Research Observer EDITOR Shantayanan Devarajan, World Banlc COEDITOR Gershon Feder,World Bank EDITORIAL BOARD Paul Collier,Oxford University Susan Collins,Georgetown University Angus Deaton, Princeton University Barry Eichengreen, University of California-Berkeley EmmanuelJimenez, World Banlc Benno Ndulu, World Banlc Howard Pack, University of Pennsylvania Luis Serven, World Bank Michael Walton, World Banlc The World Bank Research Observer is intended for anyone who has a professional interest in development. Observer articles are written to be accessible to nonspecialist readers; con- tributors examine key issues in development economics, survey the literature and the lat- est WorldBank research, and debate issuesof development policy.Articlesare reviewed by an editorial board drawn from across theBank and the international community of econo- mists. Inconsistency with Banlc policy is not grounds for rejection. The journal welcomeseditorial comments and responses,which will be consideredfor pub- lication to the extent that space permits. On occasion the Observer considers unsolicited contributions. Any reader interested in preparing such an article is invited to submit a proposal of not more than two pages to the Editor. Please direct all editorial correspon- dence to the Editor,TheWorld Bank Research Observer, 1818 H Street, NW,Washington, DC 20433, USA. The views and interpretations expressed in this journal are thoseof the authors and do not necessarily represent the views and policiesof the World Bank or of itsExecutiveDirectors or the countries they represent.The World Banlc does not guarantee the accuracy of data included in this publication and accepts no responsibilitywhatsoever for any consequences of their use. When maps are used, the boundaries, denominations, and other information do not implyon the part of the WorldBanliGroup any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. For more information, please visit the Web sites of the Research Observer at www.wbro.oxfordjournals.org, the World Bank a t www.worl.dbank.org, and Oxford University Press a t www.oxfordjournals.org. Research Observer Volume 20 Number 2 Pall 2005 Globalization, Poverty, and Inequality since 1980 David Dollar Data and Dogma: The Great Indian Poverty Debate AngusDeaton and ValerieKozel Insights on Development from the Economics of Happiness CarolGraham Housing Policy i n Developing Countries: Conjectures and Refutations 233 RobertM. BuckleyandJerry Kalaricltal Public Debt Management and Macroeconomic Stability: An Overview 259 PeterJ. Montiel Subscriptions A subscription to The World Bank Research Observer (ISSN 0257-3032) comprises 2 issues. Prices include postage; for subscribers outside the Americas, issues are sent air freight. AnnualSubscriptionRate(Volume20,2issues,2005) Academiclibraries Print edition and site-wide online access: US$103/£70 Print edition only: US$98/£67 Site-wideonline access only: US$93if.63 Corporate Print edition and site-wide online access: US$128/£85 Print edition only: US$122/£81 Site-wide online access only: US$116/£77 Personal Print edition and individual online access: US$38/£29 Please note: f.Sterling rates apply in Europe, US$ elsewhere. Readers wit11mailingaddressesin non-OECDcolintriesand in socialist economiesin transitionare eligible to receivecomplimen- tary subscriptions on reqnestby writing to the UKaddress below. There may be other subscription rates available: for a complete listing, please visit www.wbro.oxfordjournals.org/ subscriptions. Fullprepayment in thecorrect currency is required for all orders. Ordersare regarded as f m ,and payments are not refund- able. Subscriptions are accepted and entered on a complete volume basis. Claims cannot be considered more than four months akerpublication or date of order, whicheverislater. Allsubscriptionsin Canada aresubject toGST.Subscriptions in the EU may besubject toEuropean VAT. If registered, pleasesupply details to avoid unnecessary charges.Forsubscriptions that include online versions,a proportion of thesubscription price may be subject to UK VAT. Personal rates are applicable only when a subscription is for individual use and are not available if deliveryis made to a corporate address. Back issues:The current year and two previous years' issues are available from Oxford University Press. Previous volumes can be obtained from the Periodicals Service Company. 11Main Street,Germantown. NY12526. USA.E-mail: psc@periodicals.com.Tel: (518) 537-4700. Fax: (518) 537-5899. Contact information: Journals Customer Service Department, Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, UK. E-mail: jnls.cust.serv@oupjournals.org. Tel: +44 (0)1865 353907. Fax: +44 (0)1865 353485. In theAmericas,pleasecontact:JournalsCustomer ServiceDepartment. Oxford University Press, 2001 Evans Road. Cary, NC 27513, USA. E-mail: jnlorders@oupjournals.org. Tel: (800) 852-7323 (toll-free in USAICanada) or 1919) 677-0977.Fax: (919) 677-1714. In Japan,please contact:Journals Customer ServiceDepartment. Oxford University Press, 1-1-17-5F,Mukogaoka, Bunkyo-ku, Tokyo, 113-0023,Japan. E-mail: oltudaoup@po.iijnet.or.jp. Tel: (03) 3813 1461.Fax: (03) 3818 1522. Postal information:The WorldBank Research Observer(ISSN0257-3032) is published by Oxford University Press for the International Bank for Reconstmction and Developrnent/~mWORID BANK. Send address changes to The World Bunk Research Observer,JournalsCustomer ServiceDepartment, Oxford University Press, 2001Evans Road, Cary, NC 275 13- 2009. Communications regarding original articles and editorial management should be addressed to The Editor, The WorldBankResearch Observer,The WorldBank. 1818 HStreet, NW, Washington,D.C. 20433, USA. Permissions:For information on how to request permissions to reproduce articles or information from this journal, please visit www.oxfordjournals.org/jnls/pemissions. Advertising:Inquiries about advertising should be sent to Helen Pearson, Oxford Journals Advertising, PO Box 347, Abingdon OX14 lGJ,UIC. E-mail:helen@oxfordads.com. Tel: +44 (0)1235201904. Fax: +44 (0)8704 296864. Disclaimer:Statements of fact and opinion in the articlesin The World Bank Research Observerare those of the respective authors and contributors and not of the International Bank for Keconstruction and development/^^^ woRm BANK or Oxford University Press. Neither Oxford University Press nor the International Bank for Reconstruction and Develop- m e n t / ~ ~ ~BANK make any representation, express or implied, in respect of the accuracy of the material in this jour- WORLD nal and cannot accept any legal responsibility or liability for any errors or omissions that may be made. The reader should make her or his own evaluation as to the appropriateness or otherwise of any experimental technique described. Paper used:The World Bank Research Observeris printed on acid-free paper that meets the minimum requirements of ANSIStandard 239.48-1984 (Permanenceof Paper). Indexingandabstracting:The WorldBankResearch Observeris indexed andlor abstracted by ABIIINFORM,Abstracts, CAB Current ContentslSocialand BehavioralSciences,Journd of EconomicLiteraturelEconLit,PAISInternational. KePEc(Research in EconomicPapers), SocialServicesCitationIndex, and Wilson BusinessAbstracts. CopyrightOThe International Bank for Reconstruction and Development/~mW O R B~ANK 2005 Allrights reserved;no part of thispublicationmay bereproduced,storedin a retrievalsystem,or transmitted in anyformor by any means, electronic,mechanical, photocopying,recording, or otherwise without prior written permission of the publisher or a licensepermitting restrictedcopyingissued in the UK by theCopyright LicensingAgencyLtd,90 Tottenham CourtRoad, LondonW1P9HE,or in the USA by the Copyright ClearanceCenter, 222 RosewoodDrive,Danvers,MA 01923. Globalization, Poverty, and Inequality since 1980 David Dollar One of the most contentiousissues of globalizationis the effect of global economicintegra- tion on inequality and poverty. This article documentsfive trendsin the modern era of 810- balization. starting around 1980. The first trend is that growth rates in poor economies have accelerated and are higher than growth rates in rich countries for the first time in modern history. Developing countries' per capita incomes grew more than 3.5 percent a year in the 1990s. Second, the number of extremely poor people in the world has declined significantly-by 375 million people since 1981-for the time in history. The share of peoplein developingeconomiesliving onless than $1 a day has been cut in half