89125 ResearchDigest World Bank VOLUME 8 L NUMBER 2 L WINTER 2014 Shared Prosperity and the Mitigation of Poverty The World Bank Group has set in the employment of reasoning and IN THIS ISSUE two global goals for 2030. What is analytics. Shared Prosperity and the Mitigation their rationale—and what are their Over the past three decades the of Poverty … page 1 global poverty rate has declined sharp- implications for policy making? Which is more effective in reducing poverty— ly, from just below 50 percent in 1980 growth or direct welfare interventions? A new T to 17.7 percent in 2010. If all countries paper revisits the long-standing debate he World Bank Group recently continue to grow at the same rates as adopted the twin goals of ending they did over the past 20 years, with The Simple Algebra of the Shared Prosperity extreme poverty in the world by no change in income distribution, the Indicator … page 2 2030 and promoting shared prosperity global poverty rate will fall to 7.7 per- Basic algebra adds another perspective in in every society. The first goal is aimed cent by 2030. If they grow faster, at the the discussion on how to promote shared at reducing the number of people in average rates recorded in the 2000s, prosperity extreme poverty—defined as those liv- the poverty rate will fall to 5.5 percent. The Global Income Distribution: ing on less than $1.25 a day—to less These numbers suggest that growth 1988–2008 … page 3 than 3 percent of the world’s popula- alone is unlikely to get us to the 3 per- A new database shows that global inequality tion. The second focuses on raising the cent target. remains high—and that inequality within prosperity of the poorest 40 percent of Some fear that a single global tar- countries is increasing the population of every nation. World get of 3 percent could lead to attention Bank Chief Economist Kaushik Basu, in being directed to the world’s most Narrowing the Gap in India … page 4 a recent paper, examines the rationale populous countries, where large num- A comprehensive analysis of poverty in India behind these goals and their implica- bers of poor people live. Basu explains refutes the claim that reforms and growth tions for actual policy making. that while the Bank is maintaining one have failed to help the socially disadvantaged Basu analyzes past poverty trends target for the world, it is at the same and revisits the long-standing debate time deliberately trying to ensure that … page 5 Is Workfare Really Cost-Effective? over which is more effective in improv- no country is left behind. This matters Even if workfare is extremely well targeted, ing people’s livelihoods—economic tremendously, since it is in small and this matters little if the net gains to growth that serves as a rising tide to fragile states that the risk of leaving participants are small lift all boats, or direct welfare interven- chronically poor people in dire condi- page 6 Growth Still Is Good for the Poor … tions that are targeted to the poor. tions is a real and growing concern. Does the recent historical record offer Within this debate, experts argue Indeed, if all countries grow at the guidance on which macroeconomic policies about the insights provided by macro- rates they achieved during the first best support shared prosperity? economic analysis and the use of mod- decade of this century and income dis- eling to understand broad develop- tribution stays steady, 17 countries will What Is the Effect of Growth on the ment trends—as compared with those have poverty rates of more than 30 per- Distribution of Opportunities? … page 7 produced by randomized control trials cent in 2030. Even with higher growth An opportunity-sensitive approach to to gauge the efficacy of everything from rates, such countries as Burundi, the analyzing growth can help show whether conditional cash transfer schemes to Democratic Republic of Congo, and some groups are being systematically deworming efforts. Haiti will still have poverty rates ex- excluded Basu’s paper also has a subtext. ceeding 30 percent. Over the years economics has vastly Some are concerned about the improved in the use of statistics and shared prosperity goal, arguing that facts. Basu argues that where the eco- nomic policy debate has fallen short is (continued on page 8) 2 World Bank ResearchDigest The Simple Algebra of the Shared Prosperity Indicator Decomposing the shared prosperity incomes across people, the two soci- socially unsustainable in the most bru- indicator shows that over the long eties would still have the same value tal sense: if it persists, the incomes of run growth drives ever-higher levels of SP(x). In other words, it does not the poorest 40 percent shrink to zero. matter who is the first member of a But the simple decomposition of prosperity society. The shared prosperity indica- highlights the role of economic growth O tor also satisfies the weak Pareto prin- in the long run: the “sharing” runs ne of the two goals of the World ciple: if the income of every individual into the limit of perfect equality, and Bank Group’s new strategy is in a society rises—that is, if every xi only growth can then drive ever-higher to promote shared prosper- rises—the society is considered bet- levels of prosperity. Simulations in ity, defined in terms of the income ter off. SP(x) is a complete, transitive the paper illustrate how long-run growth of the poorest 40 percent of ordering over the set X of all possible compounding of growth enhances the population. The simplest indica- vectors x. shared prosperity more powerfully tor for monitoring progress toward The shared prosperity indicator than changes in the income share of this goal is the per capita income of does not satisfy the weak transfer the poorest 40 percent—even when the poorest 40 percent. The change axiom, however, because lump-sum there is a substantial positive “shift” in in this indicator can be decomposed transfers from a richer person within “sharing.” into two parts: the change in the share the bottom 40 percent of the distribu- This is probably the most useful of total income accruing to the poor- tion to a poorer one within that group element of the algebraic decomposi- est 40 percent and the growth of the would not lead to any improvement in tion: it reveals that, to a certain extent, average income of the total popula- the value of the indicator. More gen- changes in the income share of the tion. A recent paper by Rosenblatt erally, any change in the distribution poorest 40 percent are like a “shift” and McGavock explores aspects of the of income that maintains the same factor between steady states, while shared prosperity indicator, including total share accruing to the poorest 40 growth in per capita income is more a simple decomposition of the indica- percent would not affect the value of akin to a steady-state growth rate of tor in algebraic form. SP(x). shared prosperity. Interestingly, the The shared prosperity indicator has To simplify terms, we can let y(x) be empirical experience presented in the its intellectual origins in the concept the per capita income of the total pop- paper (based on an admittedly small of quintile income. Defined as the per ulation of a country with income pro- sample) shows that in many cases capita income of the poorest quintile file x, with the summation of xi’s di- growth has a larger role to play than (20 percent) of the population, quin- vided by n being simply y(x). Then “sharing” in increasing this measure of tile income was proposed as a simple SP(x) can be rewritten as: shared prosperity—even over a rela- welfare measure that is both easy to s(x) tively short time frame. calculate and easy to understand. It 0.4 draws on Rawlsian notions of promot- This expression shows that the ing the welfare of the least fortunate shared prosperity indicator is similar members of society, and also has to the Sen index of real income in the pragmatic feature of comparabil- that it is the product of a measure of ity with traditional macroeconomic inequality (the income share of the welfare measures such as per capita poorest 40 percent) and the per capita income. In developing countries, fo- income of the total population. cusing on the per capita income of the In discrete time, the percentage poorest two quintiles (40 percent) of change in the shared prosperity indi- the population could be considered cator is simply the sum of two growth more appropriate for a variety of prac- rates: the growth rate of the share of tical reasons. income accruing to the poorest 40 For notation purposes, we can let percent and the growth rate of the per SP(x) be the shared prosperity indi- capita income of the total population. cator—the per capita income of the The simplest algebraic decomposi- poorest 40 percent—for a vector x rep- tion should not be taken too literally. resenting the incomes, xi, of each per- There is a vast literature on the theory son in a country. Quintile income and and evidence of the links between David Rosenblatt and Tamara J. McGavock. the shared prosperity indicator satisfy inequality and economic growth. In 2013. “A Note on the Simple Algebra of the the criterion of anonymity: if two coun- addition, a constant negative value Shared Prosperity Indicator.” Policy Research tries have the same income vector, for the growth rate of the income Working Paper 6645, World Bank, Washington, but with a random reassignment of share of the poorest 40 percent is DC. World Bank Research Digest 3 The Global Income Distribution: 1988–2008 Asia has grown fastest, while The data show that income in- groups. The country-deciles that in Sub-Saharan Africa and the poorer equality remains high at the global 1988 were around the global median half of the population in some rich level compared with inequality