126360 FC S Co V pe u ci ntr al ie ResearchDigest Fo s a World Bank cu nd s T on op i cs VOLUME 12 NUMBER 3 SPRING 2018 Confronting the Development Challenges of Fragility, Conflict, and Violence N early half the world’s extreme consequences of conflict. One paper IN THIS ISSUE poor live in countries affected shows that domestic socioeconomic by conflict or violence. In these policies matter for global security Confronting the Development countries people not only lack secu- because disgruntled youth are sus- Challenges of Fragility, Conflict, rity; they also are more likely to lack ceptible to radicalization into violent and Violence … page 1 adequate food and basic services such extremism at home or abroad. Another as clean water and electricity, and study shows that insecurity and What Drives the Radicalization their children, if fortunate enough to violence in Uganda led to significant of Foreign Terrorist Recruits? … page 2 survive, are less likely to go to school. welfare losses for rural households. The Cost of Fear … page 3 Failure to meet these basic needs will Yet another, based on a survey of dis- lead to greater poverty while also fuel- placed people of Mali, highlights the Leaving, Staying, or Coming Back? ing unrest and more violence. role of employment and security in Migration Decisions during the Northern The World Bank has made address- decisions on whether to return. And Mali Conflict … page 4 ing fragility, conflict, and violence a paper focusing on Myanmar’s ongo- (FCV) a strategic priority. Its approach ing transition offers insights into how How Myanmar’s Reforms Are Playing Out to fragile states has evolved over the broad processes of democratization in Rural Villages … page 5 years, but the emphasis has always and globalization are changing rela- been on prevention and early action. tions between citizens and the state. Obstacles on the Road to Economic Growth In recent years the Bank has increased The remaining papers discuss poli- in the West Bank … page 6 its lending and technical assistance cies and programs aimed at recovery to FCV countries. The International and stabilization. Two papers dem- Can Public Works Help Postconflict Development Association, the part of onstrate the importance of investing Communities Escape Poverty? … page 7 the Bank that helps the world’s poorest in infrastructure and public works countries, more than doubled its com- programs for improving livelihoods. Can Mass Media Increase Women’s mitments to FCV countries in 2018, to Another paper looks at how mass me- Political Participation after Civil War? … page 8 about 20 percent of its total. The Bank dia can increase women’s participation also helps fund the disarmament, de- in the postconflict political process. Conflict and Poverty in Afghanistan’s mobilization, and reintegration (DDR) Two other papers investigate the local Transition … page 9 programs providing socioeconomic welfare effects from military and devel- support to ex-combatants. opment aid in Afghanistan and a DDR Who Benefited from Burundi’s Research provides the intellectual program in Burundi. The last paper ad- Demobilization Program? … page 10 groundwork for addressing the chal- dresses an important methodological lenges in the transition of FCV coun- issue in measuring the impact of food U.S. Food Aid and Civil Conflict Revisited: tries. While the broad nexus between aid in FCV countries. A Methodological Caution … page 11 security and development is obvious, Collectively, these studies can the underlying causes of conflict are help inform policy makers and donors Recent Policy Research Working Papers unique to each FCV situation, leading seeking to bring stability and develop- on FCV Countries and Topics … page 12 to a need for in-depth analytical work. ment to FCV countries. But much more This issue of the Research Digest remains to be done. Over the coming features recent Policy Research years we expect the World Bank to in- Working Papers that highlight chal- vest more in FCV countries and in the lenges facing FCV countries. The research to provide the necessary ana- first four focus on causes and lytical support. Stay tuned. 2 World Bank ResearchDigest What Drives the Radicalization of Foreign Terrorist Recruits? A lack of economic opportunities Some of these fighters have en- able to attract recruits from all around in countries located closer to the gaged in extreme levels of violence in the world (map 1). A total of 61 nation- Syrian Arab Republic is among the Syria and Iraq, others have perpetrated alities are represented in the sample terrorist attacks in third countries, and of foreign recruits. Admittedly, coun- factors explaining Daesh recruiting those who ultimately return to their tries closer to Syria and Iraq and with successes home country may constitute threats a larger Muslim population have more T to domestic security. The unprecedent- of their citizens joining the ranks of the he world has experienced a dra- ed levels of international recruitment terrorist organization. But a handful of matic increase in the number of by terrorist organizations make efforts countries with a small Muslim popula- terrorist attacks since 2000 and to curb radicalization at home all the tion, most of them in Western Europe, especially since 2011. More than 100 more pressing. Yet the question of are the origin of up to 12 percent of all countries were affected in 2016, with what drives radicalization into violent Daesh foreign recruits. OECD countries suffering the highest extremism is still being debated. Second, as has been documented number of casualties since the 9/11 at- A new paper by Abdel Jelil, Bhatia, in earlier studies, aspiring Daesh re- tacks. The transnational nature of ter- Brockmeyer, Do, and Joubert addresses cruits have not less but more educa- rorism has become more salient with this question using a data set that tion than the average for men in their the emergence of multinational terror contains information on 3,965 foreign country of origin. This goes against groups such as Al-Qaeda or, more recruits of Daesh. The information commonly held views that terrorism recently, the Islamic State in Iraq and includes the recruits’ age, education, originates from poverty and despera- the Levant (ISIL, also known as ISIS country of residence, and self-reported tion. Exclusive information on self- or Daesh, its Arabic acronym). The knowledge of Islam. The data come reported knowledge of Sharia and United Nations estimates that more from a leaked cache made available desired role in the organization sheds than 25,000 foreign fighters went to to journalists, who in turn shared the further light on the terrorists’ aspira- the Syrian Arab Republic and Iraq be- information with academics and policy tions: Those who choose to be admin- tween the start of the Syrian civil war research institutions. istrators are relatively more likely to in 2011 and September 2016 to fight Analysis of the data highlights in- for either Daesh or the Al-Nusra Front. teresting facts. First, Daesh has been (continued on page 12) Map 1. Countries of Origin of Daesh Foreign Recruits Number of Daesh recruits per million inhabitants 0 0.001 – 5 5 – 10 10 – 15 IBRD 43662 | MAY 2018 >15 This map was produced by the Cartography Unit of the World Bank Group. The boundaries, colors, denominations and any other information shown on this map do not imply, on the part of the World Bank Group, any judgment on the legal status of any territory, or any endorsement or acceptance of such boundaries. Note: Based on data on a cross-section of the foreign workforce of Daesh (foreign recruits who joined the ranks of the terrorist group in Syria and Iraq) during a period stretching from early 2013 to late 