WPS4013 The Role of Agriculture in Poverty Reduction An Empirical Perspective Luc Christiaensen, Lionel Demery and Jesper Kühl1, 2 Abstract: The relative contribution of a sector to poverty reduction is shown to depend on its direct and indirect `growth effects' as well as its `participation effect'. The paper assesses how these effects compare between agriculture and non-agriculture by reviewing the literature and by analyzing cross-country national accounts and poverty data from household surveys. Special attention is given to Sub-Saharan Africa. While the direct growth effect of agriculture on poverty reduction is likely to be smaller than that of non-agriculture (though not because of inherently inferior productivity growth), the indirect growth effect of agriculture (through its linkages with non- agriculture) appears substantial and at least as large as the reverse feedback effect. The poor participate much more in growth in the agricultural sector, especially in low-income countries, resulting in much larger poverty reduction impact. Together, these findings support the overall premise that enhancing agricultural productivity is the critical entry-point in designing effective poverty reduction strategies, including in Sub-Saharan Africa. Yet, to maximize the poverty reducing effects, the right agricultural technology and investments must be pursued, underscoring the need for much more country specific analysis of the structure and institutional organization of the rural economy in designing poverty reduction strategies. JEL classification: D3, O1 Keywords: Agriculture, Economic Growth, Poverty Reduction, Sub-Saharan Africa World Bank Policy Research Working Paper 4013, September 2006 The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Policy Research Working Papers are available online at http://econ.worldbank.org. 1The paper is prepared as part of a multi-country empirical investigation to re-assess the role of agriculture in poverty reduction in Sub-Saharan undertaken by the Africa Region in the World Bank and financed under the Norwegian ESSD Trust Fund. The case studies, undertaken in Ethiopia, Kenya, Madagascar and Tanzania, take a mainly micro- economic perspective, and typically utilize household and farm surveys. A companion paper synthesizes the results of these studies. 2The authors are at the World Bank and can be contacted at lchristiaensen@worldbank.org, ldemery@worldbank.org, jesper_kuehl@yahoo.dk. Comments by Martin Ravallion and Derek Byerlee helped improve an earlier draft and are gratefully acknowledged. 2 1 Introduction While it has long been recognized that economic development is inextricably linked to agriculture, there has been little consensus about its precise role. The dual economy models inspired by Lewis (1954) and popular in development economics in the 1960s and the 1970s typically featured agriculture as a backward, subsistence sector. In this view, resources were to be drawn from the unproductive agricultural sector to encourage development of the productive industrial sector. Much of the early development economics literature was thus interpreted as supporting an industrialization strategy, leading to an urban bias in development planning (Lipton, 1977), and fiscal and trade systems that systematically over-taxed agriculture (Krueger et al., 1988). A more positive view on the role of agriculture in development (especially during the early stages) emerged later, following the seminal contributions by Johnston and Mellor (1961) and Schultz (1964). They emphasized the critical contributions of the agricultural sector to growth in the non-agricultural sectors, implying that investments and policy reforms in agriculture might actually yield faster overall economic growth, even though agriculture itself might grow at a slower pace than non-agriculture. Since then several authors have found that the multiplier effects from agriculture to non-agriculture are indeed substantial, especially in Asia, but also in Sub-Saharan Africa (SSA) (Haggblade, Hammer and Hazell, 1991; Delgado et al., 1998). The experience of the Green Revolution in Asia, whereby traditional agriculture was rapidly transformed into a fast growing modern sector through the adoption of science based technology, provided further confidence in the proposition of agriculture as an engine of growth. More recently, the development community has shifted its focus from fostering economic growth per se to maximizing poverty reduction, or achieving `shared' growth--growth with a maximum pay-off in terms of poverty reduction (World Bank, 2005a). This has added a new dimension to the debate about the relative role of agriculture versus non-agriculture, as poverty reduction not only depends on the rate of overall economic growth, but also on the ability of poor 3 people to connect to that growth (i.e. the `quality' of growth). As the majority of poor people in the developing world (and especially in SSA) depend directly on agriculture for their livelihood, it is often argued that agricultural growth has a higher return in terms of poverty reduction (i.e. a higher `participation effect') than an equal amount of growth in non-agriculture. Both the growth and the participation effects continue to be hotly debated for each sector, especially in the African context. On the growth side, some contend that agricultural productivity growth is central to sustainable economic development (Mellor, 1976; Timmer, 2005). Others hold that for Africa at least, the classical intersectoral linkages no longer apply, and a pro-agriculture strategy will not deliver the overall growth necessary for rapid poverty reduction. On the participation side, the sheer weight of numbers, with the majority of poor people depending on agriculture, suggests that agriculture will deliver a greater participation effect. But it is also argued that African agricultural development will not involve the majority of poor smallholder farmers, but can only succeed among larger commercial farmers (Maxwell, 2004). The extent to which poor people would gain from a pro-agriculture strategy is questionable in this view. Understanding how these counterbalancing forces play out in terms of poverty reduction across sectors is central to the development of effective poverty reducing strategies. Yet, to further this debate, an empirical perspective is needed, focusing on three key questions: 1) Do investment and policy reform in agriculture enhance overall growth more than investment and policy reform in non-agriculture? 2) Is participation by the poor in agricultural growth on average higher than their participation in non-agricultural growth, and if so, under what conditions? 3) If a focus on agriculture would tend to yield slower overall growth, but larger participation by the poor, compared with a focus on non-agriculture, which strategy would tend to have the largest pay-off in terms of poverty reduction, and under which circumstances? These are the central issues addressed in this paper. To do so, it complements the empirical insights from the literature on historical experiences in Asia and Latin America (Ravallion and Datt, 1996, 2002; Bravo-Ortega and Lederman, 2005; World Bank, 2005b) with cross-country 4 analysis using national accounts evidence on sectoral growth combined with poverty data from household surveys. Special attention is given to SSA, though the evidence we bring to the debate covers the wider developing world. The paper begins by developing a simple conceptual framework (section 2) in which the effects of agriculture and non-agriculture on poverty are shown to arise from two principal sources: a growth effect and a participation effect. The paper then examines the direct and indirect growth effects across both sectors in sections 3 and 4, followed by an assessment of potential differences in the participation component in section 5. Section 6 synthesizes how these different effects are expected to play out in terms of poverty reduction across different groups of countries and concludes. 