since 1981, though the decline in the share living on less than $2 per day was much less dramatic. Third, global inequality has declined modestly, reversing a 200-year trend toward higher inequality. Fourth,within-country inequalityin general is not growing,thoughit has risen in several populous countries (China, India, the United States). Fifth, wage inequality is rising worldwide. This may seem to contradict the fourth trend, but it does not because there is no simplelink between wage inequality and household incomeinequality. Further- more, the trends toward faster growth and poverty reduction are strongest in developing economies that have integrated with the global economy most rapidly, which supports the view that integrationhas beena positive force for improving the lives ofpeoplein developing areas. Globalization has dramatically increased inequality between and within nations. -JayMazur(2000) Inequality is soaring through the globalization period, within countries and across countries. And that's expected to continue. OThe Author 2005. Published by Oxford UniversityPresson behalf of theInternational Balnlrfor Reconstruction and Development / m WORW BANK. All rightsreserved.For permissions,pleasee-mail:journals.perrnissions@oupjoupjournals.org. doi:10.1093/wbrolllciOO8 20:145-175 All the main parties support nonstop expansion in world trade and services although we alllinow it...makesrich peoplericher and poor peoplepoorer. -WalterSchwarz,TheGuardian We are convinced that globalization is good and it's good when you do your homeworli.. .keep your fundamentals in line on the economy, build up high levels of education, respect rule of law.. .when you do your part, we are convinced that you get the benefit. -PresidentVicenteFoxofMexico There is no way you can sustain economic growth without accessing a big and sustained market. -PresidentYoweriMuseveniofUganda We take the challenge of international competition in a level playing field as an incentive to deepen the reform processfor the overall sustained develop- ment of the economy. w o membership worlis lilie a wreclcingball,smashing whatever is left in the old edifice of the former planned economy. -Jin Liqun, Vice Minister of Finance of China There is an odd disconnect between debates about globalization in developed econo- mies and developing economies. Among intellectuals in developed areas one often hears the claim that global economic integration is leading to rising global inequal- ity-that is, that integration benefitsrich people proportionally more than poor peo- ple. In the extreme claims poor people are actually made out to be worse off absolutely (as in the epigraph from Schwarz). In developing economies, though, intellectuals and policymakers often view globalization as providing good opportuni- ties for their countries and people. To be sure, they are not happy with the current state of globalization. The epigraph from President Yoweri Museveni, for example, comes from a speech that blasts rich countries for their protectionism against poor countries and lobbies for better marliet access. But the point of these critiques is that integration-through foreign trade, foreign investment, and immigration-is basi- cally a good thing for poor countries and that rich countries could do a lot more to facilitate integration-that is, make it freer. The claims from antiglobalization intel- lectuals in rich countries, however, lead inescapably to the conclusion that integra- tion is bad for poor countries and that therefore trade and other flows should be more restricted. The first goal of this article is to document what is lmown about trends in global inequality and poverty over the long term and during the recent wave of globaliza- tion that began around 1980. Global inequality is used to mean different things in 146 The World Bank Research Observer, vol. 20, no. 2 (Fall2005) differentdiscussions-distributionamong all the citizens of the world, distribution within countries, distribution among countries, distribution among wage earners- all of which are used in this article. A second goal of the article is to relate these trends to globalization. The first section briefly discusses the growing integration of developing economies with industrializedcountries and witheach other,starting around 1980. The opening of large developingcountries,such as China and India, is arguably the most distinctive feature of this wave of globalization. The second section, the heart of the article, pre- sents evidencein support of five trends in inequality and povertysince 1980: Growth rates in poor countries have accelerated and are higher than growth rates in rich countriesfor the first time in modern history. The number of extremely poor people (those living on less than $1 a day) in the world has declined significantly-by 375 million people-for the first time in history, though the number living on less than $2 a day has increased. Global inequality has declined modestly, reversing a 200-year trend toward higher inequality. Within-country inequality is generally not growing. Wage inequality is rising worldwide. This may seem to contradict the fourth trend, but it does not because there is no simple link between wage inequality and household income inequality. The third section then tries to draw a link between the increased integration and accelerated growth and poverty reduction. Individual cases, cross-country statisti- cal analysis, and micro-evidence from firms all suggest that opening to trade and direct investment has been a good strategy for such countries as the China, India, Mexico, Uganda, and Vietnam. The conclusions for policy in the fourth section are very much in the spirit of the comments from Presidents Fox and Museveni. Devel- oping economies have a lot to do to develop in general and to make effective use of integration as part of their development strategy. Rich countries could do a lot more with foreign aid to help with that work. As Museveni indicates, access to markets in rich countriesis important. Alot of protections remain in Organisation for Economic Co-operationand Development (o~m)marketsfromthe goods and people of developing economies, and globalization would work much better for poor people if developing areas had more access to those markets. Growing Integration between Developed and Developing Economies Global economic integration has been going on for a long time. In that sense, global- ization is nothing new. What is new in this most recent wave of globalization is the David Dollar L way developing countries are integrating with rich countries. As in previous waves of integration, this change is driven partly by technological advances in transport and communications and partly by deliberate policy choices. Earlier Waves of Globalization From 1820 to 1870 the world had already seen a fivefold increase in the ratio of trade to gross domestic product (GDP) (table I). Integration increased further in 1870-19 14,spurred by the development of steam shipping and by an Anglo-French trade agreement. In this period the world reached levels of economic integration comparable in many ways to those of today. The volume of trade relative to world income nearly doubled from 10 percent in 1870 to 18 percent on the eve of World War I. There were also large capital flows to rapidlydeveloping parts of the Americas, and the ownership of foreign assets (mostlyEuropeans owning assets in other coun- tries) more than doubled in this period, from 7 percent of world income to 18 per- cent. Probably the most distinctive feature of this era of globalization was mass migration. Nearly 10 percent of the world's population permanently relocated in this period (Williamson 2004). Much of this migration was from poor parts of - Table 1. Measures of Global Integration Capital Flows Trade Flows Transport and Communications Costs (constant US$) Sea Freight Telephone Call (average ocean Air Transport (average price for a Foreign Assets1 freight andport (average revenue 3-minute call between Year World GDP (%) Tradelc~p(%) chargesper ton) perpassenger mile) New York and London 1820 - 2 - - - 1995 56.8 - - - - Source: Crafts (2000)for capital flows: Maddison (1995)and Crafts (2000)for trade flows;World Banlr (2002) for transport and communications costs. -notavailable. 