1 55933 Agricultural Price Distortions, Inequality and Poverty: Introduction and Summary Kym Anderson University of Adelaide and CEPR kym.anderson@delaide.edu.au John Cockburn Laval University and PEP jcoc@ecn.ulaval.ca Will Martin World Bank Wmartin1@worldbank.org Agricultural Distortions Working Paper 94, August 2009 This is a product of a research project on Distortions to Agricultural Incentives, under the leadership of Kym Anderson of the World Bank's Development Research Group. The authors are grateful for the distortions estimates provided by authors of the focus country case studies, for assistance with spreadsheets by Johanna Croser, Marianne Kurzweil and Signe Nelgen, for helpful comments from workshop participants, and for funding from World Bank Trust Funds provided by the governments of Japan, the Netherlands (BNPP) and the United Kingdom (DfID) and from the Australian Research Council. This paper will appear in Agricultural Price Distortions, Inequality and Poverty, edited by K. Anderson, J. Cockburn and W. Martin (forthcoming 2010). This is part of a Working Paper series (see www.worldbank.org/agdistortions) that is designed to promptly disseminate the findings of work in progress for comment before they are finalized. The views expressed are the authors' alone and not necessarily those of the World Bank and its Executive Directors, nor the countries they represent, nor of the institutions providing funds for this research project. 2 Abstract Reforms in recent decades have sharply reduced the distortions affecting agriculture in developing countries, particularly by cuts to agricultural export taxes and by some reductions in government assistance to agriculture in high-income countries, but international trade in farm products continues to be far more distorted than trade in nonfarm goods. This paper summarizes a series of empirical studies that focus on the effects of the remaining distortions to world merchandise trade for poverty and inequality, especially in developing countries. To obtain different insights into the various impacts, two global studies are undertaken using the World Bank's LINKAGE model, one multi-country study uses the Global Trade Analysis Project (GTAP) model, and ten country case studies are also included, each using a national economy-wide model. The LINKAGE model results suggest that liberalization would reduce international inequality, largely by boosting farm incomes and raising real wages for unskilled workers in developing countries, and would reduce the number of poor people worldwide by 3 percent. The analysis based on the GTAP model for a sample of 15 countries, and the ten stand-alone national case studies, all point to larger reductions in poverty, especially if only the non-poor are subjected to increased income taxation to compensate for the loss of trade tax revenue. JEL codes: D30, D58, D63, F13, O53, Q18 Keywords: Poverty, Income inequality, Global distortions to agricultural incentives, Agricultural price and trade policy reform Author contact details: Kym Anderson School of Economics University of Adelaide Adelaide SA 5005, Australia Phone +61 8 8303 4712 Fax +61 8 8223 1460 kym.anderson@adelaide.edu.au Agricultural Price Distortions, Inequality and Poverty: Introduction and Summary Kym Anderson, John Cockburn and Will Martin For decades, earnings from farming in many developing countries have been depressed by a pro-urban, anti-agricultural bias in own-country policies as well as by governments of richer countries favoring their farmers with import barriers and subsidies. Both sets of policies reduced national and global economic welfare, inhibited economic growth, and added to inequality and poverty because no fewer than three-quarters of the world's billion poorest people depend directly or indirectly on farming for their livelihood (World Bank 2007). During the past two to three decades, however, numerous developing country governments have reduced their sectoral and trade policy distortions, while some high-income countries also have begun reforming their protectionist farm policies. Partly as a consequence of those policy reforms and the associated growth of incomes in many developing countries, the number of people living on less than $1 a day nearly halved over the 1981-2005 period, and their share of the global population fell from 42 to 16 percent (Table 1). Notwithstanding that dramatic achievement in poverty alleviation, the number of extremely poor people was still almost 900 million in 2005, and it may have risen above that following the eruption of the global financial crisis that began in 2008. Moreover, most of the improvement has been in Asia (especially China), while in Sub-Saharan Africa the incidence of poverty was little lower in 2005 than in 1981, at around 40 percent (amounting to 300 million people in 2005). Despite the success of China, it still had over 100 million people on less than $1 a day in 2005, 90 percent of whom were rural. And in India the number of extreme poor remains stubbornly close to 300 million ­ and 74 percent rural, despite large subsidies to their farmers. Less pressing than extreme poverty but nonetheless still important to the welfare of individuals is the extent of income inequality. 1 In the past it was just inequality at the local level that affected individuals' utility, but the information and communication technology 1 For a review of the theoretical literature and empirical evidence on individual and societal preferences for redistribution, see Alesina and Giuliano (2009). Prasad et al. (2007) make the point that, as the number of extreme poor decline over the present century, concerns about poverty will phase out and be replaced gradually by concerns about income inequality. 2 revolution has increased awareness of income differences not only within local regions but also nationally and indeed internationally. Assessing what has happened to the world's income distribution in recent decades depends on one's focus. Milanovic (2005) points to three possibilities. One is intercountry inequality, which compares country-level average incomes where each country has an equal weight in the world distribution regardless of population size, in which case income distribution appears to have become more unequal. The second is international inequality, which still compares country average incomes but this time weighting by the populations of countries, in which case income inequality appears to have decreased although mostly due to the fast growth in populous China and India (see Bourguignon, Levin and Rosenblatt 2004, and Atkinson and Brandolini 2004). And the third possible focus is global inequality, which involves comparing individual incomes regardless of country of citizenship, thus taking into account within-country inequality which is ignored by the international inequality approach where individuals are deemed to earn their country's average income. Rapid growth in the large emerging economies has tended to offset the increase in inequality within countries and so, by this last definition, global inequality appears to have remained roughly constant since the late 1980s. 2 Given the evidence currently available, how much scope is there to further reduce poverty and inequality in the world, and in specific countries, by removing remaining distortions to incentives facing producers and consumers of tradable goods? This question is of great interest to the agricultural, trade, and development policy communities in many developing countries and in non-government organizations and international agencies, and its answer is by no means obvious. True, recent studies indicate agricultural policies are responsible for the majority of the global welfare cost of remaining distortions to goods markets; but removing those measures could affect national poverty in either direction. The answer for each country depends on its (and its trading partners') current food and agricultural policies and the earning and spending patterns of and taxes on its poor, among other things. Account also needs to be taken of three other facts. One is that the dependence of the extreme poor on agriculture for their livelihood has been declining in numerous countries, as alternative opportunities have become available outside agriculture, especially 2 A study by Sala-i Martin (2006) found that GDP per capita disparities between countries have shrunk as economies have converged. See also the analyses based on household survey data rather than GDP per capita, such as by Milanovic (2002, 2005, 2006). A recent review of the global poverty and inequality evidence is available in Ferreira and Ravallion (2008). 3 for off-farm, part-time employment. Another is that falling trade barriers have improved opportunities for farmers to specialize in a cash crop for export, increasing their potential benefits from any improvements in market access abroad for that crop's product..And a third practical reality too important not to consider in some countries is the high level of unemployment (as in South Africa) or policies that inhibit intersectoral labor mobility (as in China). Empirical studies undertaken as background for the World Trade Organization's on- going Doha round of multilateral trade negotiations suggest that in 2001, when that round was launched, policy-driven distortions to agricultural incentives contributed around two- thirds of the global welfare cost of merchandise trade barriers and subsidies (see, e.g., Anderson, Martin and van der Mensbrugghe 2006). While such empirical studies did not have access to comprehensive estimates of distortions to farmer and food consumer incentives in developing countries other than applied tariffs on imports, a more recent study that draws on a new database of distortions to agricultural incentives has confirmed that earlier result: Valenzuela, van der Mensbrugghe and Anderson (2009) suggest agricultural price and trade policies as of 2004 accounted for 60 percent of the global welfare cost of those and other merchandise trade policies. This is a striking result, given that the shares of agriculture and food in global GDP and trade are only 3 and 6 percent, respectively. The contribution of farm and food policies to the welfare cost of global trade-distorting policies for just developing countries is estimated by those authors to be even greater, at 83 percent ­ of which a little more than one-third is due to policies of developing countries themselves. Even so, the estimates of price distortions that went into that modeling study (Anderson and Valenzuela 2008, Anderson 2009) show that many developing countries protect their less- competitive farmers from import competition, so that subset of farmers might be hurt if all markets were opened. So despite much reform over the past quarter of a century in distortions to world trade, many intervention measures ­ especially agricultural ones ­ are still in place. Table 2 summarizes the extent of them in developing and high-income countries on average. It shows that the rate of assistance to farmers relative to producers of non-farm tradables has fallen by one-third for high-income countries since the latter 1980s (from 51 to 32 percent) while in developing countries it has all but disappeared (rising from -41 percent in the early 1980s to +1 percent in 2000-04). The latter trend for developing countries is partly due to the phasing out of agricultural export taxes, and partly also to assistance via import restrictions having risen over the period considered. Thus in both high-income and developing countries there is 4 now a large gap between their NRAs for import-competing and export agriculture, as well as a continuing large gap (albeit smaller than in the 1980s) between the relative rates of assistance in the two groups of countries. In light of that evidence, the above question can be expressed more specifically, for any developing country of interest, as: how important are its own policies compared with those of the rest of the world in affecting the welfare of the poor in that country, and what do agricultural policies in particular contribute to those outcomes? Clear answers to these questions are crucial to guide countries in their national policymaking and as they negotiate bilateral and multilateral trade agreements. Now is an appropriate time to address this multi-faceted question for at least two policy reasons. One is that the World Trade Organization (WTO) is struggling to conclude the Doha round of multilateral trade negotiations, and agricultural policy reform is once again one of the most