Global is at Ion and Canada

8/3/2019 Global is at Ion and Canada

1/64

C.D. Howe Institute

Institut C.D. Howe CommuniquEmbargo: For releaseMonday , Octob er 23, 2000, at 12:00 noo n

Nat ional borders st ill mat ter

despit e globalizat ion, says economist

Common perceptions about the extent of globalization are based mu ch more on myth thanon fact, says economist John Helliwell. Distance and national bord ers, he notes, havesurp risingly large an d continuing effects on the p atterns of economic life. Yet, whileliberalized international flows of goods, services, capital, and peop le present opportunitiesthat Canad ians ough t to seize, Helliwell argu es that Canada still has wide scope forauton omou s policy. It is not the forces of globalization, but Canadians own choices, basedon their own needs and aspirations, that will determine their future p rosperity and qu alityof life.

Helliwell Professor of Econom ics and McLean Chair of Canad ian Stud ies at theUniversity of British Colum bia and a Research Fellow of the C.D. How e Institute madehis remarks in th e C.D. How e Institutes ann ual Benefactors Lecture, delivered in Torontotoday.

Helliwell presents evidence about th e global, national, and local structure of economiclife, and show s that, although globalization has som e substance, it has nothing like as muchas reported in the med ia and presum ed in mu ch of the p rofessional literature. He alsoconclud es that the mu ch-discussed brain d rain from Canad a to the United States is small byhistorical standa rds and generally limited to specialized occup ations.

Helliwell also examines the effects of globalization on economic and social welfaregenerally by looking at its influence on incomes, health care, education and know ledge, andsocial capital; in each case, he conclud es, the influence of globalization seems to be small.

Helliwell says the evidence on globalization brings both good and bad new s forCanad as futu re. The good new s is, first, that d espite many increases in the strength an ddepth of international linkages over the p ast 40 years, coun tries internal economic andsocial structures rem ain mu ch tighter than is comm only believed . Second , small countriesremain as viable and vibrant as they w ere decades ago. Taken together, this evidencesuggests that globalization d oes not pose a threat to the viability and indep endence of thesmaller countr ies.

The bad n ews is that the good new s is not more widely known and hence pu t to properuse. Well-designed national and international policies, Helliwell says, would reflect the factthat, for now and the foreseeable future, geographic, social, and political distance act tomake it cheaper and safer to use familiar and trusted institutions and pathw ays. If new and


2/64

better institutions and path ways are to be built, wh ether within or across national bord ers,it will be imp ortant to do so in w ays that broaden rather than diminish the und erlyingbedrock of shared tru st.

The Benefactors Lecture, which is presented an nu ally in the fall, was spon sored thisyear by Nor and a Inc. Past lecturers includ e econom ists Paul Boothe, Thomas J. Courchene,

Pierre Fortin, Richard Harr is, Richard G. Lipsey, John McCallum, and D.G. McFetridge, andpolitical scientist Richard Simeon .

* * * * *

The C.D. Howe Institute is Canadas leading ind ependent, nonpartisan, nonprofit economic policyresearch institution. Its ind ividual and corporate members are drawn from business, labor, agriculture,un iversities, and the p rofessions.

30

For further information, contact: John Helliwell, e-mail: [email protected]; phone: (604) 822-4953;Kristine Gray (media relations), C.D. Howe Institute

phone: (416) 865-1904; fax: (416) 865-1866;e-mail: [email protected]; Internet: www.cdhowe.org

Globalization: Myths, Facts, and Consequences, Benefactors Lecture 2000, by John F. Helliwell (October 23,2000). 55 pp .; no charge (postage & hand ling extra please contact the Inst itute for d etails);ISBN 0-88806-478-0. Copies are available directly from the C.D. Howe Institute, 125 Adelaide Street East,Toronto, Ontario M5C 1L7.

C.D. Howe Institute / Institut C.D Howe Communiqu / 2


3/64

C.D. Howe Institute

Institut C.D. Howe Communiqu

Embargo : pu blier le lundi 23 octob re 2000 mid i

Malgr la m ondialisat ion,

les front ires nat ionales importent toujours,

affirme un conom iste

Lopinion gn rale lgard d e la mon dialisation et de son envergu re relve da vanta ge dumythe que des faits, affirme lconomiste John H elliwell. Dapr s lui, la d istance et les frontiresnationales ont u n effet remarquablement imp ortant et perm anent sur la structure de la vieconomiqu e de tou s les jour s. Bien que la libralisation d u flux international des biens, desservices, des capitaux et d es gens offre des possibilits don t les Canadiens d evraient tirer p rofit,M. Helliwell soutient que le Canad a est encore dot dune grand e marge dautonomie politique.Ce ne sont pas les forces de la mond ialisation, mais le propr e choix des Canad iens, mo tiv parleurs besoins et dsirs, qui dterminera leur prosp rit et leur qualit de vie futu re.

M. Helliwell, qui est p rofesseur dconom ie et titulaire de la chaire McLean d tud es

canadienn es lUniversit de la Colombie-Britann ique, et charg de recherche invit aup rs delInstitut C.D. How e, a fait ces remar ques dan s le cad re de la confrence annuelle desbienfaiteurs d e lInstitut, donne au jourd hu i Toronto.

Avanant d es preuves sur la structure mon diale, nationale et locale de la vie conomique,M. Helliwell dmon tre qu e mm e si la mond ialisation p rsente certaines qualits concrtes, ellene se rapp roche pas d e lenvergure qu e lui donn ent les mdias et une grand e partie de ladocumentation p rofessionn elle. Il conclut galement qu e lexode des cerveaux du Canad a versles tats-Unis, un sujet tant dbattu , nest que m odeste selon les normes historiques et se limitegnralem ent aux professions spcialises.

Le confrencier examine au ssi les effets de la m ond ialisation su r le bien-tre conom ique etsocial en an alysant linfluence quelle a sur les revenus, les soins d e sant, ldu cation et le

savoir, ainsi que le capital social; dans tou s les cas, il estime qu e la mon dialisation ne sem bleavoir eu quune influence modeste.M. Helliw ell explique qu e ces faits sur la mond ialisation p rsentent d e bonnes et de

mau vaises nou velles pou r lavenir du Canada. Dun e part, m algr le resserrement etlintensification d es liens internationau x au cours d es 40 dern ires annes, les structu resconomiques et sociales internes d es pays sont bien p lus herm tiques qu on ne le croitgnralemen t. De plus, les pays d e petite taille sont tou t aussi viables et vivants quils ltaientauparavant. Dans lensemble, ces facteurs suggrent que la mondialisation ne pose pas demen ace la viabilit et lind pend ance des pay s de taille mod este.


4/64

Dautr e part, ces facteurs positifs ne sont pas de notorit suffisamm ent pu blique pou r treutiliss bon escient. Des politiques n ationales et internationales bien conues, souligneM. Helliwell, devraient reposer sur le fait que, mainten ant et dans lavenir imm diat, la distancegographiqu e, sociale et politique signifie quil est moins coteux et plu s sr d adopter d esinstitutions et d es voies qu i sont familires et fiables. Si ltablissemen t d institutions et d e voiesmeilleures simp ose, quelles soient nationa les ou tran sfrontalires, il imp orte d e le faire duneman ire qui largira p lutt quelle ne rtrcira les assises de confiance mu tuelle.

La confrence des bienfaiteurs, donne chaqu e anne lautomn e, est parr aine cette annepar Nora nd a Inc.. Au nom bre des confrenciers passs, figurent Paul Boothe, Thomas J.Courchene, Pierre Fortin, Richard Har ris, Richard G. Lipsey, John McCallum et D.G. McFetridge,ainsi que le politicologue Richard Simeon.

* * * * *

LInstitut C.D. Howe est un organisme ind pendan t, non-partisan et but n on lucratif, qui joue un rleprpondrant au Canada en matire de recherche sur la politique conomique. Ses membres, individuelset socitaires, proviennent du milieu des affaires, syndical, agricole, universitaire et professionnel.

30

Renseignements : John Helliwell, courriel : [email protected]; tlphone : (604) 822-4953;Kristine Gray (relations avec les md ias), Institut C.D. How e,

tlphone : (416) 865-1904; tlcop ieur : (416) 865-1866;courriel : cdhowe@cdhow e.org; site Web : ww w.cdhow e.org

Globalization: Myths, Facts, and Consequences, Confrence 2000 des bienfaiteurs, p ar John F. Helliwell,le 23 octobre 2000, 55 p., gratuit (frais dexpdition et de TPS en su s prire d e communiquer aveclInstitut cet effet). ISBN 0-88806-478-0.

On p eut se p rocurer des exemp laires de la confrence directement aup rs de lInstitut C.D. Howe,125, rue Ad elaide East, Toronto (Ontario) M5C 1L7.

C.D. Howe Institute / Institut C.D. Howe Communiqu / 2


5/64

C.D. Howe Institute

The C.D. Howe Institute is an independent, nonprofit, research and educa-tional institution. Its goals are to identify current and emerging economicand social policy issues facing Canad ians; to analyze options for pu blic andprivate sector responses; to recommend, where appropriate, particular poli-cy options that, in the Institute's view, best serve the national interest; and tocommun icate the conclusions of its research to a dom estic and internationalaud ience in a clear, nonpartisan w ay. While its focus is national and interna-tional, the Institute recognizes that each of Canad a's regions may have a par-ticular perspective on policy issues and d ifferent concepts of what should benational priorities.

The Institute was created in 1973 by a merger of the Private PlanningAssociation of Canada (PPAC) and the C.D. Howe Memorial Foundation.The PPAC, formed in 1958 by business and labor lead ers, und ertook researchand educational activities on economic policy issues. The Foundation wascreated in 1961 to memorialize the late Rt. Hon. Clarence Decatur Howe,wh o served Canad a as Minister of Trade and Comm erce, amon g other elect-ed capacities, between 1935 and 1957. The Foundation became a separateentity in 1981.

The Institute encourages participation in and support of its activitiesfrom bu siness, organized labor, associations, the professions, and interestedindividuals. For further information, please contact the Institute's Member-ship Coordinator.

The Chairm an of the Institu te is Kent Jespersen; Jack M. Min tz is Presi-dent and Chief Executive Officer.

C.D. Howe Institute

125 Adelaide Street EastToronto, Ontario M5C 1L7

tel.: 416-865-1904; fax: 416-865-1866;

e-mail: [email protected]

Internet: www.cdhowe.org


6/64

C.D. Howe Institute

Benefactors Lecture, 2000

Globalization:Myths, Facts, and Consequences

John F. Helliwell

Professor of Economics and

McLean Chair of Canadian Studies,University of British Columbia,

and

Research Fellow,

C.D. Howe Institute

Toronto, October 23, 2000

Sponsored by

Noranda Inc.


7/64


8/64

Foreword

Globalization is the defining buzzword of our day. Enthusiasts hail a world

withou t borders, emp owering individu als and u nleashing prosperity. Detrac-tors fear unconstrained business and the erosion of democracy by interna-tional organizations. On both sides, overstatement rules.

Listening to and participating in the debate over globalization, we atthe C.D. How e Institute felt a strong n eed for m ore reliable facts and level-headed reasoning. So we asked an outstanding Canadian economist, JohnHelliwell, Professor of Econom ics and McLean Chair of Canad ian Stud ies atthe Un iversity of British Colum bia and a Research Fellow of the C.D. How eInstitute, to explain how declining costs of transportation and communica-tion are changing the world and how they are not. And we invited hisviews on how the environment for Canadian economic and social policy is

changing and, again, how it is not.In the Benefactors Lecture, 2000, Dr. Helliwell rises impressively to the

challenge. He shows that, although walls between nations are thinning,national borders still matter, and he advances several ideas about why theyare so du rable. While liberalized international flows of good s, services, cap ital,and peop le present opp ortun ities that Canadians ought to seize, Dr. Helliwellargues that Canada still has wide scope for autonomous policy. It is not theforces of globalization, but Canad ians own choices, that w ill determ ine theirfuture prosperity and quality of life.

The Institutes aim in presenting the Benefactors Lecture series is toraise the level of public debate on issues of national interest by presentingdiverse points of view. In doing so, the Institute hopes to give Canadiansmuch to think about, including information they need to exercise theirresponsibilities as citizens.

I wish to than k ou r benefactor for this years lectu re, Noranda Inc., andin particular David W. Kerr, President and Chief Executive Officer, whosesup port a lso enabled u s to make copies of the lecture available free of charge.

The text of the lectur e was copy edited by Lenore dAnjou and pr eparedfor publication by Barry A. Norris and Wendy Longsworth. As with allC.D. How e Institute pu blications, the opinions expressed here are those ofthe author, and do not necessarily represent the views of the Institut es mem-

bers or Board of Directors.

Jack M. MintzPresident and Chief Executive Officer

C.D. How e Institute


9/64


10/64

Comm entators widely and perhap s increasingly report that global-ization has sharply diminished the importance of national bor-ders, to t he extent of leaving little scope for mean ingful policies atthe national level. This view comes in at least tw o versions. One is

that the decreasing costs of transportation and communications fromships thr ough t rains and th e telegraph to the teleph one, television, cheap airtravel, and the Internet have made distance increasingly irrelevant. Thisis McLuhans global village, with implications for policies that depend onthe viewers perspective. A second, non-McLuhan v ersion of the nations d onot m atter any m ore view accepts the continuing im portance of geograph-ic distance but implicitly argues that political distance is becoming irrele-vant. Some set of assumptions of this sort lies behind the view that withinNorth America the important linkages and synergies are within geographic

regions, whether as small as Silicon Valley or Ottawas Silicon Valley Northor as large as Ontar io or Cascad ia.1

What are the p olicy imp lications of this widely presum ed w ithering ofthe nation-state? Those who agree that globalization has sharp ly diminishedthe imp ortance of national bord ers, perhaps to the point of irrelevance, havea w ide range of p olicy perspectives. At one extreme is a dow n w ith all gov-ernments view, whereby a disappearing role for national policies is muchwelcomed. An yone w ith this outlook regard s the p ossibility of supernation-al replacements for national policies as pie in the sky or as evidence of anominously encroaching w orld governm ent operating beyond citizen control.From a m ore international perspective, decreasing scope for national p oliciesis seen to create a correspond ing need for internat ionally or globally harmo-nized approaches to a rules-based system. One might think of this outlookas a world federalist interpretation of the distance does not matter anymore view.

I am grateful for commen ts on earlier versions from Bob Evans, Pierre Fortin, John McCal-lum, Jack Mintz, Finn Poschm ann , John Richards, an d Bill Robson. I am also gr ateful forcontinuing support of the underlying research by the Social Sciences and HumanitiesResearch Council of Canada, and for the research and editorial assistance of AnetaBonikowska, Aileen Battye, and David Helliwell.

1 Because geographic and p olitical distances make such a d ifference to information flows, it

is perhap s worth remind ing readers east of the Rocky Mou ntains that Cascadia is a gleamin the eye of some who live British Columbia, Washington, and Oregon. Since the politi-cal basis for such a region (wh ose name d erives from th e Cascade Mou ntains that span it)lies mainly in shared geography and perceptions that their respective national govern-ments are both ignorant and uncaring, a new nation is not in serious prospect. However,transborder conversations at the regional level are likely to focus attention on problemsparticular to the region, including intern ational border issues.


11/64

2 John F. Helliwell

A third perspective, in evidence among some of the protest groups atthe World Trade Organization (WTO) meeting in Seattle and the Interna-tional Monetary Fund (IMF) and World Bank meetings in Washington, is

that globalization is th e result of forces operating in the joint interests of rich-country capitalists and p oor-coun try oligarches, with the international finan-cial institutions acting as their lackeys. The policy implications of this per-spective range as widely as the views represented in the protests.

In this lecture, I argue that the comm on perceptions about the extent ofglobalization are based much more on myth than on fact. Distance andnational borders have surprisingly large and continuing effects on the pat-terns of econom ic life, and th ese effects both perm it and requ ire a set of pol-icy choices mu ch different from wh at w ould be considered by an yone wh othinks of the global econom y as a seamless unit.

I divide my argument into four main parts. In the first, I present evi-dence about the global, national, and local structure of economic life. Theburden of this evidence is that globalization has indeed some substance butnothing like as much as reported in the med ia and presum ed in mu ch of theprofessional literature. The second part turns to the much-discussed braindrain from Canada to the United States and concludes that the loss is histor-ically small except p erhaps for ind ividuals w ith very h igh incomes and insome highly specialized fields. The third section examines many facets ofhuman well-being; all could be affected by globalization, but its influenceseems to be sma ll. Finally, I consider the consequences for Canad ian p olicies,both domestic and international.

The Economic Separation of Nation-States

In the mid-1990s, just as John McCallum (1995) was reporting his startlingfind ing that in 1988 Canadian p rovinces had merchandise trade flows w itheach other that were 20 times greater than those between Canadianprovinces and US states of equivalent size and distance, Charles Engel andJohn Rogers (1996) were independently discovering that short-term pricelinkages between cities diminished with distance within Canada and theUnited States and were enormously less across the international border.Indeed, subsequent research u sing the Engel and Rogers data sh ows th at, forthe consum er prices they examine, there is no crossbord er linkage at any dis -tance (Helliwell 1998, 68). (These two independent findings are mutuallyconfirming. One expects to find that, where trade linkages are less dense,price linkages are also w eaker.)


12/64

Globalization: M yths, Facts, and Consequences 3

Once lodged on th e radar screen, which took some time, these findingshave led to a h ost of stud ies designed to see how general and persistent these

border effects may be. How have the border effects changed in the wake ofthe un expectedly large expansion of Canad a-US trad e following the Canad a-US Free Trade Agreem ent (FTA)? If the national bord er betw een the United

States and Canada is of so mu ch importance for merchand ise trade, what canone expect to find for services? Is capital more easily mobile? Do the resultsapp ly to other countries as well? Do they apply equ ally in all provinces and

to all indu stries? If not, do the patterns tell some consistent story?