148 The World Bank Research Observer, vol. 2 0 , no. 2 (Fall2 0 0 5 ) Europe to the Americas. But there was also considerable migration from China and India (much of it forced migration in India). While globalindicators showed consid- erable integration in 1870-1914, this was also the heyday of colonialism, and most ofthe world's people were greatly restricted in their opportunities to benefit from the expanding commerce. Global integration took a big step backward during the two world wars and the Great Depression. Some discussions of globalization today assume it was inevitable, but this dark period is a powerfulreminder that policies can halt and reverse integra- tion. By the end of this dark era both trade and foreign asset ownership were back close to their levels of 1870-the protectionist period undid 50 years of integration. The era of free migration was also at an end, as virtually all countries imposed restrictions on immigration. From the end of World War I1 to about 1980, industrialized countries restored much of the integration that had existed among them. They negotiated a series of mutual trade liberalizations under the auspice of the General Agreement on Tariffs and Trade. But liberalization of capital flows proceeded more slowly, and not until 1980 did the levelof ownership of foreign assets returned to its 1914 level. Over this period there was also modest liberalization of immigration in many industrialized countries, especially the United States. In this postwar period of globalization,many developing economies chose to sit on the sidelines. Most developing areas in Asia, Africa, and Latin America followed import-substituting industrialization strategies, keeping their levels of import protection far higher than in industrialized countries to encourage dlomestic production of manufactures and usually restricting foreign investment by multinationalfirms to encourage the growth of domesticfirms. While limiting direct investment, several developing economies turned to the expanding international bank borrowing sector in the 1970s and took on significant amounts of foreign debt. Recent Wave of Globalization The most recent wave of globalizationstarted in 1978 with the initiation of China's economic reform and opening to the outside world, which roughly coincides with the second oil shock, which contributed to external debt crises throughout Latin America and in other developing economies. In a growing number of countries in Latin America, South Asia, and Sub-Saharan Africa politicaland intellectualleaders began to fundamentally rethink development strategies. The distinctive part of this latest wave of globalization is that the majority of developingeconomies (in terms of population) shifted from an inward-focused strategy to a more outward-oriented one. This altered strategy can be seen in the huge increases in trade integration of developingareas over the past two decades.China's ratio of trade to national income David Dollar 149 has more than doubled, and countries such as Mexico, Bangladesh, Thailand, and India have seen large increases as well (figure1).But several developingeconomies trade less of their income than two decades ago, a fact that will be discussed later. The change has not been only in the amount, but also in the nature of what is traded. Twenty years ago, nearly 80 percent of developing country merchandise exports were primary products: the stereotype of poor countries exporting tin or bananas had a largeelement of truth. The big increase in merchandise exports in the past two decades, however, has been of manufactured products, so that 80 percent of today's merchandiseexports Gom developing countries are manufactures (figure2). Garments from Bangladesh, CD players from China, refrigerators from Mexico, and computerperipheralsfromThailand-these are the modernfaceof developingeconomy exports. Service exports from developing areas have also increased enormously- both traditional services,such as tourism, and modern ones, such as software from Bangalore, India. Manufactured exports from developingeconomies are often part of multinational production networks. Nike contracts with firms in Vietnam to make shoes; the "world car" is a reality, with parts produced in different locations. So part of the answer to why integration has taken off must lie with technological advances that Figure 1. Changein Tradeas a Share of DP, SelectedCountries,1977-97 (%) G China Argentina Philippines Malaysia Bangladesh Thailand India Brazil Pakistan Kenya Togo Honduras Senegal Nigeria Egypt Zambia -100 0 100 Source: World Bank (2002). 150 The World BankResearch Observer. vol. 20, no. 2 (Fall 2005) Figure2. DevelopingCountry ExportsbySector,1965-99 (% of total) Source: World Bank (2002). malieintegrated production feasible(seetable1for evidenceof the dramatic declines in the cost of air transport and international communications). But part of the answer also lies in policy choices of developing economies. China and India had almost totally closed economies, so their increased integration would not have been possible without steps to gradually liberalizetrade and direct foreign investment. Some measure of this policy trend can be seen in average import tariff rates for developing economies. Since 1980 average tariffs have declined sharply in South Asia, Latin America and the Caribbean, and East Asia and Pacific, whereas in Africa and the Middle East there has been much less tariff-cutting (figure 3).These reported average tariffs, however. capture only a small amount of what is happening with trade policy. Often the most pernicious impediments are nontariff barriers: quotas, licensing schemes, restrictions on purchasing foreign exchange for imports, and the like. China started to reduce these nontariff impediments in 1979, which led to a dramatic surge in trade (figure4).In 1978 external trade was monopolizedby a sin- gle government ministry.' Specific measures adopted in China included allowing a growing number of firms,including private ones, to trade directly and opening a for- eign exchange market to facilitate this trade. Another major impediment to trade in many developing areas is inefficient ports and customs administration. For example, it is much more expensive to ship a con- tainer of textiles from a Mombasa, Kenya, port to the East Coast of the United States David Dollar 151 Figure 3. AverageUnweightedTariff Rates,by Region,1980-98 (%) South Latin America East Asia Sub- Middle East and Europe and Industrialized Asia and the and the Saharan North Africa Central economies Caribbean Pacific Africa Asia Source: World Bank (2002). Figure 4. TradeReforms and Volumesin China,1978-2000 TradelGDP (log) Average tariff Average tariff rate 1 ! 1 o.4 , , I984 tradi; corn: , , , , , 0.1 . 11979 Open specia!economic zones to FDI, foreign exchange retention :1 I 11978 Trade monopolizedby Ministrybf ForeignEconomicRelationsand Trade 0 : 0 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 Source: Dollar and Kraay (2003). 152 The WorldBank Research Observer. vol. 20. no. 2 (Fall2005) than from Asian ports such as Mumbai, Shanghai,Bangkok, or Kaohsiung, Taiwan (China),even though Mombasa is closer (Clark and others2004). The extra cost, equivalent to an 8 percent export tax, is due to inefficiencies and corruption in the port. Long customs delays often act as import and export taxes. Developing econo- mies that have become more integrated with the world economy have reasonably well-functioning ports and customs, and their improvement has often been a delib- erate policy target. Several countries, including Kenya, trade less of their income today than 20 years ago; surely this is partly the result of restrictive trade policies, defined broadly to include inefficient ports and customs. Thus, one lcey development in this current wave of globalization is a dramatic change in the way many developingcountriesrelatetothe globaleconomy.Developing economies as a whole are a major exporter of manufactures and services-many of which compete directly with products made in industrializedcountries. The nature of trade and competition betweenrich and poorcountries hasfundamentally changed. Accelerated Growth and PovertyReduction i n Developing Econoniies Some of the debate about globalization concerns its effects on poor countries and poor people. The introduction quotes several sweeping statements that assert that global economic integration is increasing poverty and inequality in the world. But the reality is far more complex-and to some extent runs exactly counter to what is being claimed by antiglobalists. Thus, this section focuses on the trends in global poverty and inequality, and the following section links them to global integration. The trends of the last 20 years