contentious issues in those talks. The other is that poorer countries are striving to achieve their United Nations­encouraged Millennium Development Goals by 2015, the prime ones being the alleviation of hunger and poverty. It is not only farm- subsidizing rich countries that are resisting reform but also some developing countries not wishing to remove food import barriers and farmer subsidies. There are also several analytical reasons as to why now is the time to focus more thoroughly on this issue. One is that methodologies to address it have advanced at a rapid pace recently, involving microsimulation modeling based on household survey data in conjunction with economy-wide computable general equilibrium (CGE) modeling. Prominent examples include the studies in Hertel and Winters (2006) and Bourguignon, Bussolo and da Silva (2008). Household income information is increasingly important for poverty and inequality analysis because farm households and rural areas of developing countries are rapidly diversifying their sources of income beyond what agricultural land and farm labor can generate, including from part-time off-farm work and remittances (Otsuka and Yamano 2006, Otsuka, Estudillo and Sawada 2009). Hence the earlier close correspondence between net farm income or agricultural GDP and farm household welfare is fading, even in low-income countries (Davis, Winters and Carletto 2009). Second, the compilation of national household surveys that are comparable for cross- country analysis has progressed rapidly such that there are now recent surveys for more than 100 countries available at the World Bank. That Global Income Distribution Dynamics dataset (GIDD, see www.worldbank.org/prospects/gidd) has already begun to be used in conjunction with the World Bank's Linkage model of the global economy to assess global income distribution issues (e.g., Bussolo, De Hoyos and Medvedev 2008). 5 Third, the World Bank has recently compiled a very comprehensive new global database that updates and expands substantially our understanding of the distortions to agricultural incentives in developing countries in particular. 3 Those estimates have since been expressed so as to make them usable in national and global economy-wide models (Valenzuela and Anderson 2008). They differ from the usual ones employed by trade modelers of developing country policies in that they are based on direct domestic-to-border price comparisons rather than (as with the GTAP dataset, see Narayanan and Walmsley 2008) on applied rates of import tariffs and other key border measures. The present volume is a first attempt to exploit those new methodologies and databases to assess the relative impacts on national, regional and global poverty and inequality of agricultural and non-agricultural trade policies at home and abroad. Poverty is defined either in purchasing power parity terms of $1 a day per capita (extreme poverty line) and also sometimes $2 a day (moderate poverty line) or, where those indicators are not available, then the national poverty line is used. Both the incidence of poverty (the share of the population below the poverty line) and the headcount (the absolute number of poor people) are used. As for inequality, the Gini coefficient of income distribution is the key criterion adopted here. For both poverty and inequality, the national indicators are calculated where possible for farm and non-farm households separately, in addition to the national averages. In undertaking this set of studies we are acutely aware that agricultural and trade or domestic price subsidies are far from the first-best policy instruments for achieving national poverty or income distribution objectives; that is largely the prerogative of policies such as the provision of public goods or of tax/transfer measures such as the provision of social safety nets funded through general tax revenue. However, should studies such as the present ones reveal that national trade-related policies are worsening particular countries' poverty or inequality, they provide yet another reason ­ on top of the usual national gains-from-trade reason ­ for those countries to reform their policies unilaterally. Should the inequality and poverty alleviating effects of national trade-related policy reforms be contingent on the rest of the world also reforming, that provides a further reason for that country to participate actively in promoting multilateral trade negotiations under the World Trade Organization (WTO). And should global modeling studies reveal that multilateral trade reform would alleviate global inequality and poverty, it underlines the importance of bringing the WTO's Doha 3 That distortions database is documented fully in Anderson and Valenzuela (2008) and is based on the methodology summarized in Anderson et al. (2008) and detailed in Appendix A of Anderson (2009). 6 Development Agenda (DDA) expeditiously to a successful conclusion with ambitious agricultural reform commitments. A negative finding (e.g., that trade liberalization or farm subsidy cuts would increase poverty in a particular developing country) need not be a reason to shun welfare-enhancing reform, but rather to use the results to provide guidance as to where tax or social programs need to be better targeted so that all groups in society share in the economic benefits from such reform (see Ravallion 2008). Global reform results also provide bargaining power to developing countries seeking aid-for-trade side payments to alleviate any increase in poverty projected to result from multilaterally-agreed trade reform. The purpose of the rest of this chapter is to outline the analytical framework and the common empirical methodology adopted by the global and national case studies reported in subsequent chapters, to summarize and compare the modeling results from both the global and national models, and to draw some general policy implications. These findings are based on three chapters that each use a global model to examine the effects of farm and non-farm price and trade policies on global poverty and its distribution within and across many identified countries, plus ten individual developing country studies spanning the three key regions of Asia (where nearly two-thirds of the world's poor live), Sub-Saharan Africa and Latin America. Analytical framework In order to adequately capture poverty and inequality effects of price-distorting policies, careful consideration must be given to its impacts on household income and expenditure. Many farm households in developing countries rely on the farm enterprise for virtually all of their income, and in the world's poorest countries the share of national poverty concentrated in such households is large. The fact that the poorest households in the poorest countries are concentrated in agriculture means those households are likely to benefit from farm producer price increases engendered by global trade policy reform, other things equal. However, this outcome is not certain for several reasons. First, if the country provides protection from import competition for the commodities produced by the poor, their domestic prices may fall following liberalization. Second, poor farm households also spend the majority of their income on staple foods (Cranfield et al. 2003), so if food prices rise as a consequence of 7 reform then this adverse effect on household expenditure may more than offset any beneficial effect of higher earnings. The rural non-farm and urban poor, too, would be adversely affected by a rise in consumer prices of staple food. However, it is possible that a trade reform that induced a rise in food prices may also raise the demand for unskilled labor (depending on the relative factor intensity of production in the economy's expanding sectors), which ­ depending on how intersectorally mobile is labor ­ could raise the income of poor households more than it raises the price of their consumption bundle. The outcome therefore is always not going to be clear for any particular country, and certainly is an empirical matter for groups of countries because the positive and negative effects in different settings will be more or less offsetting. Some analysts have sought answers from past events, using ex post econometric or micro/household data analysis of historical data (as in the excellent set of studies in Harrison 2007), but it is not easy to find natural experiments of specific policy reforms to analyze and from which it is possible to generalize. An alternative approach ­ the one adopted for the present study ­ is to undertake ex ante analysis using economy wide models. While such models have well-recognized limitations, they are the only option available when seeking to simulate the prospective effects of removing all remaining price-distorting policies (see, e.g., Francois and Martin 2007). This is particularly so when global reform is one of the scenarios of interest, and even more so if insights into the effects on overall world poverty and inequality are being sought. The approach adopted in the present study is a variant on the path-breaking approach pioneered by Hertel and Winters (2006) in their study of the poverty consequences of a prospective Doha round agreement under the WTO. Like Hertel and Winters (2006), this study uses global models to assess the implications of global reform for poverty, plus a series of national models to allow more attention to focus on specific aspects of importance to particular countries. However, the present study contrasts with the earlier one compiled by Hertel and Winters in three key respects. First, the present study focuses on the impacts of agricultural domestic and trade policies, distinguishing them from the impacts of other merchandise trade policies. In this regard, this study is unique in making use of the new database on distortions to agricultural incentives in developing countries that has only recently become available in a format that makes it readily usable by CGE modelers (Valenzuela and Anderson 2008). Those distortion estimates for 2004 are used to represent agricultural and food policies in each of the ten national CGE models employed in the present 8 country case studies, as well as in the three global models used in Part I of this volume. A second distinction is that this study examines inequality as well as poverty. And the third difference is that the present study is able to draw on the massive data collection and modeling effort undertaken for the GIDD database, which includes data on more than a million households representing more than 90 percent of the world's population. The national CGE models are able on their own to estimate the effects of unilateral reform of agricultural or all merchandise trade-distorting policies. For the national modeler to estimate the effects of other countries' policies, however, requires input from a global model. We chose to use the World Bank's Linkage model for that purpose. It too is calibrated to 2004, based on Version 7 of the GTAP global protection database 4 apart from the replacing of its applied agricultural tariffs for developing countries with the more comprehensive set of estimates of distortion rates from the World Bank's Agricultural Distortion research project as collated by Valenzuela and Anderson (2008). As noted above, those distortion estimates suggest that, despite reforms over the past 25 years, there was still a considerable range of price distortions across commodities and countries in 2004, including a strong anti-trade bias in national agricultural policies for many developing countries plus considerable non- agricultural protection in some developing countries (see table 2 above). There are various ways of transmitting the results derived from a global CGE model such as Linkage to a single-country CGE model. Like Hertel and Winters (2006), we adopt the approach developed by Horridge and Zhai (2006). For imports, Horridge and Zhai propose the use of border price changes from the global model's simulation of rest-of-world liberalization (that is, without the focus developing country). For the focus developing country's exports, the shift in its export demand curve following liberalization in the rest of the world is given in percentage changes (following Corong, Cororaton and Cockburn 2010) by x=(1/).q where x is the percentage vertical shift in the export demand curve, is the elasticity of substitution between the exports of country i and those from other countries, and q is the percentage change in the quantity of exports under the scenario with liberalization in the rest of the world excluding the focus country. All the CGE models used in the present study are used in comparative static mode, and they assume constant returns to scale and perfectly competitive homogeneous firms and product markets. In all cases other than the very exceptional one of South Africa (and to a 4 We were fortunate in having early access to the p5 pre-release of that database ahead of the final release. Details of the latter are available in Narayanan and Walmsley (2008). 