Three years ago the World Bank said that freeing international trade of all barriers and subsidies would lift 320 million people above the $2 a day poverty line by 2015. But new World Bank projections emphasizing $1 a day poverty and based on new data and methods put the number at just 32 million people. CGD/IIE Senior Fellow William R. Cline, author of Trade Policy and Global Poverty, has been examining the Bank's new calculations and argues that the first estimate was closer to the truth.

Q: Why should these estimates matter?

A: The estimates are important both because of their inherent significance for global policy and more immediately because they could give an easy argument to the opponents of globalization by implying that even the World Bank doesn’t think global free trade would reduce poverty by much. This is a windfall for the anti-globalists.

Q: You have estimated that full trade liberalization would lift roughly 440 million people above the $2 a day poverty line. How can economists using similar tools come to such different conclusions?

A: The World Bank's first estimate was not that different: 320 million people. Its new estimate is 95 million people for the same concept ($2 per day poverty, total static and dynamic effects). This is because of lower estimates of protection in a new dataset, and because of application of lower measures than before for the percent change in poverty given a 1 percent change in income. I am still examining both of these changes, but at least initially they seem exaggerated to me.

Q: How so?

...there is a seeming change in the World Bank’s decisions on which concepts to emphasize. In some prominent publications of its new estimates, the World Bank authors do not even mention the 95 million figure and instead emphasize 32 million people lifted out of poverty, because they choose to focus on a $1 a-day definition of poverty and limit their attention only to static effects rather than including dynamic effects.

A: The new dataset says agricultural protection is much lower than in the previous dataset, but the methodological changes responsible may be questionable. One example is a shift to ignore high “above quota” tariffs entirely unless the quota is 90% filled, instead of averaging the low within-quota tariff and the high above-quota tariff. In addition, there is a seeming change in the World Bank’s decisions on which concepts to emphasize. In some prominent publications of its new estimates, the World Bank authors do not even mention the 95 million figure and instead emphasize 32 million people lifted out of poverty, because they choose to focus on a $1 a-day definition of poverty and limit their attention only to static effects rather than including dynamic effects.

Q: Your book was published in 2004. Has anything happened since then that would cause you to revise your estimates?

A: Not much has happened that would warrant a change in the estimates. It is really a question of whether the earlier datasets and methodologies were right. Although some adjustments may be appropriate, my sense is that the new World Bank estimates have gone too far. It should be kept in mind, moreover, that even the earlier estimates did not include gains to developing countries from liberalization of trade in services.

The following is the World Bank response to this critique:

Bill Cline has implied the Bank no longer views trade liberalization as important for poverty reduction, that our new numbers involve serious underestimation, and that his 440 million estimate last year is closer to reality.

On the contrary, the Bank remains committed to trade reform and the complementary policies necessary to make it effective. The conclusion we draw from our new work is that if the international community wants to use trade as a lever to reduce poverty, it has to raise its sights on global trade reform.

This is because economists cannot fully measure many of the benefits trade liberalization accurately. Most models, including our recent estimates, omit reforms in services trade; ignore the introduction of new products in response to trade openings; ignore or estimate very conservatively the increases in productivity that trade reforms stimulate; neglect increasing returns to scale; and rely on aggregations of tariffs that understate resulting potential for new trade.

To be sure, our new numbers do show a reduction of the gains from full trade liberalization. The main reason is new knowledge. Since we undertook the work Bill refers to (and which underpins his estimates), new data have become available that incorporate the effects of recent trade reforms in China, Vietnam, and elsewhere, the phasing out quotas on textiles and clothing, and, for the first time, the effects of trade preferences. All of these effects were included in previous estimate as "gains” but since they have already been realized, now enter the baseline. Bill may disagree with us about agricultural protection, but the new data and approach were selected after a very detailed debate within the modeling community--and on purely scientific grounds not with regard to the results they supported.

A second set of changes is the revision of the poverty elasticities associated with growth, together a lower baseline forecast of poverty in 2015; here too new information based upon a more comprehensive set of household surveys led us to rework our calculations. As elaborated in a recent paper ("Estimating the Benefits of Trade Reform: Why Numbers Change" (PDF) in Richard Newfarmer (ed) Trade, Doha, and Development: A Window into the Issues Washington: World Bank), about half of the drop in our estimate of trade-related poverty reduction is associated with measuring the tariff reforms and about half with the new poverty numbers.

Given these changes, one has to question Bill's assertion that "not much has changed" since 1997, the base for his 440 million calculation. In fact, a lot has changed, however politically inconvenient.

He is right, however, in pointing that our numbers are likely underestimates, if for different reasons than he cites. This is because economists cannot fully measure many of the benefits trade liberalization accurately. Most models, including our recent estimates, omit reforms in services trade; ignore the introduction of new products in response to trade openings; ignore or estimate very conservatively the increases in productivity that trade reforms stimulate; neglect increasing returns to scale; and rely on aggregations of tariffs that understate resulting potential for new trade.

Simple aggregate models like Bill's and our own are useful for headlines and we welcome working with him to improve them. At the same time, however, the main challenge is to develop better information bases on protection, the behavior and structure of markets and of the particular circumstances of poor households such as, for example, whether they are net buyers or sellers of food. The Bank is also heavily engaged in these tasks as its recent book “Poverty and the WTO” shows (Poverty and the WTO: Impacts of the Doha Development Agenda. Thomas W. Hertel and L. Alan Winters, eds., Palgrave Macmillan and the World Bank).






Contributed by William Kline, Dominique van der Mensrugghe, and Richard Newfarmer. Reprinted with permission from Center for Global Development.

To read another Global Envision article about poverty and globalization, see The Poor Like Globalization.



Return to top