2 Conceptual Framework Let Pi be any (decomposable) measure of poverty and Yi per capita Gross Domestic Product (GDP) in country i. The proportionate change in poverty in a country i can then be seen to be identical to the GDP elasticity of poverty (defined as the proportionate change in poverty divided by the proportionate change in per capita GDP)3, times the proportionate change in per capita GDP (Yi): dPi (1) Pi dPi Yi dYi Pi dYi Yi We refer to the first multiplicative term in (1) as the participation effect and the second multiplicative term as the growth effect. Not all growth processes generate an equal amount of overall growth nor an equal amount of poverty reduction (World Bank, 2000). The growth and participation effects may differ substantially across sectors. The latter has been illustrated empirically for India by Ravallion and 3Note, by using GDP growth rather than mean household income change as is common in this sort of identity, we are very much focusing on the overall growth process (not simply the growth in household income). The elasticity concept used here reflects the impact of growth on both average incomes of households and how those incomes are distributed. This is commonly referred to as the "growth elasticity of poverty" which is strictly speaking not correct. 5 Datt (1996; 2002) and for China by Ravallion and Chen (2004). To accommodate such differences, we rewrite (1) as a weighted sum of the contributions to poverty reduction of both the agricultural and the non-agricultural sectors: dPi (2) Pi q dPi Yai dYai Pi dYai Yai + (1- q) dPi Yni dYni Pi dYni Yni with a denoting agriculture, n non-agriculture, and q any constant (1 0.1). For SPG the null hypothesis cannot be rejected. In sum, agricultural growth contributes significantly more to poverty reduction across our sample of countries, at least as measured by the H and PG measures. Whether the composition of growth matters for reducing extreme poverty (as reflected in the SPG measure), is less certain. 23This is similar to other comparable approaches, for instance Ravallion and Datt (1996) and Adams (2004). 25 Interestingly, the constant term is not significantly different from zero in most of the empirical specifications reported in Table 6. With zero growth, poverty is predicted on average to remain unchanged, implying constancy on average in income/consumption inequality.24 But for the SPG measure, the constant term is significantly negative, suggesting favorable changes at the bottom of the distribution. There inequality changes appear to exert a downward pressure on poverty when there is no growth. To explore further whether the composition of growth matters for poverty reduction across the development spectrum, we split the sample in two groups of equal size based on the country's GDP per capita and run separate regressions for the low and middle-income countries.25 For the middle-income countries, the null hypothesis is rejected (see panel C of Table 6). Agricultural growth has a significantly greater impact on poverty (a being greater than n by a factor of 13.8 for the H and 10.2 for the PG measure). But as with the whole sample, the evidence of the differential impact of sectoral growth on the SPG measure is less certain. While both the coefficients on agricultural and non-agricultural growth are statistically significant (at the 10% level), they are not statistically different, despite the fact that the coefficient on agricultural growth is 7.6 times larger than the one on non-agricultural growth. 24We also applied a Gini correction to the GDP growth variables (following Ravallion, 1997) using initial year Ginis of each spell. We obtained qualitatively very similar results to those reported in Table 6. 25Taking the pooled sample, this resulted in a cut-off of US$ 1,160 GDP/capita. 26 Table 6 : Decomposition of poverty changes Middle-income All countries countries Low-income countries Coef Coef- Coef- p- Coef- p- Coef- p- ficient p-value ficient p-value ficient value ficient value ficient value A B C D E Headcount index (H) GDP/pc growth -1.68 0.00 - - - - Non agriculture pc growth -0.98 0.05 -1.57 0.01 0.12 0.89 0.12 0.91 Non-ag pc growth*SSA - - - - -0.08 0.95 Agriculture pc growth -9.35 0.02 -21.74 0.00 -6.00 0.05 -12.95 0.06 Agric pc growth*SSA - - - - 7.07 0.21 Constant -0.05 0.27 -0.06 0.22 -0.07 0.43 -0.06 0.27 -0.03 0.55 Number of observations 222 222 111 111 111 R2 0.08 0.14 0.26 0.10 0.13 Ho: test ag=nag 0.04 0.00 0.08 0.07 Ho: test nag+nag*X=ag+ag*X - - 0.31 Poverty gap index (PG) GDP/pc growth -2.03 0.00 - Non agriculture pc growth -1.32 0.04 -2.07 0.00 0.05 0.96 -0.05 0.98 Non-ag pc growth*SSA - - - - - 0.23 0.89 Agriculture pc growth -9.99 0.02 -21.15 0.01 -7.33 0.06 -9.99 0.14 Agric pc growth*SSA - - - - - 4.79 0.51 Constant -0.10 0.11 -0.11 0.09 -0.13 0.16 -0.10 0.25 -0.07 0.39 Number of observations 222 222 111 111 111 R2 0.07 0.10 0.18 0.08 0.09 Ho: test ag=nag 0.06 0.02 0.10 0.18 Ho: test nag+nag*X=ag+ag*X - - - - 0.18 Squared poverty gap index (SPG) GDP/pc growth -2.22 0,00 Non agriculture pc growth -1.60 0.05 -2.36 0.02 -0.22 0.87 -0.51 0.80 Non-ag pc growth*SSA - - - - - 0.78 0.70 Agriculture pc growth -9.36 0.07 -18.00 0.10 -7.63 0.10 -8.29 0.33 Agric pc growth*SSA - - - - 1.07 0.91 Constant -0.15 0.05 -0.16 0.04 -0.20 0.10 -0.14 0.18 -0.12 0.26 Number of observations 222 222 111 111 111 R2 0.05 0.06 0.09 0.06 0.06 Ho: test ag=nag 0.16 - 0.16 0.18 0.42 Ho: test nag+nag*X=ag+ag*X - - - - 0.13 *10%, **5%,***1% significance All estimations are corrected for heteroskedasticity with robust (cluster) Source: Own calculations based on World Bank data (2005c; 2005d) However, in the low-income countries (for all measures of poverty--see panel D of Table 6) only agricultural growth appears to affect poverty reduction--and the null hypothesis is therefore rejected. The estimated effect of non-agricultural growth on poverty is not statistically 27 significant.26 In this context it is especially worth highlighting how sectoral growth affects the poorest differently in the medium and low-income countries (as measured by changes in the SPG). While both agricultural and non-agricultural growth offer scope for extreme poverty reduction in the middle-income countries, it is only agricultural growth that appears to affect the poorest in the low-income countries. This may suggest that the poorest groups in the middle-income countries are more likely to rely also on non-agricultural activities--possibly because extreme poverty is associated with landlessness and concentrated in urban areas as in many Latin American countries which make up more than 40 percent of our middle-income group. This result also resonates somewhat with Bravo-Ortega and Lederman's (2005) finding that growth in agricultural output per worker was slightly less effective as growth in non-agricultural output per worker in raising the incomes of the poorest quintile, at least where it concerns the middle-income countries. Finally we consider the sectoral impacts in Sub-Saharan African (SSA) countries. Of the 111 observations in our low-income sample, 36 are from SSA. It is not clear, therefore, that the findings for the low-income group would necessarily apply to SSA.27 It would not be appropriate to estimate the poverty regressions separately for SSA, given the small sample. Our approach is to apply an SSA interaction term to the right-hand-side variables in the low-income country data. The results are presented in panel E of Table 6. As in case of the low-income countries, we find that only agricultural growth affects poverty reduction, while the estimated coefficients on non- agricultural growth are not statistically different from zero, supporting the critical role of agricultural growth in poverty reduction in SSA. None of the SSA interactive terms is statistically significant, indicating that the relationship between sectoral growth and poverty in low-income countries also applies to the Sub-Saharan-Africa group of countries covered in our sample. 26 The non-significance of non-agricultural growth for poverty reduction in low-income countries (and the positive sign on some of the estimated coefficients) might be the result of counteracting effects of growth within non- agriculture. Indeed, re-estimation of equation 6 using further disaggregated measures of non-agriculture into industrial and service growth, indicates that industrial growth is positively associated with poverty changes--it increases poverty, while service growth is negatively associated. Yet, consistent with the results reported here, only the coefficient on agricultural growth is statistically significant and the coefficients on industrial and service growth are neither jointly nor individually statistically significant. This holds across all poverty measures for the low-income countries. Results are available upon request from the authors. 