highlighted here are: Growth rates of developing economies have accelerated and are higher than those of industrialized countries. The number of extremely poor people (those living on less than $1a day) has declined for the time in history, though the number of peoplelivingon less than $2 a day has increased. Measures of global inequality (such as the global Gini coefficient)have declined modestly, reversing a long historical trend toward greater inequality. Within-country inequality in general is not growing, though it has risen in sev- eral populous countries (China,India, the United States). Wage inequality in general has been rising (meaning larger wage increases for skilledworlrers than for unslcilledworkers). The fifth trend may seem to run counter to the fourth trend; why it does not will be explained here. The fifth trend is important for explaining some of the anxiety about globalizationin industrialized countries. Growth Rates in Developing Economies Have Accelerated Reasonably good data on economic growth since 1960 for about 100 countries that account for the vast majority of world population are summarized in the Penn World Tables (Center for International Comparisons 2004). Aggregating data on growth rates for industrialized countries and developingeconomies for which there are data since 1960 shows that in general growth rates have declined in rich coun- tries while accelerating in developing countries (figure 5). In particular, in the 1960s growth of OECD countries was about twice as fast as that of developing areas. Per capita growth rates in rich countries have gradually declined from about 4 per- cent in the 1960s to 1.7 percent in the 1990s-close to the long-term historical trend rate of the ~ E C D countries. The rapid growth in the 1960s was still to some extent a rebound from the destruction of World War I1 as well as a payoff to eco- nomic integration among rich countries. In the 1960s and early 19 70s, the growth rate of developingeconomies was well below that of rich countries, a paradox whose origin has been long debated. The slower growth of less developed economies was a paradox because neoclassical growth theory suggested that other things being equal poor countries should grow Figure 5. DP per CapitaGrowth Rate,by CountryType,1960s-1990s (%) G " 1960s 1970s 1980s 1990s Source:Centerfor International Comparisons(2004). The World BankResearch Observer, vol. 20,110. 2 (Fall 2005) faster.This pattern finallyemerged in the 1990s,with per capita growth in develop- ing countries of about 3.5 percent-more than twice the rate of rich countries. This high aggregate growth depends heavily on several large countries that were among the poorestin the world in 1980 but that have grown wellsincethen.Ignor- ing differences in population and averaging growth rates in poor countries over 1980-2000 result in an average growth of about zero for poor countries. China, India, and several small countries, particularly in Africa, are among the poorest quintile of countries in 1980. Ignoring population,the average growth of Chad and China is about zero, and the average growth of India and Togo is about zero. Accountingfor differencesin population,though, the average growth of poor coun- tries has been very good in the past 20 years. China obviouslycarries a large weight in any calculation of the growth of poor countries in 1980, but it is not the only poor country that did well: Bangladesh, India, and Vietnam also grew faster than rich countries in the same period. Several African economies, notably Uganda, also had accelerated growth. The Number of Extremely Poor People Has Declined by 375 Million Globally The most important pointin this section is that poverty reduction in low-incomecoim- triesisveryclosely relatedto theGDPgrowthrate.The acceleratedgrowthof low-income countries has led to unprecedented poverty reduction. By poverty I mean subsisting below some absolute threshold.Most poverty analyses are carried out with countries' own povertylines,which aresetincountrycontextand naturallydiffer. China, for example, uses a poverty line defined in constant Chinese yuan. The poverty line is deemed the minimum amount necessary to subsist. In practice,esti- mates of the number of poor in a country such as China come from household sur- veys carried out by a statistical bureau. These surveys aim to measure what households actually consume. Most extremely poor people in the world are peas- ants, and they subsist to a large extent on their own agricultural output. To look only at their money income would not be very relevant, because the extremely poor have only limited involvement in the money economy. Thus measures ask house- holds what they actually consumeand attach a value to their consumption based on the prices of different commodities.So a poverty line is meant to capture a certain real level of consumption. Estimating the extent of poverty is obviously subject to error, but in many countries the measures are good enough to pick up large trends. In discussing poverty it is important to be clear on the poverty line being used. In global discussionsinternational poverty lines of either $1 a day or $2 a day, calcu- lated at purchasingpower parity,are used.For discussions of globalpoverty a common line should be applied to all countries. Chen ancl Ravallion (2004)used household survey data to estimate the number of poor people worldwidebased on the $1 a day and $2 a day poverty lines back to David Dollar 155 1981. They found that the incidence ofextreme poverty (consuming less than $1 a day) was basically cut in half in 20 years, from 40.4 percent of the population in developing economies in 1981 to 21.1 percent in 2001. It is interesting that the decline in $2 a day poverty incidence was not as great, from 66.7 percent to 52.9 percent, over the same period. Poverty incidence has been gradually declining throughout modern history, but in general population growth has outstripped the decline in incidence so that the total number of poor people has actually risen. Even in 1960-80, a reasonably pros- perous period for developingeconomies, the number of extremely poor people con- tinued to rise (figure 6).2Most striking in the past 20 years is that the number of extremely poor people declined by 375 million,while at the same time world popula- tion rose by 1.6 billion. But the decline was not steady: in 1987-93 the number of extremely poor people rose, as growth slowed in China and India underwent an eco- nomiccrisis.After1993 growth and povertyreductionacceleratedin bothcountries. The1981-2001declinein the number of extremely poor peopleis unprecedented in human history. At the same time many of those who rose above the very low $1 a day threshold are still living on less than $2 a day. The number of people living on less than $2 a day increased between 1981 and 2001 by nearly 300 million. About half the world's population still lives on less than $2 a day, and it will take several more decadesof sustained growth to bring thisfigure down significantly. Figure6. ExtremePoverty in the World. 1820-2001(millionsof peopleliving on lessthan $1 a day) Source: Bourguignonand Morrisson (2002);Chenand Ravallion(2004). The World Bank Research Observer, vol. 2 0 , no. 2 (Fall 2 0 0 5 ) Although the overall decline in extreme poverty is positive news, performance has varied by region. South Asia and East Asia and Pacific grew well and reduced poverty,but Sub-Saharan Africa had negative growth between 1981 and 2001 and a rise in poverty: the number of extremely poor peoplethere increasedfrom 164 mil- lion (41.6percent of the population) to 316 million (46.9 percent of the population). Two-thirds of extremely poor peoplestill live in Asia, but if strong growth there con- tinues, global poverty will be increasingly concentrated in Africa. Global Inequality Has Declined Modestly Global inequality is casually used to mean several things, but the most sensible defi- nition is the same as for a country: line up all the people in the world from the poorest to the richest and calculate a measure of inequality among their incomes. There are several measures, of which the Gini coefficientis the best known. Bhalla (2002) estimates that the global Gini coefficient declined from 0.67 in 1980 to 0.64 in 2000 after rising from 0.64 in 1960.Sala-i-Martin (2002) likewise finds that all the standard measures of inequality show a decline in global inequality since 1980. Both Bhalla and Sala-i-Martin combine national accounts data on income or consumption with survey-based