2014. By definition, the number of foreign recruits from Syria and Iraq is zero. World Bank Research Digest 3 The Cost of Fear Households in conflict settings created using the geo-located data mitigation strategies and are borne by suffer welfare losses from insecurity from the Northern Uganda Baseline a broader segment of the population. as well as from direct exposure to Survey 2004 and the Armed Conflict Consequently, aid and policy respons- Location and Event Data (ACLED) es should focus not just on those who violence. What is the relative size of Project. This estimation uses the dis- directly experience violence but also these costs? tance of the community from historical on the broader risk-affected popula- D attacks by the Lord’s Resistance Army. tion. These responses to the risk of espite the large and persistent These measures of risk are combined violence likely also explain some of the costs associated with conflict with measures of exposure to violence poorer postconflict outcomes, such as and, more broadly, violence, we (community and household, current in nutrition or education. Importantly, have only a limited understanding of and historical) to disaggregate the ef- since these losses occur ex ante, they the origin of these costs, particularly fects of direct and indirect exposure occur regardless of whether violence the role of insecurity (that is, the risk to violence on household per capita actually takes place. Moreover, to the of violence). The broader literature consumption. extent that perceived insecurity differs on risk and shocks suggests that the As expected, direct exposure to vio- from actual risk levels, these losses costs associated with large covariate lence causes higher losses than indi- may persist after the “official” end to shocks, such as conflict, could be very rect exposure (10 percent as compared violence, suggesting a continuum of large. But the literature focuses pri- with 6 percent at the sample mean). responses between conflict and post- marily on direct exposure to violence, But even households directly exposed conflict periods. obscuring the potential role of indi- to violence experience important The study also raises methodologi- rect exposure through insecurity. losses from perceived risk (amounting cal concerns. The effects of violence In a recent paper Rockmore ex- to 37 percent of the total losses from are typically estimated by compar- plores the relative effects of insecurity violence). Although the estimated ef- ing directly and nondirectly affected and direct exposure to violence on fect of the risk of violence is smaller households (or regions). Because inse- the welfare of rural households in than that of the direct experience of curity is not measured, the estimated northern Uganda. Because measures violence, only a modest fraction of the coefficients for exposure to violence of insecurity do not exist, the author overall population experiences vio- capture the joint effect of the risk of uses the spatial-temporal variation in lence. In contrast, virtually the entire violence and its realization. To the ex- the placement of violence to estimate population suffers from the risk of vio- tent that directly affected households spatially disaggregated measures of lence. Indeed, the results indicate that suffer from more risk on average than insecurity. The identification is based more than 60 percent of overall losses nondirectly affected households, the on the random nature of violence ex- in rural areas of northern Uganda arise effects of exposure to violence will be perienced in villages attacked by the from risk (table 1). overestimated. Lord’s Resistance Army. The results suggest that a large By using household per capita con- The community-level measures share of the overall costs from con- sumption, the study aims to capture of subjective and objective risk are flict arise from costly ex ante risk the aggregate costs that might not be apparent when looking at narrow household responses, such as labor market responses or changes in crop Table 1. Aggregate Costs of Conflict for Rural Households in Northern Uganda, 2004 portfolios. But a better understanding Preferred Full Conservative of how risk shapes these “narrow” re- model controls estimate sponses and whether they occur on the Location of losses (% of total) extensive or intensive margins is es- sential for designing more efficient and Never-attacked households 30.0 35.4 17.5 better-targeted responses in conflict Attacked households 70.0 64.6 82.5 and postconflict settings. Disaggregation of costs (% of total) Exposure to violence 37.1 28.3 43.7 Exposure to risk 62.9 71.7 56.3 Cost as % of GDP Overall 0.9 0.7 1.0 Marc Rockmore. 2016. “The Cost of Fear: The Welfare Effect of the Risk of Violence in Northern From risk 0.5 0.5 0.6 Uganda.” Policy Research Working Paper 7642, Source: Calculations based on data sources described in text. World Bank, Washington, DC. 4 World Bank ResearchDigest Leaving, Staying, or Coming Back? Migration Decisions during the Northern Mali Conflict As displaced people integrate more whether respondents were refugees decision to return supports the hy- and more in their host communities, (living in a camp) or displaced. To this pothesis that economic factors are key. the likelihood that they will return end the paper examines intention to This is underlined by the noteworthy return, a question asked of all dis- absence of evidence that perceptions decreases with time placed and refugee respondents who of personal safety were pivotal to the A had left their homes between April decision to return. Whether respon- round the world today, an es- 2012 and January 2013 and had not yet dents considered northern Mali safe or timated 65 million people are returned. whether they felt safe at home alone displaced, including 24 million The paper finds that if displaced did not significantly affect migration refugees and asylum-seekers. These people had not already returned by decisions. That said, people who per- numbers represent the second largest August 2014 (the time of the baseline), sonally experienced difficulties with refugee crisis since World War II. To the likelihood of actual return during the police were more likely to return, provide new solutions to this crisis, the 12 subsequent months was very and gun ownership was also associat- international institutions aim to com- low. The probability of transition from ed with a higher willingness to return. bine their humanitarian and develop- being internally displaced to returning While the paper finds that the ment know-how. Yet knowledge about was barely 2.4 percent and even lower decision to return is responsive to what drives the decisions of those who for refugees, at 0.3 percent. Despite economic incentives, it is remarkable are forcibly displaced is limited, not the low rate of return to the location of that over the course of the 12 monthly least because it is hard to follow peo- origin, refugees and, to a lesser extent, waves few additional households ple who are on the move or who live in IDPs expressed overwhelming interest returned. Most displacement was of precarious conditions in the proximity in returning: 93 percent of refugees short duration, and most displaced of conflict zones. and 81 percent of IDPs wanted to go people had already returned before The Listening to Displaced People back. Despite this discrepancy be- the survey. Even though the expressed Survey got around these challenges by tween stated and actual preferences, willingness to return among those who using mobile phones for interviews, the paper finds a positive correlation were still displaced was high, actual allowing it to collect information from between the two. Those