27Two countries--China and India-- between them contribute 30 observations to the low-income country sample. 28 The data reported in Table 6 give the response of total poverty to changes in the share weighted growth rates of the sectors. Estimates of the participation effects for each sector can then be obtained by simply multiplying the estimated coefficients (columns B-D, Table 6) by the sectoral shares for each country. The results (as averages per region) are reported in Table 7. We find the participation effect of agricultural growth on average across the world to be 2.2 times (=1.72/0.80) larger than the participation effect of non-agriculture. In other words, one percentage point additional growth in agricultural GDP per capita would reduce the poverty headcount on average 2.2 times more than an additional percentage point growth in non-agriculture GDP per capita. This broad finding lends support to a continued emphasis on fostering agricultural growth in reducing poverty especially given that the growth linkage effects from agriculture to non- agriculture tend to be at least as large as the reverse feedback effects. Moreover, historical evidence from both developed (Bernard and Jones, 1996) and developing (Martin and Mitra, 2001) countries indicating that agricultural productivity (as captured by TFP) has been growing at least as fast as non-agricultural productivity supports the view of agriculture as a dynamic sector with substantial growth potential which would help release labor from agriculture to non-agriculture. 29 Table 7: Decomposition of the participation effect of sectoral growth with respect to head count poverty into its share and elasticity components Region # of GDP share (%) Estimated coefficient Participation effect of countries growth on head count poverty Agric Nonag Agric Nonag1) Agric Non-agric1) Low-income group SSA 18 31 70 -6.00 - -1.83 - South Asia 4 29 71 -6.00 - -1.73 - East Asia & Pacific 6 24 76 -6.00 - -1.44 - Eastern & Central Europe 4 26 74 -6.00 - -1.57 - Latin America & Caribbean 4 19 82 -6.00 - -1.11 - Middle East & North Africa 2 15 85 -6.00 - -0.92 - ALL LOW-INCOME 38 27 73 -6.00 - -1.60 - Middle-income SSA2) 2 4 97 -21.74 -1.57 -0.76 -1.52 South Asia 0 East Asia & Pacific 2 13 87 -21.74 -1.57 -2.76 -1.37 Eastern & Central Europe 12 10 91 -21.74 -1.57 -2.07 -1.42 Latin America & Caribbean 16 10 90 -21.74 -1.57 -2.13 -1.42 Middle East % North Africa 5 14 86 -21.74 -1.57 -3.07 -1.35 ALL MIDDLE-INCOME 37 10 90 -21.74 -1.57 -2.24 -1.41 Pooled sample SSA 20 28 72 -9.35 -0.98 -2.66 -0.70 South Asia 4 29 71 -9.35 -0.98 -2.70 -0.70 East Asia & Pacific 8 22 78 -9.35 -0.98 -2.03 -0.77 Eastern & Central Europe 16 14 86 -9.35 -0.98 -1.27 -0.85 Latin America & Caribbean 20 11 89 -9.35 -0.98 -1.03 -0.87 Middle East & North Africa 7 14 86 -9.35 -0.98 -1.34 -0.84 ALL POOLED 75 18 82 -9.35 -0.98 -1.72 -0.80 1) The estimated coefficients on the effect of (share weighted) non-agricultural growth in the low-income countries are not statistically significantly different from zero. 2) The two middle-income countries are Botswana and South Africa Source: Authors' calculations based on World Bank data (2005c) While an overall focus on fostering agricultural growth in reducing world poverty appears justified from the broad average perspective, the results in Table 7 also underscore the critical need to look beyond the averages and explore the size of the participation effect across regions and countries. First, as discussed above, the elasticity of total poverty reduction with respect to sectoral GDP (i.e. the estimated coefficients) differs substantially between the middle and low-income countries where the effect of agricultural growth on overall poverty was found to be much more important. Second, the larger the share of agriculture in the total economy the more important the participation effect from agriculture. The combined effect of these two forces in the middle-income 30 countries results in the participation effect of agriculture on head count poverty being on average 1.6 times (-2.24/-1.41) larger than that of non-agriculture, while it is on average multiple times larger in the low-income countries. 6 The Relative Contribution of Agriculture and Non-Agriculture to Poverty Reduction ­ Evidence from the Recent Past From equations (4) and (6) we know that poverty reduction during a certain period can be decomposed into sectoral participation and growth effects. We estimated the participation effects of the different sectors and concluded that one percent of agricultural growth yields on average 2.2 times more poverty reduction than one percent growth in non-agriculture. While agriculture could potentially grow faster, it is likely to continue to grow at a slower pace than non-agriculture due to Engel's Law. But the indirect effects of agricultural growth on non-agriculture are substantial and likely at least as large as the reverse feedback effects. Whether investments in agriculture in a particular country would generate faster or slower overall economic growth than investments in non-agriculture is a priori not clear. This would depend on the structure of the economy and the governing institutional arrangements.28 How these potentially counterbalancing forces (potentially slower overall growth from investments in agriculture against a much larger participation effect) play out, remains an empirical question. To shed more light on this, we explore how agriculture and non-agriculture contributed to poverty reduction over the past two decades in the countries in our PovCal sample. In particular, we revisit equation (6) and estimate the relative contribution of each sector to the (predicted) change in (US1$/day) poverty incidence in these countries. To do so, we apply the estimated coefficients from the pooled data set reported in Table 6 column B to the (share weighted) sectoral GDP growth rates in our PovCal sample. The poverty spells in our sample concern the 1980-2000 period, with about two thirds of the spells occurring in the 1990s. It is especially useful to examine 28These two forces may further affect the country specific size of the participation effect as well, as will be discussed below. 31 the relative contribution of agriculture and non-agriculture to poverty reduction during this period, as it coincides with the increasing liberalization and globalization of the world economy. These evolutions might affect the feedback effects of agriculture to non-agriculture, as well as the participation effects, if globalization induced a greater correlation between domestic and international food prices and a change in the farming structures through increased vertical integration. The effect of non-agricultural growth and the constant (a measure of the effect of inequality change) is retained in the decomposition, even though their estimated coefficients are not statistically significant. Observations where the share of a sector in poverty reduction exceeds |10| are excluded, resulting in a loss of nine observations (from 222 to 213). Table 8: Sectoral decomposition of changes in headcount poverty 1) Average share to poverty No. of Non Agriculture Inequality reduction of obser- agriculture change vations Continent Sub-Saharan Africa 37 0.342 0.623 0.035 South Asia 20 0.316 0.437 0.247 East Asia & Pacific 43 0.430 0.215 0.355 East Europe & Central Asia 27 0.378 0.451 0.171 Latin America & the Caribbean 73 0.388 0.403 0.209 Middle East & North Africa 13 0.593 -0.109 0.515 Total 213 0.393 0.382 0.226 Spells with sectoral shares exceeding |10| excluded. Bolded shares are based on statistically significant regression coefficients. Source: authors' calculations The results confirm that despite its slower (direct) growth record, agriculture played a major role in the evolution of poverty during the 1980-2000 period (Table 8). On average just under 40 percent of the change in poverty incidence across the world was attributable to