9 much smaller extent for Argentina and Nicaragua), unemployment is assumed to be unaffected by the trade policy regime. These assumptions are imposed simply because of insufficient data and empirical evidence to impose alternative ones across all the countries being modeled. This use of a standard set of assumptions reduces the risk that differences across countries in results are driven by different assumptions about investment behavior, productivity growth, or the degrees of monopolistic competition, firm heterogeneity, economies of scale, or aggregate employment response to trade policy changes (see Helpman, Itskhoki and Redding 2009). The workhorse specifications that we have used almost certainly lead to underestimation of the welfare gains that would accrue from trade reform though. In particular, without dynamics the models will not generate a growth dividend from freeing up markets or from eventual productivity gains from trade. That dividend could be very substantial. 5 Moreover, since economic growth is the predominant way in which poverty is reduced in developing countries (see the literature review in Ravallion 2006), the absence of dynamics implies that the results from this study will grossly underestimate the potential poverty alleviating consequences of liberalization ­ and might in some situations indicate poverty increases when in fact they would be decreases had the growth consequences been incorporated. All the country case studies, and two of the global modeling studies in this volume, make use of household survey data in addition to a social accounting matrix (SAM). The SAM is the basis for the data in the CGE model, while the household survey data are used in microsimulation modeling. Typically the experiments are performed in two stages. The first stage involves the imposition on the national CGE model of the policy shock (either unilateral liberalization, or an exogenous shock to border prices and export demand provided by the Linkage model). This generates changes in domestic product and factor markets. The consequent changes in consumer and factor prices are then transmitted to the microsimulation model to see how they 5 See Wacziarg and Welch (2008) and Krueger (2009), as well as the collection of seminal earlier papers in Winters (2007). A paper that brings together the above ideas using a numerical open economy growth model is that by Rutherford and Tarr (2002). Their model allows for product variety, imperfect competition, economies of scale and international capital flows. It is also dynamic, so the model can trace out an adjustment path to trade reform; and it is stochastic in that it draws randomly from uniform probability distributions for eight key parameters of the model. They simulate a halving of the only policy intervention (a 20 percent tariff on imports) and, in doing so, fully replace the government's lost tariff revenue with a lump-sum tax. That modest trade reform produces a welfare increase (in terms of Hicksian equivalent variation) of 11 percent of the present value of consumption in their central model. Systematic sensitivity analysis with 34,000 simulations showed that there is virtually no chance of a welfare gain of less than 3 percent, and a 7 percent chance of a welfare gain larger than 18 percent of consumption. See also the empirical study of four developing countries in Cockburn et al. (2008). 10 alter the earnings of various household types (according to the shares of their income from the various factors) and their cost of living (according to the shares of their expenditure on the various consumer products). That in turn provides information on changes in the distribution of real household incomes and hence in inequality, and in the number of people below any chosen poverty line such as $1 a day. All country case studies in this volume ran a common set of simulations so as to compare the inequality and poverty effects in each country of own-country versus rest-of- world policies affecting markets for agricultural (including lightly processed food) goods versus other merchandise. The precise nature of the rest-of-world simulation, which employs the global Linkage model, is made clear in the next chapter and in the Appendix to this volume. The other two global studies in Part I of the book use the same 2004 global protection dataset but implement global reform shocks each using a different global model but with national household survey data attached in order to undertake microsimulations. In most cases additional simulations were also run, often to illustrate the sensitivity of the results to key assumptions pertinent to that particular case study. One in particular that the contributors to Hertel and Winters (2006) found to be important, and that is confirmed in this study as well, is the assumption about how government revenue raising would alter to make up for the loss of tariff revenue. Even though the models employed in this study are all standard perfectly competitive, constant-returns-to-scale, comparative static, economy wide CGE models, they nonetheless differ somewhat in order to capture important realities (such as labor market characteristics or data limitations) in their particular setting. However, to ensure their comparability within this volume, they all aimed to conform to a common set of factor market assumptions and closure rules in addition to using 2004 as their base and undertaking a common set of simulations using the same global distortions dataset. We know from trade theory that factor market assumptions are crucial determinants of the income distributional effects of trade policies, so all modelers assumed the following: a fixed aggregate stock of factors (including no international mobility of labor or capital or international technology transfer), with the exception of labor in the South African study where some aggregate employment responsiveness to trade policy is justified because of very high unemployment in the baseline; possibly some sector-specific capital and labor, but most capital and labor types are assumed to be intersectorally mobile with a common flexible rate 11 of return or wage; and land is assumed to be specific to the agricultural sector but mobile across the different crop and livestock activities within that sector. The key agreed macroeconomic closure rules that each case study aimed to adopt are a fixed current account in foreign currency, to avoid foreign debt considerations, and fixed real government spending and fiscal balance, so as to not affect household utility other than through traceable changes in factor and product prices and taxes. Fiscal balance is achieved by using a uniform (generally direct income) tax to replace net losses in revenue from abolishing sectoral trade taxes and subsidies. And to repeat, technologies are assumed unchanged by reform, so no account is taken of dynamic gains from trade opening and their prospective impacts through faster productivity growth on poverty and inequality. Synopsis of empirical findings: Global model results This section summarizes the results from the three global models (denoted Linkage, GIDD and GTAP). The following section then brings together the results from the ten more-detailed national case studies, before the lessons learned from both sets of analyses are drawn together. It would be surprising if all the studies came to the same conclusions, but the strength of this blend of somewhat different global and national models is that it is more likely to expose the various determinants of the measured effects in different settings than if only a single type of model was employed. Linkage model results The next chapter, by Anderson, Valenzuela and van der Mensbrugghe (2010), sets the scene for the rest of the book in that it uses the World Bank's global Linkage model (van der Mensbrugghe 2005) to assess the market effects of the world's agricultural and trade policies as of 2004. By doing so it serves two purposes. One is to provide the basis for estimating the effects of rest-of-world policies on the import and export prices and demand for the various exports of any one developing country, for use by each of the ten country case studies in Parts II to IV of this volume. The details of those results are reported in the Appendix to this 12 volume (van der Mensbrugghe, Valenzuela and Anderson 2010). The other purpose of chapter 2 is to provide estimates of various economic effects, on individual countries and country groups, so as to be able to say something about international inequality (in the Milanovic (2005) sense, taking into account the economic size of countries) and poverty (using a simple elasticities approach). The Linkage model results reported in that chapter suggest that developing countries would gain nearly twice as much as high-income countries in welfare terms if 2004 agricultural and trade policies were removed globally (an average welfare increase of 0.9 percent, compared with 0.5 percent for high-income countries ­ bottom of column 1 of table 3). Thus in this broad sense of a world of just two large country groups, completing the global reform process would reduce international inequality. 6 The results vary widely across developing countries, however, ranging from slight losses in the case of some South Asian and Sub-Saharan African countries that would suffer exceptionally large adverse terms of trade changes, to an 8 percent increase in the case of Ecuador (whose main export item, bananas, is currently heavily discriminated against in the EU market where former colonies and least developed countries enjoy preferential duty-free access). 7 Bearing in mind that three-quarters of the world's poorest people depend directly or indirectly on agriculture for their main income, and that farm sizes are far larger in high- income than in developing countries, 8 chapter 2 also looks at the extent to which agricultural and trade policies in place as of 2004 reduced rewards from farming in developing countries and thereby added to international inequality in farm incomes. It finds that net farm incomes in developing countries would rise by 5.6 percent, compared with 1.9 percent for non- agricultural value added, if those policies were eliminated (bottom of final two columns of table 3). This suggests that inequality between farm and nonfarm households in developing 6 This would continue a process that began in the 1980s, when many countries began to reform their trade and exchange rate regimes. Using the same Linkage model and database as the present study, Valenzuela, van der Mensbrugghe and Anderson (2009) found that the global reforms between 1980-84 and 2004 also boosted economic welfare in developing countries proportionately more than in high-income economies (by 1.0 percent, compared with 0.7 percent for high-income countries). 7 Even so, if one were to treat each of the 60 countries/groups of countries in that global study as able to be represented by a single household (that is, ignoring intra-country inequality), then inter-country income inequality (not taking account the differing economic size of countries) would be reduced at least slightly as measured by the Gini Coefficient, from 0.8513 to 0.8506. 8 According to the FAO (2008), less than 15 million relatively wealthy farmers in developed countries, with an average of almost 80 hectares per worker, currently are being helped, at the expense of not only consumers and taxpayers in those rich countries but also the majority of the 1.3 billion relatively impoverished farmers and their large families in developing countries who, on average, must earn a living from just 2.5 hectares per worker. 13 countries would fall. By contrast, in high-income countries net farm incomes would fall by 15 percent on average, compared with a slight rise for real non-farm value added. That is, inequality between farm and nonfarm households within high-income countries would probably increase 9 ; however, inequality between farm households in developing and those in high-income countries would decline substantially. These inequality results would not be very different if only agricultural policies were to be removed (c.f., columns 2 and 3 of table 3), underscoring the large magnitude of the distortions from agricultural, as compared with non-agricultural, trade policies. Chapter 2 also reports that unskilled workers in developing countries ­ the majority of whom work on farms ­ would benefit most from reform (followed by skilled workers and then capital owners), with the average change in the real unskilled wage over all developing countries rising 3.5 percent. However, the most relevant consumer prices for the poor, including those many poor farm and other rural households who earn most of their income from their labor and are net buyers of food, relate just to food and clothing. Hence deflating by a food-clothing price index rather than the aggregate CPI provides a better indication of their welfare change. As shown near the bottom of the final column of table 4, for all developing countries the real unskilled wage over all developing countries would rise by 5.9 percent with that deflator. That is, inequality in real incomes between unskilled wage-earners and the much wealthier owners of capital (human or physical) within developing countries would likely be reduced with full trade reform. The above results for real factor rewards and net farm income suggest that poverty, as well as international and intra-developing country inequality, could be alleviated globally by agricultural and trade policy liberalization. The authors of chapter 2 go a step further to explicitly assess reform impacts on poverty even though the Linkage model has only one single representative household per country. They do so using the elasticities approach, which involves taking the estimated impact on real household income and applying an estimated income to poverty elasticity to estimate the impacts on the poverty headcount index for each country. They focus on the change in the average wage of unskilled workers deflated by the food and clothing CPI, and assume those workers are exempt from the direct income tax imposed to replace the lost customs revenue following trade reform (a realistic assumption for many developing countries). 9 In some high-income economies, however, farm households now have higher household incomes than non- farm households (Gardner and Sumner 2007, OECD 2008). 14 Under the full merchandise trade reform scenario, table 5 reports that extreme poverty (the number of people surviving on less than US$1 a day) in developing countries would drop by 26 million relative to the baseline level of just under one billion, a reduction of 2.7 percent. The proportional reduction is much higher in China and in Sub-Saharan Africa, each falling around 4 percent. By contrast, the number of extreme poor in India (though not in the rest of South Asia) is estimated to rise, by 4 percent. 10 This follows from the estimated decline in overall income in India following trade liberalization noted in Table 3. Under the more moderate definition of poverty--those living on no more than US$2 per day--the number of poor in developing countries would fall by nearly 90 million compared to an aggregate baseline level of just under 2.5 billion in 2004, or by 3.4 percent (notwithstanding the number in India below $2 a day still increasing, but by just 1.7 percent). GIDD model results Chapter 3, by Bussolo, De Hoyas and Medvedev (2010), makes direct use of the global CGE Linkage model described above but then combines this with the newly developed Global Income Distribution Dynamics tool (Bussolo, De Hoyos and Medvedev 2008). GIDD is a framework for ex ante analyses of the income distributional and poverty effects of changes in macroeconomic, trade and sectoral policies or trends in global markets, and thus offers an alternative to the elasticity approach adopted in chapter 2. It complements a global CGE analysis by providing global micro-simulations based on standardized household surveys. This tool pools information from most of the currently available household surveys covering 1.2 million households in 73 developing countries. Household information from developed countries and Eastern Europe's transition economies completes the dataset. Overall, the GIDD sample covers more than 90 percent of the world's population. 11 In contrast with the modeling approach used in Chapter 2, the GIDD approach is able to distinguish between farm and non-farm households by the employment of the household head. However, because of differences between surveys in the coverage of household income sources, the database is unable to identify the sources of income for each household and assumes that the proportional change in income of households is driven by just the changes in wages. 10 The rise in India is partly because of the removal of the large subsidies and import tariffs that assist Indian farmers, and partly due to the greater imports of farm products raising the border price of those imports. 11 The GIDD dataset, methodology and applications are available at www.worldbank.org/prospects/gidd. 15 The key input into the micro-simulation model are results for labor income changes obtained from a variation on the Linkage model that assumes full labor mobility. 12 Two liberalization scenarios are examined: full liberalization of agricultural and lightly processed food markets without and with liberalization of nonfarm goods markets. Neither is shown to have large effects on global poverty according to GIDD. The results summarized in table 6 show the incidence of extreme poverty (US$1 per day) rising by 1.0 percent (0.5 percent from each of farm and non-farm full global trade reform). This increase in poverty is largely due to the increase in poverty in South Asia, where the number of poor people rises by 3.9 percent with complete global trade reform--a result similar to that reported in Chapter 2. Moderate poverty (US$2 per day), on the other hand is projected to fall by a similar amount (0.9 percent from agricultural reform alone and by 0.8 percent when nonfarm reform is included). These small aggregate global changes are produced by a combination of offsetting trends between farm and nonfarm households (Table 7). At the $1 a day extreme poverty level, global liberalization would raise the share of agricultural households in the world's total poor households by 1 percentage point (from 76 to 77 percent), and also the incidence of poverty among the world's agricultural households (from 32 to 33 percent), while the incidence among the world's nonfarm households would drop slightly to 8 percent. However, at the moderate poverty line of $2 a day, both agricultural and all merchandise trade liberalization globally lower the poverty incidence, by nearly 1 percent, and they reduce it for farm as well as non-farm households (compare the last two columns of table 7). There are several possible reasons as to why this sign of the effect on extreme poverty (but not on moderate poverty) differs from that in chapter 2. One is that GIDD poverty data refer to 2000 whereas the Linkage poverty numbers relate to 2004. A large share of the developing country population was bunched around the extreme poverty line in 2000 (see Figure 3.1 of Chapter 3), but by 2004 poverty had shrunk quite a bit, at least in East Asia. Another reason has to do with the fact that the GIDD results are based on changes just in labor income, rather than in income from all factors of production. In particular, by not including the effect on non-labor (especially land) income, this study understates the poverty alleviating impacts on farm households, thereby contributing to its finding that extreme poverty among farm households would increase. Furthermore, the assumption of full labor 12 While changes in labor income are the most important income change for households at or near the poverty line, accounting for changes in other sources of income may yield somewhat different results, particularly for inequality. The Linkage results used here are not identical to those in the previous chapter because, to make them compatible with GIDD, the authors had to assume labor is less than fully mobile across sectors. 16 mobility implies that unskilled farm workers "share" the gains from increased agricultural prices with their non-farm counterparts, by requiring less labor to migrate to non-farm jobs. The GIDD results suggest there could also be considerable changes in inequality following global trade reform. Indeed, table 7 shows that agricultural incomes would increase by twice as much as nonfarm household incomes in the all-goods reform scenario (0.8 percent compared to 0.4 percent) and by five times as much in the agriculture-only reform (1.1 percent compared to just 0.2 percent). While that reduction in the non-agricultural income premium on its own would reduce inequality, income dispersion within the agricultural sector is also found to increase given the different impacts of reform on income distribution in different regions, such that the final change in global inequality would be close to zero (column 1 of table 7). Chapter 3 also provides poverty and inequality results at the national and regional levels, summarized in table 8. Improvements (i.e. reductions) in these indicators are pervasive among the 19 countries of Latin America and the Caribbean and five East Asian countries. There are far fewer African examples of improvements, regardless of whether the reform scenario is agriculture-only or also includes non-farm goods, but most of the indicator changes for those countries are close to zero. It is mainly in India where extreme poverty ­ but not moderate poverty ­ worsens according to the GIDD results, which is what was also found in the previous chapter by applying poverty elasticities directly to the Linkage model results. The impact of agricultural reform on poverty in India is a vitally important subject, but one on which the existing evidence is quite mixed. In an econometric analysis of historical data, Topolova (2007) concludes that the reductions in agricultural protection associated with India's tariff reforms of the 1990s increased national poverty. By contrast, using detailed information on household incomes and expenditures, Cai, de Janvry and Sadoulet (2008) conclude that 70 percent farmers in India (86 percent of those with less than 0.2 hectares, 73 percent of those with between 0.2 and 1 ha, and 49 percent of those with more than 1 ha) would have lost from increases in the prices of staple foods during 2007-08. According to their data, this is because even the smallest farmers get only about half their income from farming. It needs to be kept in mind that the GIDD micro-simulation model assumes that household total income changes are proportional to the changes in the wage rates for their labor (agricultural or non-agricultural). While labor income is the most important income source for households at or near the poverty line, it is not the only one. Thus accounting for 17 changes in other factor returns may yield somewhat different results, especially for inequality. In the remaining studies to be reviewed, all sources of income are taken into account, so they may provide more reliable national results, albeit for a smaller sample of developing countries. GTAP model results Chapter 4, by Hertel and Keeney (2010), draws on the widely used global economy-wide model of the Global Trade Analysis Project (GTAP). It adopts the same price distortions as the other studies in this volume and runs the same scenarios, but generates its own world price changes from the GTAP model for the multilateral trade reform scenarios. Those prices changes alter border prices for the various countries in the GTAP model, a subset of which have attached to them detailed household survey data. This permits the authors to say something about poverty impacts across a range of diverse economies using an alternative internally consistent framework to that employed by Bussolo, De Hoyos and Medvedev (2010). While the number of their countries with household survey data is much smaller, the income data are richer, making it possible to capture the distributive effects of all factor income changes rather than being restricted to just labor income shocks as in chapter 3. This multi-country study focuses on 15 developing countries: five Asian (Bangladesh, Indonesia, Philippines, Thailand, and Vietnam), four African (Malawi, Mozambique, Uganda, and Zambia), and six Latin American countries (Brazil, Chile, Colombia, Mexico, Peru, and Venezuela). Overall, it concludes that removing current farm and trade policies globally would tend to reduce poverty, but primarily via agricultural reforms (table 9). The unweighted average for all 15 developing countries is a headcount decline in extreme poverty (<$1 a day) of 1.7 percent. The average fall for the Asian sub-sample is twice that, however ­ and it is in Asia where nearly two-thirds of the world's extremely poor people live (although their sample did not include China and India). Turning to their results for specific countries, it is the agricultural-exporting developing countries in the sample, namely Chile, Thailand and Vietnam, where the most poverty alleviation would occur (column 3 of table 9). The majority of the 15 countries studied experiences small poverty increases from non-agricultural reforms, although the unweighted average across the fifteen countries suggests a slight decrease, primarily due to a strong decline in Vietnam (column 2 of table 9). 18 The magnitude of their estimated extreme poverty alleviation in both Asia and Latin America is somewhat larger than the average reductions estimated for the same countries by Bussolo, De Hoyos and Medvedev (2010, table 4) using the GIDD model; and they also estimate a small reduction in poverty in Africa. 13 These GTAP results are thus closer to the Linkage model results of chapter 2. The authors explore the relative poverty-friendliness of agricultural trade reforms in detail, examining the differential impacts on real after-tax factor returns of agricultural versus non-agricultural reforms. Their analysis is extended to the distribution of households by looking at stratum-specific poverty changes. They find that the more favorable impacts of agricultural reforms are driven by increased returns to peasant farm households' labor as well as higher returns for unskilled work off-farm. They also find that liberalization of food grain markets represents the largest contribution to poverty reduction, and that removing import tariffs in those commodity markets dominates the poverty-increasing impacts of subsidy removal by high-income countries. The final column of table 9 reports the percentage change in the national poverty headcount when the poor are not subject to the income tax rise required to replace trade tax revenue following trade reform. This assumption represents a significant implicit income transfer from non-poor to poor households and thus generates a marked difference in the predicted poverty alleviation. Trade reforms go from being marginally poverty reducing in most of the 15 cases to being poverty reducing in all cases and by a considerable magnitude. It reduces the poverty rate by roughly one-quarter in Thailand and Vietnam, for example. Overall, the regional and total average extent of poverty alleviation is around four times larger in this scenario than when the poor are also assumed to be levied with income taxes to replace lost trade tax revenue. The unweighted average poverty headcount reduction for the three regions shown in the final column of table 9 are remarkably similar to the population-weighted averages by Anderson, Martin and van der Mensbrugghe (2010) reported in table 5 above with a similar tax-replacement assumption: the latter's 17 percent for Asia excluding China and India and 6.4 percent for Latin America are just slightly above Hertel and Keeney's 14 percent and 5.7 percent, while their 3.7 percent for Sub-Saharan Africa is just below the 4.5 percent obtained for the Hertel and Keeney sample. 13 An African comparison is not possible because there was only one African country common to the two sample country sets. 19 Synopsis of empirical findings: national model results We turn now to see how the results from the ten more-detailed individual country case studies compare with the above results from global models. The features of the national models are summarized in table 10. 14 Like the three global models, they focus on price- distorting policies as of 2004, even though the database for their CGE models and their household survey data typically date back a little earlier in the decade. They all include more sectoral and product disaggregation than the global models, and have multiple types of households and types of labor. All of the national studies include micro-simulations drawing on model results, as in the GIDD and GTAP global models. The national results for real GDP and household consumption suggest that GDP would increase in all 10 countries from full global trade reform, but only by 1 or 2 percent. Given falling consumer prices, real household consumption would increase by considerably more in most cases, Argentina being the notable exception (for reasons discussed below). Generally these numbers are a little larger than those generated by the global Linkage model, but they are still generally much lower than would be the case had the authors used dynamic models. They therefore share the feature of the global models of underestimating the poverty- alleviating benefits of trade reform, given the broad consensus in the literature that trade liberalization increase growth, which is in turn a major contributor to poverty alleviation. The comparative tables 11 and 12 summarize the national results for the incidence of extreme poverty 15 and income inequality, respectively, resulting from own-country or rest-of- world full liberalization of agricultural or all goods trade. Some authors ran only six of the nine simulations shown in this table, but those that ran all nine found them to sum up almost exactly, to one decimal place. We therefore have inferred the three missing results in the other country studies by assuming that the agriculture-only and nonagriculture-only results sum to the all-goods reform results. The inferred numbers are shown in italics in tables 11 14 The ten national studies are for Argentina (Cicowiez, Diaz-Bonilla and Diaz-Bonilla (2010), Brazil (Ferreira Filho and Horridge (2010), China (Zhai and Hertel 2010), Indonesia (Warr 2010a), Mozambique (Arndt and Thurlow 2010), Nicaragua (Sanchez and Vos 2010) Pakistan (Cororaton and Orden 2010), Philippines (Corong, Cororaton and Cockburn 2010), South Africa (Herault and Thurlow (2010), and Thailand (Warr 2010b). 15 Using national or $1 a day poverty lines, except for China for which results are available only at the $2 a day line. 20 and 12. In each case the total effects on poverty and inequality are subdivided into rural and urban. One should not necessarily expect the unweighted averages of the poverty results for each region to be similar to those generated by Hertel and Keeney (2010), because only half of the ten countries for which we have case studies were included among the 15 countries sampled by Hertel and Keeney. Nonetheless, the latter's unweighted averages of national poverty effects for each of the key developing country regions are reported in brackets in the last 4 rows of table 11(c), to make it easy to compare with the unweighted regional averages for our ten country case studies. In all but three of those twelve comparisons for global liberalization (agricultural, non-agricultural and non-merchandise), the projected regional average poverty reductions from global liberalization are larger from our sample of ten national case studies than from Hertel and Keeney's 15-country sample. Perhaps this suggests the poverty elasticities used in the latter study (and hence also in the Linkage model, since it generated similar results) are too small given the greater possibilities for adaptation reflected in most of the national models. 16 As for the individual country results, poverty is reduced in all ten countries by both global agricultural and, with the exception of the Philippines, non-agricultural liberalization (table 11(c)). When all merchandise trade is liberalized, the extent of reduction ranges from close to zero to about 3.5 percentage points, except for Pakistan where it is more than 6 points. 17 According to the unweighted average, more of the alleviation is due to non-farm trade reform, with the important exception of Brazil where agricultural reform is the major contributor to its large pro-poor outcome; but if the average had been weighted according to the number of people involved, agricultural reform would dominate, as with the global modeling results. The extreme Brazil result is despite the presence of tariff protection for poor import-competing farmers, and is a consequence of the increase in demand for unskilled labor following liberalization, which evidently outweighs the poverty impact of removing farm tariffs. The contribution of own-country reforms to the fall in poverty appears to be equally as important as rest-of-world reform on average, although there is some considerable cross-country divergence in the extent of this both for farm and non-farm reform. 16 Hertel and Keeney use stratum-specific poverty elasticities to map average income changes from all sources to poverty impacts. 17 The Pakistan results were generated assuming replacement of trade taxation with a rise in direct income taxes. Only urban, non-poor households pay direct taxes in Pakistan, so the removal of tariffs decreases the after- tax incomes of the urban non-poor and means the benefits of trade reform go mainly to the poor. 21 The poverty alleviation is sub-divided in parts (a) and (b) of table 11 into rural and urban sources. A glance at the final column of that part of the table reveals that rural poverty is cut much more than urban poverty in every case. That is true for both farm and non-farm trade reform, and for own-country as well as rest-of-world reform. Since the rural poor are much poorer on average than the urban poor (see Bussolo, De Hoyos and Medvedev 2010, Figure 1), this would lead one to expect trade reform to reduce inequality also. Indeed, the results at the bottom of Table 12(c) for this sample of countries show that inequality would decline in all three developing country regions following full trade liberalization of all goods, or just agricultural products, and both for own-country and rest-of- world reform. The effect of non-farm trade reform on its own is more mixed, providing another reason to urge trade negotiators not to neglect agricultural reform in trade negotiations. Rest-of-world and global agricultural reform both lead to a reduction in inequality in every country in the sample except Thailand (plus Argentina and the Philippines slightly for global reform), whereas unilateral agricultural reform reduces (or leaves constant) inequality in a small majority of countries with Argentina, China, the Philippines and Thailand being the exceptions (but the latter effects are small). Non-farm global reform increases inequality slightly in just three countries. In the cases of Indonesia and Thailand the inequality-increasing impact of non-farm reform more than offsets the egalitarian effect of farm trade reform, whereas both types of reform increase inequality in the case of the Philippines and Thailand. Inequality within the rural or urban household grouping is not altered very much by trade reform as compared with overall national inequality (compare parts (a) and (b) with part (c) of table 12). This underlines the point that trade reform would tend to reduce urban-rural inequality predominantly rather than inequality within either region. Several of the national studies investigate impacts of reforms that could complement trade reforms, most notably different approaches to deal with the elimination of trade tax revenues. If these revenues can be recouped through taxes that do not bear on the poor, then the impacts of reform for poverty reduction are more favorable. The China study focuses on the vitally important issue of reducing the barriers to migration out of agriculture, by improving the operation of land markets and reducing the barriers to mobility created by the hukou system. These measures, and international trade liberalization that increases China's market access, are found to reduce poverty such that a combination of these measures would benefit all major household groups. 22 Argentina is a special case in several respects. One is that the authors of that study (Cicowiez,, Diaz-Bonilla and Diaz-Bonilla 2010) had access only to an urban household survey, so were unable to say anything about the effect of policy reform on rural poverty or urban-rural income inequality. Secondly, Argentina imposed export taxes on farm products in late 2002 and has increased them a number of times since then. Removing them as part of a move to free trade would clearly benefit farmers and rural areas but would raise the price of food in urban areas which, other things equal, would tend to increase urban poverty (as reflected in the results in table 11(b)). Another feature of that study is that the authors allow reform to alter aggregate employment, unlike most of the other studies surveyed above, making its results less comparable with the others. Together these features cause global trade reform to reduce Argentine urban poverty and inequality but only if the country's export taxes are not included in its reform. When export taxes are eliminated as well, the results in tables 11(b) and 12(b) show that urban inequality hardly changes but urban poverty would rise. It would rise--despite the country's non-farm reform reducing urban poverty--because of the strong negative impact on the urban poor of higher food prices as a result of export tax removal. In a global reform scenario in which export taxes are left unchanged, the authors found both poverty and inequality would fall in Argentina, because it would generate less unemployment than when export taxes also are removed. What have we learned? As found in previous studies, whether based on ex post econometrics (as in Harrison 2007) or ex ante economy wide simulation (as in Hertel and Winters 2006), so this study also finds mixed results that are not easy to summarize, particularly with regard to the poverty effects. There is nonetheless a high degree of similarity in the most important sign: the estimated national extreme poverty effect of freeing all merchandise trade globally. It happens to be the effect for which there is the most overlap between the studies summarized above. Those signs, summarized in table 13, agree in all but one-seventh of the cases shown. And apart from India, there is no case where the majority of the signs indicate reform would increase poverty. This beneficial impact of full liberalization of global merchandise trade on the world's poor would come more from agricultural than non-agricultural reform; and, within 23 agriculture, more from the removal of substantial support provided to farmers in developed countries than from developing country policy reform. According to the economy wide models used in the present study, such reform would raise real earnings of unskilled workers in developing countries, most of whom work in agriculture. Their earnings would rise relative to both unskilled workers in developed countries and other income earners in developing countries. This would thus reduce inequality both within developing countries and between developing and developed countries, in addition to reducing poverty. According to the Linkage model, the number of extremely poor people in developing countries (on less than $1 a day) is estimated to fall by 2.7 percent with global opening of all goods markets, and by 4 percent in China and Sub-Saharan Africa, but to rise by 4 percent in India (or by 1.7 percent if the more moderate $2 a day poverty level is used). The GIDD model suggests that the decline in moderate poverty would be less than the Linkage model's estimate, and that extreme poverty would actually rise by 1 percent globally with full global trade reform (almost all due to India), but recall that the GIDD model only takes into account labor income effects. The 15-country results from the GTAP model are more in line with those of the Linkage results. They suggest that the poverty-reducing effect in Asia and Latin America of global reform would be twice as large as the estimates from the GIDD model, and that in Africa there would be a small decline (rather than a small rise) in poverty. The ten national case studies all find global trade liberalization to be poverty alleviating, regardless of whether the reform were to involve only agricultural goods or all goods, with the benefit coming roughly equally from reform at home and abroad. They also find that rural poverty would be cut much more than urban poverty in all ten country cases, whether from reform at home or abroad and whether or not it included non-farm goods. Global trade liberalization would reduce international inequality as between developing and high-income countries, both in total and for just farm households, according to the Linkage model. But it cannot be guaranteed that every developing country would be better off unless there is a strong economic growth dividend from reform (not captured in the comparative static modeling used in the present study). The message emerging from the GIDD analysis is less optimistic, in that it finds inequality would change little with full global reform (it falls in Latin America and rises in South Asia). This is mainly because of increased income dispersion within the agricultural sector, and despite a reduction in the farm-nonfarm household income gap. The analysis based on the GTAP model, which reinforced the findings from the Linkage model with respect to poverty, does not provide inequality effects. 24 Full trade liberalization of all goods, or just of agricultural products, also would cause inequality to decline within each of the three developing country regions covered by our sample of 10 countries, and both for own-country and rest-of-world reform. Inequality within the rural or urban household grouping would not alter much following full trade reform, suggesting that trade reform's predominant impact would be to reduce urban-rural inequality. The mechanism through which governments adapt to the fall in tariff revenue is also shown to be primordial. If it is assumed the poor do not have to bear any of the burden of replacing trade taxes, instead of sharing it proportionately, the estimated degree of poverty alleviation is about four times greater in the 15 countries studied with the GTAP model. Results from the three global analyses all indicate that removing remaining agricultural policies would have much stronger impacts on poverty and inequality than would non-agricultural trade reforms. A weighted average across the ten country case studies would probably come to a similar conclusion. This contrasts to reforms over the past three decades: Valenzuela, van der Mensbrugghe and Anderson (2009, table 13.12) estimate that global non- farm trade policy reforms between the early 1980s and 2004 boosted value added in developing country agriculture more than twice as much as global agricultural policy reforms lowered it, and so could be expected to have had a dominant impact on past alleviation of poverty and inequality. The ten national case studies also shine some light on the relative importance of domestic versus rest-of-world reform for those countries. The contribution of own-country reforms to the fall in poverty appears to be equally as important as rest-of-world reform on average, although there is some considerable cross-country divergence in the extent of this, both for farm and non-farm reform. Caveats The impacts of agricultural and other trade reform are complex, simultaneously affecting product and factor markets, government budgets and external trade. The studies included in this volume provide a broad range of ex ante modeling perspectives, including both global and national models. Considerable attention has been devoted to capturing poverty effects through the use of recent microsimulation and poverty elasticity approaches, and to using the same price distortion estimates, the same global model for getting rest-of-world border 25 shocks for the ten national models, and similar behavioral assumptions, tax replacement assumptions and model closures. Nonetheless, there is ample scope for further exploration of this issue through additional comparisons, including in the form of drilling down into each modeling result to explore its origins. Space limitations in this volume mean that such exploratory work needs to be left as an area for further research. The reforms considered here refer only to liberalization of goods trade. Freeing global trade in services is also likely to bring gains to most national economies, including their farmers. Freeing capital would add to those gains (Prasad et al. 2007), as would freeing the international movement of low-skilled labor from developing to higher-income countries (World Bank 2006). How those reforms would interact with farm and other goods trade reforms, in terms of their impacts on global poverty and inequality, is bound to be complex and so await the development of more-sophisticated global simulation models. Another key challenge that remains involves the capturing of the growth effects of liberalization and, in particular, their general equilibrium distributive (poverty and inequality) effects. This area of research has only recently begun to be addressed in the empirical literature, building on the gains made in the theoretical endogenous growth literature in the 1990s (beginning with Grossman and Helpman 1991). Existing partial equilibrium analysis strongly suggests that the trade-growth-poverty nexus is extremely important, possibly much more important than the static reallocative impacts captured in the current set of studies. There is every reason to believe that, once dynamics are included, they will reinforce the basic finding of this study that agricultural and other merchandise trade policy reform is poverty and inequality reducing. Thirdly, there is huge scope for exploring empirically the possible effects of complementary domestic reforms that could accompany agricultural price and trade policy reforms. Even in the extreme case of India, the latter reforms would probably not increase poverty if more-efficient transfer mechanisms were in place and high-payoff infrastructure investments were made. The politics of having first-best domestic policies in place are not necessarily any less complex than those associated with trade policies, however, which underscores the need for comprehensive political economy analysis that does not limit its focus just to border policy measures. Policy implications 26 The above empirical findings have a number of policy implications. First and foremost, the generally very positive poverty and inequality effects from both agricultural and other merchandise trade policy reform, whether unilateral or multilateral, should provide impetus for further liberalization of national and world agricultural and non-agricultural markets. It is well known that liberalization has both winners and losers. The basic finding in this study is that the winners would overwhelmingly be found among the poorer countries and the poorest individuals within countries. However, it is also clear that even among the extreme poor, some will lose out. Hence the merit of compensatory policies, ideally ones which reduce underinvestments in pro-growth factors such as rural human capital. At the national level, India appears to be an important example of a potential loser from global trade reform in terms of welfare, poverty and inequality. It might consider replacing its current extensive agricultural subsidies and import tariffs with targeted assistance only to the poorest farmers and rural areas (which may also help the urban poor, save government spending on fair price shops to offset the effects of tariffs on food, and reduce the adverse environmental effects of subsidies for irrigation and farm chemicals). Finally, the finding from our ten national case studies that domestic reform is as important as global reform as a way of reducing poverty suggests that developing countries should not hold back domestic reforms while negotiations in the World Trade Organization's Doha Round and other international accords continue. It also suggests that developing countries have little to gain, and potentially much to lose from a poverty alleviating perspective, from negotiating exemptions or delays in domestic reforms in the framework of WTO multilateral agreements. References Alesina, A. and P. Giuliano (2009), `Preferences for Redistribution', NBER Working Paper 14825, Cambridge MA, March (forthcoming in Handbook for Social Economics). Anderson, K. (2009), Distortions to Agricultural Incentives: A Global Perspective, 1955- 2007, London: Palgrave Macmillan and Washington DC: World Bank. 27 Anderson, K., M. Kurzweil, W. Martin, D. Sandri and E. Valenzuela (2008), "Measuring Distortions to Agricultural Incentives, Revisited", World Trade Review 7(4): 1-30, October. Anderson, K., W. Martin and D. van der Mensbrugghe (2006), "Distortions to World Trade: Impacts on Agricultural Markets and Incomes", Review of Agricultural Economics 28(2): 168-94, Summer. Anderson, K. and E. Valenzuela (2008), Global Estimates of Distortions to Agricultural Incentives, 1955 to 2007, data spreadsheets at www.worldbank.org/agdistortions Anderson, K., E. Valenzuela and D. van der Mensbrugghe (2010), "Global Poverty Effects of Agricultural and Trade Policies Using the Linkage Model", Ch. 2 in this volume. Arndt, C. and J. Thurlow (2010), "Mozambique", Ch. 10 in this volume. Atkinson, A.B. and A. Brandolini (2004), "Global World Inequality: Absolute, Relative or Intermediate", Paper prepared for the 28th Conference of the International Association for Research in Income and Wealth, Cork, Ireland, 22-28 August. Bourguignon, F., M. Bussolo and L.P. da Silva (eds.) (2008), The Impact of Macroeconomic Policies on Poverty and Income Distribution: Macro-Micro Evaluation Techniques and Tools, London: Palgrave Macmillan and Washington DC: World Bank. Bourguignon, F., V. Levin, and D. Rosenblatt (2004), "Declining International Inequality and Economic Divergence: Reviewing the Evidence Through Different Lenses", Economic Internationale 4Q: 13-26. Bourguignon, F. and C. Morrison (2002), "Inequality Among World Citizens: 1820-1992," American Economic Review 92(4): 727-44. Bussolo, M., R. De Hoyos and D. Medvedev (2008), "Is the Developing World Catching Up? Global Convergence and National Rising Dispersion", Policy Research Working Paper 4733, World Bank, Washington DC, September. Bussolo, M., R. De Hoyos and D. Medvedev (2010), "Global Income Distribution and Poverty in the Absence of Trade Distortions Using the GIDD Model", Ch. 3 in this volume. Cai, J., A. de Janvry and E. Sadoulet (2008), 'Rural Households in India: Sources of Income and Incidence of Burden of Rising Prices', mimeo, University of California, Berkeley, July. 28 Chen, S. and M. Ravallion (2008), `The Developing World is Poorer Than We Thought, But No Less Successful in the Fight Against Poverty', Policy Research Working Paper 4703, World Bank, Washington DC, August. Cicowiez, M., C. Diaz-Bonilla and E. Diaz-Bonilla (2010), "Argentina", Ch. 12 in this volume. Cockburn, J., B. Decaluwé, I. Fofana and V. Robichaud (2008), 'Trade, Growth and Gender in Developing Countries: Comparison of Ghana, Honduras, Senegal and Uganda', in M. Bussolo and R. De Hoyos (eds.), Gender Aspects of the Trade and Poverty Nexus: A Macro-Micro Approach, London: Palgrave Macmillan and Washington DC: World Bank. Corong, E., C. Cororaton and J. Cockburn (2010), "Philippines", Ch. 8 in this volume. Cororaton, C. and D. Orden (2010), "Pakistan", Ch. 7 in this volume. Cranfield, J.A.L., P.V. Preckel, J.S. Eales and T.W. Hertel (2003), "Estimating Consumer Demand Across the Development Spectrum: Maximum Likelihood Estimates of an Implicit Direct Additivity Model", Journal of Development Economics (68): 289-307. Davis, B., P. Winters and G. Carletto (2009), `A Cross-Country Comparison of Rural Income-Generating Activities', mimeo, Agricultural Development Economics Division, Food and Agriculture Organization, Rome, January. FAO (2008), FAOSTAT database, at http://www.fao.org, accessed June. Ferreira Filho, J. and M. Horridge (2010), "Brazil", Ch. 13 in this volume. Ferreira, F.H.G. and M. Ravallion (2008), `Global Poverty and Inequality: A Review of the Evidence', Policy Research Working Paper 4623, World Bank, Washington DC, May. Francois, J.F. and W. Martin (2007), `Great Expectations: Ex Ante Assessments of the Welfare Impacts of Trade Reforms', mimeo, Erasmus University, Rotterdam, January. Gardner, B. and D. Sumner (2007), U.S. Agricultural Policy Reform in 2007 and Beyond, Washington DC: AEI Press. Grossman, G.M. and E. Helpman (1991), Innovation and Growth in the Global Economy, Cambridge, Mass.: MIT Press. Harrison, A. (ed.) (2007), Globalization and Poverty, Chicago: University of Chicago Press for the NBER. Herault, N. and J. Thurlow (2010), "South Africa", Ch. 11 in this volume. 29 Helpman, E., O. Itskhoki and S.J. Redding (2009), "Inequality and Unemployment in a Global Economy", Discussion Paper 7375, Centre for Economic Policy research, London, July. Hertel, T.W., M. Ivanic, P.V. Preckel and J.A.L. Cranfield (2004), "The Earnings Effects of Multilateral Trade Liberalization: Implications for Poverty", World Bank Economic Review 18: 205-36. Hertel, T.W. and R. Keeney (2010), "Inequality and Poverty Impacts of Trade-related Policies Using the GTAP Model", Ch. 4 in this volume. Hertel, T.W. and L.A. Winters (eds.) (2006), Poverty and the WTO: Impacts of the Doha Development Agenda, London: Palgrave Macmillan and Washington DC: World Bank. Horridge, M. and F. Zhai (2006), "Shocking a Single-Country CGE Model with Export Prices and Quantities from a Rest of the world Model", pp. 94­104 in Hertel and Winters (2006). Ivanic, M. and W. Martin (2008), `Implications of Higher Global Food Prices for Poverty in Low-Income Countries', Policy Research Working Paper 4594, World Bank, Washington DC, April. Krueger, A.O. (2009), `Trade Liberalization and Growth in Developing Countries', in J. Siegfried (ed.), Better Living Through Economics, Cambridge MA: Harvard University Press (forthcoming). Milanovic, B. (2002), "True World Income Distribution, 1988 and 1993: First Calculation Based on Household Surveys Alone", Economic Journal 112(476): 51-92. Milanovic, B. (2005): Worlds Apart: Measuring International and Global Inequality, Princeton: Princeton University Press. Milanovic, B. (2006): "Global Income Inequality: What It Is and Why It Matters," Policy Research Working Paper, World Bank, Washington DC. OECD (2008), The Role of Farm Households and the Agro-Food Sector in the Economy of Rural Areas: Evidence and Policy Implications, TAD/CA/APM/WP(2008)25, Organization for Economic Co-operation and Development, Paris, 18 September. Otsuka, K., J.P. Estudillo and Y. Sawada (eds.) (2009), Rural Poverty and Income Dynamics in Asia and Africa, London and New York: Routledge. 30 Otsuka, K. and T. Yamano (2006), `Introduction to the Special Issue on the Role of Nonfarm Income in Poverty Reduction: Evidence from Asia and East Africa', Agricultural Economics 35 (supplement): 373-97, November. PovcalNet (2008), http://go.worldbank.org/7X6J3S7K90, World Bank, Washington DC, accessed May. Prasad, E.S., K. Rogoff, S.-. Wei and M.A. Kose (2007), `Financial Globalization, Growth and Volatility in Developing Countries', Ch. 11 in A. Harrison (ed.), Globalization and Poverty, Chicago: University of Chicago Press for the NBER. Ravallion, M. (2006), `Looking Beyond Averages in the Trade and Poverty Debate', World Development 34(8): 1374-92, August. Ravallion, M. (2008), `Bailing Out the World's Poorest', Policy Research Working Paper 4763, World Bank, Washington DC, November. Ravallion, M., S. Chen and P. Sangraula (2007), `New Evidence on the Urbanization of Poverty', Policy Research Working Paper 4199, World Bank, Washington DC, April. Rutherford, T.F. and D.G. Tarr (2002), `Trade Liberalization, Product Variety and Growth in a Small Open Economy: A Quantitative Assessment', Journal of International Economics 56(2): 247-72. Sala-i Martin, X. (2006), "The World Distribution of Income: Falling Poverty and Convergence, Period", Quarterly Journal of Economics 121(2): 351-97. Sanchez, M. and R. Vos (2010), "Nicaragua", Ch. 13 in this volume. Topalova, P. (2007), 'Trade Liberalization, Poverty and Inequality: Evidence from Indian Districts', Ch 7 in A. Harrison (ed.), Globalization and Poverty, Chicago: University of Chicago Press for the NBER. Valenzuela, E., D. van der Mensbrugghe and K. Anderson (2009), `General Equilibrium Effects of Price Distortions on Global Markets, Farm Incomes and Welfare', Ch. 13 in Anderson (2009). van der Mensbrugghe, D. (2005), `LINKAGE Technical Reference Document: Version 6.0', Unpublished, World Bank, Washington DC, January 2005. Accessible at www.worldbank.org/prospects/linkagemodel van der Mensbrugghe, D., E. Valenzuela and K. Anderson (2010), "Border Price and Export Demand Shocks for Developing Countries from Rest-of-World Trade Liberalization Using the Linkage Model", Appendix to this volume. Wacziarg, R. and K.H. Welch (2008), `Trade Liberalization and Growth: New Evidence', World Bank Economic Review 22(2): 187-231. 31 Warr, P.G. (2010a), "Indonesia", Ch. 6 in this volume. Warr, P.G. (2010b), "Thailand", Ch. 9 in this volume. Winters, L. A. (ed.) (2007), The WTO and Poverty and Inequality, Volumes I and II, London: Edward Elgar. World Bank (2006), Global Economic Prospects 2006: Economic Implications of Remittances and Migration, Washington DC: The World Bank. World Bank (2007), World Development Report 2008: Agriculture for Development, Washington DC: World Bank. Zhai, F. and T. Hertel (2010), "China", Ch. 5 in this volume. 1 Table 1: Global poverty and inequality, by region, 1981 to 2005 (number and percent of people on less than $1/day in 2005 PPP) 1981 1987 1993 1999 2005 Share of Index of poor (%) income who are inequality rural, (Gini 2002 coefficient) 2004a No. of people (million): Sub-Saharan Africa 157 202 247 299 299 69 n.a. East Asia and Pacific 948 598 600 425 180 85 0.37 of which China 730 412 444 302 106 90 0.36 South Asia 387 384 341 359 350 75 0.35 of which India 296 285 280 270 267 74 0.33 Latin America and Caribbean 27 35 34 40 28 34 0.52 Rest of world 9 9 15 23 22 50 n.a. WORLD 1528 1228 1237 1146 879 74 n.a. East+South Asia's share of world 87 80 76 68 60 Share of population (percent): Sub-Saharan Africa 40 42 44 46 39 East Asia and Pacific 69 39 36 24 10 of which China 74 38 38 24 8 South Asia 42 37 29 27 24 of which India 42 36 31 27 24 Latin America and Caribbean 7 8 7 8 5 WORLD 42 30 27 23 16 a Gini coefficient is the population-weighted cross-country average of national Gini coefficients in the region for the nearest available year to 2004. Source: Chen and Ravallion (2008) except for rural share (Ravallion, Chen and Sangraula 2007) and Gini coefficient (PovcalNet 2008). 2 Table 2: Nominal rates of assistance to tradable agricultural and non-agricultural products, and the relative rate of assistancea focus regions, 1980 to 2004 (percent) 1980-84 1985-89 1990-94 1995-99 2000-04 Africa NRA agric. exportables -35 -37 -36 -26 -25 NRA agric. imp-competing 13 58 5 10 2 NRA agric. tradables -14 0 -15 -9 -12 NRA non-agric. tradables 2 9 3 2 7 RRA -13 -8 -17 -10 -18 South Asiac NRA agric. exportables -28 -21 -16 -12 -6 NRA agric. imp-competing 38 63 25 15 27 NRA agric. tradables 2 47 0 -2 13 NRA non-agric. tradables 55 40 19 15 10 RRA -33 5 -16 -15 3 China and Southeast Asiac NRA agric. exportables -50 -41 -21 -2 0 NRA agric. imp-competing 1 15 3 13 12 NRA agric. tradables -35 -28 -12 5 7 NRA non-agric. tradables 21 23 20 10 6 RRA -43 -42 -26 -4 2 Latin America NRA agric. exportables -27 -25 -11 -4 -5 NRA agric. imp-competing 14 5 19 13 21 NRA agric. tradables -13 -11 4 6 5 NRA non-agric. tradables 19 17 7 7 5 RRA -27 -24 -3 -1 -1 All developing countriesc NRA agric. exportables -41 -36 -19 -6 -3 NRA agric. imp-competing 17 38 23 22 23 NRA agric. tradables -21 -16 -4 4 7 NRA non-agric. tradables 35 27 17 10 6 RRA -41 -34 -18 -5 1 High-income countries NRA agric. exportables 12 22 16 8 7 NRA agric. imp-competing 58 71 62 54 51 NRA agric. tradables 43 56 48 37 34 NRA non-agric. tradables 3 3 3 2 1 RRA 38 51 45 34 32 Source: Anderson and Valenzuela (2008), based on estimates reported in the project's national country studies. a. The RRA is defined as 100*[(100+NRAagt)/(100+NRAnonagt)-1], where NRAagt and NRAnonagt are the percentage NRAs for the tradables parts of the agricultural and non-agricultural sectors, respectively (and NRAagt is the weighted average of the NRAs for the exporting and import- competing sub-sectors of agriculture). 3 Table 3: Effects of full global liberalization of agricultural and all merchandise trade on national economic welfare and real GDP, by country and region, using the Linkage model (percent change relative to benchmark data) All sectors' Agricultural All sectors' policies policies policies Economic Agric Non-ag Agric Non-ag welfare(EV) GDP GDP GDP GDP East and South Asia 0.9 -0.3 0.7 0.5 2.9 of which China 0.2 2.8 0.2 5.7 3.0 India -0.2 -6.1 1.4 -8.3 -0.3 Africa 0.2 0.1 0.8 -0.9 0.0 Latin America 1.0 36.3 2.8 37.0 2.3 All developing countries 0.9 5.4 1.0 5.6 1.9 Eastern Europe & Central Asia 1.2 -4.4 0.3 -5.2 0.3 All high-income countries 0.5 -13.8 0.2 -14.7 0.1 World total 0.6 -1.0 0.4 -1.2 0.5 Source: LINKAGE model simulations from Anderson, Valenzuela and van der Mensbrugghe (2010). 4 Table 4: Effects of full global merchandise trade liberalization on real factor prices, by country and region, using the Linkage model (relative to the benchmark data, percent) Nominal change deflated by Real change in unskilled wages aggregate CPI deflated by: Food and Skilled Capitala Landa Aggregate Food clothing wages user cost user cost CPI CPI CPI East and South Asia 3.4 3.0 -1.8 3.2 4.6 4.8 Africa 4.7 4.3 0.1 4.4 5.8 6.9 Latin America 1.4 1.9 21.1 4.5 2.4 4.1 All developing countries 3.0 2.9 1.6 3.5 5.5 5.9 Eastern Europe & Central Asia 3.2 2.6 -4.5 1.7 4.2 4.5 High-income countries 1.0 0.5 -17.9 0.2 3.3 3.3 World total 1.3 1.2 -3.1 0.9 3.6 3.8 a The user cost of capital and land represents the subsidy inclusive rental cost. Source: LINKAGE model simulations from Anderson, Valenzuela and van der Mensbrugghe (2010). 1 Table 5: Effects of full global merchandise trade liberalization on the number of extreme poor, using the Linkage model, by region Average Baseline Change in number of Change in number unskilled headcount poor from baseline of poor from wage New levels, $1/day New levels, $2/day levels baseline levels change, Number Number reala $1/day $2/day Headcount of poor, Headcount of poor, $1/day, $2/day, $1/day, $2/day, (%) (%) (%) (%) million (%) million million million % % East Asia 4.4 9 37 8 151 34 632 -17 -52 -10.3 -7.6 China 2.1 10 35 9 123 34 440 -5 -12 -4.0 -2.7 Other East Asia 8.1 9 50 6 29 42 192 -12 -40 -30.1 -17.1 South Asia -1.9 31 77 32 454 78 1124 8 8 1.8 0.7 India -3.8 34 80 36 386 82 883 15 15 4.2 1.7 Other South Asia 4.0 29 94 26 68 92 241 -8 -7 -9.9 -2.7 Sub Saharan Africa 5.3 41 72 39 287 70 508 -11 -14 -3.8 -2.7 Latin America 4.1 9 22 8 44 21 115 -3 -6 -6.8 -4.7 Middle East & North Africa 14.3 1 20 1 3 13 40 -2 -19 -36.4 -32.7 Developing country total 5.9 18 48 18 944 46 2462 -26 -87 -2.7 -3.4 Developing excl. China 6.5 21 52 20 820 50 2022 -21 -74 -2.5 -4.7 East Europe & Central Asia 4.5 1 10 1 4 9 43 -0 -4 -6.8 -8.0 a Nominal unskilled wage deflated by the food and clothing CPI Source: LINKAGE model simulations from Anderson, Valenzuela and van der Mensbrugghe (2010). 2 Table 6: Effects of removing agricultural and all merchandise trade distortions on the number of extreme poor, using the GIDD model, by region Share of global Change in no. of poor from global trade reform of: (a)extremely poor poverty Agriculture only All merchandise (<$1 a day) (%) (million) (%) (million) (%) East Asia 24 -6.4 -2.8 -6.3 -2.8 South Asia 50 15.4 3.3 18.2 3.9 Sub-Saharan Africa 21 -1.0 -0.5 0.5 0.3 Latin America 4 -2.8 -6.9 -3.5 -8.7 Globala 100 5.0 0.5 8.9 1.0 (b)moderately and extremely poor (<$2 a day) (%) (million) (%) (million) (%) East Asia 33 -12.8 -1.6 -13.2 -1.7 South Asia 46 -3.6 -0.3 -2.0 -0.2 Sub-Saharan Africa 14 0.1 0.0 1.1 0.3 Latin America 4 -4.8 -4.6 -5.7 -5.4 a Global 100 -22.1 -0.9 -19.8 -0.8 . a Includes Middle East & North Africa, Eastern Europe & Central Asia, and high-income countries, which together account for no more than 2 percent of the world's poor. Source: Bussolo, De Hoyos and Medvedev (2010). 3 Table 7: Effects of removing agricultural and all merchandise trade distortions on global poverty and inequality of farm and non-farm households (percentage point change) Real average monthly income $1 a day $1 a day $2 a day $2 a day Gini (2000, poverty poverty poverty poverty coefficient US$ incidence share incidence share Initial levels: (%) PPP) (%) (%) (%) (%) Agricultural 0.45 65 31.5 76 73.8 70 Nonagricultural 0.63 320 8.3 24 26.7 30 All households 0.67 204 18.9 100 48.2 100 Agricultural liberalization, difference from baseline (percentage points): Agricultural 0.7 1.1a 0.86 1.1 -0.86 0.5 a Nonagricultural -0.1 0.2 -0.29 -1.1 -0.90 -0.5 All households -0.1 0.3a 0.23 0.0 -0.88 0.0 All merchandise trade liberalization, difference from baseline (percentage points): Agricultural 0.8 0.8a 1.09 1.0 -0.66 0.6 a Nonagricultural -0.2 0.4 -0.19 -1.0 -0.95 -0.6 a All households -0.0 0.4 0.39 0.0 -0.82 0.0 a Changes in average income are expressed in percentages. Source: Source: Bussolo, De Hoyos and Medvedev (2010). 4 Table 8: Effects of full global liberalization of agricultural and all merchandise trade on inequality and poverty,a using the GIDD model, by region (percentage point change) Agriculture-only reform All merchandise trade reform Gini <$1 a day <$2 a day Gini <$1 a day <$2 a day coefficient headcount headcount coefficient headcount headcount East Asia -0.72 -0.38 -0.76 -0.62 -0.37 -0.78 South Asia 0.82 1.16 -0.27 0.81 1.37 -0.15 of which: India 1.01 1.49 -0.33 1.04 1.71 -0.26 other S. Asia 0.22 0.06 -0.09 0.02 0.21 0.17 Africa -0.04 -0.23 0.02 0.06 0.11 0.25 Latin America -0.51 -0.61 -1.06 -0.65 -0.77 -1.26 World -0.1 0.23 -0.88 -0.0 0.39 -0.82 a Weighted averages across the included countries for each region. Source: Bussolo, De Hoyos and Medvedev (2010). 5 Table 9: Effects of full global liberalization of agricultural and all merchandise trade on the number of extreme poor, using the GTAP model, by country (percentage point change using $1 a day poverty line) Alternative tax Default tax replacement replacement (poor are exempt) Agriculture-only Nonagriculture- All merchandise All merchandise reform only reform reform reform Asia Bangladesh -0.3 0.5 0.3 -5.3 Indonesia -1.1 0.5 -0.6 -5.2 Philippines -1.4 0.4 -1.0 -6.4 Thailand -11.2 0.9 -10.3 -28.1 Vietnam -0.5 -5.3 -5.7 -23.6 Africa Malawi -1.6 -0.3 -1.9 -5.6 Mozambique -1.2 0.2 -1.0 -4.3 Uganda -0.0 0.1 0.1 -6.0 Zambia -0.0 0.1 0.1 -2.0 Latin America Brazil -2.5 0.4 -2.2 -10.0 Chile -4.8 0.1 -4.6 -12.3 Columbia -0.7 0.6 -0.1 -4.1 Mexico 0.8 0.4 1.1 -0.5 Peru -0.6 -0.2 -0.8 -5.2 Venezuela 0.2 0.7 0.9 -2.1 Unweighted averages: -Asia -2.9 -0.6 -3.5 -13.7 -Africa -0.7 0.1 -0.7 -4.5 -Latin Amer -1.3 0.3 -1.0 -5.7 -All 15 DCs -1.7 -0.1 -1.7 -8.0 Source: Hertel and Keeney (2010, table 5). 6 Table 10: Characteristics of the models in the global and national country studies, by country Chapter Year of SAMa Household Number of Number of Number Labor mobility protection year survey sectors products of labor inter-sectorally data year types Global Linkage 2 2004 2004 n.a. 23 23 2 Yes GIDD 3 2004 2004 c2000 23 23 4 Partial GTAP 4 2004 2001 c2000 31 6 6 Partial National China 5 2004 2002 2000 48 48 8 Partial Indonesia 6 2004 2000 1999 65 20 2 Yes Pakistan 7 2004 `00-01 '01-02 34 28 3 Farm no Philippines 8 2004 2000 2000 41 34 2 Yes Thailand 9 2004 2000 2000 65 65 2 Yes Mozambique 10 2004 2003 2002 56 56 8 No South Africa 11 2004 2002 2000 110 110 3 Unempl't Argentina 12 2004 2005 2005 24 26 6 Unempl't Brazil 13 2004 2001 2001 42 52 10 Yes Nicaragua 14 2004 2000 2001 40 40 4 Unempl't a Social accounting matrix of production and trade data. b There are 20 vintiles for each of agricultural and non-agricultural households in the summary data made publicly available on the GIDD website for each of 73 developing countries (plus a similar number for each of the high-income and transition economies identified in the model), but in the model itself the full distribution of confidential data is used (comprising 1.2 million households). Source: Global and country case studies in Parts I to IV of this volume. 7 Table 11: Impact of reform on the incidence of extreme poverty, by country (percentage point change using national or $1 a day poverty line) (a) rural poverty Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China($2/day) 58 0.3 -1.4 -1.1 0.2 -0.5 -0.3 0.5 -1.9 -1.4 Indonesia 29 0.1 -1.1 -1.1 -0.2 -3.2 -3.3 -0.1 -4.3 -4.4 Pakistan 38 -1.4 -0.1 -1.5 -6.2 -1.1 -7.1 -7.6 -1.2 -8.6 Philippines 49 0.0 -0.6 -0.3 0.6 -0.3 0.2 0.6 -0.9 -0.1 Thailand 30 0.3 -1.6 -1.3 -3.8 0.7 -3.1 -3.5 -0.9 -4.4 Mozambique 36 -1.6 0.0 -1.6 -0.5 -1.5 -2.0 -2.1 -1.5 -3.6 South Africa 17 -0.3 -0.3 -0.7 -0.8 0.0 -0.8 -1.1 -0.4 -1.4 Argentina Brazil Nicaragua 63 -0.7 0.3 -0.4 -0.6 -0.3 -0.9 -1.3 0.0 -1.3 (b) urban poverty Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China($2/day) 3 0.0 0.0 0.0 0.0 -0.1 -0.1 0.0 -0.1 -0.1 Indonesia 12 -0.1 -0.3 -0.4 -0.1 -1.7 -1.8 -0.2 -2.0 -2.2 Pakistan 20 -2.4 -0.1 -2.7 4.7 -1.4 3.1 2.3 -1.5 0.4 Philippines 19 0.8 -0.9 -0.2 1.2 -0.7 0.3 2.0 -1.6 0.1 Thailand 6 0.0 -0.8 -0.7 -3.3 0.2 -3.2 -3.3 -0.6 -3.9 Mozambique 37 -0.5 0.0 -0.5 -0.4 -1.3 -1.7 -0.9 -1.3 -2.2 South Africa 4 -0.1 -0.2 -0.3 -0.4 0.0 -0.4 -0.5 -0.2 -0.7 Argentina 13 1.3 0.1 1.5 -0.4 -0.1 -0.5 0.9 0.0 1.0 Brazil Nicaragua 27 0.3 -0.5 -0.2 -1.0 1.4 0.4 -0.7 0.9 0.2 8 Table 11 (continued): Impact of reform on the incidence of extreme poverty, by country (percentage point change using national or $1 a day poverty line) (c) total poverty Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China($2/day) 36 0.2 -0.8 -0.6 0.1 -0.4 -0.3 0.3 -1.2 -0.9 Indonesia 23 -0.0 -0.8 -0.8 -0.1 -2.7 -2.8 -0.1 -3.5 -3.6 Pakistan 31 -1.6 -0.1 -1.8 -3.6 -1.2 -4.6 -5.2 -1.3 -6.4 Philippines 34 0.4 -0.6 -0.1 0.7 -0.3 0.2 1.1 -0.9 0.1 Thailand 14 0.1 -1.1 -0.8 -3.5 0.4 -3.3 -3.4 -0.7 -4.1 Mozambique 36 -1.3 0.0 -1.3 -0.4 -1.4 -1.8 -1.7 -1.4 -3.1 South Africa 10 -0.2 -0.3 -0.5 -0.6 -0.1 -0.6 -0.8 -0.3 -1.1 Argentina Brazil 31 -0.5 -2.3 -2.8 -0.4 -0.1 -0.5 -0.9 -2.4 -3.5 Nicaragua 41 -0.1 -0.2 -0.3 -0.9 0.8 -0.1 -1.0 0.6 -0.4 Unweighted averages: -Asia 28 -0.2 -0.7 (-2.9)-0.8 -1.2 -0.8 (-0.6)-2.2 -1.5 -1.6 (-3.5)-3.0 -Africa 32 -0.8 -0.2 (-0.7)-0.9 -0.5 -0.7 (0.1)-1.2 -1.3 -0.9 (-0.7)-2.1 -Latin Am. 36 -0.3 -1.3 (-1.3)-1.6 -0.7 0.4 (0.3)-0.3 -1.0 -0.9 (-1.0)-2.0 -All 9 DCs 43 -0.4 -0.6 (-1.7)-1.0 -0.9 -0.6 (-0.1)-1.5 -1.3 -1.2 (-1.7)-2.6 a Numbers in italics for individual countries are implied assuming linearity holds; numbers do not always add because of either rounding or interaction effects Source: Country case studies in Parts II to IV of this volume plus (in the case of the unbolded numbers in brackets in the final 4 rows), from Hertel and Keeney (2010) as reported in the last 4 rows of table 1.9 above. 9 Table 12: Impact of reform on the incidence of income inequality, by country (percentage point change in Gini Coefficient) (a) rural Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China 0.32 0.0 -0.2 -0.2 0.0 0.0 0.0 0.0 -0.2 -0.2 Indonesia 0.29 0.0 0.0 0.0 0.1 0.0 0.1 0.1 0.0 0.1 Pakistan 0.26 -0.1 -0.0 -0.1 0.3 0.0 0.3 0.2 -0.0 0.2 Philippines 0.43 0.2 -0.1 0.1 0.3 0.0 0.1 0.5 -0.1 0.2 Thailand 0.33 0.0 0.5 0.5 0.4 0.0 0.4 0.4 0.5 0.9 Mozambique South Africa 0.63 -0.1 -0.1 -0.2 -0.3 0.0 -0.3 -0.4 -0.1 -0.5 Argentina Brazil Nicaragua (b) urban Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China 0.26 0.0 0.1 0.1 0.0 -0.1 -0.1 0.0 0.0 0.0 Indonesia 0.36 0.0 -0.1 -0.1 0.3 0.3 0.6 0.3 0.2 0.5 Pakistan 0.40 -0.1 -0.0 -0.1 -1.9 0.0 -1.9 -2.0 -0.0 -2.0 Philippines 0.48 0.3 -0.2 0.1 0.1 0.0 0.1 0.4 -0.2 0.2 Thailand 0.15 0.1 0.6 0.7 0.5 0.0 0.5 0.6 0.6 1.2 Mozambique South Africa 0.62 -0.1 -0.1 -0.2 -0.5 0.0 -0.5 -0.6 -0.1 -0.7 Argentina 0.50 0.3 -0.1 0.2 -0.2 -0.1 -0.3 0.1 -0.2 0.0 Brazil Nicaragua 10 Table 12 (continued): Impact of reform on the incidence of income inequality, by country (percentage point change in Gini Coefficient) (c)total Base Agriculture-only reform Nonagriculture-only reform All merchandise reform (%) Unilateral R of W Global Unilateral R of W Global Unilateral R of W Global China 0.44 0.1 -0.4 -0.3 0.0 -0.1 -0.1 0.1 -0.5 -0.4 Indonesia 0.34 0.0 -0.1 -0.1 0.2 0.2 0.4 0.2 0.1 0.3 Pakistan 0.34 -0.1 -0.0 -0.2 -3.2 -0.1 -3.1 -3.3 -0.1 -3.3 Philippines 0.51 0.3 -0.2 0.1 0.1 0.0 0.1 0.4 -0.2 0.2 Thailand 0.34 0.1 0.7 0.8 0.4 0.0 0.4 0.5 0.7 1.2 Mozambique 0.48 -1.2 -0.1 -1.3 -0.3 0.2 -0.1 -1.5 0.1 -1.4 South Africa 0.67 -0.1 -0.1 -0.2 -0.4 0.0 -0.4 -0.5 -0.1 -0.6 Argentina Brazil 0.58 -0.2 -1.4 -1.6 0.1 -0.1 0.0 -0.1 -1.5 -1.7 Nicaragua 0.53 -0.1 0.1 0.0 -0.1 -0.2 -0.3 -0.2 -0.1 -0.3 Unweighted averages: -Asia 0.39 0.1 -0.0 0.1 -0.5 0.0 -0.5 -0.4 -0.0 -0.4 -Africa 0.58 -0.7 -0.1 -0.8 -0.4 0.1 -0.3 -1.0 -0.0 -1.0 -Latin Am. 0.56 -0.2 -0.7 -0.8 0.0 -0.2 -0.1 -0.2 -0.8 -1.0 -All 9 DCs 0.59 -0.2 -0.2 -0.4 -0.3 -0.0 -0.3 -0.5 -0.2 -0.7 a Numbers in italics are implied assuming linearity holds; numbers do not always add because of either rounding or interaction effects Source: Country case studies in Parts II to IV of this volume. 11 Table 13: Direction of effects of global reform on extreme poverty, by countrya (sign of change in national population under $1 a day or national poverty line) Agriculture-only reforms All merchandise trade reform GIDD GTAP National Linkage GIDD GTAP National model model model model model model model Brazil - - - - - - Chile + - - - China - - - Colombia - - - - India + + + Indonesia - - - - - - Mexico - + - + Mozambique - - - - Nicaragua - - - - Pakistan + - + - Peru - - - - Philippines - - - - - + South Africa + - + - Thailand - - - - - - Uganda - - + + Venezuela - + - + Vietnam - - - - a These are the only countries in this study for which results are available from at least two of the models reported in the subsequent chapters. Source: Chapters in Parts I to IV of this volume.