who declared return rates during the 12 months of people who tend to be neglected in that they planned to return to north- the study were low for both IDPs and traditional surveys. The survey focused ern Mali were more likely to return in refugees. The sobering policy conclu- on refugees, internally displaced subsequent waves of the survey, and sion that follows from this is that people (IDPs), and returnees from the refugees showed a higher preference most people who remained forcibly crisis in northern Mali. Erupting in late for going back, once other explanatory displaced at the time of the study are 2011, this crisis was initially driven by variables were controlled for. Among likely to remain displaced in the fore- Tuareg groups seeking independence. those who did not want to go back, seeable future. As displaced people But over time, as opportunistic ban- the main reason was insecurity in the integrate more and more in their host dits, jihadist groups, and intercommu- North, followed by “life is easier here,” communities, the likelihood that they nal violence gained prominence, the lack of means, or business reasons. will return decreases with time. conflict created a generalized state of All groups lost jobs during the insecurity. displacement. Nevertheless, most re- Respondents were interviewed dur- turnees managed to find work again ing a face-to-face baseline interview once back, and in the last waves of between June and August 2014 and the survey they reached high employ- followed during 12 consecutive waves ment rates similar to those for IDPs through monthly mobile phone inter- in Bamako. Employment rates among views. This yielded not only one of the refugees, however, remained low and few descriptive data sets available on deteriorated over time. This may ex- forcibly displaced people, but also a plain the higher willingness to return unique longitudinal data set for study- among this group. ing migration decisions. Regression results confirm that This data set is used in a recent respondents who found employment paper by Hoogeveen, Rossi, and during displacement were less likely Sansone. Besides producing new de- to return. Households whose respon- Johannes G. Hoogeveen, Mariacristina Rossi, scriptive statistics on refugees, IDPs, dents worked during displacement and Dario Sansone. 2017. “Leaving, Staying, or and returnees, the paper analyzes were considerably less likely (up to 33 Coming Back? Migration Decisions during the how employment and safety affect the percentage points less likely) to return. Northern Mali Conflict.” Policy Research Working willingness to return, depending on The importance of employment in the Paper 8012, World Bank, Washington, DC. World Bank Research Digest 5 How Myanmar’s Reforms Are Playing Out in Rural Villages Changes in village governance, in affairs. The 2012 Ward and Village Tract Myanmar villagers were already expectations of the state, and in Administration Law changed this. The “networked” before the reforms, they connectivity are reshaping state- law contained provisions for indirectly became significantly more connected elected village tract administrators (a after the government began to roll out citizen relations in Myanmar formal state position representing sev- telecommunications reforms in 2013. M eral villages) and omitted mention of In 2009/10 only 4 percent of house- yanmar has undergone sig- village administrators. This led to rapid holds nationwide owned a mobile nificant social, political, and change. Village tract administrators phone; by April 2015, 55 percent did. economic reforms since 2011. steadily became more important than This provided people with access to Much of the academic and policy lit- village administrators, assuming greater alternative sources of information, and erature examining these reforms fo- control over important administrative community organizations in panel vil- cuses on how they have played out at functions, such as land registration and lages increasingly used social media the national level. But how have they applications for SIM cards. Pressure to mobilize in response to disaster, to played out in the rural villages where also grew on village tract administrators raise funds, and to advocate around 70 percent of Myanmar’s people live? to be more responsive to villagers and land issues. Meanwhile, migration How have local social institutions their needs; at the same time, villagers rates doubled in panel villages, further responded to new sources of conten- became more vocal and willing to lodge opening up people’s networks and tion accompanying these reforms? Is complaints with township officials. This introducing the potential for socio- the social contract in Myanmar—the was likely affected by news of success- economic and demographic change. terms on which citizens engage with ful advocacy efforts elsewhere in the The interplay between the changes one another as well as the basis on country. “Before, village leaders wielded in these three areas—village gover- which core rights and responsibilities power,” said one village leader. “Now nance, expectations of the state, and are negotiated with and upheld by the the villagers wield power.” connectivity—is reshaping state-citizen state—being rewritten? Meanwhile, significant changes relations in Myanmar and enabling In a recent paper Pursch, Wood- took place in how villagers interacted new, local social accountability mecha- house, Woolcock, and Zurstrassen use with and viewed the “everyday state.” nisms to take root. Yet it will also four years of data from a large-scale Before Myanmar’s reforms began, the likely lead to greater local complexity, longitudinal panel study on village state was coercive, with strict regula- as previously glossed-over fault lines life in Myanmar, the Qualitative Social tions determining how villagers used come to the surface. “Managing” this and Economic Monitoring (QSEM) re- land, grew crops, and engaged in po- transition requires emphasizing the search program, to examine how rural litical activities. The state provided few legitimacy of the change process and citizens’ interactions with the state services in return: the share of govern- the mechanisms for addressing the are changing. Their findings show that ment expenditure on social sectors contestation it generates. The research since Myanmar’s reforms began in was among the lowest in the world. identifies three areas where policy can 2011, how rural communities engage This began to change as Myanmar’s be strengthened to achieve this: en- with state institutions has been al- reforms took root. National reforms hancing the role of village institutions, tered significantly as a result of chang- emphasized the importance of produc- building a more systematic framework es in three areas: village governance, ing local results, with expenditure on around decentralization, and providing citizens’ expectations of the state, and social sectors increasing significantly. more timely information to communi- connectivity. Since 2012 national leg- In the QSEM panel villages, govern- ties. Investing in further grounded field islative reforms in village governance ment projects increased threefold, and research will be crucial in providing have resulted in a realignment of the approaches to service delivery began additional evidence to assist Myanmar centers of power and influence at the to include some local autonomy in citizens and public officials in shap- local level. Though the core functions implementation and decision making. ing their own evolving democratic of these institutions—maintaining so- Villagers also grew to expect more of institutions. cial order and acting as interlocutors the state, becoming more willing to between community members and ex- advocate for themselves when faced ternal actors—remain, there has been with perceived administrative barriers substantial change in how these func- or poor services. Samuel Pursch, Andrea Woodhouse, Michael tions are carried out and in what kind Greater connectivity—primarily Woolcock, and Matthew Zurstrassen. 2017. of influence the main actors have. through greater mobile connectiv- “Documenting Myanmar’s Social Transformation: At the beginning of the reform ity and migration but also through Insights from Six Rounds of Research on Liveli- period, village administrators were expanded road infrastructure and hoods and Social Change in Rural Communities.” widely perceived to be the most im- increased access to markets—also af- Policy Research Working Paper 8055, World portant actor for managing village fected social accountability. Although Bank, Washington, DC. 6 World Bank ResearchDigest Obstacles on the Road to Economic Growth in the West Bank How much does road connectivity annual obstacle deployment data complement the existing literature on affect growth? A new study uses from the United Nations Office for the market access, which typically focuses data on road closures in the West Coordination of Humanitarian Affairs on the economic restructuring expe- (OCHA) with estimates of the time rienced after a once-in-a-generation Bank to address that question cost of traversing each checkpoint at transport network extension in such T any given time during the sample pe- countries as China, India, and the he infrastructure connecting peo- riod (2005–12) collected through re- United States. The authors focus in- ple to markets typically evolves peated interviews with OCHA officials stead on “the intensive margin,” quan- only slowly and in a nonrandom starting in 2009. Using these data, tifying the economic losses and gains way. This makes quantifying the effect the authors calculate optimal travel caused by year-to-year enhancements of market access on economic perfor- routes, and attendant travel times, be- or deteriorations in intercity transit. mance a challenge. Studies examin- tween West Bank localities and West Despite these profound contextual ing the effect of investments in roads, Bank commercial centers for each differences, market access is shown to railroads, and waterways typically find year. The estimated travel times are positively predict changes in nighttime sizable returns to improving connec- combined with population count data light emissions, the preferred measure tivity. Research examining the impact to obtain a measure of market acces- of local economic performance given of changes in borders has reached sibility for each locality and each year the absence of spatially disaggregated similar conclusions. Virtually all such in the sample period, which coincided GDP measures for the West Bank. studies examine either large accumu- with a gradual alleviation of mobility This result is robust to controlling lated changes in transport infrastruc- restrictions. for conflict, arguably the most likely ture over long periods or changes in This approach accounts for the confounder of a relationship between accessibility brought about by large type and intensity of obstacles as well local economic growth and market ac- isolated shocks. as for their positioning. This is impor- cess, and to controlling for access to A new paper by van der Weide, tant, because one strategically placed external markets. Rijkers, Blankespoor, and Abrahams checkpoint can reduce accessibility According to the authors’ preferred tackles this long-standing question more severely than a multitude of estimates, a 10 percent improve- of the returns to market access from roadblocks where alternative connec- ment in market access increases local a different angle. It exploits unique tions are available. Simple count mea- output by 0.6 percent, assuming a short-run variation in travel times sures of obstacles, popularized by pre- lights-to-GDP elasticity of 0.3. A back- within the West Bank resulting from vious studies of the impact of mobility of-the-envelope calculation suggests the deployment and reconfiguration restrictions in the West Bank, fail to that in the absence of obstacles, GDP of road closure obstacles by Israel capture these interdependencies and per capita in the West Bank would to identify its impact on local GDP, so might yield misleading estimates of have been 4.1–6.1 percent higher each proxied by nighttime lights. The ob- the effect of changes in obstacles on year over the sample period. These stacles—taking the form of manned market access. estimates must be interpreted with physical barriers, including road- Moreover, the fact that a locality’s caution, however, because they are blocks, checkpoints, trenches, earth market access is affected not only by sensitive to the assumed relation- mounds, and a separation barrier checkpoints in that locality but also by ship between travel times and market wall—are part of a broader closure those deployed elsewhere in the West access and because uncertainty and regime instituted by the Israeli army Bank helps in establishing causality. distortions in the West Bank economy to safeguard the security of Israel and By isolating changes in market access may have dampened its output re- its settlers. Because the placement of that are driven not by local obstacles sponse to the alleviation of mobility obstacles is not (directly) driven by lo- but by obstacles deployed in a radius restrictions over the sample period. cal economic performance but by se- of 10–25 kilometers from the locality, curity considerations, and because the the authors are able to circumvent number, intensity, and configuration of concerns about local market access obstacles are subject to frequent and and local economic growth being unanticipated changes, this context driven by local factors that they might offers a unique quasi-natural experi- not be able to capture in their regres- ment by which to assess the impact of sion analysis. short-term fluctuations in market ac- The resulting data quantify the im- Roy van der Weide, Bob Rijkers, Brian Blank- cessibility on economic performance. pact of localized, temporary, and re- espoor, and Alexei Abrahams. 2018. “Obstacles on The authors build town-level versible shocks to market access in an the Road to Palestinian Economic Growth.” Policy market access measures by combin- economy no larger than the average Research Working Paper 8385, World Bank, ing road network data and U.S. county. This allows the study to Washington, DC. World Bank Research Digest 7 Can Public Works Help Postconflict Communities Escape Poverty? Governments often rely on public estimate its short-term effects on also invested in better health for their works to provide short-term household welfare. In early 2012, 276 families. The treated households in- employment during periods of communities were randomly divided creased their use of health services, es- between a treatment group (143 com- pecially for male children. On average, uncertainty. Can this policy work in munities), which received the program they undertook 12 percent more health postconflict settings? in April 2012, and a control group (133 facility visits than the control group. S communities), which did not receive And the share of boys ages 0–5 who ierra Leone was in the grips of a the program at that time. The study were taken to a health facility was 23 civil war from 1991 to 2002. Since measured program impacts by com- percent higher in the treatment house- then the government has made paring control and treatment commu- holds. In addition, treated households generating employment for youth a nities three to four months after the reported spending 16 percent more on priority. Such efforts not only can con- treatment communities received the drugs and medications than the con- tribute to economic stability; they also program. Longer-term effects could trol group. can help reintegrate ex-combatants, not be assessed because control com- Treated households also invested rebuild social trust, and build com- munities received the program after directly to secure future earnings. munity infrastructure. The need for August 2012. They increased their participation in these efforts was heightened in 2009 Poor communities eligible for par- informal savings groups by 16 percent. following the food and financial crises. ticipation were identified using esti- They also increased their investment In 2010, with the support of the World mates of poverty and food insecurity. in housing improvements by 33 per- Bank, the government of Sierra Leone The program relied on community- cent overall. And in rural areas they therefore launched a labor-intensive based targeting to identify beneficiary increased their investment in existing public works program—known as the households within these communities. businesses by 39 percent. Most cru- Cash for Work Program. The aim was This targeting helped identify the poor- cially, participating households created to provide short-term employment to est households that included at least new enterprises—mostly petty trade vulnerable youth (ages 15–35) in poor one member between ages 15 and 35 businesses—at four times the rate that communities. who was willing and able to work. The control households did. Are public works an effective anti- beneficiaries were entitled to a mini- These results demonstrate that poverty strategy in postconflict set- mum of 50 days and a maximum of 75 public works have promise as produc- tings? Theoretically, yes. Because they days of work at a wage rate of about tive safety nets in postconflict settings. directly provide stable incomes to $1.80 a day. This rate was designed They can provide immediate income struggling households, they can help to be lower than the market wage to support while also opening avenues protect—or even increase—spending discourage nonpoor applicants from for investment in the productive ca- on essentials and human capital dur- participating. pacity of poor households. Most im- ing difficult periods. But while a few Within three to four months the portantly, the results show that poor studies have shown positive effects program showed significant effects households plan for their future. They of public works on household income on households’ labor allocation and use temporary earnings to improve and expenditure, direct evidence of consumption patterns. It had sub- their quality of life not just in the pres- their antipoverty potential is scarce. stantial employment effects, beyond ent but also in the future. The debate is heightened in post- participation in public works. The conflict settings. The benefits of public share of household members who were works can be greater in these contexts not participating in the program but where youth employment and reinte- were participating in paid work was gration of ex-combatants are central 6 percentage points higher in treat- to postconflict recovery. A bonus is the ment households (28 percent) than in rebuilding of infrastructure damaged control households (22 percent). This during conflict. But public works pro- relative increase in nonprogram labor grams entail large costs and substan- force participation was stronger among tial administrative effort, both of which women and in rural areas. There was are difficult for postconflict countries. no increase in child labor. They also involve much scope for cor- Cash income among program par- ruption and mismanagement. ticipants increased by nearly three Nina Rosas and Shwetlena Sabarwal. 2016. “Can A recent paper by Rosas and times in treatment households. What You Work It? Evidence on the Productive Potential Sabarwal takes advantage of the did the households do with the addi- of Public Works from a Youth Employment Pro- phased rollout of Sierra Leone’s tional income? They increased spend- gram in Sierra Leone.” Policy Research Working Cash for Work Program to rigorously ing on food and durable goods. They Paper 7580, World Bank, Washington, DC. 8 World Bank ResearchDigest Can Mass Media Increase Women’s Political Participation after Civil War? Mass media can increase women’s a fourth weekly meeting, groups were understanding of politics and their political participation in conflict joined by staff from the implementing sense of empowerment to influence settings in the short run. But it will partner organization to listen to and political leaders); liberal norms and discuss a prerecorded program. The values; and group effects and collec- take more to reshape gender norms intervention consisted of at least 15 tive action. The study found suggestive O listening sessions over a period of five evidence in support of political ef- verall around the world, the weeks during the two months before ficacy relating to the electoral process participation rate for women in the elections. and of group effects. Little evidence national, local, and community The study then conducted a survey was found for greater political efficacy politics remains low—despite efforts on a broad range of informational, at- relating to general issues of politics to increase it. This encompasses titudinal, and behavioral outcomes to and governance or for increased liberal women’s participation in elections assess the effects of the intervention orientation. as well as in national and local on political participation as well as Overall, the study shows that expo- legislative bodies and community on gender gaps. The survey covered sure to politically relevant information decision-making committees. Even 600 individuals (two-thirds women, through mass broadcasting, even if when women do assume leadership one-third men) in 40 communities (20 brief, can boost women’s political ef- positions, their ability to influence treatment and 20 control communi- ficacy and participation in electoral decision making or voice policy prefer- ties). The small sample size was one processes. This result could help in- ences is often limited. The situation limitation in estimating impacts (that form policy makers, researchers, and is worse in war-torn settings, where is, for the results that are positive and development practitioners seeking women are disproportionately affected significant, only just under half the to enhance women’s participation by the consequences of conflict. estimates are statistically significant at in election activities through meth- One hypothesis has been that conventional levels of 1 percent and 5 ods that are cost-effective, relatively poor access to information may keep percent). simple, and easily adaptable in conflict women from participating in elec- Political participation was mea- settings. tion activities. To counter this pos- sured across five broad dimensions: But the lack of impact on broader sibility, the United Nations Mission voting (voting in the 2011 elections); political efficacy outcomes highlights in Liberia (UNMIL), in the run-up to campaigning (contributing time or the inability of these short-term inter- the 2011 general elections in that money to a political campaign, attend- ventions to shape the “soft elements,” country, launched a radio program ing a political rally); contacting (having such as political beliefs and attitudes that broadcast information about the contact with one’s representative in as well as social and gender norms. elections and about the importance of the national assembly or senate or Complementary targeted interventions participation. An experimental study with a government official or agency); are needed that can address socioeco- by Mvukiyehe assessed the effects of communing (attending community nomic constraints—such as unequal exposure to UNMIL’s radio program- meetings, speaking at such meetings, access to power and resources—that ming on women’s political behaviors contacting community leaders); and perpetuate gender gaps in social and and attitudes in the context of the protesting (complaining to radio sta- political participation. Finally, more elections. tions, participating in riots or violent research is needed on the channels of The experiment used a multistage protest). influence for these political participa- randomization process (for selecting Results suggest that access to tion outcomes as well as the extent to communities, households, and individ- UNMIL Radio had a significant positive which the design of interventions, such uals). It provided women in randomly effect on women’s political participa- as the scale and intensity of informa- selected communities with direct ac- tion at both the national and local tion campaigns, can maximize effects cess to a radio and organized them to level. Among women in the listening on political participation. participate in a series of weekly group sessions, electoral turnout increased sessions listening to UNMIL Radio’s by 9 percentage points, attendance at special electoral programs. The group community meetings by 5 percentage design was expected to offer women an points, and the probability of speaking opportunity for discussion and provide out at such gatherings by 10 percent- a space for communal forms of po- age points. litical participation. Listening groups The study also explored the possi- Eric Mvukiyehe. 2017. “Can Media Interventions were set up randomly in treatment ble mechanisms through which UNMIL Reduce Gender Gaps in Political Participation after communities. The groups met alone Radio may have influenced women’s Civil War? Evidence from a Field Experiment in three times a week to listen to a spe- political participation. The channels Rural Liberia.” Policy Research Working Paper cific electoral program in real time. For included political efficacy (women’s 7942, World Bank, Washington, DC. World Bank Research Digest 9 Conflict and Poverty in Afghanistan’s Transition Afghanistan has had strong growth A paper by Floreani, López-Acevedo, living standards in the country. The despite armed conflict. But declining and Rama is among the few studies to simulation indicates that a withdrawal aid and the withdrawal of foreign analyze the relationship between armed of international troops and a decline conflict, aid, troops, and economic well- in foreign aid flows would lead to an troops put that at risk being in Afghanistan—a relationship increase in armed conflict and poverty B that explains the puzzle of lower poverty —results that match current trends etween 2004 and 2013 rates in areas more severely affected in Afghanistan. While these forecasts Afghanistan experienced record by conflict. The paper also analyzes the should be interpreted with extreme economic growth, averaging impact of the reduction in international caution, the paper underscores the risk about 9 percent a year. Yet poverty re- troops on household welfare. that Afghanistan’s fragility challenges mained stubbornly high. Surprisingly, The paper uses a two-pronged ap- may worsen going forward. This would however, poverty rates were persis- proach. First, it analyzes and models bring a severe deterioration in living tently lower in the conflict area of the relationship between armed con- standards. Safeguarding the recent the South and Southwest (below 30 flict, troop deployment (both domestic achievements in human development percent in 2011) and higher in the and foreign), foreign aid, and house- therefore requires mitigating the nega- more peaceful North, Northeast, and hold expenditure. Second, it uses this tive welfare impact of declining foreign Central regions (between 40 and 50 model to simulate the consequences aid and international troop withdrawal. percent). of any reductions in foreign aid and Research suggests that conflict re- international troops. duces economic well-being, especially The analyses rely on a unique for poor households, but Afghanistan provincial-level data set constructed seems to defy this trend. One expla- for the paper. The data include infor- nation is the massive development mation on armed conflict, troops, aid, and military aid that has been chan- and household characteristics over the neled to strategic and conflict-ridden period 2007–14. These data are used locations. This targeting reflects the to model armed conflict at the provin- “winning hearts and minds” doctrine, cial level, depending on current and according to which development aid lagged presence of troops. The data can be instrumental in securing the also help to model the determinants of most bellicose areas of Afghanistan. consumption per capita and assess the In addition, military deployments can contribution from armed conflict, in- benefit households both directly—by ternational aid, and troop deployment. curbing the negative effects of armed The results show that the over- conflict on their welfare—and indirect- all impact of conflict on household ly—presumably because the demand expenditure is positive not only at for goods and services to support the the national level but also in each troops boosts local labor earnings. of the provinces. The impact is par- In 2014 a dramatic transition ticularly strong among the provinces started in Afghanistan, involving a in the East, South, and Southwest. reduction in the number of interna- Household expenditure is 10–18 per- tional troops and their replacement by cent higher in these provinces than in Afghan troops as well as a reduction in the province with the lowest level of foreign aid. Earlier research had sug- conflict intensity, despite the preva- gested that other sources of economic lence of armed conflict. Conversely, the growth—such as commercial agri- impact is weakest in the more peaceful culture and mining corridors—would North and the Northeast provinces. mitigate transition shocks. But in These findings help explain the puzzle general these hopeful scenarios over- of higher poverty prevalence in the looked the interrelation between con- Afghan provinces and regions less af- flict, development aid, and economic fected by armed conflict. well-being. The security situation in These results are then used to Afghanistan was already deteriorating perform a simulation exercise. This Vincent A. Floreani, Gladys López-Acevedo, and in 2012, and the intensity of conflict exercise suggests, in line with the lat- Martín Rama. 2016. “Conflict and Poverty in Af- has markedly increased since that est available evidence, that conflict ghanistan’s Transition.” Policy Research Working time. is taking a high toll on welfare and Paper 7864, World Bank, Washington, DC. 10World Bank ResearchDigest Who Benefited from Burundi’s Demobilization Program? Transfers to ex-combatants after financial, social, or behavioral con- generated a local economic boom the Burundi civil war led to welfare straints that are relaxed following the in villages hosting large numbers of gains for beneficiaries and their transfers. If beneficiaries do not face ex-combatants. Results therefore sug- such constraints, they should already gest that the 2005 economic boom villages—but only in the short run produce at their efficient scale before vanished when demobilized combat- D receiving the transfers, suggesting that ants had spent all their money. In line