growth in agricultural GDP--as much as growth in both industry and services combined. Even so, this is likely an underestimate, as the decompositions are based on contemporaneous growth rates in agriculture and non-agriculture. As a result, the contribution of agriculture to poverty reduction through its effect on growth in non-agriculture is attributed to the non-agricultural sector in this decomposition exercise. To the extent that the indirect feedback effect from agriculture to non- 32 agriculture exceeds the feedback effect from non-agriculture to agriculture, the contribution from agriculture to poverty reduction will be underestimated. In SSA two thirds of the predicted poverty change could be attributed to agriculture, underscoring the continuing critical importance of fostering agricultural growth in SSA for poverty reduction. While the evidence presented so far supports a focus on fostering agricultural growth as the starting point in designing poverty reducing strategies, especially in most low-income countries, there are differences to be expected in the size of the participation and indirect growth effects across countries depending not only on the share of the sector but also on the structure of the economy (e.g. equal versus unequal distribution of assets; a small open versus a large closed economy) and its institutional organization (e.g. functioning of labor and commodity markets). Our inequality corrected estimates of the participation effects indicate for example that a 0.20 increase in the national Gini (corresponding roughly to the difference in equality between Ethiopia and Zambia) would decrease the participation effect of agriculture in low-income countries on average from -1.60 to -0.96.29 Growth linkages on the other hand are likely to be smaller in small, open economies (such as Lesotho) with small elastically supplied non-tradable sectors, while countries such as Cameroon, Nigeria and Tanzania with large non-tradable agriculture and service sectors are likely to encounter large growth linkages from investments in tradables (Dorosh and Haggblade, 2003). The choice of the agricultural technology (e.g. focused on non tradable food versus tradable export crops; land versus labor saving) and its targeting (small versus large farms) can also substantially affect the size of the participation and indirect growth effects of technological change on poverty. Using CGE models applied to archetype economies de Janvry and Sadoulet (2002) emphasize for example that improvements in agricultural technology primarily affect poverty in SSA through their direct growth effects. This suggests that technological change should focus on small farmer production system though with an appropriate balance in enhancing productivity 29The coefficient on the share weighted Gini corrected agricultural and non-agricultural growth rates in low-income countries were estimated at -8,82 and -0,72 respectively, though the latter was not statistically significant. 33 among tradable crops (often export crops) and non-tradables (most often food staple crops) to avoid falling price effects.30 In Asia on the other hand, where the landless account for an important share of the poor and the labor markets are much better developed, technological change for maximum employment creation (i.e. land saving and not labor saving technologies) is advised for maximal poverty reduction. Benefits from agricultural technological change come mostly from linkage effects through the rest of the economy in Latin America, leading to the possibility that technological changes in the fields of large farmers can be more beneficial to the poor farmers than the direct effects derived from technological change on their own farm. 7 Concluding Remarks To analyze the role of agriculture in poverty reduction, the paper developed a simple conceptual framework in which the relative contribution of a sector to poverty reduction is shown to depend on four factors: its direct (1) and indirect (2) growth effects as well as its elasticity of total poverty to sectoral growth (3) and the sector's share in the overall economy (4) which together determine the sector's participation effect. Reviewing the evidence on the growth contributions of agriculture and the non-agricultural sector, it emerges that while the agricultural sector on average has grown more slowly than non-agriculture, this appears to be largely due to a migration out of agriculture into non-agriculture. The latter could simply be due to equilibrating labor movements out of agriculture into non-agriculture in response to more productive opportunities in non- agriculture (industrial pull). Yet, the more rapid increase in agricultural TFP historically observed in both developed and developing countries lends support to the view of agriculture as a dynamic sector with substantial growth potential whereby productivity increases in agriculture induce labor to move out of agriculture (agricultural push). In addition, both the micro and the cross-country evidence indicate that the indirect or growth linkage effects from agriculture to non-agriculture appear quantitatively large and at least as 30See World Bank (2005e) for a detailed discussion of this argument in Ethiopia. 34 large as the reverse feedback effect. While the evidence further suggests that these indirect growth effects are likely to be smaller in SSA compared with the rest of the developing world, they appear nonetheless quantitatively important. In sum, while the direct growth contribution of agriculture is on average likely to be smaller, this is often likely to be compensated by its contribution to non- agriculture growth through the linkage effects which tends to be at least as large as the reverse feedback effect. In evaluating the potential growth contribution to poverty reduction from investment in agriculture in a particular country, it is thus critical to account for both the contemporaneous direct effects as well as the lagged indirect effects. We find the participation effect from agriculture on the poverty head count on average to be 2.2 times larger than the participation effect from non-agriculture. This difference does not primarily follow from the large share of agriculture in these economies, but rather from the much larger elasticity of overall poverty to agricultural GDP than to non-agricultural GDP. This also holds for the middle-income countries where the participation effect of agricultural growth on head count poverty is on average 1.6 times larger than that of the other sectors. However, the poorest groups in these countries appear to be equally well served by growth in the agricultural and non- agricultural sectors. The difference in participation by poor people in growth in agriculture versus growth in non-agriculture is especially pronounced in the low-income countries, with the poverty gains from growth in agriculture multiple times larger on average than those from growth in non- agriculture, irrespective of the poverty measures used. These results also hold for SSA. The much stronger participation effects from agriculture in low-income countries clearly outweigh its potentially lower contribution to overall growth, lending support to a concerted focus on fostering agricultural growth in reducing poverty in these countries (including SSA). This is borne out by the more recent historical experience of the past two decades. It is also consistent with the more forward looking in-depth country specific evidence from four Sub-Saharan countries by Dorosh and Haggblade (2003) who find that investments in agriculture favor the poor more than similar investments in manufacturing. The evidence presented in this paper thus supports the 35 overall premise that enhancing agricultural productivity is a critical starting point in designing effective poverty reduction strategies, especially in low-income countries. This raises the question of how this can and should be achieved, especially in SSA, where agricultural productivity growth has been lagging. While a comprehensive treatment of this question falls beyond the scope of this paper, we conclude highlighting the following three points. First, the debate about investment in agriculture versus non-agriculture is often misleading, especially when applied to rural areas, as many public investments (especially rural roads but also the provision of education and health services) are equally necessary for the development of the farming and the (rural) non-farming sector (Fan, Hazell, and Thorat, 2000). Second, when it comes to agricultural specific interventions, the important potential of increased investment in agricultural R&D and extension, accounting for the great diversity of farming systems in SSA, cannot be sufficiently underscored. Agriculture is typically an atomistic industry. There are therefore few incentives for private investment in research. Relatedly, given the limited use of technology in African agriculture, the role of policies, market access and behavioral factors in adopting new technologies must also be further explored. Third, the poverty reducing effect of different agricultural technologies and investments depends greatly on the structure and institutional organization of the economy. Maximizing the poverty reducing effects of these investments thus requires in depth country specific understanding of the key features of the rural and urban economy (e.g. functioning of input, factor and commodity markets, proportion of net buyers of food staple crops, price and income elasticities of demand and supply of basic staple commodities). Too many poverty reduction strategies are still being developed without a proper diagnostic of basic facts of the agricultural sector, the rural economy and its linkages with the rest of the economy as well as an articulation of the implications of these findings for developing agricultural and rural sector strategies with a maximum impact on poverty. 36 References Adams, Richard H.J. (2004): "Economic Growth, Inequality, and Poverty: Estimating the Growth Elasticity of Poverty"; World Development 32(12), pp.1989-2004. Anand, Sudhir, and S.M.R. Kanbur, (1985): "Poverty under the Kuznets Process"; Economic Journal 95(0) (Supplement), pp. 42-50. Arellano, Manuel, and Olympia Bover (1995): "Another Look at the Instrumental Variable Estimation of Error-Components Models"; Journal of Econometrics 68, p.29-51. Bernard, A., and C. Jones (1996): Productivity across Industries and Countries: Time Series Theory and Evidence, Review of Economics and Statistics 78(1), pp. 135-146. Bond, Stephen. (2002): "Dynamic Panel Data Models: A Guide to Micro Data Methods and Practice" Centre for Microdata methods and Practice Working Paper 09/02, The Institute for Fiscal Studies, Department of Economics, Université Catholique de Louvain. Bourguignon, Francois, and Christian Morrisson (1998): "Inequality and Development: The Role Dualism." Journal of Development Economics, 57: 233-257. Bravo-Ortega, Claudio and Daniel Lederman (2005): "Agriculture and National Welfare around the World: Causality and International Heterogeneity since 1960"; World Bank Policy Research Working Paper 3499, February 2005. Byerlee, Derek, Xinshen, Diao, and Chris Jackson (2005): Agriculture, Rural Development, and Pro-Poor Growth, Country Experiences in the Post-Reform Era, Agriculture and Rural Development Discussion Paper 21, World Bank: Washington D.C. Chen, Shaohua, and Martin Ravallion (2004): "How Have the World's Poorest Fared Since the Early 1980s?" World Bank Research Observer 19(2), pp. 141-170. Christiaensen, Luc, Lionel Demery, and Stefano Paternostro (2005) "Reforms, Remoteness, and Risk in Africa: Understanding Inequality and Poverty during the 1990s", in R. Kanbur, and 37 A.J. Venables (eds.) Spatial Inequality and Development, UNU-WIDER Studies in Development Economics, Oxford University Press: Oxford. Delgado, Christopher L., Jane Hopkins, and Valerie A. Kelly (1998): "Agricultural Growth Linkages in Sub-Saharan Africa"; IFPRI Research Report 107; International Food Policy Research Institute, Washington D.C. de Janvry, Alain, and Elisabeth Sadoulet (1996): "Growth, Inequality and poverty in Latin America: a causal analysis, 1970­94"; Working Paper 784, Department of Agricultural and Resource Economics, University of California at Berkeley, California, USA. de Janvry, Alain, and Elisabeth, Sadoulet (2002): "World Poverty and the Role of Agricultural Technology: Direct and Indirect Effects" Journal of Development Studies 38(4), pp. 1-26. Dorosh, Paul, and Steven Haggblade (2003): "Growth Linkages, Price Effects and Income Distribution in Sub-Saharan Africa", Journal of African Economies 12(2), pp. 207-235. Fan, Shenggen, Peter, Hazell, and Sukhadeo Thorat (2000): "Government Spending, Growth and Poverty in Rural India", American Journal of Agricultural Economics 82(4): pp. 1038-51. Fan, Shenggen, C., Chan-Kang, and A., Mukherjee (2005): "Rural and Urban Dynamics and Poverty: Evidence from China and India" International Food Policy Research Institute, Washington D.C., mimeograph. Gollin, Douglas, Stephen Parente, and Richard Rogerson (2002) "The Role of Agriculture in Development", American Economic Review 92 (May): pp. 160-164. Granger, Clive W.J. (1969) "Investigating Causal Relations by Econometric Models and Cross- Spectral Methods", Econometrica 37(3), pp.424-438, August 1969. Haggblade, Steven, Jeffrey Hammer, and Peter Hazell. 1991. "Modeling Agricultural Growth Multipliers"; American Journal of Agricultural Economics 73 (2): pp.361-374. 38 Hayami, Yujiro and Vernon Ruttan (1985) Agricultural Development: An International Perspective; Baltimore : Johns Hopkins University Press. InterAcademy Council (2004): Realizing the Promise and Potential of African Agriculture: Science and Technology Strategies for Improving Agricultural Productivity and Food Security in Africa; Report, Amsterdam, June. Irz, Xavier, and Terry Roe (2005) "Seeds of Growth? Agricultural Productivity and The Transitional Dynamics of the Ramsey Model", European Review of Agricultural Economics, 32 (2): pp. 143-165. Johnston, Bruce, and John Mellor, 1961, "The Role of Agricultural in Economic Development", American Economic Review, No 4, pp. 566-93. Jorgenson, Dale W., Frank M. Gollop, and Barbara M. Fraumeni (1987) "Productivity and U.S. Economic Growth"; Harvard Economic Studies, vol. 159 Cambridge, Mass.: Harvard University Press. King, Robert P. and Derek Byerlee (1978): "Factor Intensities and Locational Linkages of Rural Consumption Patterns in Sierra Leone" American Journal of Agricultural Economics 60(2), pp. 197-206, (May 1978). Krueger, Ann, Maurice Schiff, and Alberto Valdes (1988): "Agricultural Incentives in Developing Countries: Measuring the Effect of Sectoral and Economywide Policies"; World Bank Economic Review 2-3, pp.255-271. Lewis, William Arthur (1954): "Economic Development with Unlimited Supply of Labor"; Manchester School of Economic and Social Studies 22, May, 139-191. Lewis, P., W. Martin, and C. Savage (1988) "Capital and Investment in the Agricultural Economy" Quarterly Review of the Rural Economy 10(1), pp. 48-53. 39 Lipton, Michael (1977): Why Poor People Stay Poor: Urban Bias and World Development, Temple Smith, London, and Harvard University Press. Loayza, Norman and Claudio Raddatz. (2005): `The Composition of Growth Matters for Poverty Alleviation' World Bank (mimeo, September). Martin, Will, and Devashish Mitra (2001) "Productivity Growth and Convergence in Agriculture versus Manufacturing", Economic Development and Cultural Change 49(2), pp.402-422. Maxwell, S. (2004) Launching the DFID consultation "New Directions for Agriculture in Reducing Poverty", Department for International Development. Available at http://dfid-agriculture- consultation.nri.org/launchpapers/simonmaxwell.html Mellor, John (1976): The New Economics of Growth: A Strategy for India and the Developing World; Ithaca, NY, USA; Cornell University Press. Mitchell, T.D., T.R. Carter, P.D. Jones, M. Hulme, and M. New (2003): "A comprehensive set of high-resolution grids of monthly climate for Europe and the globe: the observed record (1901-2000) and 16 scenarios (2001-2100)"; Journal of Climate: submitted. North, Douglas (1959): "Agriculture in Regional Economic Growth", Journal of Farm Economics 41(5), pp.943-951. Perry, G., D. de Ferranti, D. Lederman, W. Foster and A. Valdes (2005): Beyond the City: the Rural Contribution to Development--Complete Report; World Bank, Washington D.C. Ravallion, Martin (1997) `Can High Inequality Countries Escape Poverty?' Economic Letters 56(1), pp.51-57. Ravallion, M. (2001). "Growth, Inequality and Poverty: Looking Beyond Averages" World Development 29(11), pp.1803-1815. 40 Ravallion, Martin, and Monica Huppi (1991) `Measuring Changes in Poverty: A Methodological Case Study of Indonesia during an Adjustment Period.' World Bank Economic Review, Vol. 5 (January) pp. 57-82. Ravallion, M., and G. Datt (1996): "How important to India's Poor is the Sectoral Composition of Economic Growth?"; World Bank Economic Review 10(1), pp.1-25. Ravallion, M., and G. Datt (2002): "Why has Economic Growth been More Pro-Poor in Some States of India than Others?" Journal of Development Economics 65, pp. 381-400. Ravallion, M., and S. Chen (2004): "China's (Uneven) Progress Against Poverty" World Bank Policy Research Working Paper No. 3408. Schultz, Theodore W. (1964): Transforming Traditional Agriculture, New Haven, Conn., Yale University Press. Thirtle, C., L. Irz, L. Lin, V. McKenzie-Hill and S. Wiggins (2001) "Relationship between Changes in Agricultural Productivity and the incidence of poverty in developing countries" Report commissioned by DFID, London, UK. Thirtle, Colin, Lin Lin, and Jenifer Piesse (2003): "The Impact of Research-Led Agricultural Productivity Growth on Poverty Reduction in Africa, Asia and Latin America" World Development 31-12, pp.1959-1975 Timmer, Peter (1997): "How Well Do the Poor Connect to the Growth Process?" Harvard Institute for International Development. United States Agency for International Development. Timmer, Peter (2005): "Agriculture and Pro-Poor Growth: What the Literature Says." Draft paper, Agricultural and Rural Development Dept., World Bank. Windmeijer, Frank (2005): "A Finite Sample Correction for the Variance of Linear Efficient Two- Step GMM estimators" Journal of Econometrics 126, pp.25-51. World Bank (2000): The Quality of Growth, Oxford University Press: Oxford. 41 World Bank (2005a): Meeting the Challenge of Africa's Development: A World Bank Group Action Plan, World Bank: Washington D.C. World Bank (2005b): Agriculture, Rural Development and Pro-Poor Growth, World Bank: Washington D.C.. World Bank (2005c): World Development Indicators and Global Development Finance Database, World Bank: Washington D.C.; April 2005 World Bank (2005d): PovcalNet; http://iresearch.worldbank.org/PovcalNet/jsp/index.jsp, May 2005. World Bank (2005e) "Well-Being and Poverty in Ethiopia: The Role of Agriculture and Agency" Report No. 29468-ET, World Bank: Washington D.C. 42 Appendix A1: Specification and estimation of agriculture and non-agriculture linkages In particular, in our specification non-agricultural GDP per capita (Ynit) in country i at time t is assumed to depend on both levels of per capita non-agricultural GDP in previous periods as well as per capita agricultural GDP now31 and in the past. In addition, we consider a vector Xit of country-specific exogenous explanatory factors, yielding: Kn Ka Yit = 0 + Yit-kk + Yit-kk + Xit 3 + ci + it n n 1 a 2 (A1) k=1 k=0 where ci reflects unobserved country specific characteristics that determine the sectoral output, and it a white noise error term. Per capita agricultural GDP can be similarly expressed as: Ln La Yit = + Yit-ll + Yit-ll + Xit + fi + it a 0 n 1 a 2 3 (A2) l=1 l=1 with fi a time invariant country-specific unobserved effect and it a white noise error term.32 The specifications (A1) and (A2) are assumed to capture the full correlations between (per capita) non- agricultural and agricultural GDP, and it and it are therefore assumed to be uncorrelated. Equations (A1) and (A2) constitute our model of intersectoral growth linkages, where agricultural and non-agricultural GDP are interdependently determined in a dynamic process. Through the substitution of equation (A2) into equation (A1), we can obtain a reduced form expression for non-agricultural growth: Ka Kn Ka Ln Ka La Yit = 0 + n 0k + Yit-kk + 2 n 1 Y n 1 2 a 2 2 it-k-llk + Y it-k -ll k k =0 k=1 k =0 l=1 k=0 l=1 (A3) (f ) Ka Ka + X 2 it-k k + Xit3 + 3 2 i +it k + ci + it k =0 k=0 which can be further reduced to: 31The level of contemporaneous agricultural GDP is included since a good agricultural harvest can induce an immediate higher agricultural demand for goods from the non-agricultural sector. 32Concurrent non-agricultural GDP is omitted on the right-hand side, as the agricultural cycle imposes a lagged effect of any non-agricultural stimuli to agricultural output growth. Increased demand for agricultural products can only lead to a higher agricultural production at the next harvest period. We abstract from the potential increased demand for agricultural inputs reflected in concurrent non-agricultural growth. 43 P Q Ka Yit = + Yit-p 1p + Yit-p + n 0 n a 2 3 (A4) p X it-kk + hi + vit p=1 p=1 k=0 This single equation now constitutes a dynamic relationship between non-agricultural GDP and lagged levels of agricultural and non-agricultural GDP which we estimate in first differences and levels using the Arellano and Bover (1995) system GMM estimator. The lagged levels of non- agricultural GDP are correlated with the unobserved country specific effects hi and OLS is therefore inconsistent. First-differencing of equation (A4) eliminates the country-specific effect hi yielding: P Q Ka Yit = Yit-p1p + Yit-p + Xit-kk + vit n n a 2 3 (A5) p p=1 p=1 k=0 The assumed feedback mechanism between changes in agricultural and non-agricultural GDP introduces another correlation between vit and the lagged changes in agricultural and non- agricultural GDP (Yit-1 and Yit-1). To ensure consistent estimates we follow Arellano and Bover a n and use further lagged levels as instruments in the first difference equation (A5) and lagged differences to instrument Yit-1 and Yit-1 in the levels equation (A4) in effect assuming E(Yit-1, hi+ a n a vit)=E( Yit-1, hi+ vit)=0 for t=3, ..., T. The additional moment conditions following from the level n equation help address the potential problem of weak instruments, and thus low efficiency, that afflicts the Arellano-Bond estimation when the data generation process of the variables of interest approaches a unit root (Bond, 2002). A reduced form along the lines of equations (A3-A5) can also be constructed for agricultural GDP(Yit ) and its change (Yit ). a a 44 Appendix A2: Povcal data overview Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End Albania 1997-2002 0.62 0.23 0.07 0.04 0.02 0.04 29.12 28.14 Algeria 1988-1995 1.75 1.16 0.64 0.24 0.49 0.09 40.14 35.33 Argentina-Urban 1986-1992 0.29 0.09 0.17 0.01 0.20 0.00 44.51 45.35 Argentina-Urban 1992-1996 0.09 1.14 0.01 0.18 0.00 0.05 45.35 48.58 Argentina-Urban 1996-1998 1.14 7.69 0.18 3.61 0.05 2.27 48.58 52.82 Argentina-Urban 1998-2001 7.69 3.33 3.61 0.48 2.27 0.09 52.82 52.24 Azerbaijan 1995-2001 10.94 3.67 2.62 0.63 1.01 0.20 35.99 36.50 Bangladesh 1984-1986 26.16 21.96 5.99 3.92 1.96 1.08 25.88 26.92 Bangladesh 1986-1989 21.96 33.75 3.92 7.72 1.08 2.44 26.92 28.85 Bangladesh 1989-1992 33.75 35.86 7.72 8.77 2.44 2.98 28.85 28.27 Bangladesh 1992-1995.5 35.86 28.61 8.77 6.04 2.98 1.87 28.27 33.00 Bangladesh 1995-2000 28.61 26.81 6.04 5.31 1.87 1.49 33.00 31.79 Belarus 1988-1993 0.09 0.00 0.07 0.00 0.11 0.00 22.76 21.60 Belarus 1993-1995 0.00 1.43 0.00 0.52 0.00 0.41 21.60 28.76 Belarus 1995-1997 1.43 0.03 0.52 0.01 0.41 0.00 28.76 25.62 Belarus 1997-1998 0.03 0.00 0.01 0.00 0.00 0.00 25.62 27.95 Bolivia-Urban 1986-1991 20.08 5.68 6.17 0.76 2.50 0.14 51.68 42.04 Bolivia 1997-1999 20.43 14.38 9.66 5.35 6.09 2.68 58.46 44.68 Botswana 1986-1994 33.30 30.66 12.54 12.72 6.09 6.89 54.21 66.70 Brazil 1981-1984 11.80 15.21 2.97 4.09 0.99 1.44 57.57 57.88 Brazil 1984-1985 15.21 15.75 4.09 4.64 1.44 1.79 57.88 59.52 Brazil 1985-1987 15.75 11.90 4.64 3.36 1.79 1.25 59.52 59.31 Brazil 1987-1989 11.90 9.00 3.36 2.01 1.25 0.59 59.31 63.42 Brazil 1989-1990 9.00 14.04 2.01 4.27 0.59 1.70 63.42 60.68 Brazil 1990-1993 14.04 8.27 4.27 2.01 1.70 0.65 60.68 59.82 Brazil 1993-1995 8.27 10.53 2.01 3.88 0.65 1.88 59.82 61.51 Brazil 1995-1996 10.53 6.86 3.88 1.37 1.88 0.36 61.51 59.98 Brazil 1996-1997 6.86 8.96 1.37 2.09 0.36 0.65 59.98 59.05 Brazil 1997-1998 8.96 9.94 2.09 3.15 0.65 1.32 59.05 60.66 Brazil 1998-2001 9.94 8.17 3.15 2.09 1.32 0.71 60.66 59.25 Bulgaria 1989-1994 0.05 0.29 0.06 0.23 0.16 0.44 23.43 24.32 Bulgaria 1992-1996 