within had the fastest growth over the next countries. Between 1988 and 2008 the 20 years, with their incomes almost countries have lagged behind global Gini coefficient declined only doubling. Almost 90 percent of these T from around 72 percent to 70 percent. people came from Asia. People who here has been renewed interest The most important were around the 90th in recent years in tracking the source of inequality is The most important global percentile in global distribution of income. income differences be- source of inequality 1988 experienced the This presents formidable measure- tween countries. While slowest growth. Almost ment challenges, because it requires this source of inequality is income differences all of them came from assigning the correct income to every has declined, thanks between countries. mature economies (the person in the world. A recent paper to the strong growth in 27 member states of the by Lakner and Milanovic uses a newly some parts of Asia, it While this source European Union and compiled and improved database, still explains more than of inequality has the world’s other high- based on national household surveys 75 percent of global in- income countries). from 1988 to 2008, to present new evi- equality as measured by declined, it still As a result of these dence on global inequality. the Theil-L index. explains more than differences in income The data are drawn mostly from the Global inequality growth rates across the World Bank’s PovcalNet database and may be even higher. As 75 percent of global world, the global distri- the Luxembourg Income Study. Every a recent literature based inequality bution and its regional person in the world is assigned the on tax unit records has composition underwent average income of his or her income documented, household profound changes over decile. Incomes are measured on a surveys might fail to measure top in- the 20-year period. Figure 1 shows a per capita basis and adjusted for dif- comes accurately. As a result, surveys clear rightward shift of the Chinese ferences in purchasing power across might underestimate inequality and distribution—that is, toward higher in- countries. could fail to notice changes in inequal- comes. It also shows that Sub-Saharan ity at the very top of the distribution. Africa accounts for an increasing share The paper offers a of the world’s poorest people. The up- Figure 1. The Global Distribution of Income, Logarithmic Scale first preliminary anal- ward movement of China changed the Population-weighted density of log income ysis of how this might overall shape of the global distribu- 0.4 1988 affect estimates of tion from a twin-peaked to an almost 0.3 global inequality. It single-peaked distribution. The share approximates top of the world’s population living on be- 0.2 incomes by the gap tween PPP$2 and PPP$16 a day (or be- between national ac- tween PPP$730 and PPP$5,840 a year) 0.1 counts consumption increased from some 23 percent to 40 and survey means, percent. 0 in combination with a Pareto-type impu- 50 100 150 200 300 400 500 750 1k 2k 3k 4k 6k 8k 10k 15k 20k 30k 40k 50k Annual per capita income (2005 PPP$) tation of the upper tail. The estimated Population-weighted density of log income 0.4 global Gini is much 2008 higher, at nearly 76 0.3 percent in 2008, and the downward trend 0.2 almost disappears. Looking below 0.1 the surface, the paper shows how 0 the 20-year income Christoph Lakner and Branko Milanovic. 2013. growth rate between 50 100 150 200 300 400 500 750 1k 2k 3k 4k 6k 8k 10k 15k 20k 30k 40k 50k “Global Income Distribution: From the Fall of Annual per capita income (2005 PPP$) 1988 and 2008 var- the Berlin Wall to the Great Recession.” Policy China India Rest of world ies across indi- Research Working Paper 6719, World Bank, Sub-Saharan Africa Mature economies vidual country-decile Washington, DC. 4 World Bank ResearchDigest Narrowing the Gap in India Poverty has declined more sharply in rural areas, and in urban areas same direction between successive for socially disadvantaged groups (figure 1). This reduction in poverty pairs of years). While the poverty than for upper-caste groups in India was achieved alongside accelerating rate was higher for Muslims than for growth rates between 2004/05 and Hindus in the majority of the states in recent decades 2009/10. in 2009/10, the reverse was true in M Poverty fell more for scheduled Bihar, Tamil Nadu, Madhya Pradesh, uch confusion has arisen in castes and scheduled and Karnataka. An policy debates in India about tribes than for any other During the anomaly is the marginal whether growth has helped caste group between postreform era in increase in the poverty the poor—and if yes, by how much 2004/05 and 2009/10, rate between 2004/05 and over what period—and whether narrowing the gap in India, growth has and 2009/10 in Bihar for some social and religious groups are poverty