What Is the Current Effect on

Markets for Goods and Services?

First, wh at of the effects of the FTA? My most r ecent a ttemp ts to assess the

post-FTA evolution of border effects for merchandise trad e between Canadaand the United States, using d ata revised and extended since those of McCal-lum (1995), show a post-FTA reduction of about one-third, consistent with

the rise of Canad a-US trad e flows by abou t 50 percent relative to the valu esthat w ould have been expected to follow from increases in p opu lation andincome in the two countries. As for the pattern of adjustment, little change

appeared until 1990; following came a fairly sharp increase of north-southtrade densities from 1991 through 1993 and rough constancy for 1994

through 1996. Thus, in 1996 the typical Canadian province traded twelvetimes as mu ch with anoth er Canadian p rovince as with a US state of similarsize and distance.

If the border did not matter at all, then interprovincial trade flows

would exactly match those between provinces and states. Thus, Ontariowould trade ten times as much with California as it does with BritishColumbia, since California is mor e than t en times larger than British Colum -

bia, wh ile the tw o are both roughly the same d istance from Onta rio. Yet tradebetween Ontario and British Columbia is actually larger than that betweenOntario and California.2

2 To avoid possible confusion, I should point out that goods shipped from Ontario throughBritish Colum bia to foreign mar kets, whether in the Un ited States or overseas, are exclud -ed from the d ata on trade between Ontario an d British Colum bia. Likewise, SouthKorean cars imported thr ough Vancouver on their way to Toronto are not included in thestatistics on exports from British Columbia to O ntario.


13/64

4 John F. Helliwell

The Market for Goods

The main basis for the interprovincial trade data is a survey of manufactur-

ers asking them to divide their shipments by ultimate destination. Thesedata, combined with transportation data for specific commodities, are used

to allocate each provinces interprovincial exports to the interprovincialimports of each of the other p rovinces. Canada th us h as more complete andconsistent measu res of internal trade than does any other coun try. 3

This fact, combined with Canad as large size and closeness to the Un it-ed States, makes it easy to separate the important effects of distance from

those of national borders. For most countries, foreign markets are, on aver-age, much farther away th an are d omestic markets, so that the border v ari-

able is correlated w ith the d istance variable, thus creating possible difficul-

3 Most of the studies of the border effect between the United States and Canada compareinterprovincial with province-state trade, since no comparable data exist for interstatetrade in the United States. Two recent stud ies attemp t to fill in the picture by th e use of the1993 US Commodity Flow Survey (United States 1996). Hillberry (1998; 1999) combinesspecial tabulations of these data with estimates of surface shipments from US states toadjacent Canad ian pr ovinces. This methodology p rodu ces an estimate of the bord er effectin 1993 of 20.9, about 70 per cent larger than the 12.3 estimate based on 1993 Canadian datafor interprovincial and province-state shipments (Helliwell 1998, 22). By contrast, vanWincoop (2000) combines the Canadian data for province-state trade with the US Com-mod ity Flow Survey d ata to estimate a 1993 border effect of only 1.3. He argu es that H ill-berrys estimate of transborder shipments represents a severe undercount and thereforean ov erestimate of the size of the bord er effect. Hillberry argues, on th e other h and , that,

by u sing comparable sources for both interstate and state-province shipments, he is reduc-ing the likely extent of data mismeasu rement.The fact that th e Hillberry results are significantly larger than th ose based on the Cana-

dian d ata suggest that undercounting may ind eed be larger for export shipments than forthose moving interstate. However, the van Wincoop results have even more trou bling fea-tures. They imply that the density of interprovincial shipments in Canada is almost tentimes greater than th at of interstate shipmen ts in the United States. Van Wincoop arg uesthat, in theory, this result may occur because the greater size of the US economy meansthat prod ucers and consumers do not have to go so far afield to get the range of prod uctchoice they need. Yet, despite some evidence of greater mobility of goods, people, andcapital among Canad ian pr ovinces than amon g US states, none of the evidence is of any-thing like the sort to match data sugg esting that interprovincial trade flows are ten timesas intense as those amon g states.

If van Wincoops interp retation is correct and bord er effects are thus m uch larger for

small economies than for large ones, then the large post-FTA increases in transbordertrade would have been accompanied by large reductions in interprovincial trade. How-ever, although Helliwell, Lee, and Messinger (1999) find some evidence of post-FTA tradediversion from provinces to states, it is nothing like large enough to support van Win-coops hypothesis. Thus, my preliminary judgment is that, when compared with the Sta-tistics Canada estimates for either interprovincial or province-state trade, the US Com-mod ity Flow Survey da ta materially un derstate the levels of interstate merchand ise trad e.


14/64


ties in distinguishing the two effects. For Canada and the United States,things are rather different, since Canada can be seen as the top row of aneleven-row rectangle with the United States represented by the lower ten

rows. The top row also sags south in its popu lous midd le, since more thanhalf of the Canad ian pop ulation lives south of the 49th parallel, nestled w ith-in the industrial heartland of North America. As a result, the average dis-tance between Cana dian p rovinces is almost exactly the same as the averagedistance between the provinces and the 30 states used in the trade-flowsstud y, which includ e the border states and the largest of the rest (McCallum1995; Helliwell 1996c; 1997; 1998). Thus, the effects of distance are easy todistingu ish from those of the border itself.

Much of the attention on the McCallum and subsequent stu dies focus-es on the effects of the bord er, but, as Grossman (1997) and Haz ledine (2000)emphasize, the effects of distance are equally striking. Distance has a dra-matic trade-reducing effect, with each 1 percent increase in distance beingassociated w ith an appr oximately 1.5 percent reduction in interp rovincial orprov ince-state merchand ise trade. This distance effect is quite separate fromthe border effect.

For merchandise trade, the crossborder and within-Canada distanceeffects are equ ally large (Helliwell 1998, 68) so the bord er effects are not dueto per kilometer transport costs being higher for transborder shipments.More to the p oint, the distance effects are so great that they cannot representjust shipping costs, whether domestic or international. There is also no evi-dence of reduction in the distance effect over the 1988 to 1996 period (for

wh ich the interprovincial and p rovince-state trade d ata are available). Thus,when the time comes to consider the reasons for large border effects, weshould be looking for factors that could explain equ ally w ell why d istancestill has such a large separating effect.

The Market for Services

No one keeps d ata for services that are counterp arts to the interprovincialand province-state trade data underlying the estimates for merchandisetrade. Data for total interprovincial and Canada-US trade in services can,however, be used in conjunction with the results in hand for merchandise

trad e to give an ap pr oximate estimate of the bord er effects for services.These effects app ear to be between tw o and three times as large as thosefor merchandise trade, with values of about 30,4 and show no apparent

4 That is, 30 times less than one wou ld expect in the absence of an international border. Forsimplicity, I henceforth rep ort the size of border effects simply as n um erals.


15/64

6 John F. Helliwell

decline over the 198896 sam ple p eriod (Helliwell 1998, 38). It is temp ting todismiss these results noting that services in general are m ostly not trad ed,so what else would one expect to find? but that conclusion is too facile

since the comp arison is not between the trad ed and nontrad ed services butbetween two groups of traded services, those sold interprovincially andthose sold across the Canada-US bord er. As already noted, this large bordereffect is not simply a distance effect in disguise, since the average inter-prov incial distances are as great as those between p rovinces and stat es. Thus,the border effects for traded services are larger than those for traded mer-chandise, a result that should be borne in mind when trying to develop ageneral explanation for the size and preva lence of bord er effects.

Evidence from Other Countries

Attempts to estimate border effects for other groups of countries are bedev-iled by the lack of subn ational trad e and distance data. In the absence of fullspatial disaggregation of produ ction and shipment data, the current p roce-dure in most studies is to use input-output data for domestic shipments(equal to total shipments minus those exported) and to develop someapp roximation for d omestic distances. Wei (1996) approximates th e internaltrade distances by using one-quarter of the distance from a country to itsnearest trad ing partner, a p rocedure also u sed by H elliwell (1998).5

Subsequ ent research find s that Weis distance approximation generally

un derestimates, often by a large amount, the n um bers derived from p roce-du res more closely groun ded in analytic geography and the actual distribu-tion of popu lation and economic activity within each countr y. The latest esti-mates of border effects for countries in the European Union (EU) arebetween 6 and 10 (Nitsch, forthcoming; Chen 2000) or between 12 and 19(Head an d Ma yer, forthcoming), abou t 10 or mor e for coun tries of the Organ-isation for Economic Co-operation and Development (OECD) in general(Helliwell 1998, 51), and m uch larger still for developin g coun tries (ibid., 55).

5 Wei (1996) obtains small estimates of bord er effects for countries of the Organisation forEconomic Co-operation and Development, which Helliwell (1998) finds to be due to an

inapp ropr iate definition of third-party trad e potential and to the convention Wei adop tedof reporting border effects applicable to the few countries sharing common language, acommon border, and membersh ip in th e Europ ean Union. N itsch (2000) objects to Weisprocedure on the grounds that it falsely involves the size and shape of the neighboringcountry in the calculation of domestic distance; Helliwell and Verdier (1999) and Chen(2000) criticize it as inconsistent with the logic of the underlying gravity model and asignor ing available informa tion about the size an d d istribution of internal economic activity.


16/64


The estimated levels of nontariff barriers cannot explain the pattern ofbord er effects across ind ustries in th e EU (Head and Mayer 2000). This find-ing sup port s the view I take here: that bord er effects reflect national tastes

and networks, rather th an policy-related barriers.

How Mobile Is Capital across Borders?

If domestic goods markets are much tighter than international ones, oneexpects to find th e same thing true of capital markets. The common ly mad eassumption of perfect international mobility of capital requires the preva-lence of a single global real interest rate, wh ich, in tu rn, requ ires that p ricesmove to m aintain pu rchasing p ower p arity between national currencies, anassumption that is rejected strongly by the Engel and Rogers (1996) results

already qu oted. This insight raises the likelihood th at Feldstein and Hor ioka(1980) are correct in their original inference tha t h igh correlations betw eennational savings an d dom estic investment imply low international m obilityof capital.6

One w ay of checking this interp retation against comp eting theories is tocompare savings-investment correlations across countries with those forsubn ational units within the same coun try. If Feldstein and H orioka are rightin concluding that a large positive correlation between savings and invest-men t at the national level signals low international capital mobility, then oneexpects to find that the correlation is smaller, or even vanishes entirely,across provinces or regions within a country. Helliwell and McKitrick (1999)

combine Canadian provincial savings and investment rates with nationaldata from other OECD countries and find that th e strong correlation at thenational level is completely absent among the provinces, thus providingstrong support for the Feldstein and Horioka interpretation. Positive corre-lations between subnational savings and investmen t rates, using somew hatless comp lete data than is available for the Canadian pr ovinces, are similar-ly absent in Japan (Dekle 1996), the United States (Sinn 1992), and the Unit-ed Kingd om (Bayou mi an d Rose 1993).

Other types of evidence also indicate low international linkages ofnational cap ital markets. Investors systematically prefer to hold hom e-countryrather than foreign equities (French and Poterba 1991; Baxter and Jermann

6 A large subsequent literature explores what has become known as the Feldstein-Horiokapu zzle. This literatur e continues to find high correlations, of the sort discovered by Feld-stein and Hor ioka, in both cross-section and pan el data sets. While their emp irical find inghas been confirmed many times, there remains some skepticism about their conclusion,since some theoretical mod els with p erfect capital mobility still give rise to such correlations.


17/64

8 John F. Helliwell

1997); there is no evidence that savers use international capital markets tosmoo th th eir consu mp tion in the face of income shocks (Backus, Kehoe, andKydland 1992); and exchange rates are highly volatile and apparently unre-

lated to changes in fundamentals beyond a relatively weak long-run ten-den cy toward pu rchasing power p arity (Meese and Rogoff 1983; Baxter andStockm an 1989).

Why Do National Borders

Separate Markets So Strongly?

Despite the continuing p revalence of theoretical mod els that assum e perfectinternational mobility of goods and capital, the empirical evidence outlinedabove is starting to receive more theoretical attention. Obstfeld and Rogoff(2000) list six of the empirical findings given above (the border effects intrade, the Feldstein-Horioka puzzle, the home bias in equity holdings, thelack of consum ption sm oothing a cross coun tries, the bord er effects in prices,and the volatility of exchange rates) as the major unresolved p uzzles of inter-national macroeconomics. Seeking to find the simplest extensions to stan-dard theory that wou ld account for these puzzles, the authors argue that thecombination of significant internat ional transactions costs and p rices that aresticky (or fairly rigid) in term s of dom estic currency is su fficient to explainall of them .

Two More Puzzles

The idea of tackling the w hole set of related puzz les together is, in my view,exactly the right way of proceeding. I do not think Obstfeld and Rogoff(2000) go qu ite far enou gh, but som e slight extensions of what th ey prop oseare likely to do the trick. What m ore is likely to be requ ired? First, a recogni-tion of two pu zzles beyond their list of six: to the exceptionally strong effectsof distance in red ucing the d ensity of econom ic relations, and to the fact thatforward exchange rates are even worse than spot exchange rates as predic-tors of the future value of a currency. 7

Extending the puzzle list invites an extended interpretation of the rea-sons for border effects. Obstfeld and Rogoff rightly focus on transactionscosts, but they would do better to consider a broader range. The costs ofdoing business depend on the extent of networks of association and trust.

7 Obstfeld and Rogoff (2000) note the latter fact but leave its explanation as a hom eworkassignment for readers.


18/64


Recommend ations carry more weight w hen they come from p eople who areknown and trusted, and frequency of contact provides a natural flow ofinformation that is u seful in choosing futu re trade and investment partners.

This extend ed list of determ inants of tran sactions costs lessens the exclusiveimportance of political borders and of legal impediments to transborderflows. It also suggests a theoretical explanation that applies equally well tonational border effects and to large d istance effects.

Distance separ ates markets and d iminishes the frequency of casual con-tacts among traders, making it efficient for buyers, sellers, and investors toconsider local options before proceeding farth er afield . Exactly the same casecan be made about national borders, a coincidence suggesting that the bor-der and distan ce effects are likely to be explicable by the sam e general theo-ry (even though the reasons tastes and contacts change across borders maydiffer in detail from the reasons they change with distance).

While recognizing that goods and capital markets are nationally dis-tinct, Obstfeld an d Rogoff try to hold on to the assum ption th at financial cap-ital is perfectly mobile. They wish to keep this assump tion in part because itkeeps the theory simpler, and in part because of the extensive evidence thatcovered interest arb itrage is fairly perv asive.8

Recognition that the forward exchange rate is worse than the spotexchange rate as a pr edictor of the futu re spot rate show s, however, that thecovered arbitrage condition does not support the common extension touncoveredarbitrage. On the contrary, the evid ence from the p ricing of securi-ties of the type used for covered interest arbitrage (usually short-term bank

dep osits and short-term governm ent securities) is that a change in n ationalinterest rates indu ced by a change in monetary policy leads first to a changein the forward exchange premium or discount of just the size required tomaintain covered interest parity. (The exchange rate may also change butnever by enough to give the extent of overshooting required to sup port theusual assumption of uncovered interest rate parity.) In short, the variablethat adjusts freely to maintain covered interest parity after a change in anational interest rate is the forward exchange rate. This interest-respon sive-ness of the forward rate in turn contributes to its bad record as a pred ictor of

8 Covered interest arbitrage means that the interest rates on comparable securities issued by

the same or an equivalent issuer in anoth er currency w ill be priced so that the d ifferencebetween the two interest rates is exactly equal to the forward exchange premium or dis-count betw een the two curr encies. Thu s, banks offering Canad ian and US dollar dep ositsdo so at interest rates that differ by roughly the amount of the forward exchange premi-um between the two currencies. This covered interest parity is prevalent for at least somecategories of investments, leading m any analysts, including Obstfeld and Rogoff, to inferthat financial capital is perfectly mobile.


19/64

10 John F. Helliwell

the future spot rate. The reason it can move so much and so freely is thatinvestors are prep ared to m ove very little capital on an un covered basis to tiedown the forward rate to the path determined by fundamentals. Thus, the

bad pred ictive power of the forward exchange rate provides one m ore pieceof evidence of international market segm entation.

With the set of explanations extended to include a broader range oftransactions costs that rise with distance as well as across borders, we areable to explain all of the six puzz les Obstfeld and Rogoff examine, p lus th elarge effects of distance and the poor performance of the forward exchangerate as a pred ictor of futu re exchange rates.

Yet Another Puzzle

It is tempting to try the same framework on another p uzzle: Why d o differ-ent ways of estimating the extent to which goods in different countries aresubstitutable give such dramatically different answers? Traditional econo-metric estimates, which are based on the extent to w hich trad e flows respon dto changes in relative prices, tend to suggest quite small elasticities of sub-stitution betw een dom estic and foreign good s. How ever, the fact that nation-al border effects and distance effects are so large su ggests that local pr odu ctsmay be qu ite good substitu tes for those coming from other countr ies or fromgreater distances within the domestic economy.

Reconciliation lies in recognition of the market-separating effects of dis-tance and n ational borders. When good s have to cross space or bord ers, one

result is a no-trading zone of prices that tends to diminish the extent towh ich trad e flows respon d to price changes and to lead to sm all estimates ofpr ice elasticities. On th e other han d , a shift of relative pr ices within the sam elocal market can lead to a change in spending patterns that yields muchhigher estimates of the extent to w hich goods are substitutes for one another.

That local and distant products are fairly good substitutes for oneanother helps to explain w hy d istance and border effects can have so m uchinfluence on trad e flows, relative to the measur ed costs of transp ort, withou tleading to large losses of income or welfare. It also helps to explain whysome tests of traditional Heckscher-Ohlin trade theory in which trade is

inspired only by differences in comparative advan tage work m uch betterto explain trade patterns within national economies than between them(Davis et al. 1997; Trefler 1995). In the international case, borders intervene,and geographic, social, political, and cultural d istances are greater, all situa-tions that reduce the advantages of exploiting what might, at lesser cost,seem a worthwhile difference in resource endowments. These reductions


20/64


limit th e emp irical power of the trad itional theory to explain internationalda ta, comp ared to its greater ability to explain trade flows amon g the regionswithin a national economy.

One is tempted one to say that the only thing wrong with traditionalinternational trade th eory is the word international. It w orks w ell to explaintrade an d specialization within n ations and fails only wh en it neglects thefact that information flows diminish, tastes diverge, and transactions costsrise as borders and distance intervene.

How Much of the Border

Effect Is a Currency Effect?

Thus far, I have u sed the cost advan tage of operating w ithin established andwell-und erstood networks to explain wh y econom ic densities fall so rap idly

with both distance and national borders. In this explanation, the bordereffect adds to the distance effect because a number of networks are denserwithin nations than across national boundaries. To some extent, the same istrue of migration an d oth er sources of personal contacts (a ma tter for its ownsection to follow).

Another possible reason for the large size of national border effects isthe p redom inance of national curren cies. Thus, political borders are genera l-ly also currency borders, making it hard to separate the effects of bordersthem selves from those of national currencies.

I note elsewh ere (Helliwell 2000a) the scientific sign ificance of the adop-tion of the eur o in 1999 by some bu t not a ll members of the EU. This partialadoption of the euro is as close as ever happens in ma croeconom ics to a con-trolled experiment because the EU coun tries that have not accepted the newcurrency are in all other respects full members of the union, so that thechanges in trade intensities between euro and non-euro EU countries willprovide a fairly pure measure of the extent to which national border effectsare due to the use of separate national currencies. The evidence must begiven time to accum ulate. In th e meantime, we m ust rely on less direct andless satisfactory evidence.

Most previous attempts to assess the trade-reducing and possiblygrow th-redu cing effects of exchange rate volatility either sp lit exchange rate

regimes into fixed and flexible or use som e measu re of the volatility of realor nom inal exchange rates (see the survey by Ct 1994). A recent stud y byAndrew Rose (2000), however, uses bilateral trade flows among 186 coun-tries to attempt to evaluate the effects of an exchange rate union separatelyfrom th ose of exchange rate volatility. Bilateral trad e flows are m ore inform-ative than other evidence since they increase the number of observations,


21/64


allow for the gravity m odels adjustm ent for the effects of size and d istance,and make it possible to determine which flows would be affected by theadop tion of a comm on currency by some trading p artners.

Rose produces striking results. In addition to the effects of exchangerate volatility on trade, which he estimates to be significantly negative butnot very large in magnitude, he finds that his sample of common-currencycountries trade with each other three times more than do oth er coun tries ofsimilar size and distance. If this result should prove applicable to largercountries adop ting a comm on currency, it suggests that, after some period ofadjustment, trade flows among the euro countries would triple relative tothose between eu ro and non-euro EU countries. For example, trade betweenFrance and Italy, wh ich is cur rently smaller than that between France and theUnited Kingdom, w ould become three times as large.

These effects are so sizable as to m ake Rose cautiou s in h is interpreta-tion, suggesting that h is results should at least be treated as a signal that theeffects of a common currency on trade may well be larger perhaps muchlarger than has p reviously been thought wh en estimated on the basis ofthe effects of exchange ra te volatility on tra de. For examp le, typical estimatesof the effects of exchange rate volatility on Canada-US trade suggest that,withou t it, trade between the tw o countries would rise by about 6 percent, asmall fraction of the increase that took place after the adoption of the FTAand dwarfed by Roses estimate of a 300 percent increase. Thus, Roses esti-mate is large enou gh to n ote. Moreover, his research is careful enough to betaken seriously; he allows for the possible confounding effects of common

colonial ties, comm on langu age, and comm on bord ers.The first question that a rises in the read ers mind mu st be, wh ere are all

the comm on-currency areas w hose trad e effects are being assessed? Are theyexamples of areas that were recently formed and thus allow analysts to sep-arate the effects of the curr ency un ion from all of the other tr ade-creating fac-tors that m ay have led to the choice of a common curren cy in the first place?The answ er to the first question is that all the common -cur rency examples inthe data are tiny countries or dependencies, mostly islands, and the com-mon -cur rency area trade flows relate to their trad e with each other and w iththe single larger and more-established country whose currency they haveadop ted. The answer to the second question is that the stud y does not sepa-rate the effects of currency adoption since almost all of the currency unionswere in action throu ghou t the samp le period Rose uses for his analysis. Thisfact underscores the importance of the euro as the first adoption of a com-mon currency among a group of large countries with previously well-estab-lished trad ing patterns.


22/64


While we wait for the results of the euro to show themselves, whatimp ortance shou ld w e give to the Rose results as a guid e to the likely effectsof a comm on currency? On the one han d, they seem so large relative to the

existing m easures of the consequences of the establishmen t of a trading blocas to be incredible. For example, the EU itself, after more than 40 years oftrade creation, is estimated to have made trade among pairs of membercountries about 50 percent larger than that between other pairs of countriesof similar size and distance (Helliwell 1998, ch. 3). This calculation includesthe effects of tariff removal, the single market, the establishment of European-level rules and institutions, and the exchange rate stability provided by theEuropean Monetary System (EMS). The Rose estimate suggests that theseEU-expand ed trade flows will now increase threefold am ong ad opters of theeuro. Comp ared to the effects of the rest of the EU structur e and institutions,this estimate for the euro seems too large.

On th e other hand , the remaining border effects in merchand ise tradeare also surp risingly large. When seen in th is light, the Rose estimates do n otapp ear qu ite so large. For examp le, the estimates of bord er effects among th eEU countries suggest that they may still be in the range of 6 to 10 (Nitsch,forthcoming; Chen 2000). The Rose estimate of the effects of a common cur-rency as increasing trade by a factor of three thus implies that somewherebetween on e-third and one-half of the remaining bord er effect migh t be elim-inated by the adoption of a common currency. By the same token, the Roseestimate of the effects of a common currency would reduce the mid-1990sestimate of the border effect between Canad a and the United States from 12to abou t 4 (= 12/ 3). Seen in the context of the large rem aining bord er effects,the Rose estimates do not appear quite so large. Judgments about their real-ism m ust d epend on an assessment of the r eliability of the original estimatesplus some opinion about th e importan ce of the other likely causes of bordereffects in trad e.

My own jud gment on the Rose estimates is that they are likely, throu ghno fault of his, to be substan tial overestimates because they probab ly captu rethe effects of trade patterns established for reasons entirely unrelated to theuse of a common currency. More specifically, most of the common- cur-rency examp les in the Rose data reflect geograph ic, political, and resource-based reasons that have led to high trad e levels and to th e use of a comm on

currency. The Rose sample has seven or eight p ivot countries, each of wh ichhas a number of small dependencies that use its currency. For example, theNew Zealand dollar is used by Cook Islands, Niue, Pitcairn Islands, andTokelau, all small territories associated w ith N ew Zealand. The US dollar isused by American Samoa, Guam, US Virgin Islands, Puerto Rico, NorthernMarianas, British Virgin Islands, Turks and Caicos Islands, Bahamas, Liberia,


23/64


Marshall Islands, Micronesia, Palau, and Panam a. In all these cases, militaryor tourist reasons make the causality likely to run from trade to a commoncurrency, rather than the other w ay around . The same argum ent app lies to

Denmark and the Faroe Islands and Greenland and to all the islands usingthe Australian dollar. (An exception to this general pattern is provided by thecommu nity of 15 former French colonies in Africa th at share two differentAfrican francs, each of wh ich is linked to the French franc.)

Although Rose attemp ts to allow separately for colonial or dependencystatus, almost all of the trade flows h e stud ies are between a trad ing countryand tiny territories with little or no independent economic structure. Thedata th us p rovide a weak basis for jud ging wh at might h app en if establishedtrad ing countries adopted the same currency. People who argu e strongly fora comm on currency do so because they think that it would provide imp or-

tant cost savings and generate mu ch additional trade. They are thus natu-rally inclined to think that a large portion of any border effect in merchan-dise trade w ould likely d isappear if a comm on currency were adop ted. I andothers are more inclined to await the results of the euro experiment.

Migration and the Brain Drain

If border effects are very large for movem ents of good s, services, and capi-tal, one shou ld not be su rp rised to find that th e same is true for people. Eco-nomic theorists often assume that labor and natural resources are geograph-

ically rooted to th e spot w hile goods, services, and capital are, to one extentor another, mobile among nations. A century or more ago, this assumptionwas decidedly false; passports and their matching citizenship were almostun known , and each agricultural or indu strial crisis and each new rum or ofgold d iscovery led to m igrations that w ere massive by any m easure. Duringthe 1850s alone, 2 percent of the p eople of Britain m oved to Au stralia, quin-tupling the latters population in the process. Especially in immigrant cul-tures like those of North America and Australasia, international migrationstill seems very m uch the norm .

I remember, when working in the 1950s as a student for the Buildingsand Grounds Department of the University of British Columbia (UBC), dis-covering that one of my colleagues, a Scottish trad esman , had been brou ghtup 12 miles from the sea but h ad never seen it before setting sail for Canadaas a young adult. That story of local fixity and far-flung mobility struck methen as unusual. Just how unusual was it? What are the historical contextand mod ern facts for Canad ian internal and international migration?


24/64


Fifteen years later, the Bank of Canad as RDX2 mod eling team estimat-ed equ ations for m igration flows between Canada and the United States aspart of a project aimed at explaining macroeconomic linkages between the

two countries (Helliwell et al. 1971). These equations show ed average sou th-bound flows three times as large as northboun d ones and foun d th em to besignificantly responsive to differences in income levels and unemploymentrates.

Two more decades later, in the course of research for my BrookingsInstitution book on the effects of national borders (Helliwell 1998), I usedCanadian and US census d ata to stud y the extent to which residents of thetwo countries had m oved w ithin and between the tw o coun tries. Somewh atto my surprise, given the melting pot and frontier-searching images of theUnited States and the greater extent to wh ich Canad ian policies tend to su p-port incomes and employment in the poorer parts of the country, I foundthat interprovincial migration was greater than interstate migration andmor e responsive to differences in income and emp loyment p rospects (Helli-well 1996b). I also found that interp rovincial migration was a hun dr ed timesmor e likely than migration to a p rovince from a US state of similar size, d is-tance, and income level (Helliwell 1997) and northbou nd and southbou ndmigrations were greatly asymmetrical, with individuals born in Canada farmore likely to have moved to the United States than vice versa, even aftertaking into account differences in populations and income levels.

Given this background , when Ottawas Expert Panel on Skills asked m ein late 1998 to assess the d ata u nd erlying the 1990s brain d rain from Cana-da to the United States, I was expecting to find that national borders stillmatter a lot for m igrants. I also anticipated sh arply increasing southbou ndflows during the 1990s, given the unusually large income and employmentgaps between the tw o countries du ring the decade.

Two other changes during the 1990s also led me to expect much largerflows. The first was an increase in income inequality that was much greaterin the Un ited States than in Canada, d ue m ostly to larger and faster-risingUS salary rewards for higher education and, in general, for incomes at thetop end of the scale. The second was th e mu ch greater incentive for and easeof work-related migration between th e two coun tries. In the w ake of the FTAand its successor, the North American Free Trade Agreement (NAFTA), bilat-

eral trade and investment increased dramatically, leading to increased tem-porary and longer-term movements of staff and management. In addition,the trad e agreements introdu ced new categories of temporary visas that ren-dered migration to the United States very simple for Canadians with d egreesin han d and jobs to go to. These new categories, especially the NAFTA (TN)visa, which permit a series of renewals or replacements, have become an


25/64


entry method of choice for temporary workers and for some longer-termmigrants as well.

What Do the Data Show?

What d oes one find wh en checking the d ata? The story needs to be told inseveral time scales and sep arately for different skill levels and occup ationalgroupings.

Looking first at the longer term, the US and Canadian censuses providethe longest systematic record of transborder migration. Figure 1 shows forthe d ecennial censuses the nu mber of residents of each country wh o w ereborn in the other, measured as a percentage of the Canadian p opu lation. Thedifference between the two lines reveals that southbound migration hasalways been three to four times larger than northbound, although the gap

has shown some tendency to shrink over the course of the past century.The most striking feature of these da ta, however, is the extent to w hich

both m igration flows h ave shrun k. At the beginning of the tw entieth centu-ry, the nu mber of Canadian-born individu als living in the United States wasalmost 20 percent of the total population of Canad a. At the beginning of thetwen ty-first centu ry, it is about 2 percent, after an und ulating bu t fairly reg-ular slide over the previous hu nd red years.

This falloff raises a puzzle: as international linkages have in generalgrown tighter, at least over the past half-century, why have those born inCanada and the United States tended to remain at home? One reason, ofcourse, is that both countries have tended to be targets rather than sou rces ofmigration and have been attractive places to remain. Also, we m ust rem em-ber that g lobalization, as it is now th ough t of, is really re-globalization, as thefirst half of the tw entieth century, scarred by wars an d d epression, witnessedsharp reductions in international trade and capital movements. Economichistorians are still trying to d ecide if current levels of international trad e andinvestmen t, relative to GDP, are as great as those a centur y ago. For migra-tion, however, the story d iffers; the whole century h as seen increasing atten-tion paid to nationality and citizenship, with more and more screening ofwou ld-be migrants. From the Canad ian perspective, at least, the body d rainhas been steadily declining over the century.

Figure 2 shortens the time horizon, showing the more recent data inabsolute numbers, supplemented for the 1990s by estimates from the USCurrent Pop ulation Survey (CPS). These data include perman ent and tem-porary migrants and, for the years covered by the CPS, are based on a sam-ple, rather than a 100 percent census count. The figure also shows officialprojections, made in 1990, for the number of Canadian-born living in the


26/64

Globalization: Myths, Facts, and Consequences 17

United States in the 1990 and 2000 census years. The actual 1990 census,

when tabulated, already showed a shortfall below the projection, and theCPS estimates for individual years during the 1990s show a general patternof decline greater than that of the forecasts, which were based on 1980s lev-els of net migration and the expected death rate of long-past migrants.

The pattern in the figure is somewhat surprising: that the 1990s CPSnumbers fall below those projected in 1990 does not square with the wide-spread reporting of increased sou thbound m igration during the past decade,with well-publicized reports of sharply increasing numbers of Canadiansobtaining NAFTA-based temporary m igration status in the United States, or

with the fact that unem ployment rates, income gap s, tax gaps, and the effectsof the FTA have all been such as to lead one to expect a sharp increase insouthbound migration du ring the 1990s.9

Figure 1: Canadian-Born Living in the United States

and US-Born Living in Canada, 19002000

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000

25

20

15

10

5

0

Sources: Canad ian and US decennial censuses; estimated Canad ian pop ulation for 1900, 1910,and 1920 using census data from 1901, 1911, and 1921.

9 The CPS collects data each year as part of its March survey. The year-to-year fluctuationsin the nu mbers are du e partly to the fact that the survey sam ple changes from year to year.It is also possible that the calculated stand ard dev iations for these series about 85,000,according to Schmid ley and Robinson (1998, table A3) are too low, since they assum e arand om distribution of the Canad ian-born in th e United States, while migration studieshave long shown that immigrants follow p athways blazed for them by their predecessors.

Thus, the CPS may have failed to find a representative sample of the Canadian-born intheir H ollywood , Silicon Valley, academ ic, medical, and Wall Street haunts.

%

ofCanadianpopulation

Canadian-born livingin the United States

US-born livingin Canada


27/64


The US full 2000 censu s will settle this m atter. In the m eantime, it is stillreasonable to estimate that the total number of Canad ian-born now living inthe United States, on either a temporary or a permanent basis, is unlikely tobe greater than it w as ten years ago. Despite an u nusually powerful constel-lation of forces encouraging southbound migration and some ind ication ofresurgence in the final years of the 1990s, it so far looks unlikely to have beenlarge enough to offset the long-established dow nw ard trend.

Similar pattern s exist for those with higher education, although they are

always more likely to be migrants, whether within the country or interna-tionally, and are an increasing share of the total and migrant populations ashigher edu cation becomes the norm rather than the exception.

Does a Brain Drain Exist?

What about the brain drain, the expression used to cover the loss of the bestand the br ightest Canadians? This issue led the Expert Panel on Skills (2000)

to ask me to survey the evidence, persuaded the UBC Alum ni Association tosurvey its US-resident graduates to find out w hy they had left Canada (Hel-

Figure 2: Canadian-Born Living in the United States, 19802000,

Actual versus 1990 Projections

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000

850

800

750

700

650

600

550

500

Note: The d ata used here include p erman ent and temp orary migrants; those for 1994 through1999 are based on a sample, rather than a full census count.

Sources: US 1980 and 1990 censuses; US Cur rent Popu lation Survey; p rojections from Long etal. 1991, p. 70, table 2.

thousands

census an dCPS da ta

projectionsmad e in 1990


28/64


liwell and Helliwell 2000a; 2000b), and induced the Conference Board ofCanada to conclude that the brain drain of Canadian professionals to theUnited States is real and cannot be ignored and that by 1997 the annual

number of permanent and non-permanent emigrants to the United Stateshad increased to 98,000 (Iqbal 1999).

Two recent C.D. How e Institute stud ies have also examined the subject,one concluding that the brain drain is real and is costing Canada (Devoretzand Laryea 1998) and the more recent that the south ward brain flow is morewor risome th an the stat istics at first su ggest (Schwan en 2000, 17).

What sort of data are of most use in assessing the issue at first and th enat second look? In d iscussions, some people tell me th at nu mbers are irrele-vant since shining stars are leaving, and the loss of a single star imposes acost that cannot be offset by any number of impor ted d rones, even high-techdrones with three degrees. Thus, any accounting based on conventionalqualifications may fail to be convincing. However, some combination ofqualifications and skill levels provides at least som e basis for seeing wh at ishappening.

Schwanen (ibid., table 3) provides an especially useful attem pt to estimatethe flows of science and engineering workers into and out of Canada an d th eUnited States. He show s that, wh ile both coun tries are increasing their stocksof trained workers in these occupations, the Canadian stock is increasingtw ice as fast as that of the United States. The faster grow th in Cana da ap pliesto all sources, including new domestic graduates and both permanent andtemporary immigrants. How ever, Canada continues to show here, as for itspopulation in total, emigration rates higher than those of the United Statesso that, although the Canadian net stock is increasing faster than that of theUnited States, the difference between the two countries is smaller than sug-gested by th e gross flows of new grad uates and new imm igrants.

Another useful source of data on the size and structure of the braindrain is information on the m ovement of graduates from Canadian u niver-sities. Two recent studies of the migration patterns of Canadian graduatesare ava ilable: a Statistics Canad a su rvey of all of the 1995 grad uates of Cana-dian universities, and the records of residence of a large majority of livingUBC graduates. The Statistics Canada data as supplemented by a specialsurvey of those in the graduating classes of 1995 who moved to the United

States (Frank and Blair 1999) provide a precise snapshot of the migrationpatterns for an entire cohort of graduates. The UBC data are drawn from asingle institution, but they cover graduates from more than 75 years. To theextent th at the tw o bodies of data can be show n to be consistent, they aremu tually enriching: the UBC data p rovide eviden ce of trend s, and their rep-resentativeness of what is hap pen ing in the coun try as a wh ole is reflected in


29/64


the Statistics Canada data. Helliwell and Helliwell (2000a; 2000b) show aremarkable consistency between data from the two sources, with the UBCfigures differing from the national figures in ways that reflect the universi-

tys western location and relative concentration of research and graduateeducation. Overall, the share of UBCs 1995 graduates living in the UnitedStates is about the same as for Canada as a whole; this finding results fromthe offset of the slightly larger Canadian -based grou p of bachelors grad uatesby the migration of UBC PhDs, who represent a larger proportion of totalgraduates for UBC than for Canada as a whole and are more likely to bemigrants both before and after their gradu ate education.

One of the interesting differences between the UBC and the nationaldata relates to the health professions, especially nursing. A widely reportedfeature of the Statistics Canada repor t is that fully one-fifth of the US-boun dmembers from the class of 1995 were nursing graduates. But the data fromUBC show that almost non e of its nursing grad uates had moved to the Unit-ed States. The reason for this difference is proba bly that the sup ply of nu rsesdep end s crucially on the m anagem ent of the health care systems, which dif-fer a great deal from province to province in the nature and especially thetiming of the p olicies und ertaken. Several prov inces, but not British Colum-bia, were sharply reducing health care and nursing budgets in 1995, andsince retrenchments are usually implemented in the form of freezes on newhires, the nu rsing class of 1995 in some p rovinces faced a jobless market intheir home provinces. The fact that so many Canadian nurses went fromOntario to US states with growing health care needs also probably explainswhy Texas and Florida were especially important targets for the nationalgraduates, but not for the UBC cohort. UBC nurses tended to remain inBritish Columbia; migrating graduates in other disciplines tended to con-centrate in Washington, California, and Massachusetts, wh ich are all centersfor higher education and high technology.

Figur e 3 traces a long series for UBC bachelors grad uates, includ ing allthose who later acquired advanced degrees, split among four locations:British Columbia, the rest of Canada, the United States, and the rest of thewor ld. The share resident in t he United States tracks very closely the declineshown in the total census d ata. Although Figures 1 and 3 captu re the sametrend s, they measu re different things, since the UBC graduates are all in the

subpop ulation with u niversity d egrees, a group that is always quite mobile,and are classified by th eir places of residen ce as of the end of the 1990s, withthe time axis measuring the year of graduation. As of the later 1990s, theshare of UBC graduates living in the United States was not the 2 percentapp licable to the 1990s grad uates or to the census d ata, but a w eighted av er-age of the figu res app licable to all vintages of the schools grad ua tes, which


30/64

Globalization: Myths, Facts, and Consequences 21

is 3.2 percent. (The num ber is skewed toward the lower end of the range, the

size of UBC grad uating classes has been sharp ly increasing.)Figure 4 show s the d istribution of UBC PhDs, illustrating the fact (alsoreflected in the Statistics Canada data for Canada as a whole) that the pro-portion who take positions in the United States is far higher for doctoralgraduates than for bachelors graduates. This pattern deserves more atten-tion, especially as some interp ret it as evidence that Canada is losing its bestand brightest. What is missing from that analysis is an aw areness of the dif-ferent structure and clientele of different degree programs. While mostprovincial universities willingly accept students from elsewhere in Canada

and overseas, the vast majority of undergraduates were born and raisedwithin the p rovince, a strong factor in their subsequent choice of residence.Most universities advise their best undergraduates to head elsewhere

for grad uate w ork, to broaden their exposure and experience, while recruit-ing the best graduate students they can find , searching both nationally andglobally. UBCs PhD programs have grown much faster than its undergrad-uate p rograms. Fewer than half the current stock of PhD students are Cana-dian citizens, and the rest come from more than a hundred different coun-tries. It is true, as show n in Figure 4, that abou t 15 percent of UBC PhDs take

positions in the United States, somewhat more than for Canada as a w hole.That 15 percent is greater than the 7 percent share of PhD students who come

Figure 3: Current Location of Bachelors Graduates

from the University of British Columbia100

80

60

40

20

0

1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995

British Colu mbia

Rest of Canad a

United States

Rest of wo rld

Source: Upd ated from H elliwell and Helliwell 2000a.

percentageofgraduatingclass

year of graduation


31/64


from the United States, reflecting not only the large size of the US market for

PhDs but also the reality that many students who come from overseas to doPhD w ork in Canada consider subsequent job opportun ities in m any coun-tries. The number of UBC PhDs who end up working in BC or elsewhere inCanada is actually greater than the nu mber w ho come from Canad a, so thatthe universitys PhD program increases the net inflow of trained brains.

Why Do Individuals Stay or Migrate?

Thus far, I have described the data for movements of bodies and minds.What makes them stay or move? The role of taxes, salary levels, and jobopportunities is much discussed, but we have too little in the way of goodevidence. Equations explaining the d istribution of UBCs bachelors, masters,and PhD graduates among states and provinces show very large bordereffects and significant effects from differences in distance and averageincomes. The pattern for graduates at the three levels is intriguing. As thelevel of degree rises, the measu red effects for d istance and income both d rop,revealing that h igher and more specialized training gives students a thinnerbut m ore far-flung range of career op tions. To stay within ones specialty, onehas to go w here the jobs are, and these positions may often be far aw ay. Andas the research becomes m ore specialized, ones colleagues are more likely to

Figure 4: Current Location of PhD Graduates

from the University of British Columbia

0

20

40

60

80

100

1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997

British Colum bia

Rest of Canad a

United States

Rest of world

Source: Upd ated from H elliwell and Helliwell 2000a.

p

ercentageofgraduatingclass

year of graduation


32/64


be found through out th e world, p roviding contacts and job possibilities that,in their turn , are mor e far-ranging.

Har d evid ence on the effects of taxes is in short supp ly. A stud y by the

Conference Board of Canada (Iqbal 1999) apparently finds that tax rates havea large effect on the nu mber of temporary and perman ent US visas issued toCanadians. Unfortunately, the visa series is heavily compromised by theinclusion of renew als, and the rising Canad ian tax rates were correlated w iththis growing series of visas. Survey evidence (Frank and Blair 1999; Helli-well and Helliwell 2000b) tends to show that migrants list job opportunitiesfirst, salaries second, and taxes lower among the reasons for moving, butthese response are impressions and are hard to convert into nu merical esti-mates of what w ould happ en if there were changes in the level or structureof Canadian tax rates. Also suggestive is the Statistics Canad a evaluation ofthe numbers of previously Canadian taxfilers who move their official resi-

den ce for tax pu rposes to th e United States (Zhao, Drew , and Mur ray 2000).As might be expected, these movers are heavily skewed tow ard those withhigh incomes, since (as discussed in the next section) the high end is wherethe income and tax gaps have recently been largest.

A recent stu dy by Don Wagn er (2000) uses an imaginative matching ofUS and Canadian census data to use individual data to try to isolate theeffects from both incomes an d tax rates. His preliminary resu lts show signif-icant effects of differences in both incomes and tax rates. If his equ ations areused to answer the question, how many fewer Canadian residents wouldhave m oved to th e United States since 1990 had the tw o countries had iden-tical tax structures, the answer is that equality of average tax rates wouldhave reduced the southbound migration of degree holders by 10 percentfrom the alread y small base show n in Figure 2. This result aw aits confirma-tion, but it is, in th e mean time, the best evid ence available on th e effects ofincome tax differences on migration between Canada and the United States.

Zhao, Drew and Murray (2000) report data showing an upward trendduring the 1990s in the number of taxfilers leaving Canada, with the distri-bution skewed to the upper end. Since tax residence can, in some cases, bechanged w ithou t altering employm ent or ph ysical residen ce, this shift of taxbase is likely to be a consequence of the high-income tax gaps that Wagnerfind s to h ave had significant effects on 1990s migra nts.

What other d ata are available to measure the size and structure of thebrain d rain? Canadian alum ni of the H arvard Business School have checkedthe add resses of different cohorts and find that an increasing shar e is remain-ing in the United States.10 To some extent, this rise is probably in respon se

10 Personal communication.


33/64


to income and tax differences and especially to the super-hot top-end jobmarkets in the United States.

I suspect that another special factor is at play here: business schools,

especially the elite ones, have increasingly become high-level employmentagencies, with the stud ents paying h efty fees that are rep aid to th em by theirsubsequent employers. For the students, the business school is a good wayto make new contacts and change jobs, while employers use the schools astalent spotters. If evidence should show this pattern to be increasinglyimp ortan t for business schools, each will be increasingly likely to attra ct stu-dents w ho w ish to find em ployment in th e market it serves. Thus, would -beMBAs who w ant to w ork in Canada w ill be more likely to go to a Canad ianbusiness school, while those wanting to find employment in the UnitedStates w ill be much m ore likely to head there for their MBA stud ies. This ten-dency would alter, but probably not reverse, the long-standing fact that stu-dents taking education abroad are frequently drawn back to their homecountries, either immediately on completion of their education or at somelater stage in their careers.

Globalization and Welfare

It is time to make at least a brave attemp t to bring together a broader rangeof evidence on economic and social welfare. Research on these linkages isstill fragmentary , but the issues raised are imp ortant enou gh to justify a tripthrough some of the thickets. I consider separately here incomes, health,education and knowledge, and social capital. All are affected by globaliza-tion in complex wa ys, and all have dual roles in econom ic and social welfare.

If one takes a narrow view of economic welfare, the main measure ofsuccess is some measure of average real incomes, presumably broadenedfrom gross national product per capita to allow for inclusion of frequentlyunm easured components. In this narrow view, health, edu cation and knowl-edge, and social capital are important on ly to the extent that th ey play instru -mental roles in helping to achieve or maintain high levels of per capitaincomes. If, in contrast, one adop ts a broad er measu re of welfare, wheth er itis based on composite indicators like the United Nations Development Pro-grammes (UNDPs) Human Development Index, various other extended

measu res of econom ic and social pr ogress (Osberg and Sharp e 2000), or self-assessed measures of happ iness, then both the levels and the d istribution ofemployment and of incomes, edu cation and knowledge, health, and socialcapital have direct as well as instrumental importance. Where any of thebroader ap proaches are taken, then attainment of better levels and distribu-tions of outcomes in th ese other aspects of life are given value for their own


34/64


sake in add ition to w hatever direct economic advantages they m ay bring intheir wake.

In what follows, I try to deal with h ow globalization has influenced the

levels, distributions, and consequences of each of these variables in turn,paying special attention to whether and how it has altered the extent towhich national or local outcomes remain responsive to local policies andinstitutions. Only with some view on these facts can we consider the m eritsof alternative policy responses at the national level and to make predictionsabout the pr ospects for Canada and Canadians.

Incomes

Globalization, via increasing international trade and investment, is oftencharged w ith responsibility for grow ing income inequality w ithin the Un it-

ed States and w ith losses of jobs and incomes amon g those w ith lesser skills.On the other hand, openness is also credited with giving poorer countriesthe chance to catch up to the levels of efficiency, and hence to the averageincome levels, already foun d in the richer indu strial coun tries. For examp le,Sachs and Warner (1995) pr esent eviden ce showing that, if developing coun-tries are divided into two grou ps, one open an d the other closed to foreigngoods, services, and ideas, the past 30 years have seen growth and conver-gence among the open economies and no evidence of convergence amongthose that rem ained closed. Yet openness m ay eventually have d iminishingreturns (or else we wou ld find much h igher levels of incomes per capita in larg-er countries, given th e strong p ersistence of border effects already d escribed).

Furthermore, openness itself cannot be a sufficient condition forgrowth, or else Russia, Ukraine, and other countries of the former USSRwould have converged quickly to Western levels of efficiency, instead ofstagnating or worse during the 1990s. Their experience shows that many ofthe institutions and social structures taken for granted in most OECD coun-tries are critically imp ortant su pp orts for econom ic developm ent, as w ell asfor other components of economic and social health. Where an institutionalvacuum exists, openn ess may simp ly attract and foster organized crime andother forms of criminal exploitation, which always bloom more easily thando legitimate business and society where social and legal norm s are lacking.

The debate between globaphobes and globaphiles is not settled, and therelative import ance of trad e and skills-biased technical chan ge as sources ofgrow ing inequ ality of incomes is still not established. For the United States,the strongest point in favor of the skills-biased technical change argumenthas been that the level and changes in trade are not large enough to havecaused the changes in income d istribution.


35/64


I agree with that conclusion, but my reasoning differs from that of theusual proponents. In the light of the evidence presented above about thecontinuing importance of national economies, I am inclined to look for

domestic causes and consequences of changes in inequality. If I am righthere, then I expect to find substantial differences among countries experi-ences and to be able to find plausible explanations for these differences.There are indeed large international differences but also some common ele-ments in w hat has h app ened to various kinds of inequality, with the Un itedStates as the outlier and Canada in middle ground, both before and afterallowing for the important effects of taxes and transfer payments.

Income taxes and transfer paym ents are generally higher and more p ro-gressive in Canada th an in the United States. App lying each coun trys socialsafety net p rovisions to ind ividual d ata for both countries, Blank and Han-

ratty (1993, 210) estimate that app lying Canad ian pr ogram s to the US pop u-lation in the m id-1980s wou ld have cut the US after-tax poverty rate by m orethan h alf, while applying the US program param eters to Canad as pop ula-tion wou ld have increased th eir poverty rate by more than half.

The two systems have somewhat converged since that time, but itremains true that the combination of taxes and transfers works to produceafter-tax inequality that is significantly less in Canada than in the UnitedStates. Wolfson and Murphy (2000) compare the inequality of market earn-ings and of family disposable income in Canada and the United States forthe 197485 and 198597 periods. They find increases in US earningsinequality in both p eriods, and the increases were mu ch larger for men th an

for women. For both men and wom en, the measures of earnings inequalitywere lower in Canada than in the United States in 1974 and the subsequentincreases w ere smaller, so that the int ercountry gaps grew from 1974 to 1997,when they were largest for higher-earning men.

That earnings inequality should have risen faster in the United Statesthan in Canada is partly explained by the results of Murphy, Riddell, andRomer (1998), who n ote that the ratio of earnings of u niversity grad ua tes rel-ative to those of high school grad uates rose sharply in the Un ited States andfell slightly in Can ada d uring th e 1980s and 1990s. They successfully explainthe d iffering path s of the education p remium in terms of a relative supp lyvariable (the increase in the supply of those with higher education has beengreater in Can ada than in the United States). Lemieux (1993) offers anoth erreason for the rising inequality of earnings in the United States, especiallyamong men. He estimates that the drop in US unionization rates du ring hissamp le period explains more than tw o-fifths of the increase in wage inequal-ity in the United States relative to that in Cana da.


36/64


Gottschalk and Smeeding (1997) confirm that earnings inequality issmaller in Canada than in the United States and shows less evidence ofincrease, but that both countries have m ore earnings inequality than is found

in most oth er OECD countr ies. A similar story can be told abou t the ratio ofexecutive compensation to the pay of production workers, which is muchhigher and has risen much faster in the United States than in Canada(Abowd and Kaplan 1999). The midway designation of Canada applies heretoo, as both countries have relative executive compensation levels that arehigher than those typically found in other OECD countries.

For family disposab le income, Canadian inequality fell both from 1974to 1985 and from 1985 to 1997, while it rose dur ing both period s in the Unit-ed States (Wolfson and Murphy 2000). This difference confirms the resultsreport ed earlier suggesting that the inequ ality-redu cing effects of the tax andtransfer systems are mu ch larger in Canad a than in the United States.

Os

Global is at Ion and Canada

Documents

Transcript of Global is at Ion and Canada