0.33 1.76 0.30 0.97 0.63 1.16 30.80 35.04 Bulgaria 1994-1995 0.29 3.86 0.23 1.37 0.44 0.67 24.32 31.13 Bulgaria 1995-1997 3.86 0.50 1.37 0.14 0.67 0.09 31.13 26.38 Bulgaria 1997-2001 0.50 4.73 0.14 1.39 0.09 0.56 26.38 31.91 Burkina Faso 1994-1998 51.41 44.85 19.52 14.42 9.28 6.27 50.71 46.85 Burundi 1992-1998 45.24 54.56 13.83 22.69 5.66 12.67 33.33 42.39 Cameroon 1996-2001 32.45 17.10 9.05 4.09 3.30 1.38 46.82 44.55 Chile 1987-1989 6.20 4.92 1.01 1.09 0.22 0.40 56.43 57.88 Chile 1989-1990 4.92 6.19 1.09 2.12 0.40 1.15 57.88 56.49 Chile 1990-1992 6.19 1.15 2.12 0.20 1.15 0.06 56.49 55.75 Chile 1992-1994 1.15 0.81 0.20 0.08 0.06 0.01 55.75 54.79 Chile 1994-1996 0.81 0.00 0.08 0.00 0.01 0.00 54.79 57.47 Chile 1996-1998 0.00 0.85 0.00 0.11 0.00 0.02 57.47 56.65 Chile 1998-2000 0.85 0.97 0.11 0.18 0.02 0.06 56.65 57.61 China 1981-1982 23.02 13.70 5.51 2.89 1.90 0.87 30.95 28.53 China 1982-1983 13.70 10.48 2.89 1.96 0.87 0.52 28.53 28.28 China 1983-1984 10.48 7.67 1.96 1.24 0.52 0.28 28.28 29.11 China 1984-1985 7.67 6.78 1.24 1.13 0.28 0.27 29.11 28.95 China 1985-1986 6.78 7.49 1.13 1.45 0.27 0.40 28.95 32.41 45 Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End China 1986-1987 7.49 6.39 1.45 1.15 0.40 0.33 32.41 32.38 China 1987-1988 6.39 6.13 1.15 1.04 0.33 0.31 32.38 33.01 China 1988-1989 6.13 9.73 1.04 2.15 0.31 0.71 33.01 35.15 China 1989-1990 9.73 7.96 2.15 1.45 0.71 0.41 35.15 34.85 China 1990-1991 7.96 8.52 1.45 2.08 0.41 0.85 34.85 37.06 China 1991-1992 8.52 7.13 2.08 1.61 0.85 0.63 37.06 39.01 China 1992-1993 7.13 8.27 1.61 1.79 0.63 0.54 39.01 41.95 China 1993-1994 8.27 7.58 1.79 2.00 0.54 0.74 41.95 43.31 China 1994-1995 7.58 5.65 2.00 1.55 0.74 0.75 43.31 41.50 China 1995-1996 5.65 2.97 1.55 0.81 0.75 0.42 41.50 39.75 China 1996-1997 2.97 3.35 0.81 0.58 0.42 0.15 39.75 39.78 China 1997-1998 3.35 2.16 0.58 0.24 0.15 0.04 39.78 40.33 China 1998-1999 2.16 2.24 0.24 0.27 0.04 0.05 40.33 41.61 China 1999-2000 2.24 3.34 0.27 0.64 0.05 0.18 41.61 43.82 China 2000-2001 3.34 2.96 0.64 0.51 0.18 0.12 43.82 44.73 Colombia 1980-1988 7.85 4.48 2.92 1.31 1.58 0.58 59.13 53.11 Colombia 1988-1989 4.48 2.45 1.31 0.59 0.58 0.24 53.11 53.59 Colombia 1989-1991 2.45 2.82 0.59 0.76 0.24 0.32 53.59 51.32 Colombia 1991-1995 2.82 3.12 0.76 0.36 0.32 0.06 51.32 57.22 Colombia 1995-1996 3.12 5.28 0.36 1.03 0.06 0.27 57.22 56.96 Colombia 1996-1998 5.28 8.26 1.03 3.29 0.27 1.91 56.96 60.66 Colombia 1998-1999 8.26 8.18 3.29 2.21 1.91 0.79 60.66 57.92 Costa Rica 1981-1986 14.81 7.32 5.91 3.16 3.17 1.87 47.49 34.48 Costa Rica 1986-1990 7.32 5.24 3.16 1.34 1.87 0.46 34.48 45.66 Costa Rica 1990-1993 5.24 4.11 1.34 1.36 0.46 0.74 45.66 46.28 Costa Rica 1993-1996 4.11 3.57 1.36 1.09 0.74 0.54 46.28 47.08 Costa Rica 1996-1997 3.57 1.85 1.09 0.51 0.54 0.25 47.08 45.88 Costa Rica 1997-1998 1.85 6.94 0.51 3.41 0.25 2.24 45.88 51.30 Costa Rica 1998-2000 6.94 2.01 3.41 0.66 2.24 0.39 51.30 46.60 Cote d'Ivoire 1985-1986 4.71 0.00 0.59 0.00 0.11 0.00 41.21 38.62 Cote d'Ivoire 1986-1987 0.00 3.28 0.00 0.34 0.00 0.06 38.62 40.43 Cote d'Ivoire 1987-1988 3.28 7.46 0.34 1.37 0.06 0.40 40.43 36.89 Cote d'Ivoire 1988-1993 7.46 9.88 1.37 1.86 0.40 0.55 36.89 36.91 Cote d'Ivoire 1993-1995 9.88 12.29 1.86 2.41 0.55 0.71 36.91 36.68 Cote d'Ivoire 1995-1998 12.29 15.53 2.41 3.82 0.71 1.42 36.68 43.75 Croatia 1998-1999 0.07 0.23 0.07 0.24 0.15 0.54 26.82 27.71 Croatia 1999-2000 0.23 0.09 0.24 0.06 0.54 0.09 27.71 31.33 Croatia 2000-2001 0.09 0.08 0.06 0.05 0.09 0.06 31.33 31.10 Czech Republic 1988-1993 0.00 0.00 0.00 0.00 0.00 0.00 19.40 26.55 Czech Republic 1993-1996 0.00 0.12 0.00 0.24 0.00 1.05 26.55 25.82 Dominican Republic 1986-1989 8.61 3.85 2.89 0.55 1.46 0.12 47.78 50.46 Dominican Republic 1989-1992 3.85 1.55 0.55 0.55 0.12 0.38 50.46 51.36 Dominican Republic 1992-1996 1.55 1.76 0.55 0.38 0.38 0.14 51.36 48.71 Dominican Republic 1996-1998 1.76 0.00 0.38 0.00 0.14 0.00 48.71 47.44 Ecuador 1994-1995 28.89 20.21 8.47 5.77 3.27 2.28 46.55 43.73 Ecuador 1995-1998 20.21 15.00 5.77 5.97 2.28 3.42 43.73 53.39 Egypt, Arab Rep. 1991-1995 3.97 2.58 0.53 0.31 0.13 0.07 32.00 32.60 Egypt, Arab Rep. 1995-2000 2.58 3.08 0.31 0.42 0.07 0.11 32.60 34.41 El Salvador 1989-1995 21.35 25.05 12.20 10.06 10.50 5.70 48.96 49.86 El Salvador 1995-1996 25.05 25.26 10.06 10.35 5.70 5.79 49.86 52.25 El Salvador 1996-1997 25.26 21.40 10.35 7.87 5.79 3.95 52.25 50.79 El Salvador 1997-1998 21.40 21.39 7.87 7.94 3.95 3.89 50.79 52.17 El Salvador 1998-2000 21.39 31.07 7.94 14.07 3.89 8.57 52.17 53.27 Estonia 1988-1993 0.05 0.98 0.05 0.40 0.09 0.34 22.97 39.50 46 Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End Estonia 1995-1998 0.35 0.08 0.09 0.02 0.05 0.01 30.06 37.64 Ethiopia 1982-1995 32.73 31.25 7.69 7.95 2.72 3.00 32.42 39.96 Ethiopia 1995-2000 31.25 22.98 7.95 4.82 3.00 1.63 39.96 30.01 Gambia 1992-1998 53.69 26.49 23.27 8.76 13.28 3.77 47.80 50.23 Georgia 1996-1997 1.74 1.21 0.96 0.43 1.09 0.31 37.13 36.08 Georgia 1997-1998 1.21 1.62 0.43 0.14 0.31 0.02 36.08 37.38 Georgia 1998-1999 1.62 2.59 0.14 0.85 0.02 0.53 37.38 38.05 Georgia 1999-2000 2.59 2.83 0.85 0.88 0.53 0.52 38.05 38.85 Georgia 2000-2001 2.83 2.71 0.88 0.93 0.52 0.62 38.85 36.90 Ghana 1988-1989 46.51 45.45 16.06 15.27 7.51 6.99 35.35 35.99 Ghana 1989-1992 45.45 47.24 15.27 16.40 6.99 7.54 35.99 38.10 Ghana 1992-1999 47.24 44.81 16.40 17.28 7.54 8.72 38.10 40.71 Guatemala 1987-1989 47.04 34.85 22.47 16.84 13.63 10.49 58.26 59.60 Guatemala 1989-2000 34.85 15.95 16.84 4.60 10.49 1.74 59.60 59.87 Guyana 1993-1998 8.14 2.98 1.95 0.60 0.63 0.16 51.55 44.58 Honduras 1986-1989 33.74 34.22 13.67 14.33 7.15 7.72 55.09 59.49 Honduras 1989-1990 34.22 37.83 14.33 16.84 7.72 9.64 59.49 57.36 Honduras 1990-1992 37.83 28.33 16.84 11.80 9.64 6.42 57.36 54.51 Honduras 1992-1994 28.33 23.66 11.80 9.52 6.42 5.09 54.51 55.22 Honduras 1994-1996 23.66 24.96 9.52 9.12 5.09 4.37 55.22 53.72 Honduras 1996-1998 24.96 23.84 9.12 11.62 4.37 7.47 53.72 56.30 Honduras 1998-1999 23.84 20.74 11.62 7.45 7.47 3.51 56.30 56.24 Hungary 1987-1989 0.06 0.03 0.08 0.01 0.26 0.02 20.96 25.05 Hungary 1989-1993 0.03 0.19 0.01 0.16 0.02 0.31 25.05 27.94 Hungary 1993-1998 0.19 0.38 0.16 0.32 0.31 0.66 27.94 24.44 India 1983-1986 52.70 48.29 16.32 14.23 6.83 5.75 32.06 33.68 India 1986-1987 48.29 45.88 14.23 12.52 5.75 4.68 33.68 33.08 India 1987-1988 45.88 49.45 12.52 14.05 4.68 5.39 33.08 32.93 India 1988-1989 49.45 40.80 14.05 10.42 5.39 3.73 32.93 31.84 India 1989-1990 40.80 42.06 10.42 11.09 3.73 4.04 31.84 31.21 India 1990-1992 42.06 51.08 11.09 14.98 4.04 5.89 31.21 34.31 India 1992-1994 51.08 45.13 14.98 12.04 5.89 4.41 34.31 31.52 India 1994-1995 45.13 50.62 12.04 13.92 4.41 5.16 31.52 36.32 India 1995-1996 50.62 41.85 13.92 10.44 5.16 3.61 36.32 32.86 India 1996-1997 41.85 44.21 10.44 44.21 3.61 4.52 32.86 37.83 India 1993-1999 42.13 35.60 10.81 8.45 3.87 2.74 31.52 32.46 Indonesia 1984-1987 37.30 28.08 10.18 6.10 3.94 1.78 34.15 33.12 Indonesia 1987-1990 28.15 20.62 6.10 3.93 1.78 1.05 33.12 33.18 Indonesia 1990-1993 20.62 17.39 3.93 2.70 1.05 0.56 33.18 34.36 Indonesia 1993-1996 17.39 13.93 2.70 2.16 0.56 0.49 34.36 36.45 Indonesia 1996-1998 13.93 26.33 2.16 5.44 0.49 1.70 36.45 38.36 Indonesia 1998-1999 26.33 14.74 5.44 2.29 1.70 0.54 38.36 31.73 Indonesia 1999-2000 14.74 7.19 2.29 1.04 0.54 0.26 31.73 30.33 Indonesia 2000-2002 7.19 7.51 1.04 0.91 0.26 0.18 30.33 34.30 Iran, Islamic Rep. 1986-1990 1.53 1.61 0.32 0.44 0.12 0.23 47.42 43.60 Iran, Islamic Rep. 1990-1994 1.61 0.49 0.44 0.11 0.23 0.05 43.60 43.00 Iran, Islamic Rep. 1994-1998 0.49 0.26 0.11 0.04 0.05 0.01 43.00 44.10 Jamaica 1988-1989 5.02 3.42 1.38 0.32 0.67 0.04 43.16 42.02 Jamaica 1989-1990 3.42 7.72 0.32 3.28 0.04 1.96 42.02 32.90 Jamaica 1990-1991 7.72 4.10 3.28 0.45 1.96 0.07 32.90 41.11 Jamaica 1991-1992 4.10 6.65 0.45 1.03 0.07 0.24 41.11 38.39 Jamaica 1992-1993 6.65 4.92 1.03 1.32 0.24 0.65 38.39 35.67 Jamaica 1993-1996 4.92 3.15 1.32 0.74 0.65 0.33 35.67 36.43 Jamaica 1996-1999 3.15 1.68 0.74 0.43 0.33 0.21 36.43 44.22 47 Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End Jamaica 1999-2000 1.68 0.41 0.43 0.06 0.21 0.02 44.22 38.82 Jordan 1987-1992 0.00 0.55 0.00 0.12 0.00 0.05 36.06 43.36 Jordan 1992-1997 0.55 0.36 0.12 0.10 0.05 0.06 43.36 36.42 Kazakhstan 1988-1993 0.02 0.37 0.01 0.00 0.01 0.00 25.74 32.65 Kazakhstan 1996-2001 1.87 0.11 0.32 0.02 0.10 0.01 35.32 31.30 Kenya 1992-1994 33.54 26.54 12.82 9.03 6.62 4.50 57.46 44.54 Kenya 1994-1997 26.54 22.81 9.03 5.92 4.50 2.10 44.54 44.93 Kyrgyz Republic 1988-1996 0.00 19.90 0.00 9.62 0.00 6.19 26.01 52.30 Kyrgyz Republic 1993-1997 8.03 1.57 3.28 0.29 1.82 0.10 53.70 40.50 Kyrgyz Republic 1997-1998 1.57 0.22 0.29 0.02 0.10 0.01 40.50 35.98 Kyrgyz Republic 1998-1999 0.22 0.73 0.02 0.18 0.01 0.09 35.98 34.60 Kyrgyz Republic 1999-2000 0.73 1.97 0.18 0.21 0.09 0.04 34.60 30.27 Kyrgyz Republic 2000-2001 1.97 0.86 0.21 0.10 0.04 0.02 30.27 29.03 Lao PDR 1992-1997 7.75 26.33 1.00 6.30 0.23 2.24 30.40 37.00 Latvia 1988-1993 0.03 0.00 0.04 0.00 0.14 0.00 22.49 26.99 Latvia 1993-1995 0.00 1.56 0.00 1.28 0.00 2.46 26.99 30.99 Latvia 1995-1996 1.56 0.94 1.28 0.08 2.46 0.01 30.99 31.60 Latvia 1996-1997 0.94 1.06 0.08 0.11 0.01 0.01 31.60 31.69 Lesotho 1987-1993 30.34 43.14 12.66 20.26 6.85 11.84 56.02 57.94 Lesotho 1993-1995 43.14 36.43 20.26 18.98 11.84 12.42 57.94 63.16 Lithuania 1988-1993 0.08 6.78 0.05 0.87 0.08 0.15 22.48 33.56 Lithuania 1993-1994 6.78 5.90 0.87 1.14 0.15 0.30 33.56 37.33 Lithuania 1996-1998 1.07 0.59 0.50 0.26 0.51 0.24 32.36 32.16 Lithuania 1998-2000 0.59 0.53 0.26 0.18 0.24 0.13 32.16 31.85 Madagascar 1980-1993 49.18 46.31 19.74 17.64 10.21 9.02 46.85 46.12 Madagascar 1993-1997 46.31 58.22 17.64 23.47 9.02 12.00 46.12 45.97 Madagascar 1997-1999 58.22 66.03 23.47 29.42 12.00 16.35 45.97 41.80 Madagascar 1999-2001 66.03 61.03 29.42 27.90 16.35 15.70 41.80 47.45 Malaysia 1984-1987 1.96 1.20 0.40 0.17 0.14 0.04 48.63 47.04 Malaysia 1987-1989 1.20 0.93 0.17 0.14 0.04 0.04 47.04 46.17 Malaysia 1989-1992 0.93 0.43 0.14 0.03 0.04 0.00 46.17 47.65 Malaysia 1992-1995 0.43 1.03 0.03 0.11 0.00 0.02 47.65 48.52 Malaysia 1995-1997 1.03 0.17 0.11 0.02 0.02 0.00 48.52 49.15 Mali 1989-1994 16.46 72.29 3.92 37.39 1.40 23.09 36.51 50.50 Mauritania 1987-1993 46.67 49.37 20.77 17.83 12.29 8.58 43.94 50.05 Mauritania 1993-1995 49.37 29.45 17.83 9.48 8.58 4.33 50.05 37.33 Mauritania 1995-1996 29.45 29.11 9.48 9.29 4.33 4.19 37.33 37.71 Mauritania 1996-2000 29.11 25.93 9.29 7.56 4.19 2.95 37.71 39.03 Mexico 1984-1989 13.95 8.32 3.38 2.54 1.09 1.15 46.26 55.14 Mexico 1989-1992 8.32 15.77 2.54 4.13 1.15 1.43 55.14 50.31 Mexico 1992-1995 15.77 8.39 4.13 2.39 1.43 1.00 50.31 53.73 Mexico 1995-1996 8.39 6.46 2.39 1.51 1.00 0.53 53.73 51.86 Mexico 1996-1998 6.46 7.98 1.51 2.07 0.53 0.75 51.86 53.11 Moldova, Rep. 1988-1992 0.00 7.33 0.00 1.35 0.00 0.33 24.14 34.32 Moldova, Rep. 1997-1998 1.86 19.77 0.41 5.72 0.18 2.48 30.99 37.77 Moldova, Rep. 1998-1999 19.77 32.24 5.72 9.93 2.48 4.32 37.77 36.86 Moldova, Rep. 1999-2001 32.24 21.78 9.93 5.67 4.32 2.22 36.86 36.18 Mongolia 1995-1998 13.92 27.02 3.06 8.08 0.98 3.40 33.20 30.27 Morocco 1985-1991 2.04 0.14 0.70 0.03 0.50 0.01 39.19 39.20 Morocco 1991-1999 0.14 0.56 0.03 0.08 0.01 0.02 39.20 39.46 Nicaragua 1993-1998 47.94 44.71 20.41 16.64 11.19 8.23 50.33 55.13 Niger 1992-1995 41.73 60.56 12.46 33.96 5.29 23.73 36.10 50.61 Nigeria 1986-1993 65.72 59.19 29.62 29.25 16.71 18.27 38.68 44.95 Nigeria 1993-1997 59.19 70.24 29.25 34.93 18.27 21.24 44.95 50.56 48 Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End Pakistan 1987-1990 49.63 47.76 14.84 14.57 6.04 6.04 33.35 33.23 Pakistan 1990-1993 47.76 33.90 14.57 6.35 6.04 3.01 33.23 34.22 Pakistan 1993-1997 33.90 6.82 8.45 0.99 3.01 0.27 34.22 27.43 Pakistan 1997-1999 6.82 13.36 0.99 2.36 0.27 0.71 27.43 32.99 Panama 1979-1989 0.00 11.81 0.00 5.39 0.00 3.26 48.74 56.57 Panama 1989-1991 11.81 11.81 5.39 5.20 3.26 3.03 56.57 56.82 Panama 1991-1995 11.81 7.38 5.20 2.57 3.03 1.19 56.82 57.06 Panama 1995-1996 7.38 7.92 2.57 2.81 1.19 1.32 57.06 56.31 Panama 1996-2000 7.92 7.20 2.81 2.28 1.32 0.95 56.31 56.56 Paraguay 1990-1995 4.93 19.36 0.85 8.27 0.23 4.65 39.74 59.13 Paraguay 1995-1997 19.36 16.50 8.27 8.21 4.65 5.40 59.13 57.72 Paraguay 1997-1998 16.50 15.88 8.21 7.95 5.40 5.27 57.72 56.52 Paraguay 1998-1999 15.88 14.86 7.95 6.80 5.27 4.26 56.52 56.85 Peru 1986-1994 1.14 9.40 0.29 2.00 0.14 0.57 45.72 44.87 Peru 1990-1996 1.35 8.88 0.48 3.02 0.34 1.65 43.87 46.24 Peru 1996-2000 8.88 18.07 3.02 9.14 1.65 6.28 46.24 49.82 Philippines 1985-1988 22.76 18.20 5.34 3.57 1.67 0.93 41.04 40.63 Philippines 1988-1991 18.20 19.77 3.57 4.23 0.93 1.20 40.63 43.82 Philippines 1991-1994 19.77 18.36 4.23 3.85 1.20 1.07 43.82 42.89 Philippines 1994-1997 18.36 14.40 3.85 2.85 1.07 0.75 42.89 46.16 Philippines 1997-2000 14.40 15.48 2.85 2.98 0.75 0.76 46.16 46.09 Poland 1985-1987 0.21 0.16 0.14 0.10 0.20 0.13 25.16 25.53 Poland 1987-1989 0.16 0.10 0.10 0.08 0.13 0.14 25.53 26.89 Poland 1989-1998 0.10 0.46 0.08 0.25 0.14 0.30 26.89 31.60 Poland 1993-1996 4.11 0.09 1.54 0.05 0.79 0.06 32.39 32.66 Poland 1998-1999 0.46 0.61 0.25 0.54 0.30 1.05 31.60 32.90 Romania 1989-1992 0.25 0.34 0.29 0.23 0.79 0.36 23.31 25.46 Romania 1992-1994 0.34 2.81 0.23 0.76 0.36 0.43 25.46 28.20 Romania 1998-2000 1.39 2.14 0.89 0.59 1.29 0.33 31.19 30.25 Russian Federation 1993-1996 6.08 6.97 1.17 1.70 0.30 0.56 48.34 46.15 Russian Federation 1996-1998 6.97 12.66 1.70 3.46 0.56 1.27 46.15 48.67 Russian Federation 1998-2000 12.66 6.14 3.46 1.19 1.27 0.31 48.67 45.62 Senegal 1991-1994 45.38 22.32 19.96 5.68 11.18 2.14 54.14 41.28 Slovak Republic 1988-1992 0.00 0.00 0.00 0.00 0.00 0.00 19.54 19.49 Slovak Republic 1992-1996 0.00 0.49 0.00 0.06 0.00 0.01 19.49 25.81 Slovenia 1987-1993 0.02 0.04 0.02 0.04 0.06 0.10 23.60 29.18 South Africa 1993-1995 10.02 6.28 1.42 0.56 0.27 0.07 59.33 56.59 South Africa 1995-2000 6.28 10.71 0.56 1.74 0.07 0.37 56.59 57.77 Sri Lanka 1985-1990 9.39 3.82 1.42 0.67 0.27 0.23 59.33 30.10 Sri Lanka 1990-1996 3.82 6.56 0.67 1.00 0.23 0.26 30.10 34.36 Thailand 1981-1988 21.64 17.85 5.40 3.64 1.78 1.00 45.22 43.84 Thailand 1988-1992 17.85 6.02 3.64 0.48 1.00 0.05 43.84 46.22 Thailand 1992-1996 6.02 2.20 0.48 0.14 0.05 0.02 46.22 43.39 Thailand 1996-1998 2.20 0.00 0.14 0.00 0.02 0.00 43.39 41.36 Thailand 1998-1999 0.00 2.02 0.00 0.06 0.00 0.00 41.36 43.53 Thailand 1999-2000 2.02 1.93 0.06 0.05 0.00 0.00 43.53 43.15 Trinidad and Tobago 1988-1992 2.25 3.95 0.20 0.99 0.02 0.43 42.60 40.27 Tunisia 1985-1990 1.67 1.26 0.34 0.33 0.13 0.17 43.43 40.24 Tunisia 1990-1995 1.26 1.02 0.33 0.19 0.17 0.07 40.24 41.66 Tunisia 1995-2000 1.02 0.32 0.19 0.07 0.07 0.03 41.66 40.81 Turkey 1987-1994 1.49 2.35 0.36 0.55 0.18 0.24 43.57 41.53 Turkey 1994-2000 2.35 0.87 0.55 0.21 0.24 0.10 41.53 40.03 Turkmenistan 1988-1993 0.00 20.65 0.00 5.30 0.00 1.84 26.39 35.38 Uganda 1989-1992 39.18 30.54 15.00 9.60 7.57 4.22 44.36 43.19 49 Country Spell Headcount ($1) Pov. gap ($1) Pov. gap squ. ($1) Gini Start End Start End Start End Start End Uganda 1992-1996 30.54 23.68 9.60 5.95 4.22 2.16 43.19 37.40 Uganda 1996-1999 23.68 26.85 5.95 7.74 2.16 3.20 37.40 43.11 Ukraine 1988-1992 0.06 0.02 0.07 0.01 0.19 0.01 23.31 25.71 Ukraine 1992-1999 0.02 2.92 0.01 0.62 0.01 0.26 25.71 28.96 Ukraine 1995-1996 2.06 0.00 0.64 0.00 0.39 0.00 39.29 33.18 Uruguay 1981-1989 0.91 0.58 0.50 0.34 0.56 0.40 43.65 42.33 Uruguay 1989-1996 0.58 0.56 0.34 0.18 0.40 0.12 42.33 43.76 Uruguay-Urban 1996-1998 0.56 0.61 0.18 0.20 0.12 0.12 43.76 45.18 Uruguay-Urban 1998-2000 0.61 0.20 0.20 0.05 0.12 0.02 45.18 44.56 Uzbekistan 1988-1993 0.00 3.28 0.00 0.46 0.00 0.11 24.95 33.27 Uzbekistan 1998-2000 19.16 17.32 8.12 4.26 4.70 1.86 45.35 27.03 Venezuela, RB 1981-1987 7.52 6.60 1.46 1.04 0.38 0.22 55.82 53.45 Venezuela, RB 1987-1989 6.60 2.95 1.04 0.87 0.22 0.45 53.45 44.08 Venezuela, RB 1989-1993 2.95 2.66 0.87 0.58 0.45 0.22 44.08 41.68 Venezuela, RB 1993-1995 2.66 9.43 0.58 2.86 0.22 1.31 41.68 46.84 Venezuela, RB 1995-1996 9.43 14.69 2.86 5.62 1.31 3.17 46.84 48.79 Venezuela, RB 1996-1997 14.69 9.65 5.62 2.88 3.17 1.27 48.79 48.80 Venezuela, RB 1997-1998 9.65 14.31 2.88 6.58 1.27 4.08 48.80 49.53 Vietnam 1993-1998 14.63 3.08 2.55 0.48 0.65 0.10 35.68 35.40 Yemen, Rep. 1992-1998 3.55 10.21 1.11 2.30 0.67 0.85 39.45 33.44 Zambia 1991-1993 64.64 73.57 38.91 42.66 28.78 29.55 60.05 52.61 Zambia 1993-1996 73.57 72.63 42.66 37.75 29.55 23.88 52.61 49.79 Zambia 1996-1998 72.63 63.65 37.75 32.65 23.88 21.07 49.79 52.60 Zimbabwe 1991-1995 33.32 56.12 8.96 24.17 3.34 13.04 39.42 50.12 Note: Start- or end-years specified with a 0.5-digit were rounded up for the calculation of the spell length.