rates between been very good Hindus and in Gujarat being left behind. Journalists and non- the socially disadvan- for Muslims, though governmental organizations remain taged and upper-caste for the poor, and these unexpected deeply skeptical that growth has been groups. This pattern especially so for those trends were reversed in good for groups that were disadvan- provides clear evidence 2011/12. taged over long periods in the past. to refute the claim that in disadvantaged To the extent that In a recent paper Panagariya and reforms and growth groups a relationship exists Mukim ask simple questions about the have failed to help the between inequality and evolution of poverty in the postreform socially disadvantaged poverty, it suggests that era in India. Using data from expendi- or that they are leaving these groups more unequal states have lower levels ture surveys, they look at how poverty behind. A continuation of this trend, of poverty. Kerala offers the most dra- levels have changed over the past few aided by further reforms and higher matic example: its Gini coefficients in decades across six dimensions: over growth rates, would help eliminate the both rural and urban areas are among time, across states, in rural and urban difference in poverty rates between the highest, and its poverty rates in areas, across social groups, across re- the disadvantaged and the more both regions among the lowest. ligious groups, and on the basis of dif- privileged. The extent of poverty matters— ferent poverty lines. The results show Comparisons between states reveal while many people in India are poor, that growth has been very good for the that those with large scheduled caste some live in extreme poverty. The poor, and especially so for those in dis- and scheduled tribe populations face a authors emphasize the importance advantaged groups. greater task in combating poverty. This of this distinction, especially in de- Indeed, the authors find that pov- point is brought out most forcefully by bates about redistribution—and argue erty in India declined between 1993/94 a comparison of Punjab and Kerala. that the aim should be to delve more and 2009/10 along each of the six di- These two states had very similar pov- deeply into these and other issues in mensions. Poverty fell for each social erty rates by social group in 2009/10. the future. The hope is that by provid- group (scheduled castes, scheduled But because the poverty rates for the ing poverty estimates along different tribes, and other backward castes), for scheduled castes are higher than those dimensions, their research and other each religious group (Hindu, Muslim, for other groups in both states, and ongoing efforts will help ensure that Christian, and others), in every state, because the scheduled castes account future policy debates are based on for a much larger share fact. Figure 1. Poverty Rates for Selected Groups in India of the population in Percent Punjab, that state has a 70 significantly higher over- 60 all poverty rate. 50 Religious groups 40 replicate the broad pattern seen for social 30 groups. Between 1993/94 20 and 2009/10 poverty 10 fell in every single state 1983 1987/88 1993/94 2004/05 2009/10 for both Hindus and Scheduled tribes Scheduled castes Muslims, though the Muslims Average (for all groups) Arvind Panagariya and Megha Mukim. 2013. change was not always “A Comprehensive Analysis of Poverty in India.” Source: Estimates based on expenditure survey data from the National Sample Survey Office. Note: The average refers to the average poverty rate across the 17 states on which the paper monotonic (that is, it Policy Research Working Paper 6714, World reports. was not always in the Bank, Washington, DC. World Bank Research Digest 5 Is Workfare Really Cost-Effective? With participants’ forgone earnings standard evaluation methods rely on wage rate for casual market labor. On factored in, workfare may be less comparing means for program partici- average, forgone earnings amount to cost-effective against poverty than pants with those for a selected com- about a third of the workfare wage rate. parison group of nonparticipants. Factoring in these household- other options A recent paper by Murgai, specific opportunity costs, the authors I Ravallion, and van de Walle revisits find that the extra earnings from the ndia’s Mahatma Gandhi National these questions for the Mahatma workfare program in Bihar had less im- Rural Employment Guarantee Gandhi National Rural Employment pact on poverty than would the same Scheme, established through a Guarantee Scheme in Bihar. The state gross expenditure used to provide 2005 law, is explicitly aimed at reduc- is among the poorest in India, with either a uniform transfer to everyone ing rural poverty. It promises 100 days a rural poverty rate of 55 percent in (whether poor or not) or a uniform of work a year per rural household to 2009/10 and a rural unemployment transfer to all those holding a govern- those willing to do unskilled manual rate (18 percent) that is twice the ment-issued ration card intended for labor at the statutory minimum wage national average. Its rate of underem- poor families. This also holds when the announced for the program. The work ployment is likely to be even higher. authors allow for 10 percent leakage in requirement is meant to ensure pro- This is the type of poor labor-surplus the transfer schemes, though if leakage poor targeting. By attracting only economy in which workfare has been turned out to be much larger than that, those in genuine need, and encourag- seen as having much promise for fight- workfare would have a greater effect on ing a return to the regular workforce ing poverty. poverty. when help is no longer needed, the The authors ask whether the self- The authors conclude that even in program aims to incentivize behaviors targeting aspect of the workfare pro- this poor labor-surplus rural economy, that solve the problem of knowing gram is sufficiently pro-poor to justify the much-vaunted self-targeting who is genuinely “poor” and who is it as an efficient means of transferring mechanism that is achieved by im- not. money to poor people. Could it be posing work requirements does not The available evidence suggests that the information constraints are tilt the balance in favor of unproduc- that the program is quite well targeted so severe and unemployment so high tive workfare over options using cash to poor rural households. But even that the self-targeting mechanism us- transfers with little or no targeting. For excellent targeting matters little if the ing work requirements tilts the balance workfare to be more cost-effective than net gains to participants are small. in favor of workfare even if the work other options, it would have to work While work requirements can ensure were to produce nothing of value? Or better. Reforms would need to reduce good targeting for workfare programs, are the latent forgone incomes too the substantial unmet demand for this probably comes at a cost to poor large even in this poor labor-surplus work, close the gap between stipulated participants—notably in forgone earn- economy? The authors employ a novel wages and wages received, and ensure ings. To minimize such costs, work- survey-based method to measure that workfare is productive—that the fare tends to be advocated in places these costs at the individual level and assets created are of value to poor and times with high unemployment, use these cost measures in estimat- people. And cost-effectiveness would such as in lean agricultural seasons ing the poverty effects of the workfare need to be reassessed at the higher and during macroeconomic crises. program relative to those of various levels of funding that these changes Advocates often assume that workers counterfactuals. would imply. would be idle in the absence of the The survey asked sampled partici- program, and conclude that the net pants about their expected employ- gain is the workfare wage. ment and earnings had they not been Yet even with reasonably high taking part in the program, providing unemployment overall, the forgone individual-specific information on income of workfare participants is un- forgone opportunities. While acknowl- likely to be zero—and its magnitude edging that counterfactual survey clearly matters in whether workfare is questions can be difficult, the authors a cost-effective policy choice. Little found that response rates were high is known about these hidden costs. and the answers make sense. The few available estimates of mean The results indicate that workfare forgone income for participants range participants in Bihar, though poor, Rinku Murgai, Martin Ravallion, and Domi- from 25 percent of workfare earnings are not drawn solely from the pool of nique van de Walle. 2013. “Is Workfare Cost- (in the Indian state of Maharashtra) the unemployed. Many report forgone Effective against Poverty in a Poor Labor-Surplus to 50 percent (in Argentina). To ar- earnings, but these vary greatly, rang- Economy?” Policy Research Working Paper 6673, rive at estimates of forgone earnings, ing from roughly zero to close to the World Bank, Washington, DC. 6 World Bank ResearchDigest Growth Still Is Good for the Poor Do some macroeconomic policies decompose the income growth of the broader literature to matter directly for and institutions support shared poorest quintiles into average income inequality, such as primary school en- prosperity directly—and not just by growth and increases in the share of rollment, inequality in educational at- income accruing to the poorest quin- tainment, government expenditure on supporting overall growth? tiles. In their core specification, which education and health, and agricultural T regresses the average income growth productivity. he