofthis farm work is done by hand. But manu-facturing is different. The standardized and

mechanized methods of production in man-ufacturing leave little scope for the substi-tution of unskilled workers for capital, andan increase in the global unskilled or semi-skilled workforce in manufacturing is likelyto require more physical capital. Thus thehuge increase in the effective global supplyof unskilled workers seemed sure to lowerthe wage rates of workers without highschool educations, and sure to increase thecompensation for natural resources and forskilled workers with college educations andabove, but has unclear effects on the mid-dle range of human skills (high school grad-uates or less) and on the return to physicalcapital.17 According to this view, countrieswell endowed in natural resources andhuman capital benefit as the Chinese andIndians and Indonesians and Africans areintegrated into the global economy, butcountries with large endowments ofunskilled workers suffer from the increasedcompetition. Within countries, expect a riseof inequality as the unskilled suffer fromthe hot breath of Chinese competition.

There is a third entirely optimistic view.According to these optimists, we should beexpecting the global labor pool to be lev-elized mostly from below, with dramaticgains in incomes in the poor countries but nodownward pressure on incomes among therich. According to this view, the low incomesamong the left-out two thirds came not fromthe absence of trading opportunities butmostly from technological and institutionalbackwardness. According to this view, it isnot mobility of goods that matters most; it isthe mobility of ideas. If one allowed com-mercial exchanges between East and West,ideas would flow along with the goods. Then

17 The very low real rates of return on financial invest-ments in the wake of the 2001 recession are whatGreenspan has called a “conundrum.” Part of the expla-nation is that global GDP growth is more concentrated inhigh savings countries in Asia, but another part has to dowith the surprising weakness in demand for loanablefunds from all those low-skilled, low-wage workers thathave been added to the global local market.


1980 and 2000 Globa l Income Distributio n


0

10,000

20,000

30,000

40,000

0% 20% 40% 60% 80% 100%

Cumulative Population

2000

Japan

China India

1980

U.S.

Figure 7. Global Labor Pools in 1980 and 2000

Rea

l Per

Cap

ita G

DP,

$19

95

the Western way of living would move East,and North and South as well, and Westerntechnology and Western market competitionand Western business organizations wouldraise the living standards of all those whohad been left behind.

So what actually happened and why?Figure 7 compares the 1980 global incomedistribution with the year 2000 distribution.The year 2000 distribution isn’t flatter at all.While it is true that there was substantialincome growth in the shallow (poor) end ofthe pool, most importantly in India andChina, there was also very substantialincome growth at the wealthy end. Indeed,half of global GDP growth in this periodoriginated in four countries: two wealthycountries (United States and Japan) and twopoor countries (China and India). Butincome growth didn’t occur uniformly. Theglobe’s middle class was left behind, with noincome growth over those two difficultdecades between the seventeenth percentileand the thirty-sixth percentile.

A Great Equalization thus did occur—it justdidn’t apply to the wealthiest 18 percent ofcountries. The wealthiest countries managedto prosper in the face of manufacturing jobs

being tugged into the poor part of the globewhile the middle class countries stagnated.

What is the difference between thewealthy and middle-income countries? Ibelieve that the answer is that the wealthyand the poor countries have different“domains of competition.” Inside of manu-facturing, there is a segment of footloosemundane labor-intensive activities and aset of innovative and/or capital-intensiveactivities that are firmly rooted. The foot-loose standardized products can be accu-rately described in documents (blueprintsand engineering specifications and words)and the finished products can be easilyinspected to determine if they meet thespecifications.

The footloose standardized products aresold in competitive global markets, whichcontrol the prices and the wages as well. But,because of market power, the prices of theinnovative capital-intensive products are setby their manufacturers, who thus have con-siderable leeway in setting wages and work-ing conditions. Think Ely Lilly today or FordMotor Company in 1965.

If your country’s prosperity in 1980 depend-ed on attracting the mundane footloose



manufacturing activities most notably inapparel, footwear, textiles, and consumer elec-tronics, your best hope for economic growthwas to begin in poverty. The global competi-tion for these footloose jobs was and is hope-lessly intense. That is the story of the globe’smiddle class. Their prosperity depended onattracting footloose manufacturing jobs fromthe high-wage countries but they couldn’tcompete successfully for those functionsagainst China and other very-low-wage coun-tries.

However, the high-income countries in1980 prospered in the two subsequentdecades because their domain of competi-tion was mostly in the rooted functions andbecause they could derive substantial eco-nomic gains from transferring the relativelysmall numbers of footloose jobs to the low-wage countries. That is not to say that every-one in the high-income countries gained.On the contrary, those workers who foundtheir jobs contested by low-wage, far-awayworkers either had to find a way of escapingthat competition through training and edu-cation or they had to accept lower wagesand harsher working conditions than theymight otherwise have experienced.

In other words, we have been here before.Newspaper anecdotes regarding the “off-shoring” of mundane service activities likecall centers and not-so-mundane activitieslike writing software have raised againalarms about a Great Equalization. But wehave heard “The World is Flat” before and itdidn’t turn out that way. That doesn’t meanto fall asleep at the switch. The liberaliza-tions that created a global labor market formundane manufacturing caused a massivechange in the competitive landscape, andthose countries and individuals who wereprepared for the change prospered whilethose who did not had harder times. The vastimprovements in telecommunications andthe Internet have also caused a massivechange in the competitive landscape, whoseeffects are just beginning to be evident. Wewould be wise to learn from the first Great

non-Equalization to prepare for the next.The lesson of the first is that infrastructureand workforce quality can create deep rootsthat hold the best jobs firmly in place.

7.3 Trade in Products Is a NeighborhoodExperience

7.3.1 The Gravity Model: Location,Location, Location

There is very little that we economistsfully understand about global trade but thereis one thing that we do know—commercedeclines dramatically with the distance. It’snot a small world.

The distance effect on international com-merce is described by what is known as thegravity model, which is one of the first mod-els estimated by economists,18 and possiblythe only important finding that has fully with-stood the scrutiny of time and the onslaughtof econometric technique. According to thisempirical model, commerce between anytwo countries is proportional to the productof the masses (GDPs) divided by the distancebetween them raised to approximately 0.9.19

The message of this gravity model is that theglobe is not nearly as small as newspapersand business school curricula suggest.

You can see the powerful reality of thegravity model in figure 8, which is a scatterdiagram comparing the intensity of WestGerman trade in 1985 with the distance toits trading partner. On the vertical axis is ameasure of trade intensity: trade with thepartner divided by partner GNP. On the hor-izontal axis is the distance from Germany tothe partner. Both scales are logarithmic. If

18 See Leamer and James Levinsohn (1995) for a briefreview.

19 Per the meta-analysis of Anne-Célia Disdier andKeith Head (2005). The distance elasticity depends onthe product. See Leamer (1994) and James E. Rauch(1999), who offers an explanation in terms of trust andunderstanding by grouping products: “OrganizedExchange Goods,” “Reference Priced Goods,” and“Differentiated Commodities.” The elasticity is higherfor some commodities and lower for others.



0%

1%

10%

100%

100 1,000 10,000 100,000

Tra

de /

Part

ner

GN

P

Distance in Miles to German Partner

Figure 8. West German Trading Partners, 1985

you didn’t think that distance matters muchfor international commerce, this figureshould convince you otherwise. There is aremarkably clear log-linear relationshipbetween trade and distance. An estimateddistance elasticity of –0.9 means that eachdoubling of distance reduces trade by 90percent. For example, the distance betweenLos Angeles and Tijuana is about 150 miles.If Tijuana were on the other side of thePacific instead of across the border inMexico and if this distance were increased to10,000 miles, the amount of trade woulddrop by a factor of 44. Other things heldconstant, expect the amount of commercebetween Shanghai and LA to be only about2 percent of the commerce between Tijuanaand LA.

But, you must be imagining, the force ofgravity is getting less, much less. In 1997,Frances Cairncross, a journalist with theEconomist, anticipated Friedman’s TheWorld is Flat by proclaiming in her booktitle The Death of Distance,20 and she fol-lowed that with The Death of Distance 2.0

20 The Death of Distance: How the CommunicationsRevolution Is Changing our Lives, by Frances Cairncross,(2.0 from Harvard Business School).

in 2001, a paperback version with 70 per-cent more material because “In the threeyears since the original Death of Distancewas written, an extraordinary amount haschanged in the world of communicationsand the Internet.”21 The facts suggest oth-erwise. In my own (Leamer 1993a) study ofOECD trade patterns, I report that thisdistance elasticity changed very littlebetween 1970 and 1985 even with the con-siderable reduction in transportation andcommunication costs that were occurringover that fifteen year time period. Disdierand Head (2005) accurately title theirmeta-analysis of the multitude of estimatesof the gravity model that have been madeover the last half-century: “The PuzzlingPersistence of the Distance Effect onInternational Trade.” They find “the esti-mated negative impact of distance on traderose around the middle of the century andhas remained persistently high since then.This result holds even after controlling formany important differences in samples andmethods.”

The distance effect on trade has notdiminished even as transportation costs and

21 http://www.deathofdistance.com/.



communication costs have fallen. “How canthat be right?” shoppers may ask them-selves, “There surely are a lot more Asiangoods on the shelves and racks of stores inthe United States than there were a coupleof decades ago. We have the impressionthat Shanghai is sitting just beyond thebreakwater of the LA harbor.” Keep inmind that the gravity model doesn’t dependonly on distance; it has trade proportionalto the product of the GDP’s. It is the prod-uct of the GDPs that accounts for theincrease in trans-Pacific trade, not a declin-ing effect of distance. Clearly if all the eco-nomic mass of the globe were concentratedon a single location, there wouldn’t be anytrades across space. According to the gravi-ty model, the greatest amount of globaltrade would occur if GDP originated half inone location and half at the opposite side ofthe Earth. That is a pretty good descriptionof the emerging distribution of global GDP.Thus, while it is true that Asian productsare crowding the shelves of stores in LosAngeles, according to the gravity model,that comes from economic growth in Asia,not a decline in the effect of distance. Whatis happening is that growth of theeconomies in Asia is creating tradingopportunities that did not exist before. Theglobe is not shrinking. Economic activity isdispersing.

If you want to see the emerging econom-ic mass in Asia, I recommend WilliamNordhaus’s spinning globe that illustrates bycolor-coding the locations from which glob-al GDP originates: http://www.econ.yale.edu/~nordhaus/homepage/homepage.htm.

This spinning globe dramatically revealsthe very strong clustering of economic activ-ity. Even within the United States there aredistinct clusters of economic activity. Myversion of this clustering phenonenom isconveyed by figure 9,22 which says it matters

22 From Leamer (1997).

23 The measure of distance to markets is Di = (ΣjwjDij

−.6)−1/.6,where wj = GDPj/ΣGDPj.

who your neighbors are. On the horizontalaxis is a measure of distance to global GDPsuggested by the gravity model.23 On thevertical axis is per capita GDP. This figureindicates that distance to markets has a dra-matic effect on GDP per capita. The fall-offin income levels when distance goes from1000 to 2000 miles is very great. After 3000miles there is hardly a country with a decentper capita GDP.

In 1960, only Australia and New Zealandwere able to escape the curse of being faraway. Maybe these English-speakingCommonwealth countries are closer toBritain and the United States than geographysuggests. By 1990, two other countries hadescaped the force of gravity, namelySingapore and Taiwan. Part of the explana-tion for their ability to escape gravity may begravity itself. There is scarcity value in beingdifferent and the gravity model suggests thatscarcity needs to be measured locally. Forthat reason, capital accumulation has a biggerimpact if your neighbors are capital-scarce.Indeed, studies of Asian growth by AlwynYoung (1995) and Jong-Il Kim and LawrenceJ. Lau (1996) both argue that it was factoraccumulation not growth in total-factor-pro-ductivity that accounts for the Asian miracles.These countries have had relatively rapidcapital accumulation and also relatively largeelasticity of output with respect to capital, theformer partly induced by the latter.

Though we know that geography has had alarge and persistent effect on trade andprosperity, we don’t understand why. I aminclined to think that there are two possibleexplanations. One possibility is that theglobe is not getting smaller at all. The con-tracts that are necessary to exchange mostgoods require trust and understanding, butthere have been hardly any improvements inthe technologies for creating trust andunderstanding over long distances. We are



Taiwan

New Zealand

0

5000

10000

15000

20000

0 1000 2000 3000 4000 5000 6000 7000 8000

Access to Markets and Per Capita Income: 19 90

Distance to GDP, 1990

Per C

apita

Inc

ome,

199

0

Australia

Singapore

Access to Markets and Per Capita Income: 1960

0

5000

10000

15000

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Distance to GDP, 1960

Per C

apita

Inc

ome,

196

0

Australia, NewZealandVenezuela

TrinidadUruguay

Mauritius

Figure 9. Geographical Clustering of High Income Countries

still animals and, like dogs that don’t recog-nize the image of another dog on TV, we can-not feel fear and doubt or their opposites,trust and understanding, unless we are in thesame physical space.

Speaking of understanding and distance, ina remarkable paper, Blum and Avi Goldfarb(2006) find that gravity applies even to the

Internet—U.S. surfers favor foreign websitesclose to the United States.

For websites that do not involve a financialtransaction, the distance effect is smaller thanthe average of the studies in Disdier andHead (2005), with the distance elasticity equalto 0.9 percent. For websites that involve afinancial transaction, the distance elasticity is



24 My colleague, Sebastian Edwards, wisely made thispoint to me.

25 For more on the persistence of the distance effect,see Disdier and Head (2005). See also David Hummels(2001) and Alan V. Deardorff (2003) for discussion of therole of timeliness in trade patterns.

1.8 percent, therefore larger than the averageeffect in the studies cited above. For taste-dependent websites, the distance elasticity isequal to 3.25 percent (p. 386).

Blum and Goldfarb interpret this as a tasteeffect, with consumers clustered togetherwith similar tastes and with far-away suppli-ers not really understanding what distantcustomers desire. “North American music,for example, is very different from Indianmusic. Similar stories apply to games andpornography” (p. 385).

The other possibility why the distanceeffect has been so persistent is that,although the earth has been getting small-er, that doesn’t change much the geographyof commerce. In the relentless search forthe least-cost location, it doesn’t matter if Ais 50 percent better than B, or 5 percent, or0.5 percent. Still A is the preferred loca-tion. Thus, as the cost of doing businessover long distances diminishes, the goodsare delivered cheaper and faster, but fromthe same sources.24 Here we again need tomake clear the important distinctionbetween movement and mobility. Withfalling transportation costs, the mobilitychanges but the movement may not. Thegravity model measures movement, notmobility. For wages, it is mobility (con-testability) that matters. As the globeshrinks, wages for footloose work can con-verge because of contestability, eventhough the geography of commerce staysthe same.25

My bottom line: There are many advan-tages that children can enter this world with,including intelligence, physical power andagility, good looks, and caring parents. It alsomatters where you live.

26 “At least five definitional and methodological fac-tors contribute to the difference between U.S. and Indiandata on BPT services. First, India and the United Statesfollow different practices in accounting for the earningsof temporary Indian workers residing in the UnitedStates. Second, India defines certain services, such assoftware embedded on computer hardware, differentlythan the United States. Third, India and the UnitedStates follow different practices for counting sales byIndia to U.S.-owned firms located outside of the UnitedStates. The United States follows International MonetaryFund standards for each of these factors. Fourth, BEAdoes not report country-specific data for particular typesof services due to concerns about the quality of respons-es it receives from firms when they allocate their affiliat-ed imports to detailed types of services. As a result, U.S.data on BPT services include only unaffiliated importsfrom India, while Indian data include both affiliated andunaffiliated exports. Fifth, other differences, such asidentifying all services importers, may also contribute tothe data gap.”

7.4 How Important Is “Outsourcing,”Really?

In a recent class of sixty Executive MBAs, Iasked who personally felt the force of compe-tition for their job from India or China. Onehand was raised. Then I asked whose compa-nies were interested or engaged in movingjobs to India or China. Two-thirds of thehands were raised. I wonder if they knowmore about their own jobs than about theircoworkers.

Though there is a great deal of fuss inthe media about the movement of U.S.service jobs to India, the number of U.S.workers affected by outsourcing surelyremains low.

In response to queries from Congressabout the amount of outsourcing that theUnited States has been experiencing, theGAO (2004) issued a first report with thenot-very-promising title “Current Govern-ment Data Provide Limited Insight intoOffshoring of Services” and followed thatone up a year later, GAO (2005) with “U.S.and India Data on Offshoring ShowSignificant Differences.”

It may not be perfectly accurate, but hereis what the GAO (2005, executive summary)has to say (BPT = business, professional, andtechnical services):



The gap between U.S. and Indian data ontrade in BPT services is significant. For exam-ple, data show that for 2003, the United Statesreported $420 million in unaffiliated importsof BPT services from India, while Indiareported approximately $8.7 billion in exportsof affiliated and unaffiliated BPT services tothe United States. 26

Compared with an $11 trillion economy,those numbers are small potatoes, theIndian estimate being less than 0.1 percentof GDP, less than the GDP measurementerror by a wide margin. Dividing those twoestimates by say, $100,000 in revenue perjob, that translates into a low of 4,200 jobsto a high of 87,000. That compares with theU.S. economy that increases payroll jobson average by almost 200,000 jobs permonth.

No big problems here that I can see, thoughthere is a difference between movement andmobility.

7.5 It’s Mostly the Balance of Productivityand Demand That KeepsManufacturing Jobs from Growing

If not much in intellectual services,maybe we can find a lot of jobs lost to off-shoring of manufacturing work. Maybe,but not actually. The long-run story inmanufacturing is growth in domesticdemand that is more-or-less matched bygrowth in productivity, leaving employ-ment levels pretty constant. The increasein the number of jobs in manufacturing inthe first seven decades of the twentiethcentury came mostly from domesticdemand growing more rapidly than pro-ductivity. My hunch is that this has to dowith product innovation, which wasstronger in manufacturing prior to the1970s.

Figure 10 illustrates the employmentlevel in durable manufacturing since the1960s, and the corresponding trade deficitin durables. Figure 11 has the same for non-durables. There doesn’t seem to be muchrelationship between the deficits and the

jobs until after 2000, but dividing thosemost recent deficits by the average compen-sation per job gives an estimate of 1.5 mil-lions jobs lost in nondurables and 2 millionin nondurables. These are large, evocativenumbers.

But before we jump to trade as the driv-er, better lay out all the possibilities. Itcould be that a burst in productivity isallowing the few to do the work of themany. And it could be that domesticdemand is too weak to allow normal jobformation. The level of employment is nec-essarily equal to domestic demand timesthe ratio of production to domesticdemand divided by productivity (outputper worker):

Workers = Domestic Demand

GDP 1× ⎯⎯⎯⎯⎯⎯⎯⎯ × ⎯⎯⎯⎯⎯⎯ ,Domestic Demand GDP/Worker

where domestic demand is equal toGDP + Imports − Exports and the ratio ofGDP to domestic demand measures thefraction of demand satisfied from local sup-ply, which gets smaller as the externaldeficit widens. This is only an accountingidentity and we need to be a little careful indrawing causal conclusions from these num-bers, particularly because some of the pro-ductivity changes are induced by foreigncompetition.

The trends since 1970 in employment andthese three components of the employmentidentity are reported in table 2. In durables,domestic demand has been growing smartlyat the rate of 5.6 percent per year but theeffect of this strong demand growth onemployment has been largely offset by animprovement in productivity at the rate of5.5 percent per year. Thus demand growthnet of productivity improvement yields apotential growth in employment in durablesof 0.1 percent per year. But from that num-ber we need to subtract 0.5 percent toaccount for the trend in the sourcing of sup-ply from foreign locations, leaving a trend



−300

−200

−100

0

5.0

5.5

6.0

6.5

7.0

7.5

1970 1975 1980 1985 1990 1995 2000 2005

Bill

ions

of

$20

00M

illions of

Workers

1996 Q 1

177 Billion $Increase in Deficit1.5 million jobs

$118,000 per job

Employment: Right Scale

Nondurables Deficit: Left Scale

Figure 10. Employment in Nondurables and the External Deficit in Nondurables

0

100

8

9

10

11

12

13

1970 1975 1980 1985 1990 1995 2000 2005

Bill

ions

of $

2000

Millions of W

orkers

1998 Q1Employment: Right Scale

Durables Trade Deficit: Left Scale

$377 billion increase in Deficit2.0 million jobs

$188,000 per job

−100

−200

−300

−400

−500

Figure 11. Employment in Durables and the External Deficit in Durables

downward of employment in durables equalto –0.4 percent per year. One might betempted from this to conclude that, but forthe trade deficit, there would have beensome slight employment growth in durableswhile trade pulls that number into negativeterritory. But the larger story is strongdomestic demand growth offset by strongproductivity growth.

The story of nondurables is rather differ-ent with weaker domestic demand growth(2.6 percent) and with productivity growth(3.0 percent) that outstrips domesticdemand, leaving employment growth at –0.4percent per year, absent any globalizationeffect. The globalization effect subtractsanother 0.1 percent from that number, lead-ing to employment in nondurables declining



TABLE 2DECOMPOSITION OF EMPLOYMENT TRENDS

Trends: 1970 to 2005Durables Nondurables

Employment Growth Rate –0.4% –0.5%Consisting of

Growth in Domestic Demand 5.6% 2.6%(-) Growth in Productivity –5.5% –3.0%

+ Domestic Sourcing(GDP/Domestic Demand) –0.5% –0.1%

at the trend rate of –0.5 percent per year,about the same as durables.27

The bottom line: employment trends inmanufacturing are affected a little bit byinternational trade, but the real story isstrong productivity growth pretty muchkeeping up with the growth in domesticdemand.

Whooaa. Hold on a minute. I am mis-leading you. It is these kinds of quantity cal-culations that have led most economistsincluding Krugman and Lawrence (1993)and Lawrence F. Katz and Kevin M.

27 A continuation of these long-term trends impliesjob losses in manufacturing at roughly –0.5 percent peryear, but since the recession of 2001 we have lost fully 18percent of payroll jobs. Whence comes that? The tablebelow reports the change in the deviation from trendfrom 2000 Q1 to 2005 Q2 for employment in durablesand nondurables. This table indicates that, since 2000,employment in durables has deteriorated by 17.2 per-cent relative to trend. That decline in employment iscomposed of 7.3 percent loss of work from weakness indemand, 6.8 percent from unusually strong productivitygrowth, and 3.1 percent because of the rising tradedeficit. For nondurables, the big news is in the produc-tivity number. Job loss in nondurables of 16.3 percentrelative to trend is explainable by a surge in productivity,17.6 percent relative to trend. Some of the burst in pro-ductivity is a response to greater foreign competition andsome of the burst is a ripple effect of information tech-nology investments made in the Internet Rush.

Decomposition of Job Losses since 20002000–2005 Change in Deviation from Trend

Durables NondurablesEmployment –17.2% –16.3%Consisting ofDomestic Demand –7.3% 2.6%Productivity –6.8% –17.6%Trade (GDP/Domestic Demand) –3.1% –1.3%

Murphy (1992) to the conclusion that inter-national trade is not much affecting U.S.workers. These folks are my version ofSamuelson’s John and Jane Doe. They areignoring the difference between movementand mobility. They are ignoring the possibil-ity that manufacturing jobs may be contest-ed by low-wage foreign workers but still stayright here in the United States. That globalcontest may be met by a deterioration ofwages and working conditions in the UnitedStates, not a movement of jobs to low-wagecountries. Leamer (1998) and Lawrenceand Matthew J. Slaughter (1993) and othershave conducted empirical studies of theimpact of international trade on U.S. wagesbased on the extreme opposite assumptionthat contestability is perfect. In these stud-ies, it is assumed that it doesn’t matter if theUnited States is running a surplus in appar-el, a deficit, or has balanced trade. Whatmatters is whether the United States iscompeting in the same apparel productmarkets as the Chinese. Then U.S. apparelwages have to be set to the Chinese level,adjusted for productivity differences.

The real bottom line: we do not know thebreadth and intensity of global contestabili-ty of U.S. jobs and, until we do, we will nothave a real handle on the impact of globalcompetition on the U.S. workforce.

7.6 The United States has ExtraordinaryAdvantages in the Use of the Internet

There is one more thing that is really notflat: the Internet. The United States is the



TABLE 3RESIDENCES OF INTERNET HOSTS

Fraction ofInternet Hosts Global Fraction of

Country Population Internet Hosts Per Capita Population Global Hosts

1 United States 295,734,134 115,311,958 0.390 4.58% 67.27%2 Japan 127,417,244 12,962,065 0.102 1.98% 7.56%3 Netherlands 16,407,491 4,518,226 0.275 0.25% 2.64%4 United Kingdom 60,441,457 3,398,708 0.056 0.94% 1.98%5 Canada 32,805,041 3,210,081 0.098 0.51% 1.87%6 Brazil 186,112,794 3,163,349 0.017 2.88% 1.85%7 Australia 20,090,437 2,847,763 0.142 0.31% 1.66%8 Taiwan 22,894,384 2,777,085 0.121 0.35% 1.62%9 Germany 82,431,390 2,686,119 0.033 1.28% 1.57%

10 France 60,656,178 2,396,761 0.040 0.94% 1.40%

Source: http://www.cia.gov/cia/publications/factbook/rankorder/2184rank.html

TABLE 4RESIDENCES OF INTERNET USERS

Internet UsersPercent of Own Percent of Global Percent of Global

Country Population Population Internet Users

1 United States 53.76% 4.58% 22.59%2 China 7.20% 20.25% 13.36%3 Japan 44.89% 1.98% 8.13%4 Germany 47.31% 1.28% 5.54%5 Korea, South 60.07% 0.75% 4.15%6 United Kingdom 41.36% 0.94% 3.55%7 France 36.11% 0.94% 3.11%8 Italy 31.84% 0.90% 2.63%9 India 1.71% 16.75% 2.63%

10 Canada 49.11% 0.51% 2.29%

primary home of the Internet and, in manyways, is the center of the New Economy.Fully 67 percent of Internet hosts reside inthe United States (table 3), and 23 percentof Internet users (table 4) compared with apopulation fraction of 4.6 percent. Ofcourse, China and India, because of theirhuge populations, show up in the list of thetop ten homes of Internet users, but myadvice nonetheless is: bet on the UnitedStates.

8. The Coming of the Postindustrial Age

Finally, I want to comment on what I thinkis the big issue. It isn’t globalization or a flatworld; it’s technology and the postindustriallabor markets.

The United States is in the midst of a rad-ical transformation from industrial to post-industrial society. Some of this transition isassociated with the movement of mundanemanufacturing jobs to low-wage foreign



0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

Agriculture

Manufacturing

Agriculture

Manufacturing

Figure 12. U.S. Employment Shares in Agriculture and Manufacturing

locations, but much of it comes from thedramatic changes in technology in theintellectual services sectors. The policyresponse to the globalization force is prettystraightforward: we need to make the edu-cational and infrastructure investmentsthat are needed to keep the high-payingnoncontestable creative jobs here at homeand let the rest of the world knock them-selves silly competing for the footloosemundane contestable jobs. The response tothe technological trends that are alteringthe nature of the relationship-based jobs isnot so clear cut.

The U.S. transition from an agrarian toan industrial economy that began in theeighteenth century was put on hold duringthe Great Depression but accelerated dur-ing both World War I and World War II.Excluding the war years of 1942–45, thetransition to an industrial society, illustratedin figure 12, reached its zenith in the 1950swith 30 percent of our workforce in manu-facturing and 10 percent in agriculture. Thehigh-growth Kennedy/Johnson expansion ofthe 1960s kept the jobs in manufacturing at28 percent, but the transition to a postindus-trial society began in earnest in the 1970s.While jobs in agriculture continued todecline throughout the century, dropping

now to only about 1 percent of our work-force, there has also been a sharp drop inemployment in manufacturing in the lastthree decades, falling in the most recent data(2005) to only 11 percent of our workforce.The speed of this decline after 1970 from a28 percent share to an 11 percent share inmanufacturing is every bit as rapid as thespeed in the decline of agricultural jobs inthe first seven decades of the twentieth cen-tury.

This transition to the postindustrial agehas consequences that are at least as pro-found as the transition from agriculture toindustry. This will alter the way wealth is cre-ated and all that flows from the “means ofproduction,” including politics and socialstructures.

8.1 Marx and The Transition fromAgriculture to Industry

Studies of the transition from agrarian ageto industrial age hint at what the next tran-sition might entail. Nathan Rosenberg(1983) offers a cogent view of technologyand production in the industrial age:

Although, therefore, the manufacturing sys-tem achieved a growth in productivity throughthe exploitation of a new and more extensivedivision of labor, a rigid ceiling to the growth in



productivity continued to be imposed by limi-tations of human strength, speed and accuracy.Marx’s point, indeed, is more general: Scienceitself can never be extensively applied to theproductive process so long as that process con-tinues to be dependent upon forces the behav-ior of which cannot be predicted andcontrolled with the strictest accuracy. Science,in other words, must incorporate its principlesin impersonal machinery. Such machinery maybe relied upon to behave in accordance withscientifically established physical relationships.Science, however, cannot be incorporated intotechnologies dominated by large-scale humaninterventions, for human action involves toomuch that is subjective and capricious. Moregenerally, human beings have wills of their ownand are therefore too refractory to constitutereliable, that is, controllable inputs in complexand interdependent productive processes (p.42, my italics).Relics of by-gone instruments of labor possessthe same importance for the investigation ofextinct economical forms of society, as do fos-sil bones for the determination of extinctspecies of animals. It is not the articles made,but how they are made, and by what instru-ments, that enables us to distinguish differenteconomical epochs (Karl Marx, Capital, quot-ed by Rosenberg, p. 40).

8.2 Not All Tasks Can Be Embodied inEquipment

Thus, per Marx, we are what we operate,and what was essential about the industrialage is not what we produced but how weproduced it. During the industrial age, sci-ence and industry collaborated to embody inequipment those tasks that are repetitive,codifiable, and programmable, thus freeingthe productive process from the caprice ofhuman intervention. Mechanization of workwas not limited to manufacturing andoccurred also on the farm. But mechaniza-tion of services was much more limited.Getting a haircut in 2005 is not much differ-ent from getting a haircut in 1850. And hav-ing a will drawn up in 1970 was about thesame as having a will drawn up in 1900.

The mundane physical tasks that havebeen left to humans require a degree of dex-terity that is difficult (expensive) to achieve

with a machine, but year after year advancesin science transfer more and more of thesefunctions to machines. Meanwhile, the eco-nomic liberalizations over the last threedecades have added to the global workforcean enormous number of workers in Mexicoand Brazil and China and India and so on,offering to do the mundane physical tasks atrates of pay that are barely subsistent. Thusglobalization and technology have ganged upafter 1970 to rapidly reduce the demand formundane physical labor in the United States.

Most of the innovations of the Industrialage have made very little encroachment onintellectual tasks, mundane or otherwise. Anattorney, an architect, a teacher all did aboutthe same work in 1970 as they did in 1800.Absent innovations in production and com-munication, one might imagine a globalizedpostindustrial United States in which mun-dane physical tasks like cutting hair wouldremain only in the local nontraded sector andthe rest of the jobs would be mixtures of mun-dane-intellectual tasks (clerks), creative-intel-lectual tasks (designers and researchers andrepairmen), and social/organizing/motivatingtasks (managers).

But the microprocessor has changed thefuture of intellectual work, eliminating themundane-intellectual tasks. Think about anarchitect. In 1970, the time of a creativearchitect was partly consumed by the task ofrendering the drawings. Some of this workcould be done by assistants but the commu-nication costs were often so high that itmade more sense to have the master do thedrawings. The personal computer, however,allowed the architect to render the drawingswith great efficiency, thus freeing up time todo the creative tasks that the computer can-not ever perform. While, for mundane pro-grammable tasks, it is true that “humanbeings have wills of their own and are there-fore too refractory to constitute reliable, thatis, controllable inputs in complex and inter-dependent productive processes,” the oppo-site is true for creative tasks. It is machinesthat lack wills of their own and are therefore



too obedient to constitute reliable, that is,innovative inputs in complex and interde-pendent creative processes. Indeed, when Iteach data analysis I emphasize the constantstruggle between machine and man for con-trol of the process. We data analysts reallywant to be able to press a button and havethe computer do the work but the creativetask of drawing inferences from data alwaysrequires a heavy human input and, ifthrough laziness and seduction, we come toimagine that the computer can think, we willsurely be making major misinterpretationsof the data. When one starts to lose controland not know if one button on the computeris any different from another, it is wise toshut the computer down and go play a roundof golf. The human will be better able tomaintain control after a little time off.

8.2.1 Is a Computer a Forklift or aMicrophone?

Education may be a solution to the tem-porary and permanent income inequalityproblems caused by the increased supply ofMicroprocessors. We just need to teacheveryone how to write computer code. Thismight work, but it might not. I like to raisesome doubts by posing the rhetorical ques-tion; “Is a computer more like a forklift ormore like a microphone?” It doesn’t mattermuch who drives the forklift, but it matters alot who sings into the microphone. Thinkabout the forklift first. You might be a lotstronger than I, but with a little bit of train-ing, I can operate a forklift and lift just asmuch as you or any other forklift operator.Thus the forklift is a force for income equal-ity, eliminating your strength advantage overme. That is decidedly not the case for amicrophone. We cannot all operate a micro-phone with anywhere near the same level ofproficiency. Indeed, I venture the guess thatI would have to pay you to listen to me sing,not the other way round. And I seriouslydoubt that a lifetime of training would allowme to compete with Springsteen orPavarotti.

The effect of the microphone and massmedia have been to allow a single talentedentertainer to serve a huge customer baseand accordingly to command enormous earn-ings. This creates an earnings distributionwith a few extremely highly paid talented andtrained individuals and with the vast group ofslightly less talented working in LA restau-rants, hoping someday to hit it big. Thus,opposite to the forklift, the microphone cre-ates a powerful force for inequality. ThinkSilicon Valley, with extraordinary richesaccruing to some but with the service work-ers living in their cars because they cannotafford the homes.

A computer is both a forklift and a micro-phone. Clerks in McDonalds no longer haveto be able to read or to compute—they onlyhave to recognize the picture of a hamburgeron the cash register. That’s the forklift. Itdoesn’t much matter who punches the but-tons. Thus your intelligence advantage overme is eliminated by the computer, just asyour strength advantage was eliminated bythe forklift. But for many other operations itmatters enormously who types on the com-puter. One example is computer program-ming. The vast majority of people areincapable of producing commercially viablecomputer code. That’s the microphone. Itamplifies your natural advantages. Without acomputer, an architect’s time is partly con-sumed by mundane tasks such as renderingdrawings. A lawyer’s time is consumed writ-ing and checking sentences in wills. An econ-omist’s time is consumed making datadisplays. These mundane tasks are now trans-ferred to computer assistants, who listen infi-nitely more attentively and who carry out thetasks with much greater precision than anyhuman assistant. A talented architect with acomputer assistant can serve a muchenlarged customer base. A talented attorney,or a talented economist, or a talented radiol-ogist with computer assistants can servemuch enlarged customer bases. These talent-ed individuals command high wages whilethe less talented struggle for customers.



Computer technology seems therefore tobe taking us into a future where there are afew very talented, very well-paid peopleand the rest of us are doing the mundanecomputer-assisted tasks which don’trequire us to read, write, or even think verymuch. Just push the right button now andthen.

In other words, the information revolutionmay be a powerful force for income inequal-ity by raising the compensation for naturaltalents and also the interaction between tal-ent and training. It is the interactionbetween talent and training that is particu-larly difficult to deal with. If talent and train-ing had additive effects on earnings, thencompensatory education for the disadvan-taged could be a low-cost solution forincome inequality problems. But if trainingis much more effective for the talented, thetalented will naturally receive more of it andthe amount of compensatory training that isneeded to equalize incomes may be enor-mous and a great social waste—think of meand Pavarotti again.

9. Conclusion

Here are titles of four books. Based on thetitles alone, which do you think would sellthe most copies?

The World is FlatThe End of PovertyIn Defense of GlobalizationGlobalization and Its Discontents

The startling reality is that The World isFlat has been on the New York Times best-seller list forever and is ranked number 1 onAmazon on February 21, 2006. Meanwhile,Jeffrey Sachs’s The End of Poverty:Economic Possibilities for Our Time isranked number 515, which seems like a bignumber compared with the number 1, butBhagwati’s In Defense of Globalization isranked 20,602, and Joseph Stiglitz’sGlobalization and Its Discontents is ranked52,196.

We economists have great ideas but notgreat ways of expressing ourselves. It startswith bad titles. This raises the philosophicalquestion: When economists speak, but noone listens, did we say anything?

Thus I understand that it doesn’t muchreally matter what I think. The market hassaid that The World is Flat is a great book.But just for my own personal amusement,here comes the summary.

First of all, metaphorical titles as power-ful as The World is Flat really matter. Withthat title, readers have their fears rein-forced, needlessly. The debate about how tohandle our economic relations with othercountries has already been harmed enoughby misleading metaphors. Those who favorhigh tariffs call it “protection” as if a wolfwere lurking beyond our borders ready todevour our jobs. And those who favor lowtariffs call it “free” trade, as if paying a cou-ple of more dollars for the shirts and jeanswe buy at Wal-Mart amounted to a period ofincarceration.

Other than the metaphorical mischief,this book is long on anecdotes, interpreta-tions, and insight. It’s an eye-openermethodologically because of the clearprogress Friedman makes without benefitof the union card we call the Ph.D. inEconomics. But he doesn’t get “there”because, I think, he has little knowledge ofthe vast amount of work that has been doneby economists on these topics.

Friedman does get some of the policyresponse right:

And it requires a Great Society that commitsour government to building the infrastructure,safety nets, and institutions that will help everyAmerican become more employable in an agewhen no one can be guaranteed lifetimeemployment (p. 277).

My vision is to put every American man orwoman on a campus (p. 290).

But when Friedman calls this program“compassionate flatism” that is the flat thatbroke this camel’s back. Worse still, thisnonsense metaphor is becoming altogether



28 One obvious thing to do is to remove the burden ofhealth care from the labor contract by providing a mini-mal level of universal care. Those health care benefits thatare still provided by some firms are hard on globally con-testable jobs since foreign employers don’t pay them, andthey make employers wary of forming long-term relation-ships with prospective employees as they look into thefuture and see a mountain of health care costs. Anotherobvious role for policy is aggressive intellectual propertyprotection. We cannot have relationships betweenemployers and employees based on the knowledge assetsthey create if those knowledge assets can be stolen.

commonplace, filtering into classrooms andboardrooms. We should take better care ofour language.

On the policy front, Friedman misses thedistinction between markets and relation-ships, and thus misses the potential for poli-cy measures that might facilitate theformation of long-term relationshipsbetween workers and employers. We econo-mists do a linguistic disservice when we calla relationship-free, frictionless outcome“perfectly competitive.” The Luddites had itright when they complained that there isnothing perfect about that outcome at all.Frictions are our friends. Frictions give usthe peace of mind that we will still have ourjobs when we wake up tomorrow. Frictionsreduce the chances that one party will try to“hold-up” the other, absconding with thelion’s share of the mutual benefits. Frictionsthus give us the confidence to make the rela-tionship-specific investments from whichgreat returns can flow. It’s the friction we call“falling in love” that allows the humanspecies to flourish. I am not endorsing theFrench solution of permanent jobs for all,which creates a forced one-sided “marriage”between a willing worker and an unwillingemployer without even the benefit of a first“date.” That makes French employers reluc-tant to marry workers and leaves the Frenchunemployment level unnaturally high. Butwe should be thinking long and hard abouthow we can make our “single” workers more“lovable” and what we can do to maintain the“marriages” between employers andemployees that are working.28

29 Frank Levy reminded me to include this explicitly.He might also want me to leave no doubt that, unlike ourTreasury Secretary John Snow who has defended CEOpay as “market-determined,” I think there is a differencebetween Pavarotti whose talent is demonstrable andwhose pay is commensurate, versus the super-high-paidCEO whose talent is unclear and whose negotiated com-pensation bears little apparent relationship to the socialproduct of the activity.

Friedman also misses the point that edu-cation works well as a treatment for theincome inequality that comes fromincreased global competition, but not so-well for the talent-driven income inequalityof the postindustrial age.29 Beyond the factthat no amount of education will allow me tosing like Pavarotti, the well-paid tasks of thepostindustrial era involve subtle job-specifictacit knowledge that is impossible to teachwith the lecture/library work/exam style ofuniversity classes, which works best to trans-mit codifiable knowledge and rule-baseddecisionmaking. That’s training, not educa-tion. For tacit knowledge, it’s not that expe-rience is the best teacher. It’s the onlyteacher. That, by the way, brings us back tothe need for frictions. Who would haveinvested in twenty years of formal educationand another ten years of on-the-job trainingto become an economist if they thought thejob might disappear when they had finallymastered it?

The final third of Friedman’s book is aninsightful and lucid discussion of the stresspoints between the Muslim and theJudeo–Christian world. (You are free toobject that I am not competent to reviewthat material.)

But: Physically, culturally, and economical-ly the world is not flat. Never has been, neverwill be. There may be vast flat plains inhabit-ed by indistinguishable hoi polloi doing mun-dane tasks, but there will also be hills andmountains from which the favored will lookdown on the masses. Our most important giftsto our offspring are firm footholds on thosehills and mountains, far from the flat part ofthe competitive landscape. Living in the



United States helps a lot and will continueto. But those footholds will increasinglyrequire natural talent. As a by-product ofour search for personal pleasure, we pro-vide our children highly loaded dice to rollat the genetic craps table. Beyond the all-important luck of the genetic draw, it takesthe kind of education that releases ratherthan constrains their natural talent. Wesend our children to good private schoolsand then on to UCLA. The rest is up tothem.

I am sorry to say it that flat way. It’s not anapt metaphor, even though it is a powerfulone.

“When I use a word,” Humpty Dumpty said inrather a scornful tone, “it means just what Ichoose it to mean—neither more nor less.”

“The question is,” said Alice, “whether youCAN make words mean so many differentthings.”

“The question is,” said Humpty Dumpty,“which is to be master—that’s all” (LewisCarroll, Through the Looking Glass, Chapter 6).

REFERENCES

Anderson, James E., and Eric van Wincoop. 2004.“Trade Costs.” Journal of Economic Literature,42(3): 691–751.

Autor, David H., Lawrence F. Katz, and Melissa S.Kearney. 2006. “The Polarization of the U.S. LaborMarket.” NBER Working Papers, no. 11986.

Bhagwati, Jagdish. 1958. “Immiserizing Growth: AGeometric Note.” Review of Economic Studies,25(3): 201–05.

Bhagwati, Jagdish. 2004. In Defense of Globalization.Oxford: Oxford University Press.

Bhagwati, Jagdish, and Marvin H. Kosters, eds. 1994.Trade and Wages: Leveling Wages Down?Washington, D.C.: AEI Press.

Bhagwati, Jagdish. 2005. “A New Vocabulary forTrade.” The Wall Street Journal, 4 August.http://yaleglobal.yale.edu/display.article?id=6109.

Blum, Bernardo S., and Avi Goldfarb. 2006. “Does theInternet Defy the Law of Gravity?” Journal ofInternational Economics, 70(2): 384–405.

Blum, Bernardo S., and Edward E. Leamer. 2004.“Can an FTAA Suspend the Law of Gravity and Givethe Americas Higher Growth and Better IncomeDistributions?” In Integrating the Americas: FTAAand Beyond, ed. A. Estevadeordal, D. Rodrik, A.Taylor, and A. Velasco. Cambridge: HarvardUniversity David Rockefeller Center for Latin

American Studies; distributed by Harvard UniversityPress, 539–72.

Borjas, George J., and Valerie A. Ramey. 1995.“Foreign Competition, Market Power, and WageInequality.” Quarterly Journal of Economics, 110(4):1075–110.

Cairncross, Frances. 1997. The Death of Distance: Howthe Communications Revolution Is Changing OurLives. London: Orion Business Books.

Davis, Donald R., and David E. Weinstein. 1996.“Does Economic Geography Matter forInternational Specialization?” NBER WorkingPapers, no. 5706.

Deardorff, Alan V. 2003. “Time and Trade: The Role ofTime in Determining the Structure and Effects ofInternational Trade, with an Application to Japan.”In Analytical Studies in U.S.–Japan InternationalEconomic Relations, ed. R. M. Stern. Cheltenham,U.K. and Northampton, Mass.: Elgar.

Diamond, Jared. 1997. Guns, Germs, and Steel: TheFates of Human Societies. New York: W. W. Norton& Company.

Disdier, Anne-Célia and Keith Head. 2005. “ThePuzzling Persistence of the Distance Effect onBilateral Trade.” Working Paper.

Easterly, William. 2001. The Elusive Quest for Growth:Economists’ Adventures and Misadventures in theTropics. Cambridge and London: MIT Press.

Frankel, Jeffrey, Ernesto Stein, and Shang-jin Wei.1995. “Trading Blocs and the Americas: The Natural,the Unnatural, and the Super-Natural.” Journal ofDevelopment Economics, 47(1): 61–95.

Freund, Caroline L., and Diana Weinhold. 2004. “TheEffect of the Internet on International Trade.”Journal of International Economics, 62(1): 171–89.

GAO Report to Congressional Committees. 2004.“Current Government Data Provide Limited Insightinto Offshoring of Services.” Washington, D.C.:Government Accountability Office.

GAO Report to Congressional Committees. 2005.“U.S. and India Data on Offshoring Show SignificantDifferences.” Washington, D.C.: GovernmentAccountability Office.

Goolsbee, Austan. 2000. “In a World without Borders:The Impact of Taxes on Internet Commerce.”Quarterly Journal of Economics, 115(2): 561–76.

Harrigan, James. 2004. “Specialization and theVolumes of Trade: Do the Data Obey the Laws? InHandbook of International Trade, Volume 1, ed. K.Choi and J. Harrigan. London: Basil Blackwell,85–118.

Heckscher, E. F. 1919. “The Effect of Foreign Tradeon the Distribution of Income.” Ekonomisk Tidskrift,21: 497–512.

Hotelling, Harold. 1929. “Stability in Competition.”Economic Journal, 39(153): 41–57.

Hufbauer, Gary Clyde. 2001. “Globalization Facts andConsequences.” Institute for International Economics.

Hummels, David. 2001. “Time as a Trade Barrier.”Unpublished.

Jones, Ronald W. 1967. “International CapitalMovements and the Theory of Tariffs and Trade.”Quarterly Journal of Economics, 81(1): 1–38.

Katz, Lawrence F., and Kevin M. Murphy. 1992.



“Changes in Relative Wages, 1963–1987: Supply andDemand Factors.” Quarterly Journal of Economics,107(1): 35–78.

Kemp, Murray C. 1966. “The Gain from InternationalTrade and Investment: A Neo-Heckscher–OhlinApproach.” American Economic Review, 56(4):788–809.

Kim, Jong-Il, and Lawrence J. Lau. 1996. “The Sourcesof Asian Pacific Economic Growth.” CanadianJournal of Economics, 29: S448–54.

Krugman, Paul. 1991. “Increasing Returns andEconomic Geography.” Journal of PoliticalEconomy, 99(3): 483–99.

Krugman, Paul. 1995. “Technology, Trade, and FactorPrices.” NBER Working Papers, no. 5355.

Krugman, Paul, and Robert Z. Lawrence. 1993. “Trade,Jobs, and Wages.” NBER Working Papers, no. 4478.

Lawrence, Robert Z., and Matthew J. Slaughter. 1993.“International Trade and American Wages in the1980s: Giant Sucking Sound or Small Hiccup?”Brookings Papers on Economic Activity, 1: 161–210.

Leamer, Edward E. 1968. “Location Equilibria.”Journal of Regional Science, 8(2): 229–42.

Leamer, Edward E. 1993a. “U.S. Manufacturing andan Emerging Mexico.” North American Journal ofEconomics and Finance, 4(1): 51–89.

Leamer, Edward E. 1993b. “Wage Effects of aU.S.–Mexican Free Trade Agreement.” In TheMexico–U.S. Free Trade Agreement, ed. P. M.Garber. Cambridge and London: MIT Press,57–125.

Leamer, Edward E. 1996. “Questions, Theory andData.” In Foundations of Research in Economics:How Do Economists Do Economics?, ed. S. G.Medema and W. J. Samuels. Advances in EconomicMethodology series. Cheltenham, U.K. and Lyme,N.H.: Elgar, 175–90.

Leamer, Edward E. 1997. “Access to Western Markets,and Eastern Effort.” In Lessons from the EconomicTransition: Central and Eastern Europe in the 1990s,ed. S. Zecchini. Dordrecht: Kluwer AcademicPublishers, 503–26.

Leamer, Edward E. 1998. “In Search ofStolper–Samuelson Linkages between InternationalTrade and Lower Wages.” In Imports, Exports, andthe American Worker, ed. Susan Collins.Washington, D.C.: Brookings Institution Press,141–203.

Leamer, Edward E. 2004. “Immiserizing Outsourcingand Trickle-Up Outsourcing Managing U.S.Knowledge Assets.” Unpublished.

Leamer, Edward E., and James Levinsohn. 1995.“International Trade Theory: The Evidence.” InHandbook of International Economics, Volume 3, ed.G. M. Grossman and K. Rogoff. Handbooks inEconomics, vol. 3. Amsterdam; New York andOxford: Elsevier, North-Holland, 1339–94.

Leamer, Edward E., and Peter Schott. 2005. “The Rich(and Poor) Keep Getting Richer.” Harvard BusinessReview, 83(4): 20.

Leamer, Edward E., and Michael Storper. 2001. “TheEconomic Geography of the Internet Age.” Journalof International Business Studies, 32(4): 641–65.

Levy, Frank, and Richard J. Murnane. 2004. The New

Division of Labor: How Computers Are Creating theNext Job Market. Princeton: Princeton UniversityPress; New York: Russell Sage Foundation.

Levy, Frank, and Ari Goelman. 2005. “Offshoring andRadiology.” In Brookings Trade Forum: 2005:Offshoring White-Collar Work, ed. S. Collins and L.Brainard. Washington, D.C.: Brookings InstitutionPress, 411–23.

Lösch, August. 1938. “The Nature of EconomicRegions.” Southern Economic Journal, 5(1): 71–78.

Lösch, August. 1954. The Economics of Location: APioneer Book in the Relations between EconomicGoods and Geography. Translated from the SecondRevised (1944) Edition by William H. Woglom withthe Assistance of Wolfgang F. Stolper. New Haven:Yale University Press.

Mundell, Robert A. 1957. “International Trade andFactor Mobility.” American Economic Review, 47(3):321–35.

Olin, Bertil. 1933. Interregional and International Trade.Reprinted by the Harvard University Press, 1967.

Rauch, James E. 1996. “Trade and Search: SocialCapital, Sogo Shosha, and Spillovers.” NBERWorking Papers, no. 5618.

Rauch, James E. 1999. “Networks versus Markets inInternational Trade.” Journal of InternationalEconomics, 48(1): 7–35.

Rauch, James E., and Vitor Trindade. 2005. “Necktiesin the Tropics: A Model of International Trade andCultural Diversity.” NBER Working Papers, no.11890.

Redding, Stephen, and Anthony J. Venables. 2003.“Geography and Export Performance: ExternalMarket Access and Internal Supply Capacity.”NBER Working Papers, no. 9637.

Rosenberg, Nathan. 1983. Inside the Black Box:Technology and Economics. Cambridge: CambridgeUniversity Press.

Ruffin, Roy J. 1984. “International Factor Movements”In Handbook of International Economics, Volume 1,ed. R. W. Jones and P. B. Kenen. Handbooks inEconomics series, no. 3. New York; Amsterdam andOxford: North-Holland, 237–88.

Sachs, Jeffrey D. 2003. “Institutions Don’t Rule: DirectEffects of Geography on Per Capita Income.” NBERWorking Papers, no. 9490.

Sachs, Jeffrey D. 2005. The End of Poverty: EconomicPossibilities for Our Time. New York: Penguin Press.

Samuelson, Paul A. 2004. “Where Ricardo and MillRebut and Confirm Arguments of MainstreamEconomists Supporting Globalization.” Journal ofEconomic Perspectives, 18(3): 135–46.

Steele, Cherie, and Arthur Stein. 2002. “Communi-cations Revolutions and International Relations.” InTechnology, Development, and Democracy: Inter-national Conflict and Cooperation in the Infor-mation Age, ed. Juliann Emmons Allison. Albany:SUNY Press, 25–54.

Stiglitz, Joseph E. 2002. Globalization and ItsDiscontents. New York and London: Norton.

von Thünen, Johann H. 1826. Der isolirte Staat inBeziehung auf Landwirtschaft und Nationalökonomie,oder Untersuchungen über den Einfluss, den dieGetreidepreise, der Reichtum des Bodens und die



Abgagen auf den Ackerbau ausüben, Vol. 1.Weber, Alfred W. 1909. Theory of the Location of

Industries. Chicago: University of Chicago Press.Williamson, Oliver E. 1975. Markets and Hierarchies:

Analysis and Antitrust Implications. New York:

Free Press.Young, Alwyn. 1995. “The Tyranny of Numbers:

Confronting the Statistical Realities of the East AsianGrowth Experience.” Quarterly Journal ofEconomics, 110(3): 641–80.

a flat world

Documents