data on distribution. Deaton (2004) discusses the problems of using national accounts data for studying poverty and inequality, noting among other things that the growth rates in national accounts data for China and India are arguably overestimated. This bias would tend to exaggerate the decline in global inequality over the past 25 years. Hence, there is a fair degree of uncertainty about the magnitude of the estimated decline in global inequality. For historical perspective, Rourguignon and Morrisson (2002) calculate the global Gini coefficient back to 1820. Although confidence in these early esti- mates is not high, they illustrate an important point: global inequality has been on the rise throughout modern economic history. Bourguignon and Morrisson estimate that the global Gini coefficient rose from 0.50 in 1820 to about 0.65 around 1980(figure 7). Sala-i-Martin (2002) estimates that it has since declined to 0.61. Other measuresofinequality such as mean log deviation show a similar trend, ris- ing until about 1980 and then declining modestly after (figure 8). Roughly speak- ing, the mean log deviation is the percent difference between averageincome in the world and the income of a randomly chosen individual who represents a typical per- son. Average per capita income in the world today is around $5,000,but the typical person lives on 20 percent of that, or $1,000. The advantage of the mean log devia- tion is that it can be decomposed into inequality between countries (differencesin per capita income across countries) and inequality within countries. This decompo- sition shows that most inequality in the worldcan be attributed to inequality among David Dollar 157 Figure 7. GlobalGini Coefficient,1820-1997 Source: Bourguignon and Morrisson (2002):Sala-i-Martin(2002). countries. Global inequality rose from 1820 to 1980, primarily because already relatively rich countries (thosein Europe and North America)grew faster than poor ones. As noted in the discussion of the first trend, that pattern of growth was reversed starting around 1980, and the faster growth in such poor countries as Bangladesh, China, India, and Vietnam accounts for the modest decline in global inequality since then.4 (Slowgrowth in Africa tended to increase inequality, faster growth in low-income Asia tended to reduce it, and Asia's growth modestly out- weighed Africa's.) Thinking about the differentexperiencesof Africa and Asia,as in the last section, helpsgive a clearer pictureof what is likely to happen in the future. Rapid growth in Asia hasbeen a forcefor greater globalequality becausethat iswhere the majorityof the world's extremely poor people lived in 1980-and they benefited from growth. But if the same growth trends persist, they will not continue to be a force for equal- ity. Sala-i-Martin (2002) projects future global inequality if the growth rates of 1980-98 persist: global inequality will continue to decline until about 2015, after which global inequality will rise sharply (seefigure 8). A large share of the world's poor people still lives in India and other Asian countries, so that continued rapid growth there will be equalizing for another decade or so. But increasingly poverty will beconcentrated in Africa,so that if slow growth persists there, globalinequality will eventuallyrise again. The World Bank Research Observer, vol. 20, ?lo.2 (Fall2005) Figure 8. Global HouseholdInequality, 1820-2050 (meanlog deviation) Source: Bourguignon and Morrisson (2002):Sala-i-Martin (2002). Within-Country Inequality Is in General Not Growing The previous analysis shows that inequality within countries has a relatively small role in measures of global income inequality. But people care about trends in ine- quality in their own societies (arguably more than they care about global inequality and poverty). So a different question is what is happening to income inequality within countries. One common claim about globalization is that it leads to greater inequality within countries and thus fosterssocial and politicalpolarization. To assess this claim Dollar and Kraay (2002) collected income distribution data from more than 100 countries, in some cases going back decades. They found no general trend toward higher or lower inequality within countries. Focusing on the share of income going to the bottom quintile, another common measure of inequal- ity, they found increases in inequality for some countries (for example, China and the United States) in the 1980s and 1990s and decreases for others. They also tried to use measures of integration to explain the changes in inequality that have occurred, but none of the changes were related to any of the measures. For example, countries in which trade integration increased showed rises in inequality in some David Dollar 159 cases and declines in others (figure 9). They found the same results for other measures, such as tariffrates and capital controls. Particularly in low-income coun- tries, much ofthe import protection benefitedrelatively rich and powerfulgroups,so that integration with the global market went hand in hand with declinesin income inequality. It is widely recognized that income distribution data have a lot of mea- surement error, which makes it difficult to identify systematic relationships, but given the available data, there is no robust evidence that integration is systemati- cally related to higher inequality within countries. There are two important caveats to this conclusion. First, inequality has risen in several very populous countries, notably China, India, and the United States. This means that a majority of citizens of the world live in countries in which inequality is rising. Second, the picture of inequality is not so favorable for rich countries in the past decade. The Luxembourg Income Study, using comparable, high-quality income distribution data for most rich countries, finds no obvious trends in inequal- ity through the mid- to late 1980s. Over the past decade, through, inequality has increased in most rich countries. Becauselow-skilled workers in these countries now compete more with workers in developing economies, global economic integration can create pressure for higher inequality in rich countries while having effects in poor countries that often go the other way. The good news from the Luxembourg IncomeStudy is that "domesticpoliciesand institutions still have large effectson the level and trend of inequality within rich and middle-income nations, even in a glo- balizing world....Globalization does not force any single outcome on any country" Figure9. Correlationbetween Changein GiniCoefficient andChangeinTrade as a Shareof GDP - c .- m g 0 .- .- c a .- L3 a, m m 6 Change in trade to GDP Source: Dollar and Kraay (2002). 160 The World Bank Research Observer. vol. 20, no. 2 (Fa112005) (Smeeding 2002, p. 179). In other words, some rich countries have maintained stable income distributions in this era of globalization through their social and eco- nomic policies (on taxes, education, welfare,and the like). Wage Inequality Is Rising Worldwide Much of the concern about globalization in rich countries relates to worlcers, wages, and other labor issues. The most comprehensive examination of globaliza- tion and wages used International Labour Organization data from the past two decades (Freeman and others 2001). These data look across countries at what is happening to wages for very specificoccupations (forexample, bricklayer, primary schoolteacher, nurse, autoworker). The study found that wages have generally been rising fastest in more globalized developing economies, followed by rich countries, and then less globalizeddeveloping economies (figure 10).More global- ized developing economies are the top third of developing economies in terms of increased trade integration over the past 20 years (Dollar and Kraay 2004). Less globalized developing economies are the remaining developing economies. The fastest wage growth is occurring in developing economies that are actively increasing their integration with the global economy. Although the general rise in wages is good news, the detailed fmdingsfrom Free- man and others (2001)are more complex and indicate that certain types of worlcers benefit more than others. First, increased trade is related to a decline in the gender Figure 10. Wage Growth by CountryType,1980s-1990s(%) Less globalized Rich countries More globalized developing countries developing countries Source: Freeman and others (2001). David Dollar wage gap. More trade appears to lead to a more competitive labor market in which groups that have been traditionally discriminated against-women, for example- fare especiallywell (Oostendorp2002).Second,the gains from increased trade appear to be larger for sltilled worlters. This finding is consistent with other work showing a worldwide trend toward greater wage inequality-that is, a larger gap between pay foreducated worlters and pay forless educated and unsldled workers. Galbraith and Liu (2001), for example, find a worldwide trend toward greater wage inequality among industries. Wages in skull-intensive industries, such as aircraft production, have been going up fasterthan wages in low-skillindustries, such as garments. If wage inequality is going up worldwide, how can income inequality not be rising in most countries?There are several reasons. First, in the typical developing econ- omy wage earners make up a small share of the population. Even unskilled wage worlters are a relatively elite group. Talte Vietnam,for example, a low-income coun- try with a survey of the same representative sample of households early in liberaliza- tion (1993)and five years later. The majorityof households in the country (andthus in the sample) are peasants. The household data show that the price of the main agricultural output (rice) went up dramatically while the price of the main pur- chased input (fertilizer)actually went down. Both movements are related directly to globalization because over the survey period Vietnam became a major exporter of rice (raisingits price)and a major importer of fertilizerfrom cheaper producers (low- ering its price). Poor families faced a much bigger wedge between rice's input price and output price, and their real income went up dramatically (Benjamin and Brandt 2002). So, one of the most important forces acting on income distribution in this low-income country had nothing to do with wages. Several rural households also sent a family member to a nearby city to work in a factory for the first time. In 1989 the typical wage in Vietnamese currency was the equivalent of $9 a month. Today, factory workers malting contract shoes for U.S. brands often make $50 a month or more. So the wage for a relatively unskilled worker has gone up nearly fivefold. But wages for some skilled occupations, for example, computer programmers and English interpreters, may have gone up 10 times or more (Glewweand others 2004).Thus, a careful study of wage inequality is likely to show rising inequality. But how wage inequality translates into household inequality is very complex. For a surplus worker from a large rural household who obtains a newly created job in a shoe factory, earnings increase from $0 to $50 a month. If many new wage jobs are created, and if they typically pay much more than people earn in the rural or informal sectors, a country can have rising wage inequality but stable or even declining income inequality. (The Gini coefficient for household income inequality in Vietnam actually declined between 1993 and 1998, according to Glewwe 2004b.) In rich countries most household income comes from wages, but household income inequality and wage inequality do not have to move in the same direction. If 162 The World Bank ResearchObserver, vol. 2 0 , no. 2(Fall2005) there are changes in the way that people partner and combine into households, household inequality can rise even if wage inequalitystays the same. Another point about wageinequalityand householdincomeinequalityrelevant to rich countries is that measures of wage inequality are often made before taxes are talcen out of earn- ings.Ifthe country has a strongly progressiveincome tax, inequalitymeasuresfrom household data (which are often made after taxes are taken out of earnings) do not have to followpretax wageinequality.Tax policy can offsetsome of the trends in the labor market. Finally, households can respond to increased wage inequality by investing more in their children's education. A higher economic return to education is not a bad thing, as long as there is equal access to education for all. Vietnam saw a tremen- dous increase in the secondary school enrollment rate in the 1990s-from 32 per- cent in 1990-91 to 56 percent in 1997-98 (Glewwe 2004a). This increase partly reflects society's and the government's investment in schools (supported by aid donors)and partly reflects households' decisions.If little or no return to education is perceived (thatis,no jobs at the end of the road),it is much harder to convincefami- lies in poor countries to send their children to school. Where children have decent accessto education, a higher skill premiumstimulates a shift of the labor force from low-skill to higher-skilloccupations. It should also be noted that there has been a largedeclinein child labor in Vietnam since the country started integrating with the global market. There is ample evi- dence that child labor is driven primarily by poverty and educational opportunities. Child labor is more prevalent in poor households, but between 1993 and 1998 it declined for all income groups (figure 11). The change resulted from the fact that everyone was richer than they were five years earlier and from the expansion of schoolingopportunities. From this discussion of wage trends, it is easy to see why some labor unions in rich countries are concerned about integration with developingeconomies. It is difficult to provethat integrationisincreasingwage inequality,but it seemslikelythat integration is onefactor.Concerningthe immigrationside of integration,Borjasand others (1997) estimate that flows of unslulled labor into the United States have reduced wages for unskilledlabor by 5 percentfromwheretheyotherwisewould be.Immigrantswho find new jobs earn much more than they did before (10times as much, accordingto World Bank 2002), but their competition reduces the wages of U.S. workers already doing such jobs. Similarly,imports of garments and footwear from countriessuch as Bangladeshand Vietnamcreate jobsfor worlcersthat payfarmore than other opportu- nitiesin thosecountriesbut put pressureon unskilledwages in rich countries. Thus overall the era of globalizationhas seen unprecedentedreductionof extreme poverty and a modest declinein global inequality.But it has put real pressureon less skilled workers in rich countries-a key reason why the growing integration is con- troversialin industrializedcountries. Figure 11. Child Labor and Household Consumption Levels in Vietnam,1993 and 1998 I I I I I I I I I I 6.54 6.64 6.74 6.84 6.94 7.04 7.14 7.24 7.34 7.44 7.54 Per capita household consumption1993 (log scale) Source:Edmonds (2001). Is There a Link between Integratio~iand Poverty Reduction? To keep track of the wide range of explanations that are offered for persist- ent poverty in developing nations, it helps to keep two extreme views in mind. The first is based on an objectgap: Nations are poor because they lack valuable objects like factories, roads, and raw materials. The second view invokes an idea gap: Nations are poor because their citizens do not have access to the ideas that are used in industrial nations to generate economic value... Each gap imparts a distinctive thrust to the analysis of development policy. The notion of an object gap highlights saving and accumulation. The notion of an idea gap directs attention to the patterns of interaction and communication between a developingcountry and the rest of the world. -PaulRomer(1993) Developingeconomies have become more integrated with the global economy in the past two decades, and growth and poverty reduction have accelerated. A natural question is whether there is a link between the two. In other words, could countries 164 TheWorld Bank Research Observer. vol. 20. no. 2 (Fall2005) such as Bangladesh, China, India, and Vietnam have grown as rapidly if they had remained as closed to foreign trade and investment as they were in 1980?This can- not be answered with scientificcertainty, but several different types of evidence can be brought to bear on it. It is useful to begin with what to expect from economic theory. As the quote from Romer suggests, traditional growth theory focuses on accumulation and the "object gap" between poor countries and rich ones. If increasing the number of factories and workplaces is the only important action, it does not matter whether the environ- ment is closed or dominated by the state. This model was followed in the extreme by China and the Soviet Union, and to a lesser extent by most developing economies, which followed import-substituting industrialization strategies throughout the 1960s and 1970s. The disappointing results from this approach led to new thinking by policymakers in developing areas and economists studying growth. Romer was one of the pioneers of the new growth theory that emphasized how innovation occurs and is spread and the role of technological advance in improving the stan- dard of living. Different aspects of integration-sending students abroad to study, connecting lo the Internet,allowing foreign firms to open plants,purchasing the lat- est equipment and components4an help overcome the "idea gap" that separates poor countries from rich countries. What is the evidence on integration spurring growth?Some of the most compel- ling evidence comes from case studies that show how this process can work in par- ticular countries. Among the countries that were very poor in 1980, China, India, Uganda, and Vietnam provide an interesting range of examples: China China's initial reforms in the late 1970s focused on the agricultural sector and emphasized strengthening property rights, liberalizing prices, and creating internal markets. Liberalizing foreign trade and investment were also part of the initial reform program and played an increasingly important role in growth as the 1980s proceeded (seefigure 4).The role of international links isdescribedin a case study by Eckaus (1997,pp. 415-37): China's foreign trade began to expand rapidly as the turmoil created by the Cultural Revolution dissipated and new leaders came to power. Though it was not done without controversy, the argument that opening of the econ- omy to foreign trade was necessary to obtain new capital equipment and new technology was made official policy....Most obviously, enterprises created by foreign investors have been exempt from the foreign trade plan- ning and control mechanisms. In addition, substantial amounts of other types of trade, particularly the trade of the township and villageenterprises David Dollar 165 and private firms, have been relatively free. The expansion of China's par- ticipation in international trade since the beginning of the reform move- ment in 1978, has been one of the most remarkable features of its remarkabletransformation. India It is well lmown that India pursued an inward-oriented strategy into the 1980s with disappointing results in growth and poverty reduction. Bhagwati (1992, p. 48) crisplystates the main problems and failures of the strategy: I would divide them into three major groups: extensive bureaucratic cbn- trols over production, investment and trade; inward-looking trade and for- eign investment policies;and a substantial public sector, going well beyond the conventional confines of public utilities and infrastructure. Under this policy regime India's growth in the 1960s (1.4 percent a year) and 1970s (-0.3 percent) was disappointing. During the 1980s India's economic perfor- mance improved, but this surge was fueled by deficit spending and borrowing from abroad that was unsustainable.In fact,the spending spreeled to a fiscaland balance of payments crisis that brought a new, reform government to power in 1991.Srinivasan (1996,p. 245) describesthe key reform measures and their results: In July 1991, the government announced a series of far reaching reforms. These included an initial devaluation of the rupee and subsequent market determination of its exchange rate, abolition of import licensing with the important exceptions that the restrictions on imports of manufactured con- sumer goods and on foreigntrade in agriculture remained in place, convert- ibility (withsome notable exceptions) of the rupee on the current account; reduction in the number of tariff lines as well as tariff rates; reduction in excise duties on a number of commodities; some limited reforms of direct taxes; abolition of industrial licensing except for investment in a few indus- tries for locational reasons or for environmental considerations, relaxation of restrictions on large industrial houses under the Monopolies and Restric- tive Trade Practices (MRTP) Act; easing of entry requirements (including equity participation) for direct foreign investment; and allowing private investmentin someindustrieshitherto reservedfor publicsectorinvestment. In general, India has seen good results from its reform program, with per capita income growth above 4 percent a year in the 1990s. Growth and poverty reduction have been particularly strong in states that have made the most progress liberalizing the regulatory framework and providing a good environment for delivery of infra- structure services (Goswamiand others 2002). 166 The World Bank Research Observer. vol. 20, no. 2 (Fall 2005) Uganda Uganda has been one of the most successful reformers in Africa during this recent wave of globalization,and its experience has interesting parallelswith Vietnam's.It, too, was a country that was quite isolated economically and politically in the early 1980s. The roleof trade reform in itslarger reformcontext isdescribedin Collierand Reinildca (2001,pp. 30-39): Trade liberalization has been central to Uganda's structural reform program....In 1986 the NRM government inherited a trade regime that included extensive nontariff barriers, biased government purchasing, and high export taxes, coupled with considerable smuggling.The nontariff bar- riers have gradually been removed since the introduction in 1991 of auto- matic licensing under an import certification scheme. Similarly, central government purchasing was reformedand is now subjectto open tendering without a preference for domestic firms over imports....The average real GDP growth rate was 6.3 percent per year during the entire recovery period (1986-99) and 6.9 percent in the 1990s. The liberalization of trade has had a marked effect on export performance. In the 1990s export volumes grew (at constant prices) at an annualized rate of 15 percent, and import volumes grew at 13 percent. The value of noncoffee exports increased five- fold between1992 and 1999. Vietnam The same collection that contains Eckaus's (1997) study of China also has a case study of Vietnam, analyzing how the country went from being one of the poorest countries in the 1980s to being one of the fastest growing economies in the 1990s (Dollarand Ljunggren1997, pp. 452-55): That Vietnam was able to grow throughout its adjustment period can be attributed to the fact that the economy was being increasingly opened to the international market. As part of its overall effort to stabilize the econ- omy, the government unified its various controlledexchange rates in 1989 and devalued the unified rate to the level prevailing in the parallel market. This was tantamount to a 73 percent real devaluation: combined with relaxed administrative procedures for imports and exports, this sharply increased the profitabilityof exporting. This...policy produced strong incentives for export throughout most of the 1989-94 period. During these years real export growth averaged more than 25 percent per annum,and exports were a leadingsector spurring the expansion of the economy.Rice exports were a major part of this success in David Dollar 167 1989; and in 1993-94 there was a wide range of exports on the rise, including processed primary products (e.g., rubber, cashews, and coffee), labor-intensive manufactures, and tourist services.. ..In response to stabili- zation, strengthened property rights, and greater openness to foreign trade, domestic savings increased by twenty percentage points of GDP, from nega- tivelevels in the mid-1980s to 16 percent of GDP in 1992. Are These Individual Country Findings Generalizable? These cases provide persuasive evidence that openness to foreign trade and invest- ment-coupled with complementary reforms-can lead to faster growth in develop- ing economies. But individual cases always beg the question, how general are these results? Does the typical developing economy that liberalizes foreign trade and investment get good results?Cross-country statistical analysis is usefulfor looking at the general patterns in the data. Cross-country studies generally find a correlation between trade and growth. To relate this to the discussion in the first section, some developing economies have had large increases in trade integration (measured as the ratio of trade to national income),and others have had small increases or even declines. In general, the countries that had large increases also had accelerations in growth. The group of developing economy globalizersidentified by Dollar and Icraay (2004) had population-weighted per capita growth of 5 percent in the 1990s,com- pared with 2 percent in rich countries and-1 percent for other developing countries (figure12). This relationship between trade and growth persists after controlling for reverse causality from growth to trade and for changes in other institutions and policies (Dollarand Icraay 2003). Athird type of evidence about integration and growth comes from firm-levelstud- ies and relates to the epigraph from Romer. Developing economies often have large productivity dispersion across firms making similar things: high-productivity and low-productivity firms coexist, and in small markets there is often insufficient com- petition to spur innovation. Aconsistent finding of firm-levelstudies is that openness leads to lower productivity dispersion (Haddad1993; Haddad and Harrison 1993; Harrison 1994). High-cost producers exit the market as prices fall; if these firms were less productive or were experiencing falling productivity, their exits represent productivity improvements for the industry. Although the destruction and creation of new firmsis a normal part of a well-functioning economy, attention is sirnply too often paid to the destruction of firms-whichmisses half the picture. The increase in exits is only part of the adjustment-granted, it is the first and most painful part- but if there are no significantbarriers to entry, there are also new entrants. The exits are often front loaded, but the net gains over time can be substantial. Wacziarg (1998) uses 11episodes of trade liberalization in the 1980s to examine competition and entry. Using data on the number of establishments in each sector, 168 The World Bank Research Observer, vol. 20. no. 2 (Fall2 0 0 5 ) Figure12. PerCapita GDP GrowthRates,by CountryType,1990s(%, basedon GDP in purchasing power parity terms) ' Other developing globalizers countries Source: Dollar and Kraay (2004). he calculates that entry rates were 20 percent higher in countries that liberalized than in countries that did not. Thisestimatemay reflectother policiesthat accompa- nied trade liberalization,such as privatization and deregulation,so this is likely to be an upper bound of the impact of trade liberalization. However, it is a sizable effect and indicates that there is plenty of potential for new firms to respond to the new incentives. The evidence also indicates that exit rates may be significant,but entry rates are usually of a comparable magnitude. Plant-leveldata from Chile,Colombia, and Morocco spanning several years in the 1980s when these countries initiated trade reforms indicatethat exit rates range from 6 percent to 11 percent a year and entry rates from 6 percent to13 percent. Over time the cumulative turnover is quite impressive,with a quarter to a third of firms having turned over in four years (Rob- erts and Tybout 1996). The higher turnover of firms is an important source of the dynamic benefit of openness. In general, dying firms have falling productivity and new firms tend to increase their productivity over time (Aw and others 2000; Liu and Tybout 1996; Roberts ancl Tybout 1996). Aw and others (2000) find that in Taiwan (China) within a five-year period the replacement of low-productivity firms with new, higher-productivity entrants accounted for half or more of the technologiical advance in many Taiwaneseindustries. Although these studies shed somelight on why open economiesare more innova- tive and dynamic, they also show why integration is controversial. There will be more dislocationin an open,dynamic economy-with some firms closingand others David Dollar 169 starting up. If workers have good social protection and opportunities to developnew skills,everyone can benefit.But without these policiesthere can be some big losers. Surveys of the literature on openness and growth generally find the totality of the evidence persuasive. Winters (2004, p. F4), for example, concludes: "While there are serious methodological challenges and disagreements about the strength of the evidence, the most plausible conclusion is that liberalisation generally induces a temporary (but possibly long-lived)increase in growth. A major component of this is an increase in productivity." Similarly, economic historians Lindert and Williamson (2001, pp. 29-30) sum up the different pieces of evidence linking integration to growth: "The doubts that one can retain about each individual study threaten to block our viewof the overall forest of evidence. Even though no one study can estab- lish that openness to trade has unambiguously helped the representative Third World economy, the preponderance of evidence supports this conclusion." They go on to note the "empty set" of "countries that chose to be less open to trade and factor flowsin the 1990s than in the 1960s and rose in the global living-standard ranks at the same time. As far as we can tell, there are no anti-global victories to report for the postwar Third World. We infer that this is because freer trade stimulates growth in Third World economies today,regardless of its effects before1940." Making Globalization Work Better for Poor People So far, the most recent wave of globalization starting around 1980 has been associ- ated with more rapid growth and poverty reduction in developing economies and with a modest decline in global inequality. These empirical findings from a wide range of studies help explain what otherwise might appear paradoxical: opinion sur- veys reveal that globalization is more popular in poor countries than in rich ones. In particular, the Pew Research Center for the People and the Press (2003) surveyed 38,000 people in 44 countries in all developingregions. In general, there was a pos- itive view of growing economic integration worldwide. But what was striking in the survey was that views of globalization were distinctly more positive in low-income countries than in rich ones. Although most people expressed the view that growing global trade and business ties are good for their country, only 28 percent of people in the United States and Western Europe thought that such integration was "very good." By contrast, the share who thought integration was very good was 64 percent in Uganda and 56 percent in Vietnam. These countries stood out as particularly proglobalization, but respondents from developing economies in Asia (37 percent) and Sub-Saharan Africa (56 percent) were also far more likely to find integration "very good" than respondents from rich countries. Conversely, a significant minority (27 percent) in rich countries thought that "globalization has a bad effect" on their country, 170 The World BankResearch Observer, vol. 2 0 , no. 2 (Fall 2 0 0 5 ) compared with a negligible number of households in developing economies in Asia (9 percent)or Sub-Saharan Africa (10percent). Developing economies also had a more positive view of the institutions of globaliza- tion. Some 75 percentof householdsin Sub-Saharan Africa thought that multinational corporationshad a positiveinfluenceon their country,compared with only 54 percentin rich countries.Views ofthe effectof the International MonetaryFund, the World Bank, and the WorldTrade Organization (wro)were nearly as positive in Ph9ca (72 percentof householdssaid they had a positiveeffecton their country).By contrast,only 28 percent of householdsin Africa thought that antiglobalizationprotestorshad a positive effect on their country. Views of the protestors were more positive in the United States and WesternEurope(35 percentsaid the protestorshad a positiveeffecton their country). Although global economic integration has the potential to spur further growth and povertyreduction, whether this potential is realizeddependson the policiesof develop- ing economies and the policies of industrialized countries. True integration requires not just trade liberalization but also wide-ranging reforms of institutions and policies, as the cases of China and India illustrate so clearly. Many of the countries that are not participating very much in globalization have serious problems with the overall investment climate, for example, Kenya, Myanmar, Nigeria, and Pakistan. Some of these countries also have restrictive policies toward trade. But even if they liberalize trade, not much is likelyto happen without other measures. It is not easy to predict the reform paths of these countries. (Consider the relative successes cited here: China, India,Uganda, Vietnam. In each case their reform was a startling surprise.) As long as there are locations with weal