share of the population in of the poorest quin- Using Bayesian mod- absolute poverty has fallen tiles on overall average The historical record el averaging as a tool for sharply in the developing world income growth, they of the past four systematic data analy- over the past three decades, from 52 find a slope coefficient sis, the authors find lit- percent in 1980 to 21 percent in 2010. very close to—and not decades provides tle evidence that any of Much of this reduction has been due significantly different little guidance on the policies and institu- to rapid growth in large and initially from—one, indicating tions reflected in these poor developing countries such as that the incomes of the which macroeconomic variables are signifi- China and India. These past successes poor increase on aver- policies support cantly correlated with are making the World Bank’s austere age at the same rate growth in the incomes threshold for absolute poverty of $1.25 as overall incomes. shared prosperity of the poor beyond any a day less relevant for many develop- Moreover, a standard as distinct from direct effect of these ing countries. This has led the World variance decomposi- variables on growth it- Bank to put a new institutional em- tion indicates that aggregate growth self. This of course does phasis on tracking shared prosperity, three-fourths of the not imply that there are defined in terms of the growth rate of cross-country variation in the income no policy interventions that might pro- the incomes of the poorest 40 percent growth of the poorest 40 percent of the mote an increase in the income share of households, and to publicly commit population is due to growth in average of the poorest quintiles. But it does to the twin goals of ending extreme incomes. These results underscore the tell us that the historical record of the poverty and fostering shared prosper- importance of overall growth for im- past four decades provides little guid- ity in the developing world. provements in living standards among ance on which macroeconomic policies This new emphasis on shared pros- the poorest in societies. support shared prosperity as distinct perity naturally raises two key ques- Although growth in average in- from aggregate economic growth. tions. First, to what extent does shared comes accounts for most of the in- These findings suggest both good prosperity differ from simple prosperity, come growth in the poorest quintiles, and bad news for promoting shared which could be defined as overall in- it is nevertheless important to investi- prosperity. The good news is that poli- come growth? Second, are there mac- gate the role of macroeconomic poli- cies that promote economic growth roeconomic policies and institutions cies in driving the remaining variation. will on average also raise the incomes that can directly support shared pros- If, for example, there are some policy of the poor, thereby promoting shared perity other than through their effects combinations that increase the share prosperity. The bad news is that we on overall growth? of income accruing to the poor while cannot identify specific macroeconom- Dollar, Kleineberg, and Kraay ad- sustaining a given growth rate, these ic policies that are particularly condu- dress these two questions in a recent would naturally be preferred from the cive to promoting shared prosperity paper, elaborating on earlier work by standpoint of promoting shared pros- other than through their direct effects Dollar and Kraay in “Growth Is Good perity. But if trade-offs existed in the on economic growth. This underscores for the Poor” (Journal of Economic Growth sense that pro-poor policies might the challenge in defining practical ac- 7 [2002]: 195–225). The authors find a lead to lower average growth rates, tions and approaches to tackle the strong equiproportionate relationship policy makers would have to be aware goal of promoting shared prosperity. between overall growth and average that these effects can offset each other income growth in the poorest quintiles. in pursuit of the goal of promoting The reason is that changes in the in- shared prosperity. come share of the poorest quintiles— The authors investigate how the the difference between prosperity and income share of the poorest quintiles shared prosperity—are on average correlates with a variety of country- small and uncorrelated with aggregate level variables commonly thought growth. to matter for growth, such as trade David Dollar, Tatjana Kleineberg, and Aart The findings are based on analysis openness and macroeconomic and Kraay. 