isarmament, demobilization, DDR allowances would be consumed. with this interpretation, the evolution and reintegration (DDR) pro- Greater demand should generate a of prices between 2006 and 2010 was grams are an essential part of temporary economic boom at the local negatively affected by the presence of peace-building interventions. These level. But if beneficiaries do face finan- CNDD-FDD ex-combatants. programs provide a series of services cial, social, or behavioral constraints In the short run the positive direct and financial aid to former combat- that are relieved by the transfers, DDR effect of the DDR program shows that ants, allowing them to reintegrate into allowances would be invested in pro- the program indeed provided a safety civilian society. Financed by an array ductive assets. In this case the impact net to ex-combatants. The positive of sources, including multilateral and on the consumption and assets of ben- spillovers further show that the pro- donor funding, such programs have eficiary households should be positive gram not only benefited beneficiaries; been implemented in more than 30 and long lasting, though the increase it also benefited communities where countries in the past two decades. in consumption should be rather ex-combatants returned. This second More than 2.5 million people have small. Spillovers on nonbeneficiaries effect was likely to ease the return of benefited from some form of demo- at the community level would be ex- ex-combatants to their villages, and bilization worldwide since 1989, at an pected to be marginal. this supports the reinsertion program average annual cost of more than $600 Whether the consumption or in- implemented in postconflict countries million. vestment scenario prevails is an em- as a short-run strategy to reduce the In Burundi a DDR program demo- pirical question, one that the authors risk of relapse into conflict. bilized and reintegrated 30,000 former seek to assess. Their results show that But results also suggest that the combatants between 2005 and 2010. demobilization grants had a positive DDR program was not sufficient for The peace process involved two waves and significant impact on economic creating a virtuous circle toward eco- of demobilization targeting two ma- outcomes for FNL beneficiaries and nomic development, because the jor rebel groups, with the program indicate positive spillovers for non- positive impact of cash transfers was benefiting the CNDD-FDD from 2005 beneficiary households in the short short-lived. The return to a precarious and the FNL from 2010. Burundi com- run. These results suggest that the economic situation may encourage batants were granted two allowances consumption scenario prevailed in ex-rebels to again take up arms and worth about $550 each, one in cash Burundi. When demobilized FNL com- engage in violence. Indeed, recent em- and one in kind (such as vocational batants returned home, they spent pirical evidence suggests that demo- training or start-up material for a a large part of their allowances on bilized combatants played a key role small business). Each allowance was immediate needs, such as food and in the violence that affected the 2010 approximately equivalent to the an- other basics. The demand for locally and 2015 elections in Burundi. These nual earnings of a farm household or produced goods rose, which generated conclusions should be taken into 18 months’ salary for a soldier in the a short-run economic boom in villages consideration by policy makers when army—thus a considerable amount. hosting numerous ex-combatants. implementing DDR processes in other What were the welfare effects of Consistent with these observations, countries. this cash transfer program? Combining the authors find that 2010 prices were panel data of households collected higher in such villages. in 2006 and 2010 with official records For demobilized CNDD-FDD com- from the National Commission for batants, benefiting from the 2005 Demobilization, Reinsertion and wave of the DDR program, the analy- Reintegration, a paper by D’Aoust, sis identifies a downward trend in Sterck, and Verwimp evaluates the consumption and nonfood spending program’s impact on the welfare of in the period 2006–10. Similarly, the Olivia D’Aoust, Olivier Sterck, and Philip Ver- the households hosting a demobilized density of demobilized combatants wimp. 2016. “Who Benefited from Burundi’s combatant and assesses spillovers on who benefited from this first wave of Demobilization Program?” World Bank Eco- nonbeneficiary households. the program is negatively correlated nomic Review, published electronically July 5. According to economic theory, the with outcomes of interest. It is rea- https://doi.org/10.1093/wber/lhw033. (Also issued impact of cash and in-kind transfers sonable to assume that the first wave as Policy Research Working Paper 7732, World depends on whether beneficiaries face of the program, like the second one, Bank, Washington, DC, 2016.) World Bank Research Digest11 U.S. Food Aid and Civil Conflict Revisited: A Methodological Caution A recent paper calls into question correlated with the total quantity vary much year to year. This poses a causal methodology used in many of food aid shipped abroad by that two problems. First, since production studies, including one that links U.S. country, but are not otherwise likely shocks are likely correlated in adjacent to be directly associated with conflict. years, levels of production in a given food aid and conflict Because U.S. wheat production does year cannot be treated as random B not vary across recipient countries, relative to years just before or after. ased on anecdotal evidence, they construct a shift-share variable In addition, factors such as climate research has long hypothesized based on the proposition that addi- or commodity prices might jointly af- a causal relationship between tional food aid flows disproportionate- fect wheat yields and global conflict. food aid and conflict—though without ly to frequent, regular recipients. The Controlling for these joint influences agreement on the expected direction interaction of U.S. wheat production through common year effects across of the effect. If aid reduces poverty and with regularity of aid receipt becomes countries that do or do not regularly creates stability during emergencies, it analogous to a continuous difference- receive aid is not valid. If countries may reduce strife. Alternatively, if aid in-differences treatment. Higher U.S. that do not receive aid are not prone to resources are easily stolen or manipu- wheat production leads to more aid conflict, this strategy effectively means lated, food shipments might prolong (the treatment) that is disproportion- using countries where conflict rarely conflict. Establishing the true relation- ately sent to regular recipients (the occurs to model the determinants of ship is notoriously difficult, primarily treated group) rather than the least conflict. because of the problem of reverse regular recipients (the control group). To test for spurious correlation, causality. If aid shipments are dispro- Using data on direct shipments of CB propose a new type of placebo portionately directed to places more wheat food aid in 1971–2006, NQ show test based on randomly assigning the prone to civil unrest, a positive corre- that ordinary least squares estimates identity of the aid recipients within lation observed between the quantity indicate a negative, statistically insig- each year. They find that the positive of food aid received and the likelihood nificant relationship between aid and relationship between aid and conflict of conflict would not necessarily imply conflict, while the IV estimates show becomes much stronger when aid re- that food aid causes conflict. that U.S. wheat production and aid cipients are randomly assigned annual A recent paper by Christian and shipments are positively and statisti- food aid volumes than it is in the true Barrett examines the methodologi- cally significantly associated with con- data. This finding calls into question cal solution proposed by a celebrated flict only among regular recipients of the claimed causal relationship. study reporting a relationship be- aid. They argue that the most plausible Finally, CB use Monte Carlo analy- tween aid shipments of wheat from explanation is that food aid prolongs sis to show that if both production and the United States and civil conflict conflict. conflict are strongly trending variables in recipient countries. In a context of Christian and Barrett (CB) show that correlate over the study period, chronic global food emergencies, the that this conclusion is based on flaws the shift-share strategy can identify a prospect that food aid might have both in the understanding of how U.S. positive relationship between aid and serious unintended consequences food aid policy changed over the study conflict even if the true causal rela- naturally excites much policy interest. period and in the statistical reasoning tionship is one in which aid prevents But the study, by Nathan Nunn and that seems to have gone previously conflict. Nancy Qian (“US Food Aid and Civil unnoticed in the literature using simi- These findings invalidate the con- Conflict,” American Economic Review 104, lar methods. clusion that food aid causes prolonged no. 6 [2014]: 1630–66), is vulnerable On the policy side, CB point out conflict, while sounding a note of cau- to an underappreciated source of bias that the U.S. wheat price support tion for a popular empirical strategy that more broadly affects a popular policy ended by the mid-1990s. If this and offering a new tool for diagnosing econometric strategy for establishing policy was driving aid allocations, issues with this panel data IV estima- causal relationships in nonexperimen- as NQ hypothesize, the relationship tion strategy. tal panel data. between production and aid should Nunn and Qian (NQ) try to obviate attenuate in the last decade of the NQ the reverse causality problem by using data. Surprisingly, the opposite seems a type of instrumental variables (IV) to be true, meaning that price support estimation method sometimes called policy does not provide a basis for the a “shift-share instrument,” popular natural experiment proposed by NQ. in the applied economics literature. CB next turn to a diagnosis of the Paul Christian and Christopher B. Barrett. 2017. Their IV strategy is based on the idea statistical reasoning underlying the “Revisiting the Effect of Food Aid on Conflict: A that wheat production shocks in the shift-share methods used by NQ. Methodological Caution.” Policy Research Working United States are strongly positively Annual wheat production does not Paper 8171, World Bank, Washington, DC. 12World Bank ResearchDigest Recent Policy Research (continued from page 2) to 42 additional Daesh recruits. This Working Papers on FCV is a strong effect that is found to be Countries and Topics restricted to countries at a relatively 7757 Pirates of Somalia: Crime and Deterrence on What Drives the short distance from Syria. In countries the High Seas Quy-Toan Do, Lin Ma, and Claudia Ruiz Radicalization of Foreign farther away, unemployment may still lead to radicalization, with radicalized 7778 Dictators Walking the Mogadishu Line: How Men Become Monsters and Monsters Terrorist Recruits? individuals engaging in local acts of Become Men Shaun Larcom, Mare Sarr, and Tim Willems violence rather than traveling to far- 7797 Combining Mentoring Programs with Cash Transfers for Adolescent Girls in Liberia: away Syria or Iraq. Baseline Report have a tertiary education, suicide fight- Thus the main implication of the Kelly Hallman, Elizabeth Kelvin, Berk Özler, Juliette Seban, Erica Kuhlik, Cooper Alton, ers are relatively more likely to have a study’s findings is that domestic so- Joseph Kamara, and Sarah Goodman secondary education, and fighters are cioeconomic policies have an impact 7826 Minimizing Ecological Damage from Road Improvement in Tropical Forests relatively more likely to have a primary on global security because disgruntled Susmita Dasgupta and David Wheeler education only. In addition, religious youth are susceptible to radicaliza- 7970 Recovery from Conflict: Lessons of Success Hannes Mueller, Lavinia Piemontese, and knowledge is low overall among the tion into violent extremism at home Augustin Tapsoba recruits and associated with higher or abroad. Cross-border terrorism has 7984 Agent Banking in a Highly Under-Developed levels of education. turned security into a global public Financial Sector: Evidence from the Democratic Republic of Congo Finally, exploiting individual-level good, the provision of which requires Robert Cull, Xavier Gine, Sven Harten, and education information for these fight- policies coordinated at a multilateral Anca Bogdana Rusu 8032 Can Public Works Programs Reduce Youth ers, the authors link the unemploy- level. Crime? Evidence from Papua New Guinea’s ment rate among workers of a given Urban Youth Employment Project Oleksiy Ivaschenko, Darian Naidoo, David education level and in a given country Newhouse, and Sonya Sultan with the size of the cohort of Daesh 8036 They Got Mad Skills: The Effects of Training on Youth Employability and Resilience to the recruits from that same education level Ebola Shock and country. The analysis suggests that Nina Rosas, Maria Cecilia Acevedo, and Mohamed Abdel Jelil, Kartika Bhatia, Anne Samantha Zaldivar higher unemployment rates are a push 8075 Predicting Conflict Brockmeyer, Quy-Toan Do, and Clément Joubert. factor toward radicalization, especially 2018. “Unemployment and Violent Extremism: Bledi Celiku and Aart Kraay in countries at a shorter distance from 8231 How Much Oil Is the Islamic State Group Evidence from Daesh Foreign Recruits.” Policy Producing? Evidence from Remote Sensing Syria; an increase of one percentage Research Working Paper 8381, World Bank, Quy-Toan Do, Jacob N. Shapiro, Christopher D. Elvidge, Mohamed Abdel-Jelil, Daniel P. point in the unemployment rate leads Washington, DC. Ahn, and Kimberly Baugh 8301 Insecurity and Industrial Organization: Evidence from Afghanistan Joshua Blumenstock, Tarek Ghani, Sylvan Herskowitz, Ethan B. Kapstein, Thomas Scherer, and Ott Toomet 8345 Water When It Counts: Reducing Scarcity through Irrigation Monitoring in Central Mozambique Paul Christian, Florence Kondylis, Valerie Mueller, Astrid Zwager, and Tobias Siegfried Papers can be downloaded at www.worldbank.org/research. The World Bank Research Digest is a quarterly publica- The Research Digest is financed by the Bank’s Re- Editorial Committee: Shanta Devarajan, David tion disseminating findings of World Bank research. search Committee and managed by DECSO, the Strat- Rosenblatt, Aslı Demirgüç-Kunt, Roberta Gatti, The views and interpretations in the articles are those egy and Operations unit of the Development Economics Hiau Looi Kee, Chorching Goh, and Shiva S. Makki of the authors and do not necessarily represent the Senior Vice Presidency (DEC). The Research Digest is (managing editor). Editor: Alison Strong; production: views of the World Bank, its Executive Directors, or the not copyrighted and may be reproduced with appropri- Roula Yazigi. For information or free subscriptions, countries they represent. ate source attribution. send email to research@worldbank.org or visit http://www.worldbank.org/researchdigest. The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper