112736 S er t Po pe y v ci an ResearchDigest al d Is In su e World Bank e qu on al it y VOLUME 11 NUMBER 2 WINTER 2017 Estimating International Poverty Lines Is the World Bank’s international in a common year, inflation data for IN THIS ISSUE poverty line of $1.90 a day both several years are needed to update robust and relevant? Recent their values. The estimated value of Estimating International Poverty $1.90 (expressed in 2011 dollars ad- Lines … page 1 analysis suggests that it is justed for purchasing power parity, or T PPP) requires an average of 14 years Global Inequality between he share of people living in ex- of inflation data from each of the 15 1988 and 2008 and the treme poverty, as measured by countries in the reference group, some “Elephant Graph” … page 2 the international poverty line of which have had lapses in the con- Challenges in Counting the estimated by the World Bank, has be- sistency of their inflation measures. World’s Hungry … page 3 come one of the most prominent indi- Errors in inflation data can similarly cators for assessing progress in global change the poverty classification of Incentive Effects of Antipoverty economic development. It has been a tens of millions of people. Programs … page 4 central indicator for the Millennium Another concern is whether na- Development Goals and is now an tional poverty lines from past decades Importing High Food Prices important one for the Sustainable are too frugal or outdated for assess- by Exporting … page 5 Development Goals. Since the dollar- ing societies’ views of basic needs. Are a-day poverty line was introduced in old poverty lines relevant for today’s The Decision to Invest in Child 1990, the guiding concept for estimat- international poverty line? Quality over Quantity … page 6 ing the international poverty line has In a recent paper Jolliffe and Prydz been to base the line on a typical val- construct a new set of national poverty Supporting Parents to Improve ue of a subsample of national poverty lines that help inform many issues re- Preschool Outcomes … page 7 lines for some of the world’s poorest lated to global poverty measurement, Can Conditional Cash Transfers countries. The most recent World Bank including analysis on the relevance Improve Child Nutrition? … page 8 international poverty line, $1.90 per and robustness of the $1.90 inter- person per day, is based on the simple national poverty line. They derive a Globalization, Comparative average of national poverty lines from set of national poverty lines that are Advantage, and Fertility the 15 poorest countries from a sam- more recent, represent more coun- Decisions … page 9 ple of 74 national poverty lines dating tries, and directly correspond with from 1985 to 2005. national poverty counts as reported Experimental Justice Reform … page 10 The use of 15 national poverty lines by the World Bank for monitoring the that are more than 14 years old on Millennium Development Goals and Recent World Bank Research average creates concerns for several the Sustainable Development Goals. Publications and Papers reasons. One worry is that they provide The authors’ approach is based on Poverty and Inequality … page 12 weak support for the international on estimating national poverty lines poverty line: small changes in the by combining national poverty head- composition of the 15-country refer- counts with corresponding consump- ence group can result in large changes tion and income distributions as re- in the estimated international poverty ported in the World Bank’s PovcalNet line, altering the poverty status of tens database (a public research tool for of millions of people. international poverty estimation). The age of the lines is also a con- cern. To state the value of these lines (continued on page 11) 2 World Bank ResearchDigest Global Inequality between 1988 and 2008 and the “Elephant Graph” The global growth incidence curve distribution. The paper is the first seen faster growth than the rest of the (or “elephant graph”) reveals the analysis of global inequality that tries global distribution and thus occupy contrast between fast growth in Asia to adjust for missing top incomes. It higher global percentiles in 2008 than approximates top incomes by the gap in 1988. The quasi-non-anonymous and slow growth in the West between national accounts consump- GIC in figure 1 controls for such A tion and survey means, in combination churning as well as for differences in recent paper by Lakner and with a Pareto-type imputation of the population growth rates. Specifically, Milanovic presents new evi- upper tail. This results in an estimated this curve keeps a country-decile fixed dence on global inequality. global Gini index that is much higher, at its 1988 position and shows what The paper uses a newly compiled and at nearly 76 in 2008, and the downward growth it experienced over the next 20 improved database based on national trend almost disappears. years. Even after controlling for such household surveys from 1988 to 2008 Differences in average income churning in the global distribution, the and with data drawn mostly from the between countries remain the most growth rate dips around the 80–85th World Bank’s PovcalNet database and important source of global inequal- global percentile. the Luxembourg Income Study. Every ity. While this source of inequality has There were some 207 million person in the world is assigned the declined, thanks to the strong growth people around the 80–85th percen- average income of his or her income in some parts of Asia, it still explains tile in 1988. About 172 million (more decile. Incomes are measured on a around three-quarters of global in- than 80 percent) were from today’s per capita basis and adjusted for dif- equality as measured by the Theil-L OECD countries. Even if the formerly ferences in purchasing power across index. On the other hand, (population- Communist economies are dropped, countries. weighted) inequality within countries there are still 145 million people (or 70 The data show that global income has increased, both in levels and as a percent) from the “old rich” economies. inequality remains high compared share of global inequality. While figure 1 shows relatively with inequality within countries: be- The aspect of the paper receiving large gains for the part of the distribu- tween 1988 and 2008 the global Gini most attention has been the anony- tion around the median, these gains index declined only slightly, from mous global growth incidence curve are measured in relative (percentage) around 72 to 70. But global inequality (GIC) shown by the solid blue line in terms. Precisely because global in- may be even higher than this measure figure 1, which has come to be known comes are distributed very unequally, indicates. Household surveys often fail as the “elephant graph.” The figure the absolute gains are much greater to measure top incomes accurately— reflects the income growth rate for dif- for higher percentiles. While aver- and as a result, might underestimate ferent percentiles of the global distri- age per capita income increased by inequality and fail to notice changes bution between 1988 and 2008. Growth PPP$25,000 for the top 1 percent be- in inequality at the very top of the has been distributed such that it was tween 1988 and 2008, it increased by fastest for percen- only PPP$400 at the global median. Figure 1. Growth Incidence, 1988–2008 tile groups just Thus 44 percent of the increase in Annual growth rate of real PPP income (%) above the global global income between 1988 and 2008 3.0 median (at around went to the top 5 percent of the world Anonymous Quasi-non-anonymous 3 percent a year), population. 2.5 followed by the global top 1 per- 2.0 cent, while the 80th percentile saw very little growth. 1.5 Because of churning in the 1.0 global distribu- tion, the country 0.5 composition of the global percentile Christoph Lakner and Branko Milanovic. 2016. 0 groups tracked by “Global Income Distribution: From the Fall of the 0 10 20 30 40 50 60 70 80 90 100 the anonymous Berlin Wall to the Great Recession.” World Bank Percentile in the global income distribution GIC can change Economic Review 30 (2): 203–32. For recent Note: The anonymous GIC shows the growth rate for each ventile (as well as the top 1 percent) of over time. For ex- discussion of the elephant graph, see http://blogs. the global distribution, ignoring changes in country composition. For 63 countries observed in 1988 and 2008, the quasi-non-anonymous GIC plots growth for a country-decile against its position in the ample, many Asian worldbank.org/developmenttalk/rebuttal-elephant- 1988 global distribution. decile groups have graph-discussion. World Bank Research Digest 3 Challenges in Counting the World’s Hungry Variations in the way household variation in the design of household supervised personal diary with de- surveys measure food consumption consumption surveys around the tailed cross-checks across household lead to big differences in estimates world. The variation occurs in many members. Because the study is a ran- dimensions, including these: domized within-village experiment of hunger •  The method of data capture, that has good covariate balance and G which is typically either diary or a re- extends over an entire calendar year, lobal numbers on the preva- call questionnaire. the results are not subject to seasonal lence of hunger are widely •  The reference period over which or community-level confounders. The followed, not least because consumption is measured, which can only difference in population hunger ending hunger is the second of the 17 vary from one day to one week to one numbers should be due to the particu- Sustainable Development Goals. But month. lar survey design administered. And both the standard approach to esti- •  The degree of commodity detail, what a difference it is. mating these numbers and an alterna- which can range from a handful of ag- The average estimated daily per tive that has been increasingly pro- gregate commodities to more than 400 capita (kilo)calorie intake ranges from moted face measurement challenges. relatively detailed commodities. 1,794 to 2,677, depending on the sur- These raise questions about the cross- Using a consumption survey experi- vey module. As a consequence, the country validity of both approaches as ment in Tanzania, the authors study estimated prevalence of hunger, based currently practiced, according to find- the implications of survey design for solely on HCES data, ranges from 19 ings in a recent paper by De Weerdt, the measurement of hunger. It turns to 68 percent (table 1). This wide range Beegle, Friedman, and Gibson. out that hunger estimates are even in prevalence translates into a differ- The use of household survey data is more sensitive to survey design than ence of more than 23 million people in fundamental in both approaches. The total consumption estimates are, be- Tanzania, a country with a population standard approach combines mean cause most of the design differences in of 45 million according to the 2012 per capita food consumption derived global practice are due to differences census. from national food balance sheets in how food consumption is measured. And it’s not just prevalence esti- (FBS) with estimates of the distribu- For example, a review of consumption mates that are influenced by survey tion of caloric intake (caloric variance) surveys from more than 100 countries design. The targeting of the hungry will across households that are based on finds that while the most common also be affected. The hunger profiles household survey data. The alternative recall period for food consumption suggested by each module differ be- relies solely on household consump- is seven days, the seven-day recall is cause the hunger indicator varies sig- tion and expenditure surveys (HCES), used in only 31 percent of the surveys. nificantly with household size, wealth, measuring hunger directly as a func- The study explores the net effect education, and other measures. For tion of the observed total household of reporting error arising from seven example, relative to the gold standard, food consumption (in relation to esti- different, very common designs of recall surveys underestimate hunger as mates of household caloric need). consumption survey, including one the household grows richer, but over- For both the FBS and HCES that the authors consider to be the estimate it as the household increases methods, one challenge is the wide survey “gold standard”—an intensively in size. Therefore, simple mean cor- rection factors for each module would not be sufficient when comparisons Table 1. Survey Design and Estimates of Calorie Availability need to be made across population Hunger prevalence (%) groups with different characteristics. Hybrid This study’s analysis could potentially Mean Direct estimate from be used to make corrections, though kilocalories estimate from FBS and HCES it would be difficult to measure and Type of HCES design per capita HCES data data control for all relevant characteris- Long list of 59 food items; 14-day recall 1,794 68 25 tics. Perhaps the bigger challenge is Long list of 59 food items; 7-day recall 2,129 48 23 to determine the extent to which any correction factor estimated from this Short list of 17 food items; 7-day recallª 2,066 48 23 study can apply to different contexts Long list of 59 food items; usual 12-month recall 1,909 59 26 and countries (for example, settings 14-day household diaries with frequent visits 2,412 27 20 with different income levels or staple foods). 14-day household diaries with infrequent visits 2,517 23 21 14-day individual diaries with frequent visits 2,677 19 21 a. Because the 17 foods account for 77 percent of the food budget, calorie availability is scaled up by 1/0.77. (continued on page 11) 4 World Bank ResearchDigest Incentive Effects of Antipoverty Programs Does China’s targeted antipoverty practice have either calculated the some insights for policy reform in the program trap participants in poverty transfers or taxes implied by the for- light of the literature on optimal taxa- by creating disincentives for them to mal rules or calculated conditional tion. In theory, the Dibao program’s means of actual transfers or taxes at design is ideal for protection from earn additional income? each level of net income. It is well rec- poverty but bad for promotion out of T ognized that behavioral responses can poverty because it imposes a 100 per- argeted social policies aimed at invalidate either method. The authors cent marginal tax rate on poor partici- reducing poverty are widespread study the bias in statistical estimates pants—a poverty trap. But the authors today, both among rich countries of the benefit withdrawal rate induced find no sign of this in the data. Indeed, and, increasingly, among emerging by latent incentive effects and income the benefit withdrawal rate appears to middle-income economies. Concerns measurement errors. They also identify be quite low when compared with the about their incentive effects have long a third source of bias (not previously range of values suggested by optimal been prominent. Do they create disin- discussed in the literature): correlated tax studies. Incentives for promotion centives for participants to earn more incidence heterogeneity. For example, appear to be strong, but performance income, leading to a poverty trap? on moral grounds program administra- in protection is weak. This long-standing debate is rel- tors may resist reducing the benefits The authors argue that the reason evant to a major new antipoverty of the poorest family when its income may be found in local implementation program in China. Introduced in the rises slightly. The extent to which this practices. Local agents implicitly put 1990s, the Minimum Livelihood happens will naturally vary with the a far higher weight on promotion than Guarantee program—popularly known amount of local discretion. implied by the program’s design. But as the Dibao program—covered about The authors offer a new approach the data indicate heterogeneity in this, 21 million people in urban areas and to estimating average benefit inci- with richer cities tending to put higher 54 million in rural areas by 2013. dence that can be implemented with weight on protection and poorer cities The Dibao program is aimed at pro- essentially the same data as prevail- appearing to have a greater problem viding locally registered urban house- ing methods of nonbehavioral benefit with participant capture. One possible holds whose per capita income falls incidence analysis, but without ignor- explanation is that richer cities have below a predetermined local poverty ing incentive effects and measurement more professional local administra- line with a transfer payment sufficient errors. In a departure from past work tors, less prone to participant capture. to bring their income up to that line. they focus directly on a key policy pa- The key policy conclusion is that The program’s design implies that rameter, the benefit withdrawal rate. the Dibao program is unlikely to pro- participants who have a small increase The key assumption is that incentive vide a strong disincentive for partici- in nonprogram income would face an effects and classical measurement er- pants to earn extra income. Incentive equal reduction in program receipts— rors affect only certain lagged income effects appear to be more serious than a 100 percent benefit withdrawal rate components but that these still have presumed by standard nonbehavioral (or marginal tax rate). Incentives to es- predictive power for isolating exog- benefit incidence analysis but still cape poverty would be weak or absent. enous variation in total income net of much less severe than basic incentive But there are many reasons why the transfers or taxes. This justifies an in- theory would suggest. Indeed, the find- actual benefit withdrawal rates of an strumental variables estimator for the ings suggest that reforms to the pro- antipoverty program might differ from mean benefit withdrawal rate. gram should strive for a higher benefit the nominal rate. In the Dibao program To implement this approach, the withdrawal rate in local implementa- one possible reason is the scope for authors use a specially designed and tion, alongside expanded coverage. local discretion and innovation. commissioned survey to study the In a recent paper Ravallion and Dibao program. The results suggest a Chen study the Dibao program with sizable bias in the benefit incidence the goal of assessing whether it has picture that is implied by either the created a poverty trap—whether it op- formal rules or the usual statistical erates in practice in the ways that its practice of calculating conditional formal rules suggest. The paper focus- means at different net incomes. The es on the problem of estimating the estimated mean benefit withdrawal mean benefit withdrawal rate, given by rate is much lower than the program’s the average rate at which transfer re- formal rules suggest, yet about twice Martin Ravallion and Shaohua Chen. 2015. ceipts respond to differences in house- that implied by the standard statistical “Benefit Incidence with Incentive Effects, Measure- hold income. approach. ment Errors and Latent Heterogeneity: A Case Most methods of calculating the By focusing on the key parameter Study for China.” Journal of Public Econom- benefit withdrawal rate found in for policy design, the authors also offer ics 128: 124–32. World Bank Research Digest 5 Importing High Food Prices by Exporting In Lao PDR, which uses export to Thailand and Vietnam, the coun- Thailand and Vietnam, and thus sig- controls to keep rice prices stable, try’s main trading partners. The study nificantly increase incomes from agri- prices are more likely to rise after a provides a number of insights. First, culture and stimulate modernization Lao PDR imports rice price changes of the sector. Indeed, there is emerging good harvest than a bad one in the long run by exporting its rice to evidence that many poor households R Thailand and Vietnam. Second, rice also become better off in the medium apidly rising staple food prices prices in Lao PDR are considerably term as the benefits from the agri- are a challenge in many develop- lower than in Thailand and Vietnam, cultural expansion eventually trickle ing countries: they directly affect despite the evidence of market inte- down, likely in the form of greater de- poor households, which often spend gration. Third, and most interestingly, mand for farm workers and higher rural more than half their income on food, good harvests and large exports result wages. But the benefits of this indirect and they increase the risk of food in higher domestic prices later in the effect are of course uncertain. Some riots and political unrest. Rice trade year. This is consistent with export intermediate solution that protects the policy is therefore often not guided by controls being relaxed after good har- poor may therefore be advisable. Cash economic principles alone, and many vests, presumably to prevent sharp or food transfers are one policy option decisions tend to be highly politicized. price falls, triggering a surge in exports for protecting vulnerable households In the Lao People’s Democratic early in the season that leaves stocks against rising food prices. Republic, one of the least developed depleted ahead of the next harvest. However, trade liberalization may countries in Southeast Asia, glutinous The use of export controls with the not be politically feasible because of (sticky) rice consumption provides aim of maintaining stable and afford- its distributional consequences, as evi- almost 70 percent of calorie and pro- able prices is common in developing denced by the number of governments tein intake. In 2010 the price of rice in countries with a potentially exportable that use trade policies to influence the country increased by more than staple food. In Lao PDR this policy cre- food prices. Adopting a variable export 50 percent, leading the government ates two regimes: a no-trade regime in tax with transparent rules may be a vi- to impose a ban on the export of rice. which prices are determined primarily able second-best policy option. Ideally, The ban lasted only until February by domestic factors (there is also ille- once economic growth has reduced the 2011, yet it was widely criticized by gal trade) and a trade regime in which dependence of most poor households economists and multilateral organiza- they are close to the (higher) inter- on glutinous rice, export restrictions tions. Evaluating whether this criti- national prices. When there is a good can be fully dismantled. cism was justified or not requires a harvest and an opportunity to export, complete picture of how staple food millers (the main traders) wish to sell prices are determined. as much as possible. Stocks therefore Lao PDR has an extensive set of tend to decline fast in a surplus mar- controls on international rice trade, ket because of the easing of controls, at both the national and provincial which produces a larger than normal level. The central government has the price increase at the end of the season. responsibility for maintaining stable Why do millers not behave as if rice prices and can both impose trade they expect a price spike at the end of bans and release stocks to prevent the season and thus store rice during price increases. In general, however, the year? One conjecture is that the decisions about international trade are scope for storage is limited by uncer- made at the provincial level, though tainty resulting from a combination of this is not allowed by the World Trade segmented markets, the irregularity of Organization. Because prices in neigh- the export opportunities, and the po- boring countries are higher than in tential risk of government intervention. Lao PDR, a temporary removal of an Millers get the opportunity to sell large export ban in one province is likely to quantities of rice abroad only occa- generate both increased exports and sionally, and because they do not know interprovincial trade. how long it will last, they sell as much In a recent paper Durevall and van as they can. der Weide evaluate the drivers of glu- Lifting the trade restrictions per- Dick Durevall and Roy van der Weide. 2016. tinous rice prices and the reason for manently would then seem to be a “Importing High Food Prices by Exporting: Rice recent price spikes in Lao PDR, using natural policy option. The expectation Prices in Lao PDR.” Journal of Agricultural data on prices, exports, and harvests. is that this would raise domestic rice Economics. Published electronically August 4. The focus is on the role of exports prices by reducing the price gap with doi:10.1111/1477-9552.12179. 6 World Bank ResearchDigest The Decision to Invest in Child Quality over Quantity Family size matters for household between these students and their Living Standards Survey (2002, 2006, investment in education in Vietnam: richer peers. and 2008). parents with fewer children spend Moreover, in countries where the Estimation results indicate that private-school sector is almost non- each additional sibling reduces a rural more per child on tutoring existent (at least at the pre-tertiary household’s investments in a child’s O level), such as Vietnam, private tu- schooling as measured through a vari- ver the past four decades there toring represents a type of flexible ety of indicators: it reduces education has been much study of the re- household education investment that expenditure by 0.4 standard devia- lationship between household is most likely to be the equivalent of tions and tutoring expenditure by 0.5 choices on the quantity and quality of household investment in private edu- standard deviations; it decreases the children. The hypothesis driving the cation in other contexts. Vietnam also child’s probability of being enrolled literature is that parents make trade- offers a particularly interesting case in tutoring by 32 percentage points; offs between the number of children study because it has undergone rapid and it cuts the average time spent they bear and the “quality” of those change in fertility and educational at- on tutoring by 74 hours a year and children—shorthand for the amount tainment, with fertility falling sharply the number of years of tutoring by of investment that parents make in at the same time that educational at- 1.4. With respect to the differences their children’s human capital. If this tainment has been rising rapidly. The between investments in tutoring and hypothesis is true, it has implications government has paid much attention regular school, estimation results for policies aimed at increasing eco- to family planning and has promul- show that one more sibling reduces by nomic growth and reducing poverty. gated policies over the past 50 years 31 percentage points the probability For example, this could motivate poli- encouraging (and, for government em- of attending tutoring (unconditional cy makers to work on policies aimed at ployees, requiring) families to restrict on whether the child is enrolled in helping couples avoid unwanted births their number of children to one or school or not); reduces by 243,000 or subsidizing birth control. two, but very few if any studies rigor- dong (about $15) the amount spent on A recent paper by Dang and Rogers ously investigate the quantity-quality education net of tutoring expenditure; investigates a new measure of house- tradeoff for this country. and reduces by 8 percentage points holds’ investment in their children: To identify the causal impacts of the share of tutoring expenditure in private tutoring (or extra classes) in family size on household investment education expenditure and by 20 per- mainstream subjects at school in in private tutoring in rural Vietnam, centage points the years attending which children are tested. Private the authors use as an instrument the tutoring as a share of completed years tutoring is now widespread in many distance from the household’s com- of schooling. countries and may be an especially mune to the nearest family planning These results provide considerable good measure of a household’s deci- center. In contrast to those used in support for the quantity-quality trade- sion to invest voluntarily in children’s most previous studies, this instrumen- off in the Vietnamese context. The human capital—compared with enroll- tal variable allows them to study the analysis also suggests that compared ment, for example, which may also effects of family size for families with with traditional indicators such as reflect exogenous factors such as com- one child or more. The authors also enrollment, data on tutoring may be a pulsory schooling laws. Evidence in- provide the most comprehensive em- more illuminating indicator of parents’ dicates that private tutoring improves pirical investigation to date of differ- willingness to invest in the quality of students’ academic performance in a ent aspects of household investment education of their children and can of- number of countries. in private tutoring for each child. fer a promising avenue of research in Given the rapid expansion of edu- On the conceptual front, the au- other contexts. cational attainment around the devel- thors explicitly investigate the nexus oping world, the tradeoffs that house- between private tutoring and regular holds make between the quantity and school—particularly the absolute and quality of children may increasingly relative differences between house- manifest themselves outside the for- hold investments in regular school mal education system. Indeed, this is and those in private tutoring—which likely in developing and richer coun- has not been examined before. They tries alike. In response to the widening also slightly extend the standard inequality in the United States, for ex- Becker-Lewis quantity-quality tradeoff Hai-Anh Dang and Halsey Rogers. 2016. “The ample, there have been calls for more framework to provide further insights Decision to Invest in Child Quality over Quantity: summer and extracurricular programs that can then guide the empirical Household Size and Household Investment in boosting the skills of low-income stu- analysis, which uses data from three Education in Vietnam.” World Bank Economic dents, to help level the playing field rounds of the Vietnam Household Review 30 (1): 104–42. World Bank Research Digest 7 Supporting Parents to Improve Preschool Outcomes caregivers read to children or played with them. For informal preschools in resource- which are estimated to serve more Teacher training alone (or with poor settings, integrating support of than half a million children in around monthly stipends for retention) did not parenting can be more effective than 5,000 communities across the country. improve children’s outcomes, despite These centers generally have poor- significant improvements in the class- simply improving classroom quality quality facilities, lack basic play and room environment and teacher behav- L learning materials, and operate for a iors. Moreover, a rich battery of child ack of adequate preparation is few hours each weekday morning—and assessments, conducted 36 months among the key risk factors for they are run by teachers who are typi- after baseline, showed no treatment poor performance in primary cally untrained and unpaid and have a effects among the six- to eight-year-old school. In recognition of this, the low level of education. To support early children in any treatment arm, indicat- Sustainable Development Goals call childhood development and learning, ing a substantial fade-out of program for all children, by 2030, to have ac- the government decided to improve impacts in the integrated intervention. cess to quality early childhood devel- the supply of play and learning ma- The findings echo those of Chile’s opment, care, and preprimary educa- terials in these centers and provide Un Buen Comienzo, which also found tion so that they are ready for primary teachers with additional training and that improvements in classroom quali- education. mentoring. ty did not translate into improvements A common approach to trying to The study designed an experiment in child-level outcomes at the end of improve children’s outcomes centers in which the control group received the two-year teacher training interven- on increasing enrollment in preschool only a standard kit of supplies from tion. Both the study in Chile and the programs or trying to improve the UNICEF, while a second arm also re- one in Malawi highlight the difficulty of quality of existing programs. Research ceived teacher training and mentor- converting program-induced improve- has shown that being assigned to ing. In a third arm the trained teach- ments in classroom quality into better higher-quality classrooms in kindergar- ers were assigned to receive a small child outcomes. ten can modestly increase test scores monthly stipend during the first school The trial in Malawi combined the among children. But an evaluation of year following the intervention, in- two main existing approaches to early Chile’s Un Buen Comienzo (A Good tended to increase retention and moti- childhood investments—preschool Start), a large-scale, randomized study vation among these otherwise unpaid quality improvements and parenting of an effort to improve the quality of workers. In the fourth and final arm support—which had not previously preschool education, showed that of the trial the school-based teacher been tested together. It found promis- while the program led to significant training program was complemented ing evidence that this approach can improvements in many classroom by a 12-module, group-based, parent- improve children’s outcomes over and characteristics and teacher behaviors, ing education program for the primary above teacher training—at least in the it had no effects on children’s language caregivers of the children enrolled at short run. or literacy skills. the center. This model is less costly In addition, the trial design allowed Another approach revolves around than stand-alone home-visiting pro- the study to identify the causal effect support of parents. A study in Mexico grams, and because the newly trained of exogenously improving classroom found positive effects on child devel- teachers and mentors in the commu- and parenting quality on child devel- opment from group-based parent- nity deliver the group-based parenting opment outcomes. The analysis sug- ing support, as did a similar study in training, it can be easily and inexpen- gests that in this context, improving Colombia that used a home-visiting sively scaled up. classroom quality had only a negligible approach. Primary child outcomes improved impact while improving parenting In a recent study using a cluster- at the 18-month follow-up, but only quality had significant and large ef- randomized controlled trial, Özler, in the treatment group receiving the fects. The results suggest that group- Fernald, Kariger, McConnell, Neuman, integrated intervention—with teacher based parenting support can have sig- and Fraga tested the effectiveness of training and parenting education. nificant benefits for child development teacher training at informal schools in Children in this group had significantly in an informal preschool setting, but a resource-poor setting. The aim was higher scores in an assessment of lan- that the early benefits faded over time. to assess whether such school-based guage skills, and they exhibited more interventions are more effective in im- pro-social behaviors than both the Berk Özler, Lia C. H. Fernald, Patricia Kariger, proving early childhood development control group and the teacher-training- Christin McConnell, Michelle Neuman, and Edu- and primary school readiness when only group. The gains at the child level ardo Fraga. 2016. “Combining Preschool Teacher combined with group-based parenting from the added parenting education Training with Parenting Education: A Cluster- training. were accompanied by substantial im- Randomized Controlled Trial.” Policy Research The study focused on community- provements in family care indicators, Working Paper 7817, World Bank, Washington, based childcare centers in Malawi, such as how many times a day primary DC. 8 World Bank ResearchDigest Can Conditional Cash Transfers Improve Child Nutrition? Evidence from the Philippines of growth faltering—that is, children Results show that the CCT program suggests that conditional cash under age two—and those who may led to a 15 percent improvement in transfer programs can reduce severe be less responsive to interventions at height-for-age and a 40 percent reduc- health care facilities. Including older tion in severe stunting in children. In malnutrition among children children, and therefore those whose considering some behavioral changes C height is less responsive to nutritional that may explain these gains in nutri- onditional cash transfer (CCT) interventions, may mask the program tional outcomes, the authors docu- programs are a central pillar of impact when the period of study does ment improvements in the utilization social protection policy in de- not include the first months of the of perinatal care, though no changes veloping countries. These programs child’s life. But the issue is not only in the rate of institutional deliveries, typically make the receipt of benefits one of research design; the quality of presence of skilled birth attendants, or conditional on such behaviors as health services may also limit the ex- breastfeeding practices. In addition, use of health services or school at- tent to which the use of services trans- poor children ages 0–3 in treated vil- tendance—with the aim of fostering lates into a nutritional impact. lages were more likely to have received income growth through investments In a recent paper Kandpal, age-appropriate health services, in- in health and schooling while also ad- Alderman, Friedman, Filmer, Onishi, cluding growth monitoring and immu- dressing existing poverty. and Avalos use data from an experi- nizations, in the previous six months. In the two decades since Mexico’s mental evaluation of the Philippine Infant and young child feeding practic- PROGRESA program introduced cash CCT program to explore the effects of es also appear to have been affected, transfers linked to participation in CCTs on child nutrition. They focus on with young children in treated areas health services for low-income house- children in the vulnerable age group of being more likely to consume eggs and holds, the impact of such targeted 0–3 years in a high-malnutrition set- dairy than children in control areas. social protection programs has been ting. The Philippine program provides The Philippine CCT program may assessed on a range of outcomes. cash transfers to poor households have been effective at improving child Virtually all those that have included with children ages 0–14 conditional on nutrition for several reasons. Increased incentives to use health services have investments in child health and educa- income from the cash transfer may led to greater participation in preven- tion as well as use of maternal health have facilitated an improvement in tive health and nutrition activities. services. Eligible households that diet diversity, while the program condi- On average, however, these programs satisfy compliance conditions receive tions, by promoting timely perinatal have not led to similar improvements a combination of health grants and care and regular growth monitoring in nutritional outcomes for children. In education grants every two months. and health service use, may have led particular, they have generally failed to The grants range from $11 to $32, de- to increased diet diversity and clinical reduce stunting, a measure of chronic pending on the number of eligible chil- counseling. Finally, the information child malnutrition: an analysis of 17 dren in the household. The maximum provided through the enforcement of programs that combined conditional monthly transfer represents 23 percent program conditions may have com- and unconditional cash transfers of beneficiaries’ household incomes. bined with the additional cash to shows an effect size that is neither The authors estimate treatment result in greater investments in chil- statistically significant nor biologically effects in 65 treated and 65 control dren’s welfare and improved health meaningful. villages, using data collected after 31 and sanitation. There are many possible reasons for months of implementation. To capture this disconnect between the success the range of potential impacts, they of CCT programs in improving health analyze a number of outcomes: stan- care utilization and their limited im- dardized measures of child height-for- pact on nutritional outcomes for young age and weight-for-age; stunting and children. One reason is that many severe stunting, which reflect deficits evaluations have focused on programs in height-for-age; and underweight and in Latin American countries with a severely underweight status, which relatively low prevalence of stunting or reflect low weight-for-age. They also underweight. Another is that the evalu- assess the impact on perinatal care, ations often are conducted for a rela- institutional delivery, presence of tively short period and therefore can- skilled birth attendants, breastfeeding Eeshani Kandpal, Harold Alderman, Jed Fried- not capture cumulative effects. In addi- practices, immunization, growth moni- man, Deon Filmer, Junko Onishi, and Jorge tion, most studies include all children toring and deworming, care-seeking, Avalos. 2016. “A Conditional Cash Transfer Pro- under age five and thus include both and children’s intake of foods that are gram in the Philippines Reduces Severe Stunting.” those in the period of greatest risk rich in protein. Journal of Nutrition 146 (9): 1793–800. World Bank Research Digest 9 Globalization, Comparative Advantage, and Fertility Decisions Countries that have a comparative fertility. The authors’ main thesis thus refined petroleum products and nu- advantage in sectors with a greater combines the hypothesis that fertil- clear fuel. The natural-resource-based demand for female labor also have ity is affected by women’s opportunity sectors are among the least female in- cost of time with the insight that this tensive, with a share of female workers lower fertility opportunity cost is higher in countries in total employment of 11–13 percent. A with a comparative advantage in in- This explains why countries with major ttempts to understand popu- dustries that have a higher demand for shifts toward natural resources show a lation growth and the deter- female labor. reduction in their female labor needs minants of fertility date as far The authors provide empirical of exports. back as the late 1700s, when Thomas evidence for this phenomenon us- This pattern is more general. Malthus wrote An Essay on the Principle ing industry-level export data for 61 Analysis controlling for income, overall of Population. In more recent times, manufacturing sectors in 145 countries trade openness, and regional location postulating that fertility decisions over five decades. They combine the shows a pronounced negative relation- are influenced by women’s opportu- classification of industries according ship between fertility and female labor nity cost of time, economists have to their intensity of female labor use needs of exports. incorporated choice over fertility into with data on countries’ export shares Fertility is an economic decision, growth models in order to understand to construct, for each country and time and like all economic decisions it has the joint behavior of population and period, a measure of its female labor long been considered an appropriate— economic development throughout needs of exports. This index captures the and important—subject of analysis history. Most analyses examine indi- degree to which a country’s compara- by economists. It is now clear that in- vidual countries in a closed-economy tive advantage is in female-intensive ternational trade—or, more precisely, setting. But in an era of ever-increas- sectors. The measure is used to show comparative advantage—matters for ing integration of world markets, the empirically that fertility is lower in fertility decisions. The authors’ results role of globalization in determining countries with a comparative advan- emphasize the heterogeneity of the fertility can no longer be ignored. tage in female-intensive sectors. effects of trade on countries’ industrial In a recent study Do, Levchenko, One way to provide a simple il- structures and gender outcomes. From and Raddatz examine both theoretical- lustration of the results is to compare a policy perspective, the results sug- ly and empirically the impact of com- countries with the largest positive gest that it will be more difficult for parative advantage in international and negative changes in female labor countries with technology-based com- trade on fertility outcomes. The au- needs of exports between the 1960s parative advantage in male-intensive thors’ conceptual framework is based and today. For the countries with the goods to undertake policy measures to on several observations. First, women largest observed increases in female reduce the gender gap in labor market bear a disproportionate share of the labor needs of exports, the common conditions, potentially leading to a burden in raising children. While there pattern is that they change their spe- slower pace of women’s empowerment. has been a debate on whether this cialization from agriculture-based sec- In an increasingly integrated global empirical regularity is due to nature tors to wearing apparel. In Cambodia, market, the road to female empower- or nurture, the outcome remains that for example, 80 percent of exports ment is paradoxically very specific to having a child reduces a woman’s la- in the 1960s were in food products each country’s productive structure bor supply more than a man’s. Second, sectors. By the 2000s, 85 percent of and exposure to international trade. industries vary a great deal in the gen- Cambodian exports were in wearing der mix of their labor force: some in- apparel. Since food products sectors dustries employ primarily women, oth- are right in the middle of the distribu- ers primarily men. For example, female tion of female intensity, and wearing workers account for as little as 8–9 apparel is the most female-intensive percent of employment in industries sector, this type of change in special- such as heavy machinery, and as much ization will lead to large increases in as 60–70 percent in some types of tex- female labor needs of exports. tiles and apparel. Third, differences in The largest observed decreases technology and resource endowments in female labor needs of exports are will lead to some countries having a driven by the discovery of natural re- comparative advantage in industries sources. For example, Niger was an that happen to hire a lot of women. agricultural exporter in the 1960s, with Quy-Toan Do, Andrei Levchenko, and Claudio All this being so, countries with nearly 80 percent of exports in food Raddatz. 2016. “Comparative Advantage, Inter- a comparative advantage in female- products. By the 2000s more than national Trade, and Fertility.” Journal of Devel- intensive goods will exhibit lower 60 percent of Niger’s exports were in opment Economics 119: 48–66. 10World Bank ResearchDigest Experimental Justice Reform A new, experimental approach reliably generate a seemingly impres- around inherently contentious issues to justice reform seeks to foster sive array of discrete “deliverables” (such as land and natural resources), inclusive decision-making and that can be readily photographed, and they seek to ensure its equitable counted, tracked, aggregated, and functioning through the use of regular pragmatic, locally based solutions compared. From both an accounting feedback grounded in locally generated A and an accountability perspective, data. New experimentalism incorpo- dam Smith famously declared such tasks raise few red flags and, af- rates a pragmatic orientation when de- in 1755, two decades before ter a time, enable senior managers to scribing the nature of its participants the publication of The Wealth of present a coherent, even compelling, but also positions the development Nations, that little more was required narrative to skeptical politicians and practitioner as relatively nonideologi- “to carry a state to the highest degree voters of how public resources were cal (with some commitment to liberal of opulence from the lowest barbarism spent in sensible ways in response process norms); consistently self- but peace, easy taxes and a tolerable to a clear and present development critical; and able to produce different administration of justice…” More challenge. kinds of information and translate it recently a remarkably broad set of ac- According to an assessment by into politically salient action for dif- tors, across the political spectrum, leading analysts of the field, however, ferent kinds of audiences. Such practi- have endorsed building and maintain- the actual achievements of legal trans- tioners are embedded “organizational ing “the rule of law” as a primary issue planting are fragile at best. Echoing entrepreneurs” rather than distant facing every country. Indeed, in 2010 misgivings dating back to at least the technical specialists. then World Bank president Robert 1970s, a high-profile review in 2008 A leading instantiation of this al- Zoellick argued at a major conference argued that thinking of law as a read- ternative approach is the World Bank’s that “the rule of law must be at the ily transferable technology is a seri- Justice for the Poor program (J4P). Now center of the development agenda.” ous error—because law consists of 14 years old, J4P can claim significant One might imagine that a high- institutions deeply embedded in par- achievements in terms of its global priority development goal ostensibly ticular political, economic, and social impact on the theory and practice of uniting an otherwise unlikely assort- contexts. justice reform, seeking more to influ- ment of constituencies would be Since roughly 2000 an array of dif- ence at scale rather than operate at grounded in a robust foundation of ferent organizations have sought to scale (such as through its contribution theory, method, and evidence—or at heed calls to explore a completely dif- to numerous World Development Reports least a steadily expanding record of ferent approach to justice reform—an and the work of other advocacy groups practical achievement. If anything, approach that Desai and Woolcock, in such as the World Justice Project). however, the law and development a recent article, term “new experimen- It remains to be seen whether these field has been beset from the outset by talism.” For these organizations the initial gains can be more formally rou- trenchant critique of its dominant ap- aspirational goal has been to explore tinized into everyday practice within proach, “legal transplanting,” in which the experience of justice-seeking from and beyond the World Bank, but do- the forms of institutions, codes, and diverse perspectives, incorporating in ing so will entail remaining true to procedures from one (usually high- particular the experiences and aspi- its own founding principles of regular income) country are introduced into rations of users (for example, those self-critique. another (usually low-income) country. seeking to access a prevailing justice Faced with the challenge of building system), mid-level public authority, or enhancing a legal system in the af- sovereign administrative authority, termath of a civil conflict, for example, and transnational private interests; to orthodoxy as funded by international help forge and protect a political space donors may focus on constitutional in which locally generated evidence reform, redrafting legal codes, oversee- can inform (more) equitable contests ing elections, building courthouses between those representing these mul- and jails, training police officers and tiple voices; and to engage in iterative prosecutors, and upgrading adminis- problem-solving (as opposed to solu- trative systems (for example, providing tion-selling) in response to concerns the latest software for tracking cases). nominated and prioritized by these All these tasks, importantly, can be users (as opposed to foreign experts). funded, procured, and implemented in New experimentalist approaches to Deval Desai and Michael Woolcock. 2015. “Ex- ways entirely consistent with bureau- justice reform thus diverge from ortho- perimental Justice Reform: Lessons from the World cratic imperatives back in the capital doxy in two key ways: they steward an Bank and Beyond.” Annual Review of Law cities of donor countries. They will inclusive process of decision-making and Social Science 11: 155–74. World Bank Research Digest11 (continued from page 1) Estimating International Poverty Lines Because PovcalNet harmonizes its relevance of $1.90 as an estimate of complete and recent set of national measures of well-being in per capita basic needs, the authors subsample poverty lines, both estimates result terms, the resulting national poverty the set of national poverty lines clos- in an international poverty line that lines are also in per capita terms. est in time to 2011 (the PPP reference directly corresponds to the World Bank Matching the poverty lines with report- year). This results in a sample of na- definition of extreme poverty. This ed poverty headcounts and harmoniz- tional poverty lines from 115 countries, finding suggests that the current inter- ing the lines in per capita terms helps which require on average only one national poverty line of $1.90 is robust to overcome comparability problems year of inflation data to be expressed to measurement concerns and contin- with previous databases of national in 2011 currency. Following the same ues to be a relevant marker of extreme poverty lines. guiding principle of basing the interna- poverty in the poorest of countries. The resulting database provides tional poverty line on a typical value of 864 national poverty lines covering 129 a subsample of the lowest of national countries, more than a tenfold increase poverty lines, the authors examine over the database of national poverty two subsamples of national poverty lines used to estimate the $1.90 line. lines—one from countries classified as Most of these 864 national poverty low income and the other from coun- lines, and most of the countries, are tries in the bottom quartile (based on bunched together at relatively low val- a measure of national consumption). ues. Of all the poverty lines, 37 percent The median national poverty line from are less than $3 a day and 52 percent the bottom quartile is $1.86, and the are less than $5 a day. Of those that median from the low-income subsam- Dean Jolliffe and Espen Beer Prydz. 2016. “Esti- are less than $5 a day, there is a no- ple is $1.91. mating International Poverty Lines from Compa- ticeable mass near $1.90. Despite the use of different ap- rable National Thresholds.” Journal of Econom- To assess concerns about infla- proaches for selecting the reference ic Inequality 14 (2): 185–98. doi:10.1007/ tion adjustments and the continued subsample, and of a much more s10888-016-9327-5. (continued from page 3) Challenges in Counting the World’s Hungry The focus so far has been on the design (as seen in table 1) will also survey methods and, more broadly, HCES method, but the global hun- affect the FBS-derived estimates: es- between the HCES method and the ger estimates derived from the FBS timates of the prevalence of hunger hybrid FBS-HCES method, caution is method are also subject to the vaga- range from 20 to 26 percent on the ba- warranted in drawing inferences from ries of cross-country survey design. sis of survey-design-driven differences comparisons of survey-based hunger Recall that in this method, while the in the estimated variance of the calorie estimates over time and space. mean is taken from the national food distribution. balance sheets, the variance of calorie HCES data are ubiquitous and hold consumption is measured through the potential for being a useful com- household consumption and expen- plement to existing measures of hun- diture surveys. There are numerous ger. Their appropriateness for measur- grounds for questioning the accuracy ing hunger could be improved through of the balance sheets, including the more thoughtful harmonization of Joachim De Weerdt, Kathleen Beegle, Jed Fried- degree to which they capture national survey design. But until more is done man, and John Gibson. 2016. “The Challenge of postharvest stores and losses and to undertake these design changes, Measuring Hunger through Survey.” Economic the accuracy of their root crop yields. and to understand the various sources Development and Cultural Change 64 (4): But it is clear that variation in survey of error and how they differ between 727–58. 12World Bank ResearchDigest Recent World Bank Research Publications and Papers on Poverty and Inequality Chapters in Books Ali, Daniel Ayalew, Klaus Deininger, and Marguerite Jirasavetakul, La-Bhus Fah, and Christoph Lakner. 2016. “The Duponchel. 2016. “Using Administrative Data to Assess Distribution of Consumption Expenditure in Sub-Saharan Andrews, Matt, Lant Pritchett, Salimah Samji, and Michael the Impact and Sustainability of Rwanda’s Land Tenure Africa: The Inequality among All Africans.” Policy Woolcock. 2015. “Building Capability by Delivering Regularization.” Policy Research Working Paper 7705, Research Working Paper 7557, World Bank, Washington, Results: Putting Problem-Driven Iterative Adaption (PDIA) World Bank, Washington, DC. DC. Principles into Practice.” In A Governance Practitioner’s Notebook: Alternative Ideas and Approaches, edited Brück, Tilman, Patricia Justino, Philip Verwimp, and Andrew Lakner, Christoph. 2016. “Global Inequality: The Implications by Alan Whaites, Eduardo Gonzalez, Sara Fyson, Tedesco. 2016. “Measuring Violent Conflict in Micro- of Thomas Piketty’s Capital in the 21st Century.” Policy and Graham Teskey, 123–33. Paris: Organisation for Level Surveys: Current Practices and Methodological Research Working Paper 7776, World Bank, Washington, Economic Co-operation and Development. Challenges.” Policy Research Working Paper 7585, World DC. Bank, Washington, DC. D’Souza, Anna, and Dean Jolliffe. 2016. “Coping with Food Marrero, Gustavo A., Juan Gabriel Rodriguez, and Roy Price Shocks in Afghanistan.” In Food Price Volatility Brunori, Paolo, Flaviana Palmisano, and Vito Peragine. 2016. van der Weide. 2016. “Unequal Opportunity, Unequal and Implications for Food Security and Policy, edited “Inequality of Opportunity in Sub-Saharan Africa.” Policy Growth.” Policy Research Working Paper 7853, World by Matthias Kalkuhl, Joachim von Braun, and Maximo Research Working Paper 7782, World Bank, Washington, Bank, Washington, DC. Torero, 543–81. Springer. DC. Narayan, Ambar, and Rinku Murgai. 2016. “Looking Back on Ferreira, Francisco, Sergio Firpo, and Julian Messina. 2016. Deininger, Klaus, Hari K. Nagarajan, and Sudhir K. Singh. Two Decades of Poverty and Well-Being in India.” Policy “Understanding Recent Dynamics of Earnings Inequality 2016. “Short-Term Effects of India’s Employment Research Working Paper 7626, World Bank, Washington, in Brazil.” In New Order and Progress: Development and Guarantee Program on Labor Markets and Agricultural DC. Democracy in Brazil, edited by Ben R. Schneider. Oxford: Productivity.” Policy Research Working Paper 7665, Narloch, Ulf Gerrit, and Mook Bangalore. 2016. Oxford University Press. World Bank, Washington, DC. “Environmental Risks and Poverty: Analyzing Geo-Spatial Ferreira, Francisco, and Vito Peragine. 2016. “Individual Ferreira, Francisco, Sergio Firpo, and Antonio F. Galvão. 2017. and Household Data from Vietnam.” Policy Research Responsibility and Equality of Opportunity.” In Oxford “Estimation and Inference for Actual and Counterfactual Working Paper 7763, World Bank, Washington, DC. Handbook of Well-Being and Public Policy, edited by Growth Incidence Curves.” Policy Research Working Newhouse, David, Pablo Suarez-Becerra, and Martin C. Matthew Adler and Marc Fleurbaey. Oxford: Oxford Paper 7933, World Bank, Washington, DC. Evans. 2016. “New Estimates of Extreme Poverty for University Press. Friedman, Jed, Kathleen Beegle, Joachim De Weerdt, and Children.” Policy Research Working Paper 7845, World John Gibson. 2016. “Decomposing Response Errors in Bank, Washington, DC. Journal Articles Food Consumption Measurement: Implications for Survey Pave Sohnesen, Thomas, and Niels Stender. 2016. “Is Design from a Survey Experiment in Tanzania.” Policy Douidich, Mohamed, Abdeljaouad Ezzrari, Roy van der Random Forest a Superior Methodology for Predicting Research Working Paper 7646, World Bank, Washington, Weide, and Paolo Verme. 2015. “Estimating Quarterly Poverty? An Empirical Assessment.” Policy Research DC. Poverty Rates Using Labor Force Surveys: A Primer.” Working Paper 7612, World Bank, Washington, DC. World Bank Economic Review 30 (3): 475–500. Fujii, Tomoki, and Roy van der Weide. 2016. “Is Predicted Skoufias, Emmanuel. 2016. “Synergies in Child Nutrition: Data a Viable Alternative to Real Data?” Policy Research Ferreira, Francisco, Shaohua Chen, Andrew Dabalen, Interactions of Food Security, Health and Environment, Working Paper 7841, World Bank, Washington, DC. Yuri Dikhanov, Nada Hamadeh, Dean Jolliffe, Ambar and Child Care.” Policy Research Working Paper 7794, Narayan, Espen Prydz, Ana Revenga, Prem Sangraula, Gaddis, Isis. 2016. “Prices for Poverty Analysis in Africa.” World Bank, Washington, DC. Umar Serajuddin, and Nobuo Yoshida. 2016. “A Global Policy Research Working Paper 7652, World Bank, van der Weide, Roy, Christoph Lakner, and Elena Count of the Extreme Poor in 2012: Data Issues, Washington, DC. Ianchovichina. 2016. “Is Inequality Underestimated in Methodology and Initial Results.” Journal of Economic Gill, Indermit S., Ana L. Revenga, and Christian Zeballos. Egypt? Evidence from House Prices.” Policy Research Inequality 14 (2): 141–72. 2016. “Grow, Invest, Insure: A Game Plan to End Extreme Working Paper 7727, World Bank, Washington, DC. Ferreira, Francisco, Nora Lustig, and Daniel Teles. 2015. Poverty by 2030.” Policy Research Working Paper 7892, Verme, Paolo. 2016. “Subsidy Reforms in the Middle East “Appraising Cross-National Income Inequality World Bank, Washington, DC. and North Africa Region: A Review.” Policy Research Databases: An Introduction.” Journal of Economic Goraus, Karolina, and Gabriela Inchauste. 2016. “The Working Paper 7754, World Bank, Washington, DC. Inequality 13 (4): 497–526. Distributional Impact of Taxes and Transfers in Poland.” Wydick, Bruce, Paul Glewwe, and Laine Rutledge. 2016. Policy Research Working Paper 7787, World Bank, Working Papers Washington, DC. “Does Child Sponsorship Pay Off in Adulthood? An International Study of Impacts on Income and Wealth.” Akresh, Richard, Damien de Walque, and Harounan Hallegatte, Stephane, Mook Bangalore, and Adrien Camille Policy Research Working Paper 7563, World Bank, Kazianga. 2016. “Evidence from a Randomized Evaluation Vogt-Schilb. 2016. “Assessing Socioeconomic Resilience Washington, DC. of the Household Welfare Impacts of Conditional and to Floods in 90 Countries.” Policy Research Working Unconditional Cash Transfers Given to Mothers or Paper 7663, World Bank, Washington, DC. Fathers.” Policy Research Working Paper 7730, World Bank, Washington, DC. Working Papers can be downloaded at http://econ.worldbank.org To download the World Bank Research E-Newsletter, 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: Chorching Goh (managing editor), tion disseminating findings of World Bank research. Research Committee and managed by DECDP, 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