2013. “Growth Still Is Good for the Poor.” of data covering 118 countries over political stability. They also examine a Policy Research Working Paper 6568, World the past four decades. The authors number of variables considered by the Bank, Washington, DC. World Bank Research Digest 7 What Is the Effect of Growth on the Distribution of Opportunities? Analyzing growth in terms of or excluding some socio- Figure 1. Growth Incidence Curve for Italy opportunity provides a better economic groups? Annual growth rate of per capita income (%) understanding of its distributional The authors propose 5 2002–06 two instruments to inves- 2006–10 effects tigate these questions. T The individual Opportunity he relationship between in- Growth Incidence Curve equality and growth was origi- (individual OGIC) plots 0 nally investigated through a the rate of growth of the macroeconomic approach, on the value of the opportunity basis of aggregate indicators. More set provided to individu- recently a microeconomic approach als, making it possible to −5 has flourished. This approach evalu- assess whether growth ates growth by looking at its impact on worsens or alleviates 0 20 40 60 80 100 Population percentiles (income distribution) each part of the income distribution. IOp. The type Opportunity Note: Population percentiles are obtained by ranking individuals by their income at each The main instrument is the Growth Growth Incidence Curve (type point in time. Incidence Curve, which plots the mean OGIC) plots the rate of income growth of each percentile in growth of the value of the Figure 2. Individual Opportunity Growth Incidence Curve for Italy the distribution in a given period. opportunity set of each Annual growth rate of individuals’ opportunity (%) Common to all this literature is a fo- socioeconomic group (or 4 2002–06 cus on outcomes, such as income or “type”), making it pos- 2006–10 consumption. sible to detect possible 2 A recent branch of normative eco- inequality traps. nomics, which gave rise to the Equality Embracing the EOp 0 of Opportunity (EOp) theory, proposes perspective can consider- opportunities (rather than outcomes) ably change the evalu- as the most appropriate variable for ation of growth. This −2 judging equity. Proponents of EOp can be seen in illustra- accept outcome inequalities arising tions based on data for −4 from individual choices, but not those Italy (derived from the 0 20 40 60 80 100 caused by factors beyond individual Bank of Italy’s Survey on Population percentiles (smoothed distribution) control. Household Income and Note: The smoothed distribution is obtained by substituting for each individual’s income Sharing this view, Peragine, Wealth). Viewed from the value of his or her opportunity set (measured by the average income of his or her type). Population percentiles are then computed by ranking individuals by the value of their Palmisano, and Brunori argue in a the outcome perspec- opportunity set at each point in time. recent paper that complementing tive, the growth that took the standard micro approach with an place in Italy in 2002–06 Figure 3. Type Opportunity Growth Incidence Curve for Italy analysis of the impact of growth in the was quite progressive: Annual growth rate of types’ average income (%) space of opportunity provides a bet- the GIC has a smoothly 6 2002–06 ter understanding of the distributional decreasing shape (figure 2006–10 effects of growth. This can help distin- 1). This progressivity is 4 guish between two growth processes reversed when the EOp that have, say, the same impact in re- perspective is adopted: 2 ducing poverty and inequality, though the individual OGIC in one case all members of a certain shows a slightly increas- 0 ethnic minority—or all people whose ing shape (figure 2). In parents are illiterate—experience the the most recent growth −2 lowest growth rate while in the other episode, in 2006–10, case poverty reduction is uncorrelated however, the burden of −4 with differences in ethnicity or family the recession was borne 0 5 10 15 Types background. by the more vulnerable Using this approach, the authors segments of the popula- Note: The distribution is partitioned into 18 types using information on three characteristics of the household head: parents’ highest educational attainment, parents’ highest occupational address the following questions: Is tion, as shown by the re- status, and geographic area of birth. Types are then ranked by initial average income. growth reducing or increasing the de- gressive pattern of both gree of opportunity inequality (IOp)? the individual and type And is growth systematically favoring OGIC (figure 3). (continued on page 8) 8 World Bank ResearchDigest (continued from page 1) focusing on the growth rate of the In conclusion, Basu recommends Recent Policy Research per capita income of the poorest 40 going for both pro-growth and specific Working Papers percent ignores inequality. Basu as- anti-poverty interventions, but do- serts that this criticism is misplaced. ing so with the knowledge that there 6633 Industrial Policy in the African Context Countries already track the growth rate are no easy fixes. Ending poverty and Joseph Stiglitz, Justin Lin, Célestin Monga, of the per capita income of the entire grappling with inequality will require and Ebrahim Patel 6649 Shared Prosperity: Links to Growth, population, and as soon as one sees bringing the best evidence and the best Inequality and Inequality of Opportunity the income growth rate for the poorest reasoning to bear, along with a healthy Ambar Narayan, Jaime Saavedra, and Sailesh Tiwari 40 percent next to this figure, it is clear dose of innovation. 6650 Up in Smoke? Agricultural Commercialization, whether the poorest 40 percent are Rising Food Prices and Stunting in Malawi Benjamin Wood, Carl Nelson, Talip Kilic, and falling further behind or catching up. Kaushik Basu. 2013. “Shared Prosperity and the Siobhan Murray One weakness in the shared pros- Mitigation of Poverty: In Practice and in Precept.” 6723 The Impact of High School Financial perity measure, however, is that two Education: Experimental Evidence from Brazil Policy Research Working Paper 6700, World Miriam Bruhn, Luciana de Souza Leão, nations with the same per capita in- Bank, Washington, DC. Arianna Legovini, Rogelio Marchetti, and Bilal Zia come and with the same income share 6727 Stunted Growth: Why Don’t African Firms accruing to the poorest 40 percent (as What Is the Effect of Growth on Create More Jobs? Leonardo Iacovone, Vijaya Ramachandran, and shown by Lorenz curves intersecting the Distribution of Opportunities? Martin Schmidt at the 40 percent mark) will be judged 6728 Opportunity-Sensitive Poverty Measurement identical, even though one country (continued from page 7) Paolo Brunori, Francisco Ferreira, Maria Ana Lugo, and Vito Peragine may have greater inequality every- The differences in the shapes of 6729 Intrinsic Motivation, Effort and the Call to where else. While this is a legitimate these growth curves highlight the im- Public Service Sheheryar Banuri and Philip Keefer shortcoming, Basu argues that simplic- portance of enlarging the space of in- 6736 The Challenge of Measuring Hunger ity is important if the twin goals are to vestigation in assessing distributional Joachim De Weerdt, Kathleen Beegle, Jed Friedman, and John Gibson be easily understood and widely taken dynamics. The results suggest that 6737 Sanitation and Externalities: Evidence from up. analyzing growth in terms of oppor- Early Childhood Health in Rural India Exploring growth and direct wel- tunity sheds light on many aspects of Luis A. Andres, Bertha Briceño, Claire Chase, and Juan A. Echenique fare interventions as instruments for development that traditional analytical 6740 Is Extreme Poverty Going to End? An tackling poverty, Basu makes the case tools, focusing solely on income, can- Analytical Framework to Evaluate Progress in Ending Extreme Poverty for avoiding an “either-or” approach. not capture. This opportunity-sensitive Nobuo Yoshida, Hiroki Uematsu, and Carlos The experience of countries as diverse approach provides useful insights for E. Sobrado 6742 Evaluating Aid for Trade: A Survey of Recent as China, Brazil, and India shows that policy making: it may help in targeting Studies direct interventions by the state to specific groups and identifying priori- Olivier Cadot, Ana Fernandes, Julien Gourdon, and Aaditya Mattoo reduce malnutrition or provide educa- ties in redistributive and social policies. 6748 School Resource and Performance Inequality: tion or health services can make a dif- Evidence from the Philippines Futoshi Yamauchi and Suhas Parandekar ference. Indeed, it can be argued that Vito Peragine, Flaviana Palmisano, and Paolo rather than waiting for growth to bring Brunori. 2013. “Economic Growth and Equality improvements in these areas, coun- of Opportunity.” World Bank Economic Re- Working Papers can be downloaded at http://econ.worldbank.org tries can actually use such interven- view. Published online October 14. doi:10.1093/ To download the World Bank Research E-Newsletter, tions to step up their growth rates. wber/lht030. go to http://econ.worldbank.org/research_newsletter The World Bank Research Digest is a quarterly publica- The Research Digest is financed by the Bank’s Editorial Committee: Indermit S. Gill (managing editor), tion disseminating findings of World Bank research. Research Committee and managed by DECRS, the Aslı Demirgüç-Kunt, and Shiva S. Makki. Editor: Alison The views and interpretations in the articles are those research support unit of the Development Economics Strong; production: Roula Yazigi. For information or of the authors and do not necessarily represent the Senior Vice Presidency (DEC). The Research Digest is free subscriptions, send email to research@worldbank. views of the World Bank, its Executive Directors, or the not copyrighted and may be reproduced with appropri- org or visit http://econ.worldbank.org/research_digest. countries they represent. ate source attribution. The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper