Shocks and Household Welfare in Sudan October 2019 Poverty and Equity Global Practice, Africa Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. . Copyright Statement: The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750- 8400, fax 978-750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. This paper was written by Alvin Etang Ndip (Senior Economist, GPV01) and Sering Touray (Consultant, GPV01), as part of the Sudan Programmatic Poverty Assessment (P164694). Overall guidance was provided by Pierella Paci (Practice Manager, GPV01). The authors would like to thank Aly Sanoh (Senior Economist/Statistician, GPV07) and Oleksiy Ivaschenko (Senior Economist, GSPGL) for very useful peer reviewer comments. The paper also benefited from comments from Eiman Adil Mohamed Osman (Consultant, GPV01). Vice President Hafez Ghanem Country Director Carolyn Turk Senior Director Carolina Sanchez-Paramo Practice Manager Pierella Paci Task Team Leader Alvin Etang Ndip Table of Contents List of Figures ................................................................................................................................................. i List of Tables .................................................................................................................................................. i List of Boxes ................................................................................................................................................... i Abbreviations ................................................................................................................................................ ii Executive Summary...................................................................................................................................... iii 1. Introduction .......................................................................................................................................... 1 2. Poverty, Shocks, and Household Welfare ............................................................................................. 9 2.1. Shocks in Sudan................................................................................................................................ 10 2.1.1 Heterogeneity in the Prevalence of Shocks across Space and Livelihoods................................ 14 2.1.2 Prevalence of Specific Types of Shocks ...................................................................................... 19 2.2 Vulnerability of Households to Idiosyncratic Shocks ........................................................................ 23 2.3 Covariate Shocks ............................................................................................................................... 24 2.3.1 Rainfall Shocks ........................................................................................................................... 25 2.3.2 Conflict ....................................................................................................................................... 25 3. Impact of Shocks on Household Welfare ............................................................................................ 27 3.1. Methodology.................................................................................................................................... 27 3.1.1 Data ............................................................................................................................................ 27 3.1.2 Estimation Strategy .................................................................................................................... 27 3.2 Results ............................................................................................................................................... 28 3.2.1 Impact of Shocks on Household Welfare ................................................................................... 28 3.2.2 Heterogeneity of the Impact of Shocks on Household Welfare ................................................ 32 4. Coping with and Resilience to Shocks ................................................................................................. 39 4.1 Household Coping Strategies ............................................................................................................ 39 4.2 Resilience of Sudanese Households to Shocks ............................................................................... 433 5. Conclusion and Policy Implications ................................................................................................... 466 References .................................................................................................................................................. 48 List of Figures Figure 1: Real GDP Growth, Rainfall, and Inflation in Sudan, 2000–2018 .................................................... 2 Figure 2: Shocks in Sudan: Specific Types of Shocks Reported, Effects, and Coping Strategies, 2009 and 2014/15 ....................................................................................................................................................... 13 Figure 3: Prevalence of Shocks across Households in Sudan, 2009 and 2014/15 ...................................... 16 Figure 4: Prevalence of Shocks across States in Sudan, 2009 and 2014/15 ............................................... 17 Figure 5: Household Livelihoods and Prevalence of Shocks in Sudan, 2009 and 2014/15 ......................... 18 Figure 6: Heterogeneity in the Prevalence of Shocks in Sudan: Household Livelihoods ............................ 20 Figure 7: Heterogeneity in the Prevalence of Shocks in Sudan: Consumption Per Capita Quintiles .......... 21 Figure 8: Heterogeneity in the Prevalence of Shocks across States ........................................................... 22 Figure 9: Covariate Shocks: Price Shocks and Crime................................................................................... 26 Figure 11: Shocks in Sudan: Risk-Coping Strategies, 2009 and 2014/15 .................................................... 41 Figure 12: Help Received ............................................................................................................................ 43 Figure 13: Household CSI ............................................................................................................................ 45 List of Tables Table 1: Test of Means between Households Affected by Shocks and Those Not Affected, 2009 ............ 23 Table 2: Test of Means between Households Affected by Shocks and Those Not Affected, 2014/15 ...... 24 Table 3: Estimating the Impact of Shocks on Welfare of Sudanese Households: Regression Results ....... 28 Table 4: Estimating the Heterogeneity of the Impact of Shocks on Household Consumption Per Capita . 33 List of Boxes Box 1: Sudan’s Safety Nets and Social Protection Programs ........................................................................ 5 i Abbreviations ACLED Armed Conflict Location and Event Database CHRS Center for Hydrometeorology and Remote Sensing CSI Coping Strategy Index DRM Disaster Risk Management GDP Gross Domestic Product GoS Government of Sudan IMF International Monetary Fund NBHS National Baseline Household Survey NGO Nongovernmental Organization NHBPS National Household Budget and Poverty Survey PPP Purchasing Power Parity SIP Social Initiatives Program SSN Social Safety Net ii Executive Summary The Sudanese economy has faced several shocks over the years —sometimes resulting in devastating impacts on the economy and the welfare of Sudanese households. The sources of these shocks vary— ranging from weather-related shocks such as droughts and floods to the global financial crisis and commodity price hikes. In the absence of effective social protection programs, exposure of households to frequent shocks lowers their ability to escape poverty—pushing households slightly above the poverty line back into poverty and sliding poor households deeper into poverty. Poor households are often particularly vulnerable to shocks. Household vulnerability can be thought of as a function of (a) the level of exposure to negative shocks, (b) the extent of impacts on the household, and (c) the type of coping mechanisms available to the household. Poor households (particularly those whose main livelihood is rain-fed agriculture) face significant risk of being exposed to various idiosyncratic and covariate shocks such as rain variation, pest and crop diseases, and low and volatile prices of agricultural output, among others. Where these shocks result in poor harvest, loss of livestock and other assets, damage to dwellings, and so on, households’ ability to sufficiently mitigate the immediate and long-term impact of the shocks on their welfare becomes weakened. The extent of the impacts of shocks on household welfare depends on the nature and severity of the shocks as well as households’ capacity to manage its risk of exposure to shocks ex ante and/or mitigate the impact of shocks ex post. In most developing countries, this relationship between shocks, its impact on household welfare, and households’ risk coping and management strategies is further complicated by the fact that well-functioning credit, insurance, and asset markets are often missing. Informal coping strategies often developed by households to internalize missing/imperfect credit and insurance markets, such as informal risk-sharing arrangements between households within a community, informal loans and systems of support between family members and friends, and so on, are often insufficient to completely mitigate the impact of shocks, particularly in the case of covariate shocks. Under these conditions, social protection programs can complement such informal arrangements and minimize households’ use of severe risk coping/management strategies such as selling productive assets, producing low-risk and low- return crops, minimizing expenses by reducing consumption, removing children from school, and so on. While the use of these strategies may (at least in the short term) mitigate the impact of shocks, the long- term effects on households’ resilience to future shocks and the threat of poverty traps may be large. This paper applies this framework to examine the impact of shocks on the welfare of Sudanese households and explore coping strategies typically utilized by households to mitigate the negative effects of shocks. The paper uses the 2009 National Baseline Household Survey (NBHS) and the 2014/15 National Household Budget and Poverty Survey (NHBPS) to document the main types of shocks that Sudanese households are exposed to and describe the profile of Sudanese households likely to be vulnerable and/or resilient to shocks. To complement this analysis, the paper uses the most recent round of the data collected in 2014/15 (containing information on idiosyncratic shocks) together with data on covariate shocks such as rainfall and conflict obtained from other sources to estimate the impact of shocks on household welfare. Since the impact of shocks on household welfare is likely to be multidimensional, various indicators of household welfare such as consumption, poverty status, assets, dietary quality, and diversity are considered in the paper. Results from the analysis are used to highlight the state of social protection in Sudan and discuss the need for an expansion of the existing system. iii The prevalence of shocks in Sudan is most common among poor, agricultural, and rural households. Most households affected by shocks in Sudan are poor, engaged in agriculture (particularly crop farming and animal husbandry), and live in rural areas. The latest round of data showed that despite a decrease in the overall prevalence of shocks in Sudan (from 65 percent of households in 2009 to 45 percent in 2014/15), these households continue to face significant risk of exposure to shocks—nearly 50 percent of poor and rural households, 63 percent of crop farmers, and 50 percent of households engaged in animal husbandry were affected by shocks in 2014/15. Furthermore, the shocks experienced by these households are mostly agriculture related—livestock loss, crop diseases, and floods/drought. In rural Sudan where poverty is high and agriculture is the main source of livelihood, exposure of households to shocks that deplete their assets and livestock, lower their income, and destroy their dwellings is likely to significantly affect household welfare. Existing social protection programs are mostly insufficient and/or inaccessible— only 24 percent of households affected by shocks in 2014/15 received help from family/friends, nongovernmental organizations (NGOs), or the government. As a result, most households rely on their own coping strategies—about 23 percent of households affected by shocks in 2014/15 relied on selling their assets, dissaving, or renting out their farm to mitigate the impact of the shocks. Floods/droughts have the largest negative effect on the welfare of Sudanese households. Consumption per capita of households affected by weather shocks such as flood/drought decreases by 5.3 percent on average. Further analysis using households’ poverty status shows that flood/drought shocks increase households’ likelihood of being poor—floods/droughts increase the probability of affected households being poor by 57 percent relative to those not affected. Similar negative effects of weather-related shocks are also observed on household assets, dietary diversity, and dietary quality. The reduction in households’ assets and increased likelihood of being poor for households affected by floods/droughts show the vulnerability of these households. The loss of assets, in particular, could make it harder for these households to escape poverty, thereby trapping them in a cycle of intergenerational poverty. The impact of shocks is mitigated using coping strategies. However, these gains occur at the expense of the depletion of household assets and savings, thereby weakening households’ resilience to shocks. The large negative effects of shocks on the welfare of Sudanese households (particularly those with low capacity to cope with shocks) highlight significant limitations in households’ ability to fully mitigate the impact of shocks. Strengthening the resilience of these households requires a combination of social protection programs which target vulnerable households; and economic stability and growth which create economic opportunities such as jobs and social services like education for Sudanese households. However, existing social protection programs (mainly Zakat Centers and NGO transfers to poor Sudanese households) appear to be insufficient. Furthermore, the current macroeconomic environment characterized by loss of oil revenue imposes budget constraints on government’s efforts to build resilience through social protection reforms and expand economic opportunities. Nonetheless, ongoing reforms to restore macroeconomic stability and economic growth are important first steps to strengthening resilience. In addition, integrating existing social protection programs with Disasters Risk Management (DRM) strategies is crucial to ensuring that such programs are more shock-responsive particularly for agricultural households who face significant exposure to weather related shocks which have large negative effects on their welfare. Ongoing reforms must also include the agricultural sector (particularly the promotion of climate-smart and pest resilient agricultural practices) to reduce the vulnerability of agricultural households. Agricultural institutions such as extension services complemented by existing informal and semi-formal institutions can ensure the diffusion of such practices to increase resilience and iv lower vulnerability. Given the current macroeconomic environment, the Government of Sudan (GoS) requires both financial and technical multi-sector support to implement these reforms. v 1. Introduction Major shocks have repeatedly hit the Sudanese economy, sometimes with catastrophic results. The sources of these shocks have been varied, ranging from weather-related shocks such as droughts and floods to the global financial crisis and commodity price hikes. At the macroeconomic level, the prevalence of these shocks lowers economic growth and prolongs economic recovery. As shown in Negative shocks can aggravate poverty among Sudanese households. It is estimated that 36.1 percent of Sudanese were poor in 2014/15 (national poverty line), which is 13.5 percent based on the US$1.90 (in 2011 purchasing power parities [PPPs]) per person per day, extreme poverty, and a standard typical of low-income countries and 46.1 percent based on the US$3.20 (in 2011 PPPs) per person per day, moderate poverty, and a standard typical for lower-middle-income countries (World Bank, 2019). The majority of poor Sudanese households live in rural areas and 35.5 percent of them are poor. For most of these households, agriculture is their main source of livelihood, leaving them vulnerable to weather variations such as droughts and conflict that disrupts agricultural activity. The prevalence of negative shocks often leaves vulnerable households falling into or sliding deeper into poverty. Figure 2- panel a) shows significant decrease in real consumption per capita in Sudan in the years in which households were affected by floods/drought, civil conflict and/or price hikes. Furthermore, these effects tend to be heterogenous across the consumption distribution and across space. Panel b) shows that between 2009 and 2014, only the bottom-20 percent (particularly in rural Sudan) enjoyed a small increase in consumption per capita. The bottow-40 percent and top-60 percent experienced a decline of 0.1 and 1.7 percent in real consumption per capita over the same period- with a larger decline among urban households. Minimizing the effect of shocks on the welfare of Sudanese households requires building resilience and lowering vulnerability through effective social protection programs in particular and a stable and conducive macroeconomic environment in general. Page 1 of 49 Figure 1, prevalence of natural disasters such as floods or droughts, and other shocks such as civil conflict and price hikes appear to correlate with low growth rates of real gross domestic product (GDP). The erratic rains of 2004, the combined effects of drought and civil insecurity in 2009, and late and sporadic rains in 2011 contributed to the decline in GDP growth rates between 2010- 2011 before reaching all-time low levels in 2012 after the secession of the South in 2011. Negative shocks can aggravate poverty among Sudanese households. It is estimated that 36.1 percent of Sudanese were poor in 2014/15 (national poverty line), which is 13.5 percent based on the US$1.90 (in 2011 purchasing power parities [PPPs]) per person per day, extreme poverty, and a standard typical of low-income countries and 46.1 percent based on the US$3.20 (in 2011 PPPs) per person per day, moderate poverty, and a standard typical for lower-middle-income countries (World Bank, 2019). The majority of poor Sudanese households live in rural areas and 35.5 percent of them are poor. For most of these households, agriculture is their main source of livelihood, leaving them vulnerable to weather variations such as droughts and conflict that disrupts agricultural activity. The prevalence of negative shocks often leaves vulnerable households falling into or sliding deeper into poverty. Figure 2- panel a) shows significant decrease in real consumption per capita in Sudan in the years in which households were affected by floods/drought, civil conflict and/or price hikes. Furthermore, these effects tend to be heterogenous across the consumption distribution and across space. Panel b) shows that between 2009 and 2014, only the bottom-20 percent (particularly in rural Sudan) enjoyed a small increase in consumption per capita. The bottow-40 percent and top-60 percent experienced a decline of 0.1 and 1.7 percent in real consumption per capita over the same period- with a larger decline among urban households. Minimizing the effect of shocks on the welfare of Sudanese households requires building resilience and lowering vulnerability through effective social protection programs in particular and a stable and conducive macroeconomic environment in general. Page 2 of 49 Figure 1: Real GDP Growth, Rainfall, and Inflation in Sudan, 2000–2018 350 70 Floods % OF REAL GDP GROWTH & INFLATION 60 300 High inflation 50 MILLIMETERS OF RAINFALL rates 250 40 Drought Erratic Rains Late rains 30 200 Drought High inflation rates 20 150 10 100 Conflict in 0 Civil the South insecurity -10 50 Secession of -20 0 South Sudan -30 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 YEARS Annual Rainfall (in mm) Real GDP growth-Annual percent change (alternate axis) Inflation, average consumer prices (alternate axis) Source: Authors’ calculation using data from the World Bank and International Monetary Fund (IMF). Figure 2: Real private consumption, 2008-2018 and Consumption Growth incidence by percentile and locality, May 2009 to November 2014. a) Per capita private consumption b) Consumption per capita growth growth 10% Annual growth in real per capita 4% 6% 2% consumption 0% 2% Growth rate in % 2009 2008 2010 2011 2012 2013 2014 2015 2016 2017 2018 Sudan -2% -2% Rural -4% Urban -6% -6% -10% -8% 0 50 100 Percentile of real per capita consumption -10% Source: World Bank staff calculations. Note: growth rates for 2017 and 2018 are estimates and forecasts, respectively. Note: 95- percent confidence intervals for consumption growth in panel b) indicated through dashed lines and based on 200 bootstrap replications. Page 3 of 49 Sudan’s current macroeconomic environment of low growth, high inflation and a constrained fiscal space; and Social Safety Net (SSN) programs which exist but are limited in coverage; have joint and individual implications on the resilience and vulnerability of households to shocks. The economy which continues be characterized by adverse events (such as rising inflation and currency devaluation) triggered by macroeconomic reforms following the secession of the South increases the vulnerability of households to poverty and lowers their resilience to shocks (World Bank 2018). Furthermore, the loss of oil revenue after the secession imposes budget constraints on government’s activities including financing existing social protection programs (which receive an estimated 1 percent of Sudan’s GDP) to increase access and coverage. Sudan has a comparatively well-established social protection system. There are fourteen (14) SSN programs compared to an average of fifteen (15) in the continent and twenty (20) in lower middle- income countries. These programs range from unconditional and conditional cash transfers, in kind programs, school feeding, health insurance among others; and benefit an estimated 88 percent of the Sudanese population (World Bank, Atlas of Social Protection Indicators of Resilience and Equity - ASPIRE).1 The presence of these programs is a necessary but not sufficient condition to build the resilience and lower the vulnerability of households to shocks. Furthermore, the main focus of these programs is often to reduce the extent poverty among the chronically poor and hence not fully integrated into Sudan’s Disaster Risk Management (DRM) strategy. Thus, there is the need to ensure that these programs efficiently target and effectively uplift vulnerable groups of the population out of poverty- particularly poor households who are vulnerable to various shocks and lack the capacity to mitigate their impact when they occur. For these households, access to social protection can minimize long-term effects of using extreme coping strategies. However, data shows that majority of Sudanese households (particularly the poorest) rely on costly coping strategies such as using savings and/or selling assets (World Bank 2014, 24 and 113). Furthermore, a recent report on the state of social protection in Sudan documents the uneven coverage of existing programs and provides recommendations to expand coverage, improve targeting, and evaluate their effectiveness (Kjellgren et al. 2014). Thus, reforming SSN programs to increase access and coverage as well as restoration of macroeconomic stability is needed to strengthen the resilience and reduce poverty among these households. 1 https://www.worldbank.org/en/data/datatopics/aspire Page 4 of 49 Box 1: Sudan’s Safety Nets and Social Protection Programs At the time of the NHBPS 2014/15, existing government and parastatal transfers covered only a small fraction of the poor. Food aid recipients were heavily concentrated in North Darfur, benefiting from well-targeted transfers from humanitarian actors. Government transfers and transfers expended by Zakat, a parastatal religious organization, on the other hand, covered only a small fraction of the poor and lacked accuracy in targeting. However, the receipt of transfers, especially the amounts household received, is likely underestimated. While food aid programs were reasonably targeted to the poor, government transfers, NGO/charity transfers, and Zakat transfers had comparatively low targeting accuracy. The bottom 40 percent accounted for around 60 percent of the recipients of food aid (World Bank 2018). This compares to a share in consumption expenditure among this group of around 20 percent. Hence, food aid in Sudan is progressive and pro-poor. While pro-poor and other types of transfers lack targeting accuracy, the bottom 40 account only accounted for 39.4 percent, 45.5 percent, and 46.5 percent of the beneficiaries of government programs, NGOs/charity transfers, and Zakat transfers, respectively. In recent years, the Government of Sudan (GoS) has made significant improvements in its social protection system but targeting remains a problem. The GoS established the Social Initiatives Program (SIP) in 2012. The largest component of the program is the cash transfer component, which covers over 500,000 households, each receiving SDG 250 per month. In 2016, the government launched a new social safety net program called Shamel, which includes the basic SIP component2 and added new interventions that support livelihoods; basic service delivery in water, health, and education; and strengthened community engagement. However, with increasing inflation, the value of transfers to households has been eroded. This, coupled with poor targeting and delayed payments, suggests a need to review the cash transfer program to make it more impactful. Options include increasing amounts transferred to households or redirecting funds to other safety net programs that could help provide school meals and keep children in school while at the same time supporting their nutrition. Reforming these social protection programs to effectively reduce poverty by building resilience and lowering vulnerability to shocks can be greatly informed by an understanding of the extent of the effect of shocks on household welfare. The analysis of the vulnerability of households to negative shocks is an important input in the design of social protection programs, poverty reduction strategies, disaster risk management policies, among others. Results from these studies illustrate the link between poverty, social protection and disaster risk-management and hence inform the design of such policies. Typically, these studies describe: (a) households’ level of exposure to negative shocks, (b) the extent of impact of such shocks on households, and (c) the type of coping mechanisms available to households. Furthermore, these studies also provide detailed insights into the nature of shocks (whether idiosyncratic or covariate) as well as differences in the risk of exposure to shocks. Idiosyncratic shocks affect individual households- such as sickness/death, loss of harvests, livestock or other assets, fire outbreaks, robbery/assault etc. Covariate shocks on the other hand affect all households with a given community- such as floods/drought, civil conflict, price hikes, etc. This distinction is particularly important in developing countries such as Sudan where formal insurance markets are missing or inaccessible and informal systems of support such as risk- sharing arrangements and informal savings and credit associations are commonly used by affected 2Cash transfers, noncontributory health insurance, microfinance, rural women’s empowerment, support to the disabled and orphans, and the national student welfare fund. Page 5 of 49 households to mitigate the impact of shocks. Since covariate shocks affect all households within affected communities, informal strategies cannot effectively cushion the effect of such shocks leaving households more vulnerable. Households' risk of exposure is typically dependent on factors such as their location and/or occupation. For instance, weather and the probability for natural disasters are often correlated with location through attributes such as elevation, proximity to water sources, etc., and the type of livelihood activities (agriculture versus non-agriculture) driving differences in risk of exposure to such shocks across households. Further to the analysis of the nature of shocks and risk of exposure, the description of the availability of ex-ante risk mitigation and ex post coping strategies (including access to social protection programs) is equally important in drawing insights to better inform interventions. Differences in the impact of shocks across households and communities may be driven by differences in the state of infrastructure, access to technologies, insurance, and practices to lessen the impact of a shock. For instance, irrigation infrastructure or certain farming practices could mitigate the impact of a mild drought on a farmer’s harvest. When a shock affects a household negatively, the types of ex post coping strategies available to it will determine its level of resilience, that is, whether it can bounce back or not. This could be in the form of savings or loans, assistance from family and friends, or formal social protection programs. When such options are not available, households must resort to reducing consumption or selling productive assets, both of which can have longer-term negative impacts on welfare. To mitigate the impact of shocks, households can ex ante choose to diversify their economic activities through a combination of agricultural and nonagricultural activities, the production of multiple or drought-resistant crops, and so on. However, these strategies are often insufficient to fully mitigate the impact of the shocks—particularly among households with low assets and high vulnerability to shocks. (Dercon 2002). Understanding the extent to which these strategies are available to and used by households across communities as well as their (in)effectiveness are important inputs for designing and reforming social protection programs. However, a knowledge gap (driven largely by the data constraints) exists in our understanding of the prevalence of shocks and the strategies used to mitigate them by Sudanese households. Addressing this gap is an important first step towards reforming existing SSNs. This can be done by using data that enable identifying households that are vulnerable to shocks based on their profile, such as their source of livelihood or location, which increases their risk of exposure to shocks and their access to social protection. In the case of Sudan, these efforts are constrained by the lack of data (Kjellgren et al. 2014). The lack of data on measures of household welfare, prevalence of shocks, and coping strategies used to mitigate the impact of shocks significantly constrained the analysis of household vulnerability and resilience to shocks and its implications on the effects of shocks on their welfare. Recently, the Central Bureau of Statistics began conducting a household survey collecting data on these variables—the National Baseline Household Survey (NBHS) in 2009 and the National Household Baseline and Poverty Survey (NHBPS) in 2014/15. Both datasets contain information on household demographics, assets, consumption/expenditure, shocks experienced, coping strategies used, and agricultural activities making it possible to describe the prevalence of idiosyncratic shocks and the types of coping strategies used to mitigate them in each period and discuss changes overtime. However, it is not be possible to examine the dynamics in the effects of shocks on household welfare at household level. This is because both surveys have difference sampling frameworks in addition to political and geographic differences between the two Page 6 of 49 periods- one being before the secession and the other being after. However, using cross-section regression techniques on the latest survey, it is possible to examine the effects of shocks on the welfare of Sudanese households to draw insights for policy. This paper applies this framework to examine the impact of idiosyncratic and covariate shocks on the welfare of Sudanese households. Data on idiosyncratic or household level shocks such as floods/drought, sickness, crop/livestock loss, death of family member, fire, robbery, etc. are self-reported and provided in both surveys. We complement this data with information on covariate shocks such as flood/drought and conflict3 obtained from additional data sources. In particular, state-level rainfall and conflict data is extracted from geo-referenced data overtime. Deviations of observed rainfall and conflict in specific years (such as the years for which there is household data) from historical levels can be constructed and used as indicators of covariate shocks at state-level. For instance, households in a given state are likely to experience floods (droughts) are when higher (lower) than usual rainfall is observed. Similarly, the extent to which the presence and intensity of conflict in a given state constitute a shock to households in the state can be defined based on deviations from historical trends. The paper begins by documenting the incidence of these shocks among households in Sudan. In this section, we describe the exposure of Sudanese households to shocks disaggregating them into space (rural/urban as well as administrative units such as states), livelihoods, and into individual types of shocks.In addition, we also examine the coping strategies used to mitigate shocks and/or ensure resilience as well as the availability of or access to social protection programs and financial services. The objective of this exercise is to (at least descriptively) build a profile of households who are likely to be vulnerable to shocks based on the types of shocks reported by households and their characteristics such as their occupation/livelihood, location, and assets/wealth. In this analysis, we discuss potential consequences (as it relates to poverty more broadly and social protection in particular) of exposure to shocks and the type of coping strategies available and used by households. The paper also examines the multidimensionality of the effect of shocks by considering several indicators of household welfare. Given that shocks affect several aspects of household welfare, we estimate the effect of shocks on various indicators of household welfare such as consumption/aggregate expenditure (which can be further disaggregated into food and non-food consumption/expenditure), dietary diversity, poverty status, and measures of wealth such as assets owned by the household. The objective of this exercise is to examine the extent to which shocks affect household welfare while accounting for the fact that such effects are multidimensional and likely to be heterogenous due to differences in household wealth, coping strategies, occupation, demographics, geography, and availability of social protection programs, among others. Furthermore, we examine the extent of this heterogeneity in the impact of shocks across households by relating the incidence of shocks to regions in Sudan, household livelihood, household index of coping strategies, households’ access to credit, and households’ receipt of support from informal risk-sharing arrangements, nongovernmental organizations (NGOs), and the government. The concluding section focuses on drawing insights from the results to inform policy. The results from the analyses of the prevalence of shocks among households in Sudan and its impact on their welfare can be used to highlight the extent of the vulnerability (or resilience) of Sudanese households to shocks, which 3 A separate paper examined the effect of price shocks on the welfare of Sudanese households. For this reason, this paper only focuses on flood/drought and conflict as covariate shocks. Page 7 of 49 can be used to inform social protection reforms. In the concluding sections, we highlight the implications of the results such as the profile of households most vulnerable to shocks, various coping strategies (including social protection programs) available to and used by households to mitigate shocks and build resilience. Using these findings to the extent possible, we discuss the implication of the results on state of social protection in Sudan—its effectiveness where it used or the possible consequences of its inaccessibility to vulnerable households, the complementarity (or otherwise) that exists between social protection and informal/semiformal coping strategies, and so on. This section also discusses possible reforms while being conscious of the constraints (budget and otherwise) imposed by the current macroeconomic environment and narrow fiscal space. It is expected that these insights will stimulate further discussions on possible social protection reforms in Sudan to reduce vulnerability and sustain gains in poverty and inequality reductions. In summary, the paper aims to examine the following: • To describe the prevalence of idiosyncratic and covariate shocks among households in Sudan as well as describe the profile of households most vulnerable to shocks. • To estimate the effect of shocks on household welfare using measures of consumption/expenditure, food security, dietary diversity, and wealth/assets. • To describe the different types of coping strategies (including social protection programs) used by Sudanese households to mitigate the impact of shocks and their implication on their resilience to reoccurring shocks. • To emphasize, throughout the analysis, the spatial distribution of the prevalence, vulnerability, and impact of shocks as well as household livelihood zones and availability of or access to social protection programs. • To highlight the policy implications of the results, particularly as it relates to the state of social protection and possible areas of reform where feasible (given the limited resources and pressing challenges the country faces) to reduce vulnerability and build resilience of households in Sudan. Page 8 of 49 2. Poverty, Shocks, and Household Welfare The analysis of the vulnerability of households to shocks and its impact on their welfare is particularly important for a country such as Sudan where many households face the risk of being exposed to idiosyncratic and covariate shocks. Most of these households who face significant risk exposure to shocks also lack the capacity to mitigate the impact of such shocks when they occur. Empirical studies have shown that this is largely because they are often the poorest households- particularly in developing countries (Jalan and Ravallion (1999). As a result, they face significant risk of consumption and/or income fluctuations and poverty traps.. In Sudan, recent data shows striking similarities in the key attributes of the profiles of affected households and the stylized facts described above. For instance, in 2009, the poorest households—the bottom 20 percent of the distribution—in Sudan had low asset levels (see the annex of the World Bank [2011]). More than 60 percent of them are engaged in agriculture-related activities, and hence they have high exposure to agroclimatic-related shocks. In 2009, 21 percent of the poorest households in Sudan were affected by droughts or floods; 28 percent were affected by crop diseases; and 34 percent reported stolen livestock as a shock (World Bank 2011). Added to these, other covariate shocks such as the presence of conflict in and around Sudan is likely to affect economic activities of households, thereby increasing vulnerability.4 The exposure of households (particularly the poorest) to the risk of shocks may have long-lasting negative effects on household welfare. Understanding the nature of the impact of shocks on household welfare has been the subject of several studies. These studies provide insights on how shocks affect households by identifying the households most affected by shocks, the extent to which various indicators of household welfare are affected by various shocks, the duration of the such impacts, and the effectiveness of coping strategies in mitigating the impact of shocks. Using consumption as a measure of welfare, several studies have shown a negative effect of shocks. For instance, Jacoby and Skoufias (1998) as well as Dercon and Krishnan (2000), among others, observed that poor weather may affect household welfare by lowering household income, which also lowers consumption. Other studies such as Skoufias and Quisumbing (2005) examine the impact of shocks by disaggregating household consumption into food and non-food consumption. They showed that food consumption is better ensured than non-food consumption, particularly against idiosyncratic shocks. The reduction in non-food consumption, they argued, can be interpreted as a strategy to better ensure food consumption. Aside from consumption/income, shocks may also significantly affect other measures of household welfare such as assets, health, and vulnerability to future poverty (Hoddinott 2006); Hoddinott and Kinsey (2001); and Porter and Hill (2015). In addition to various indicators of welfare, the impact of shocks may also vary across shocks—idiosyncratic shocks (those affecting individual households such as death of a household member, theft of livestock or crops, and so on) are often better insured than covariate shocks (those affecting all households in a community such as floods and droughts) (Harrower and Hoddinott 2005). Access to social protection and safety net programs and households’ capacity to effectively cope with shocks are likely to influence the impact of shocks on household welfare. Most households in developing countries, where credit, asset, and insurance markets are missing/incomplete and/or social protection and safety net programs are unavailable, cope with shocks by using strategies that may trap them into 4The most recent data on these indicators were provided by the 2014/15 NBHS, which is the main dataset used in this paper. Using this dataset, we discuss recent trends in poverty and vulnerability in Sudan with emphasis on changes following the secession of the South later in the paper. Page 9 of 49 poverty (World Bank 2014, 24 and 113). Typically, these households resort to selling productive assets— livestock, land, and so on— (Deaton 1992; Dercon 2002); using savings (Paxson 1992); investing in low- risk, low-return crop choices and asset portfolios (Rosenzweig and Binswanger 1993); and increasing labor supply by removing their children from school (Jacoby and Skoufias 1997; Kochar 1998; Morduch 1995), among others. The use of these strategies may have significant consequences on households’ poverty status and national poverty eradication achievements. For instance, where the use of these strategies depletes households’ stock of assets, savings, human capital, and so on or is insufficient to fully cushion the impact of shocks (Morduch 1995; Townsend 1995), households may be trapped in poverty, resulting in a cycle of intergenerational poverty. See Dasgupta (1993), Dercon (1996), and Dercon (2004) among others for empirical evidence on the effects of these strategies. Similarly, households marginally above the poverty line may be at risk of sliding back into poverty when affected by a shock. Jointly, these effects could lower national achievements in the fight against poverty. To accelerate poverty reduction, it will be important to identify vulnerable households and help them become more resilient to common risks and shocks that may push them back into poverty. Effective measures should aim to (a) lower exposure to shocks by promoting the use of irrigation systems and drought-resistant seeds and diversification into less risky economic activities and (b) improve coping mechanisms, such as cash transfer programs or other financial instruments, to complement existing informal safety nets. To inform these reforms, there is a need to empirically examine the vulnerability of Sudanese households through the lens of (a) the level of exposure to negative shocks, (b) the extent of impacts of shocks on household welfare, and (c) the type of coping mechanisms available to the household—including access to social protection programs. It is also useful to examine these caveats along the lines of different types of shocks—covariate and idiosyncratic shocks—as well as factors which affect the extent of exposure to shocks such as location and occupation or livelihood. This paper uses this framework to examine the effect of shocks on the welfare of Sudanese households. By estimating the impact of various shocks on measures of household welfare as well as the strategies used to cope with shocks, a profile of households most vulnerable to shocks can be built through their characteristics—occupation/livelihood, gender of head, poverty status, and location. We consider both self-reported idiosyncratic shocks such as death/sickness of household members, loss of livestock, crops, or assets, and so on and covariate shocks such as deviation of rainfall from historical average to indicate flood/drought, conflict, and so on. In addition to identifying the characteristics of households most vulnerable to shocks, the paper also aims to highlight the shocks with most impact on household welfare while accounting for the heterogeneity of the impact across households based on wealth, livelihood, location, and access to safety nets/social protection programs. 2.1. Shocks in Sudan This section discusses the exposure and vulnerability of Sudanese households to shocks beginning with the 2009 NBHS followed by the 2014/15 National Household Budget and Poverty Survey (NHBPS) , which gives a more recent picture. It describes the main types of shocks reported by households overall as well as the differences in the prevalence of shocks across rural/urban areas, states, household livelihoods, and quintiles of household per capita consumption. To complement this description of exposure to shocks across households, this paper profiles households most likely to be vulnerable to shocks based on statistical differences in the characteristics of households affected by shocks and those not affected. Page 10 of 49 In both surveys, households were asked about the shocks that severely affected them during the previous five years from a list of preidentified categories of shocks. These include floods/drought, crop disease, livestock loss, severe illness, damage to household dwelling, water shortage, death of a household member, and so on. In addition to identifying the shocks they experienced, households were also asked about the estimated effect of the shock in monetary value as well as the coping strategies used to mitigate the impact of the shock. We use this information to estimate the effect of shocks on household welfare. This section focuses on describing the exposure of Sudanese households to shocks and profiling vulnerable households. Agriculture-related shocks are most common in Sudan. Livestock loss and crop disease shocks affected 26 percent and 19 percent of Sudanese households in 2009, respectively. Other shocks commonly reported by Sudanese households include floods/droughts and illness/accident-related shocks. Nearly 15 percent of households reported these shocks in 2009 (see Figure 3b ). Since households were asked to identify the shocks that severely affected them in the previous five years from a list of common shocks, households can be affected by and can report more than one shock. Majority of Sudanese households report having been affected by a single shock, particularly in 2014/15 when 60 percent of households were affected by only one shock. In 2009, however, more households were affected by more than one shock—27 percent of households were affected by two shocks (compared to 24 percent in 2014/15) and 16 percent were affected by three shocks (compared to 11 percent in 2014/15 (see Figure 3a). In the absence of effective coping strategies and/or social protection programs, the effect of shocks on the welfare of households is likely to be large, particularly for households facing multiple shocks. According to households’ own estimates of the impact of shocks, health-related shocks such as death of a household head and severe illness/accident and loss of livestock are the most severe shocks in Sudan, particularly in 2014/15. On average, households affected by severe illness/accident or death of the household head in 2014/15 reported an estimated effect of more than SDG 11,000 (equivalent to USD 240)5. Households affected by loss of livestock reported an average impact of more than SDG 10,000 (equivalent to USD 220) (see Figure 3c). The severity of health-related shocks and livestock loss is perhaps driven by the loss of income from economic inactivity due to illness/accident or death of the household head and loss of assets in the case of livestock loss-related shocks. As a result, these shocks result in significant welfare loss, particularly for households where these activities and assets form a significant component of household income or wealth. Despite being available to relatively few households affected by shocks, help received is a common coping strategy among Sudanese households. Less than 35 percent of households affected by shocks in 2009 and less than 25 percent of those affected by shocks in 2014/15 reported that they relied on help received to cope with shocks (see Figure 3d). Although this indicates the presence of some form of social protection provided by family/friends, NGOs, and the government, only a handful of households affected by shocks have access to it. Furthermore, although there is little or no empirical evidence on the effectiveness of these systems of support in Sudan (at least to our knowledge), other studies have indicated that relying on these arrangements may further expose households to risk of welfare loss and 5 This is based on a exchange rate of USD 1= SDG 45.11 obtained from the Central Bank of Sudan’s website - https://cbos.gov.sd/en/exchange-rates. Compared to measures of household welfare such as consumption per capita, these effects are fairly large. For instance, the average consumption per capita for the richest households in 2014 was SDG 12,776. Page 11 of 49 poverty trap given that help received may be (at best) insufficient to fully mitigate the impact of idiosyncratic shocks or (at worst) unavailable in the case of covariate shocks (Harrower and Hoddinott (2005). Therefore, there is a need for social protection in Sudan to target households who do not have access to these programs while also complementing the semiformally and informally designed quid pro quo risk-sharing arrangements for households who have access to it. Another common coping strategy for households affected by shocks is the sale of assets or use of savings. More than 20 percent of households affected by shocks in both 2009 and 2014/15 relied returns from the sale or rent of their assets or use of their savings to mitigate the impact of shocks (see Figure 3d). This category of coping strategies includes the use of cash savings and sale of tools, crops, animals, farm land, and so on. The use of assets and savings in response to shocks may further trap poor households into poverty, thereby lowering gains in poverty reduction in Sudan as with other developing countries. Furthermore, the state of asset markets in developing countries—which are often poorly developed (Dercon 2002), thereby affecting returns from the sale of assets—may further affect the effectiveness of these responses to shocks. In the absence of social protection programs, limitations in the access to semiformal and informal risk- sharing arrangements are likely to trigger the use of severe coping strategies such as selling assets and dissaving. Thus, these results highlight the need for effective social protection programs to build resilience and lower vulnerability, particularly for households with low assets and/or those most vulnerable to covariate shocks. Information about the heterogeneity in the type of risk-coping strategy used across different shocks faced by households will be useful in the design of these programs. Page 12 of 49 Figure 3: Shocks in Sudan: Specific Types of Shocks Reported, Effects, and Coping Strategies, 2009 and 2014/15 (a) Number of Shocks Reported (b) Specific Types of Shocks Reported Death of Hh. Head 2% More than 5 1% 1% Livestock Loss 15% 26% 1% Crop Disease 8% 5 Shocks 19% 3% Severe Illness/Accident 7% 18% 4 Shocks 3% Floods/Drought 15% 5% 15% Death of a hh. Member 6% 13% 3 Shocks 11% 6% 16% Dwelling Damaged 12% Severe Water Shortage 3% 8% 2 Shocks 24% 27% Fire 6% 8% Robbery/Assult 5% 6% 1 Shock 60% 47% Other Shocks 2% 3% 0% 20% 40% 60% 80% 0% 10% 20% 30% 2014/15 2009 2014/15 2009 (c) Self-reported Effects of Shocks in SDGs (d) Coping Strategies Used Received Help from Death of Hh. Head 11,760 Family/Friends, NGOs or 24% 33% Gov't Severe Illness/Accident 11,014 2,315 Sell Assets/Use 23% Other Shocks 10,800 Savings/Rent Farm 25% 2,746 Livestock Loss 10,268 7% 2,653 Increase Labor Supply 13% Robbery/Assult 10,030 2,086 Borrowed Money 9% Fire 9,197 10% 2,296 Death of a hh. Member 7,364 1,612 16% Other 10% Floods/Drought 6,554 2,213 5,351 Reduced Expenses or 9% Crop Disease 1,511 3% Consumption Dwelling Damaged 3,783 1,451 Kids Migrate or removed 4% Severe Water Shortage 3,346 from School 1% 970 0 5000 10000 15000 0% 10% 20% 30% 40% 2014/15 2009 2014/15 2009 Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Page 13 of 49 2.1.1 Heterogeneity in the Prevalence of Shocks across Space and Livelihoods In Sudan, there is a high incidence of shocks in rural areas. Overall, idiosyncratic shocks appear to be most prevalent among households in rural Sudan. The percentage of Sudanese households affected by idiosyncratic shocks such as droughts/floods6, crop diseases, livestock loss, illness, and death of a household member is much higher in rural areas than in urban areas. In 2009, 72 percent of rural Sudanese households reported having been affected by these shocks compared to 51 percent of urban households and 64 percent nationally. A similar trend is also observed in 2014/15 where 49 percent of rural households reported at least one shock compared to 36 percent in urban areas and 45 percent nationally (see Figure 4a). It is important to highlight that more than 90 percent of households whose main livelihood is agriculture live in rural Sudan. Therefore, the significant exposure of rural households to shocks can in part be explained by the fact that agriculture-related shocks are the most common shocks in Sudan and that a majority of rural households are engaged in agriculture-related activities such as crop farming and animal husbandry. Poor households appear to be most exposed to idiosyncratic shocks, but the difference in the prevalence rate between poor and non-poor households is small. In both 2009 and 2014/15, the prevalence of shocks was slightly higher among poor Sudanese households (defined using the national poverty line) than non-poor households and the national average. In 2009, 68 percent of poor households reported at least one shock whereas 63 percent of non-poor households and 65 percent of all Sudanese households reported at least one shock (see Figure 4c). Similar results are also observed in 2014/15—49 percent of poor households were affected by at least one shock compared to 43 percent of non-poor households. Using the distribution of consumption per capita, a similar trend is observed. Although households in the bottom 20 percent of the distribution are more exposed to shocks, the incidence of shocks among households in the top end of the distribution is also high. More than 70 percent and 50 percent of the poorest households reported at least one shock in 2009 and 2014/15, respectively, compared to nearly 60 percent and 40 percent of the richest households in 2009 and 2014/15, respectively (see Figure 4d). The high prevalence of shocks among both poor and non-poor households is an interesting observation. It will be useful to examine the extent to which this observation is explained by differences in the types of shocks, risk exposure due to differences in livelihood activities, vulnerability, and resilience between poor and non-poor Sudanese households. We discuss this further in section 2.1.2 by disaggregating exposure to shocks into specific types of shocks reported across household livelihoods and poverty status as well as differences in coping strategies used to mitigate the impact of shocks. In terms of household livelihood, shocks are most prevalent among households engaged in agriculture- related activities—particularly crop farming. More than 60 percent of households whose main livelihood is crop farming reported at least one shock in 2004/15 and up to 83 percent in 2009. A similar trend is also observed among households engaged in animal husbandry—nearly 50 percent of these households reported at least one shock in both years (see Figure 4b). Aside from the high prevalence of shocks, poverty rates for households whose main livelihood is agriculture related is equally high. More than 50 percent of crop farmers in 2009 and more than 30 percent in 2014/15 were poor. Similarly, in 2009, nearly 6 Although floods/drought are likely to affect entire communities making them covariate shocks, this aspect of weather-related shocks is captured in the section on covariate shocks- Section 2.3. Self-reported incidences of flood/drought capture the idiosyncratic aspect of such covariate shocks- mainly how it affected the household such as damage to dwelling, loss of stored food, etc due to floods; and lost seeds and harvests, etc. due to droughts. Page 14 of 49 40 percent of households engaged in animal husbandry and more than 25 percent in 2014/15 were poor.7 These characteristics further illustrate the vulnerability of agricultural households to shocks. High poverty rates among these households is likely to significantly affect their capacity to fully cushion the impact of shocks, threatening their ability to escape poverty. Furthermore, for rain-dependent farmers, for instance, exposure to the risk of variations in rain, resulting in floods/droughts, crop diseases/failure, and loss of income/livestock, may have large negative welfare effects, particularly among the poor who may lack the capacity to cope with the shocks. Crop farmers are particularly likely to be less resilient to these shocks since (unlike households involved in animal husbandry) they may not have sufficient assets such as livestock, some of which they can sell in the event of a shock. As a result, they may resort to using emergency coping strategies such as selling the next year’s seeds, tools, or farm land, which potentially traps them in poverty. Another category of households highly exposed to shocks is those that rely on pensions and remittances. Inactive labor market participation possibly due to old age and fluctuations in remittances is likely to drive vulnerability among these households. It is interesting to note that more than 65 percent of households that rely on remittances as a source of livelihood live in rural areas whereas more than 50 percent of pensioners live in urban areas. In addition, poverty rates among these households are relatively low—less than 30 percent in both years. Apart from differences in the prevalence of shocks across household likelihoods, differences in the types of shocks faced by households is another important aspect in the analysis of the vulnerability of households. Overall, it can be observed that poor rural agricultural households are highly exposed to shocks in Sudan. Furthermore, in the absence of social protection programs, low asset levels of these households and risk inherent in rain-fed agriculture potentially lower their resilience to shocks. Distinguishing the extent to which the vulnerability of agricultural households to shocks is driven by the nature of their livelihood (such as the risk of rain variation) or by the fact that they are likely to be less resilient (due to low asset levels, fluctuations in agricultural income, and so on.), though an interesting empirical question is beyond the scope of this paper. 7 Based on authors’ calculation using 2009 NBHS and 2014/15 NHBPS. Page 15 of 49 Figure 4: Prevalence of Shocks across Households in Sudan, 2009 and 2014/15 (a) Rural versus Urban (b) Household Livelihoods 80% 90% 83% 72% 79% 80% 70% 65% 70% 65% 63% 61% 61% 57% 60% 51%52% 54% 50% 60% 50% 40% 39%39% 51% 37% 37% 36% 37% 50% 49% 40% 35% 45% 30% 40% 36% 20% 10% 30% 0% 20% 10% 0% Urban Rural Sudan 2009 2014/15 2009 2014/15 (c) Household Poverty (d) Quintile of Per Capita Consumption 80% 80% 71% 68% 68% 69% 70% 65% 70% 63% 65% 62% 60% 58% 60% 49% 51% 50% 48% 45% 50% 47% 43% 45% 43% 40% 38% 40% 30% 30% 20% 20% 10% 10% 0% Non-Poor Poor Sudan 0% Poorest Poor Middle Rich Richest Sudan 2009 2014/15 2009 2014/15 Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Majority of the households affected by shocks live in the Central Darfur and Kordofan regions of Sudan. At state level, Southern Kordofan, Blue Nile, Western Darfur, and Northern Darfur have the highest incidence of shocks—more than 80 percent of households in these states reported at least one shock 2009 and more than 50 percent in 2014/15. On the other hand, low prevalence of shocks is observed in North- Page 16 of 49 Eastern Sudan. For instance, in the state of Red Sea, less than 50 percent in 2009 and less than 30 percent in 2014/15 reported at least one shock. Other states in the region such as Kassala, River Nile, and Khartoum also have low prevalence of shocks (see Figure 5). Figure 5: Prevalence of Shocks across States in Sudan, 2009 and 2014/15 (a) 2009 (b) 2014/15 Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Shocks are most prevalent in states where crop farming is the main household livelihood. Across states, crop farming and wage/salaried occupations are the most common household livelihoods. Crop farming is particularly common in the Darfur and Kordofan regions whereas wage/salaried occupations are common livelihoods in the Eastern part of the country, particularly in and around the capital city Khartoum. Differences in the prevalence of shocks across states appear to correlate with household livelihood. For instance, shocks are more prevalent in the Darfur and Kordofan regions where crop farming is the main economic activity. In addition, agriculture-related shocks such as crop and livestock losses are most common in these areas. In other states where wage/salaried occupations are the most common activity, shocks are generally less prevalent perhaps because agriculture is not a common livelihood. Households in these states are likely to be less vulnerable to fluctuations in income which agricultural households in other states may be exposed to (see Figure 6). Page 17 of 49 Figure 6: Household Livelihoods and Prevalence of Shocks in Sudan, 2009 and 2014/15 (a) Main Household Livelihood: 2009 (b) Main Household Livelihood: 2014/15 (c) Main Shocks in Sudanese States: 2009 (d) Main Shocks in Sudanese States: 2014/15 Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Page 18 of 49 2.1.2 Prevalence of Specific Types of Shocks The previous subsection provided an overview of the exposure of Sudanese households to idiosyncratic shocks. It discussed the percentage of Sudanese households affected by shocks in general, disaggregating it by rural/urban, states, household livelihood, and quintile of per capita consumption. In this subsection, we disaggregate the prevalence of shocks into the different types of shocks reported by households — drought/floods, crop diseases, livestock loss, illness, death of a household member, and so on. The objective of this disaggregation is to highlight the differences in the prevalence of different types of shocks overall as well as across subgroups of the population distinguished by household livelihoods, poverty, and states. As a follow-up to earlier observations in the data, the high prevalence of shocks among both poor and non-poor households, it will be useful to examine differences in the types of shocks that affect households across livelihoods and poverty status. Furthermore, detailed information about the prevalence of specific types of shocks across livelihoods and space as well as the type of coping mechanisms available to and used by households is particularly important to effectively design and implement social protection and poverty reduction policies. We consider these issues in this subsection of the paper. There is significant heterogeneity in the prevalence of shocks across household livelihoods. As expected, households whose main livelihood is agriculture are more exposed to agriculture-related shocks. Shocks due to loss of crops and livestock are highest among households involved in crop farming and animal husbandry—livestock loss is highest among households involved in animal husbandry. Other shocks such as illness, death of a household member, and damage to household are common among households involved in other activities such as wage/salaried jobs and pensioners (see Figure 7). Although overall the difference in the prevalence rate of shocks among poor and non-poor households is small, large differences in the types of shocks that affect them are observed between the poorest and the richest households. Exposure to agriculture-related shocks is highest among poor households. Households in the bottom 40 percent of the consumption per capita distribution have the highest incidence of shocks due to livestock loss, crop diseases, and flood/droughts. On the other hand, exposure to other shocks such as illness are higher among households on the higher end of the consumption per capita distribution. For instance, in 2014/15, 9 percent of households in the top 20 percent of the consumption per capita distribution reported severe illness/accident as a shock compared to 6 percent of households in the bottom 20 percent of the same distribution in the same time period (see Figure 8). Exposure of poor households to shocks such as livestock and crop loss shows the vulnerability of these households, which may significantly affect progress in reducing poverty and vulnerability. The prevalence of such shocks is likely to significantly lower resilience through a reduction of accumulated assets and expected income, thereby increasing the likelihood of such households being trapped in poverty. In the absence of social protection programs, loss of these assets and income, which are expected to help mitigate the impact of shocks, may have long-lasting effects on household welfare. Livestock losses are more prevalent in Western Sudan, and floods/droughts are more prevalent in the South and in parts of the East. The prevalence of shocks differs across states. Shocks related to livestock losses are most common in the Darfur region, whereas floods/droughts are most common in the Blue Nile and Sennar states. The highest percentage of households affected by livestock loss live in Western Sudan—particularly in the Darfur region (such as in Eastern and Central Darfur) and in parts of the Kordofan region (such as South Kordofan). In the South, shocks related to floods/droughts are most prevalent. In states such as Blue Nile and Sennar, floods/droughts are the most common shock despite their low percentage of prevalence (see Figure 9). Page 19 of 49 Figure 7: Heterogeneity in the Prevalence of Shocks in Sudan: Household Livelihoods 2009 60% 50% 40% 30% 20% 10% 0% Crop farming Animal Wages and Owned business Property income Remittances Pension Aid Others husbandry salaries enterprises Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks 2014/15 40% 35% 30% 25% 20% 15% 10% 5% 0% Crop farming Animal Wages and Owned Property Remittances Pension Aid Others Transfers from husbandry salaries business income Hh. members enterprises Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks Death of a hh. Head Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS. Page 20 of 49 Figure 8: Heterogeneity in the Prevalence of Shocks in Sudan: Consumption Per Capita Quintiles 2009 35% 30% 25% 20% 15% 10% 5% 0% Poorest Poor Middle Rich Richest Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks 2014/15 25% 20% 15% 10% 5% 0% Poorest Poor Middle Rich Richest Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks Death of a hh. Head Source: Authors’ calculation using 2009 NBHS and 2014/15 NHPBS. Page 21 of 49 Figure 9: Heterogeneity in the Prevalence of Shocks across States 2009 60% 50% 40% 30% 20% 10% 0% Northern River Nile Red Sea Kassala Al-Gadarif Khartoum Al-Gezira White Nile Sinnar Blue Nile Northern Southern Northern Western Southern Kordofan Kordofan Darfur Darfur Darfur Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks 2014/15 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Floods/Drought Crop Disease Livestock Loss Severe Illness/Accident Death of a hh. Member Fire Robbery/Assult Dwelling Damaged Severe Water Shortage Other Shocks Death of a hh. Head Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS. Page 22 of 49 2.2 Vulnerability of Households to Idiosyncratic Shocks The vulnerability of Sudanese households to shocks is described using the characteristics of households such as demographics, income, assets, and poverty status. This is done by identifying common characteristics of households affected by shocks and the extent to which these attributes differ between households exposed to shocks and those that are not. By means of a t-test, we obtain the statistical significance of the differences in the characteristics of households affected by shocks and those not affected, which is then used to describe the profile of vulnerable households. By combining two rounds of data, Insights from these attributes, albeit descriptive, can inform the design of social protection programs that seek to target most vulnerable households to effectively build resilience. Poor Sudanese households are most vulnerable to shocks. Overall, 42 percent of households who reported shocks are poor and as shown earlier predominantly live in rural areas. The total per capita expenditure for households facing shocks is SDG 156 on average (compared to a national average of SDG 160 and SDG 177 for households who did not report a shock). Similarly, the per capita expenditure on food in these households (SDG 98) is significantly lower than that of households not affected by shocks, which is SDG 111 on average. More than 30 percent of households affected by shocks have consumption expenditure in the bottom two quintiles, and more than 50 percent reported that their income in the month preceding the survey was between SDG 500 and SDG 999.99 and between SDG 1,000 and SDG 4,999.99 in the previous year. Other indicators of poverty—food insecurity, consumption levels, ownership of assets, and household wealth—indicate that poorer households are most vulnerable to shocks. On average, households facing shocks spent significantly less on most categories of expenditure—except health (in per capita terms)— than households not affected by shocks. The higher expenditure on health is driven in part by the occurrence of shocks relating to illness/accident and so on. Also, households affected by shocks have on average more livestock units when compared to households not affected. This difference is likely to be driven by the fact that shocks are more prevalent among agricultural households. In terms of demographics, households affected by shocks are more densely populated with slightly older household heads. However, the difference in the latter is statistically insignificant (see Error! Reference source not found. and Error! Reference source not found.). Table 1: Test of Means between Households Affected by Shocks and Those Not Affected, 2009 Variables No Shock Mean Shock Mean Mean Difference Urban (=1) 2,548 1.557 5,365 1.752 −0.196*** Poor (=1) 2,548 0.376 5,365 0.432 −0.056*** Consumption quintile 2,548 3.339 5,365 3.094 0.244*** Per capita food expenditure (in SDGs) 2,548 111.126 5,365 97.512 13.614*** Per capita education expenditure (in SDGs) 2,548 3 5,365 2.379 0.621*** Per capita health expenditure (in SDGs) 2,548 9.414 5,365 11.531 −2.117*** Per capita clothing expenditure (in SDGs) 2,548 5.723 5,365 5.363 0.360** Per capita utilities expenditure (in SDGs) 2,548 13.405 5,365 12.286 1.119*** Per capita transport expenditure (in SDGs) 2,548 16.577 5,365 12.12 4.457*** Per capita personal care expenditure (in SDGs) 2,548 6.559 5,365 5.525 1.033*** Per capita house related expenditure (in SDGs) 2,548 9.278 5,365 8.415 0.863*** Page 23 of 49 Variables No Shock Mean Shock Mean Mean Difference Per capita recreational expenditure (in SDGs) 2,548 0.95 5,365 0.848 0.102 Per capita other expenditure (in SDGs) 2,548 1.32 5,365 0.903 0.418*** Total per capital expenditure (in SDGs) 2,548 177.35 5,365 156.89 20.469*** Household asset index 2,548 42.16 5,365 36.64 5.516*** Household tropical livestock index 2,548 1.08 5,365 2.66 −1.575*** Household size 2,548 5.75 5,365 6.37 −0.626*** Male-headed household 2,548 0.83 5,365 0.82 0.01 Age of the household head 2,548 45.74 5,365 45.85 −0.11 Total income in previous month (in SDGs) 2,544 1,672.69 5,356 462.88 1,209.81 Total income in previous year (in SDGs) 2,544 7,483.45 5,356 6,524.2 959.20 5 Source: Authors’ calculation using 2009 NBHS data. *** p<0.01, ** p<0.05, * p<0.1 Table 2: Test of Means between Households Affected by Shocks and Those Not Affected, 2014/15 Variables No Shock Mean Shock Mean Mean Difference Urban (=1) 5,840 1.645 6,113 1.752 −0.107*** Poor (=1) 5,840 0.305 6,113 0.357 −0.052*** Extremely poor (=1) 5,840 0.21 6,113 0.257 −0.047*** Per capita education expenditure (in SDGs) 5,840 84.07 6,113 74.054 10.016*** Per capita health expenditure (in SDGs) 5,840 355.398 6,113 390.416 −35.018** Per capita clothing expenditure (in SDGs) 5,840 175.497 6,113 178.322 −2.825 Per capita utilities expenditure (in SDGs) 5,840 126.527 6,113 124.198 2.328 Per capita transport expenditure (in SDGs) 5,840 393.269 6,113 361.97 31.299*** Per capita personal care expenditure (in SDGs) 5,840 199.746 6,113 182.771 16.975*** Per capita house related expenditure (in SDGs) 5,840 218.84 6,113 186.886 31.955*** Per capita recreational expenditure (in SDGs) 5,840 47.559 6,113 49.777 −2.218 Per capita other expenditure (in SDGs) 5,840 39.972 6,113 34.494 5.478 Food Insecurity (=1) 5,840 0.081 6,113 0.103 −0.022*** Household size 5,840 5.506 6,113 6.162 −0.656*** Household asset index 5,840 26.638 6,113 24.063 2.575*** Household tropical livestock units 5,840 4.292 6,113 6.678 −2.385** Source: Authors’ calculation using 2014/15 NHBPS data. *** p<0.01, ** p<0.05, * p<0.1 2.3 Covariate Shocks In addition to idiosyncratic shocks, households may also be exposed to covariate shocks that potentially affect all households in a community such as a town or state. In this section, we describe the exposure of Sudanese households to covariate shocks at the state level. We consider deviation of rainfall from historical averages to identify covariate flood/drought shocks and the number of conflict/crime events. Rainfall data at the state level were obtained from the Center for Hydrometeorology and Remote Sensing Page 24 of 49 (CHRS)8. The conflict/crime data are obtained from the Armed Conflict Location and Event Database (ACLED)9. We focus on more recent data of these variables to better inform policy. 2.3.1 Rainfall Shocks Error! Reference source not found.a shows rainfall recorded in 2014 along with average rainfall between 2003 and 2015 at the state level. Rainfall in 2014 was higher than the historical average in all states in Sudan. For most states, this was a significant improvement from the amount of rainfall recorded in 2013. In states such as Central Darfur, West Kordofan, While Nile, and Sennar, the 2014 rains were 200 mm higher than the historical average. The variation in the deviation of rainfall across states can illustrate the extent to which the welfare of households is potentially affected. A large deviation in rainfall may indicate higher possibility of floods, thereby negatively affecting household welfare, whereas a slight deviation may indicate better chance of improved productivity for rain-dependent farmers. 2.3.2 Conflict Error! Reference source not found.b shows the number of conflict events along with the average number of conflict events between 2012 and 2015. For most states, the number of conflict events increased in 2014 relative to recent averages. In the Darfur region and parts of the Kordofan region as well as in Khartoum, the number of conflict events increased in 2014 reaching more than 400 events in North Darfur. However, some degree of heterogeneity exists in the type of events that are popular across states. For instance, in the Darfur region, most of these events occurred in the form of ‘violence against civilians’ whereas in Khartoum, ‘protests/riots’ dominated conflict events.10 8 See http://chrsdata.eng.uci.edu/. 9 See https://www.acleddata.com/data/ 10 The ACLED provides information on the type of conflict which is used to construct these categorizations. Page 25 of 49 Figure 10: Covariate Shocks: Price Shocks and Crime (a) Rainfall 1600.00 1400.00 1200.00 1000.00 800.00 600.00 400.00 200.00 0.00 Rainfall (2014) Avg. Rainfall (2003-2015) (b) Conflict 450 400 350 300 250 200 150 100 50 0 Number of conflict Events (2014) Avg. Number of Conflict events (2012-2015) Source: Authors’ calculation using ACLED and CHRS databases. Page 26 of 49 3. Impact of Shocks on Household Welfare 3.1. Methodology 3.1.1 Data We estimate the impact of shocks on household welfare using the latest round of the NBHS collected in 2014/15. It contains information on expenditure and consumption, welfare, poverty status, assets, and demographics of Sudanese Households. A representative sample of 13,800 households was drawn from all 18 States of Sudan, of which 11,953 households were surveyed during three rounds of data collection with a response rate of 87 percent (Central Bureau of Statistics 2017). 3.1.2 Estimation Strategy Our empirical strategy is conceptually based on the permanent income and full insurance models of consumption dynamics (Morduch 1995). These models argue that with access to insurance, credit, and/or liquid assets, prevalence of shocks (which hitherto will affect household income) will have a low or no correlation with household consumption (Bardhan and Udry 1999). In the absence of these mechanisms or where they are incomplete, households resort to various coping strategies (ex-ante and/or ex post) to manage and mitigate the impact of shocks and ensure consumption smoothing. Therefore, the effectiveness of these strategies can be examined through the extent to which exposure to shocks results in fluctuations in household consumption. Since we have one round of data, we rely on the cross-sectional variation in indicators of household welfare to understand the effect of idiosyncratic and covariate shocks on Sudanese households. In our analysis, idiosyncratic shocks are self-reported by household heads and include exposure to floods/droughts, loss of crops/livestock, illness, death of a household member, robbery/assault, and damage to household dwelling. We combine these data with those on rainfall (deviation from the long- run average), price (deviation of the predicted from the actual price level), and incidence of conflict/crime as proxies for covariate shocks. Using this approach, various indicators of household welfare such as consumption, education, asset accumulation, and nutrition levels are used to account for the multidimensionality of the impact of shocks. In the basic model, we estimate the following specification: ln( ) = + + + where is an indicator of household welfare such as consumption, food consumption score (FCS), dietary diversity, and poverty status for household in community/state ; is the vector of household characteristics such as demographics, assets, and employment for household in community/state ; is a vector of shocks; is a vector of community characteristics or fixed effects; and is the error term. Our main parameter of interest in this specification is . The extent to which shocks affect the welfare of Sudanese households can be examined by assessing whether is significantly different from zero. While this specification allows us to estimate the effect of shocks on household welfare, it does not allow us to capture the heterogeneity in the impact of shocks across space and households—how does the Page 27 of 49 impact of shocks vary across regions and household livelihoods? This is particularly useful for the analysis of vulnerability and resilience of households to shocks. Therefore, we augment this basic model by relating exposure to shocks with regions, household livelihoods, and other household characteristics related to vulnerability and resilience (such as intensity of using coping strategies and access to credit and help) as follows: ln( ) = + + ′ ′ + + The extent to which is significantly different from zero illustrates the degree of heterogeneity in the effect of shocks across household characteristics. We examine these results through the lens of vulnerability and resilience of Sudanese households to shocks. 3.2 Results 3.2.1 Impact of Shocks on Household Welfare In this subsection, we discuss the estimated impact of the effect of shocks on household welfare. To capture the multidimensionality of the impact of shocks, we consider various indicators of household welfare such as consumption per capita, the FCS, assets, poverty, and dietary diversity. The FCS is calculated by weighting the different food groups (cereals, meat/fish, fruits, oils, sugar, and so on) consumed by the household. This indicator combines indicators of dietary diversity, food frequency, and nutritional importance of food groups. Food groups such as meat/fish, eggs, and so on are allocated higher weights whereas food groups such as sugars, coffee/tea, mineral water, and so on receive lower weights. Therefore, higher values of the FCS imply that households consume more healthier food items. The household asset index is constructed by weighting different assets (such as motor vehicles, mobile phones, television, radio, and so on) owned by households. It is an indication of household wealth and accumulation of assets. Higher values indicate that households have more assets. Low dietary diversity is identified from the number of food groups consumed by the household. Households who consumed less than 7 of the 13 food groups have low dietary diversity. The coping strategy index (CSI) is a measure of the severity in the use of coping strategies by households faced with a shock, and it is constructed by weighting strategies used to mitigate the impact of shocks by the frequency of their use —typically the number of days in a week. Higher values of the CSI indicate an intense use of coping strategies when faced with a shock, hence a higher likelihood of food insecurity.11 Table 3: Estimating the Impact of Shocks on Welfare of Sudanese Households: Regression Results Ordinary Least Squares Logit (1) (2) (3) (4) (5) Log (Consumption Asset Household Low Dietary Variables FCS per Capita) Index Poverty Diversity Shock typea: Flood/drought −0.0530*** −1.933*** −4.915** 0.276** 0.606* (0.0140) (0.701) (2.111) (0.129) (0.349) Crop/livestock loss 0.00696 −0.190 -3.048* -0.0310 −0.427 11In constructing the FCS, CSI, and dietary diversity indicators, we relied on guidance provided by the World Food Programme (see https://www.humanitarianresponse.info/en/operations/afghanistan/document/guidance-note-calculation-household- food-security-outcome-indicators) Page 28 of 49 Ordinary Least Squares Logit (1) (2) (3) (4) (5) Log (Consumption Asset Household Low Dietary Variables FCS per Capita) Index Poverty Diversity (0.0137) (0.608) (1.609) (0.122) (0.388) Health shocks 0.0387** 1.229* -1.077 -0.152 −0.271 (0.0155) (0.698) (1.989) (0.141) (0.466) Dwelling damaged/robbery/fire -0.0151 0.678 -3.324 -0.0293 −0.294 (0.0141) (0.686) (2.027) (0.131) (0.410) Other shocks −0.00748 −0.210 −7.519*** 0.0322 −0.301 (0.0302) (1.445) (2.372) (0.235) (0.747) Covariate shocksb Log (number of conflict events in −0.418*** 13.22*** 0.874 3.671*** −2.403** 2014) (0.0506) (2.424) (4.096) (0.488) (1.204) Log (deviation of 2014 rainfall 0.549*** −23.62*** −1.621 −4.460*** 3.504** from historical average) (0.0726) (3.475) (6.233) (0.709) (1.777) Number of shocks reported 0.0353*** 0.697** 0.101 −0.259*** −0.228 (0.00781) (0.309) (0.631) (0.0678) (0.154) Household CSI −0.00591* 0.0574 0.396 0.0430 0.0428 (0.00345) (0.154) (0.379) (0.0301) (0.0668) Household size −0.0915*** 0.881*** −0.305 0.609*** −0.303*** (0.00213) (0.0853) (0.269) (0.0187) (0.0550) Household dependency ratio −0.0374*** 0.333* 1.886*** 0.216*** 0.0549 (0.00372) (0.179) (0.581) (0.0359) (0.0884) Household head age −0.00374** 0.0160 0.180 0.0302** −0.0460 (0.00153) (0.0687) (0.172) (0.0138) (0.0303) Household head age2 4.89e-05*** −0.000389 −0.00135 -0.000306** 0.000527* (1.50e-05) (0.000674) (0.00181) (0.000135) (0.000303) Male-headed household (=1) −0.00270 −1.463*** 3.247*** 0.0335 0.0366 (0.0118) (0.525) (1.191) (0.104) (0.232) Rural area (=1) -0.0984*** -6.174*** −0.771 −0.853*** 0.397* (0.00913) (0.420) (1.018) (0.0859) (0.222) Household head education level c Not stated −0.0298*** −2.661*** −1.667 0.219** 0.828*** (0.0107) (0.511) (1.295) (0.0965) (0.298) Primary education 0.0443*** 0.148 −0.0938 −0.325*** 0.0809 (0.0125) (0.603) (2.455) (0.117) (0.471) Secondary education 0.0730*** 1.570*** 1.124 −0.531*** 0.956** (0.0126) (0.602) (1.633) (0.122) (0.410) Post-secondary education 0.201*** 5.363*** 12.52*** −1.121*** 1.235** (0.0201) (0.854) (3.265) (0.195) (0.602) Khalwa −0.00546 −0.273 −2.583 −0.106 0.657 (0.0143) (0.675) (1.806) (0.130) (0.401) Household received help (=1) −0.0187** 1.682*** −3.911*** 0.179** −0.161 (0.00923) (0.438) (1.179) (0.0814) (0.212) Household borrowed money (=1) −0.00220 −1.083*** 0.0873 −0.194*** 0.0107 (0.00726) (0.336) (1.393) (0.0659) (0.179) Extreme poverty (=1) −0.451*** −9.647*** −4.395*** 1.961*** (0.00843) (0.491) (1.014) (0.230) Food insecure (=1) −0.257*** −4.434*** 1.142 0.641*** Page 29 of 49 Ordinary Least Squares Logit (1) (2) (3) (4) (5) Log (Consumption Asset Household Low Dietary Variables FCS per Capita) Index Poverty Diversity (0.0122) (0.617) (0.973) (0.232) Quintile of household asset indexd Poor 0.0449*** 4.483*** 6.642*** −0.268*** −0.498** (0.0107) (0.487) (0.457) (0.0928) (0.198) Middle 0.0804*** 5.753*** 12.87*** −0.777*** −0.894*** (0.0111) (0.502) (0.548) (0.101) (0.271) Rich 0.129*** 8.129*** 19.74*** −1.231*** −0.635** (0.0116) (0.529) (0.902) (0.106) (0.284) Richest 0.274*** 11.17*** 78.71*** −2.158*** −1.446*** (0.0138) (0.597) (2.164) (0.128) (0.425) Quintile of household livestock unitse Middle −0.0552*** −1.598*** −1.627 0.418*** −0.0475 (0.0102) (0.454) (1.036) (0.0940) (0.217) Rich −0.0448*** −0.492 −2.804*** 0.338*** −0.227 (0.0100) (0.460) (1.022) (0.0919) (0.230) Richest −0.0192* −0.199 −0.107 0.166* 0.0668 (0.0110) (0.503) (1.310) (0.0992) (0.225) Diversity of income activities −0.00280 0.854** 2.648 −0.0658 −0.186 (0.00731) (0.332) (1.986) (0.0630) (0.218) Household livelihoodf Animal husbandry 0.0982*** −1.387 −1.417 −0.415** 0.338 (0.0225) (0.905) (2.150) (0.182) (0.330) Wages and salaries 0.0203* 2.749*** −4.899*** −0.495*** −0.532* (0.0104) (0.481) (1.377) (0.0954) (0.275) Owned business enterprise 0.0492*** 2.220*** 2.769* −0.509*** −1.088*** (0.0114) (0.525) (1.547) (0.101) (0.319) Property income 0.0415* 0.275 2.619 −0.655*** −0.626 (0.0217) (0.910) (2.920) (0.188) (0.623) Remittances 0.127*** 2.968*** 2.366 −0.653*** −0.0940 (0.0221) (0.871) (3.003) (0.176) (0.399) Pension 0.0474 −1.730 19.63 −0.533 −1.318 (0.0420) (1.494) (30.07) (0.447) (0.965) Others -0.0267 −2.138*** −1.064 −0.176 −0.227 (0.0172) (0.821) (1.635) (0.148) (0.330) Constant 8.374*** 98.00*** −1.922 2.966** −8.949** (0.145) (6.926) (15.37) (1.441) (3.743) State fixed effects Yes Yes Yes Yes Yes Observations 11,596 11,596 11,596 11,596 11,105 R-squared 0.684 0.357 0.354 0.341 0.249 Notes: a. Base category for shock type = No shock reported; b. Data on conflict was obtained from the ACLED (https://www.acleddata.com/data/), whereas the rainfall data were obtained from the CHRS (http://chrsdata.eng.uci.edu/); c. Base category for the household head education level = No education; d. Base category for quintile of household assets = Poorest; e. Base category for quintile of household livestock units = Poor; f. Base category for household livelihood = Crop farming. Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. Page 30 of 49 Weather shocks have the most significant impact on Sudanese households. Although Sudanese households are exposed to various shocks, flood/drought-related shocks have the most adverse effects. By several measures of household welfare, households affected by flood/drought-related shocks experience significantly lower levels of welfare. Relative to households not affected by shocks, exposure to floods/droughts reduces consumption per capita, the FCS (a measure of healthy diets), and the asset level. Furthermore, households affected by floods/droughts are more likely to be poor and have low dietary diversity. Other shocks such as illness and death of a household member have a surprisingly a positive effect on household consumption per capita. This differential impact can possibly be explained by the fact that flood/drought-related shocks are more likely to be covariate in nature whereas health- related shocks are to a certain degree likely to be more idiosyncratic in prevalence. This distinction is likely to have profound effects in the estimated impact of shocks in communities where formal credit and/or insurance markets are missing and therefore households rely on semiformal institutions and informal arrangements to cope with shocks. Under these arrangements, incidence of shocks that are covariate in nature such as floods/droughts are less likely to be mitigated given that other households face similar shocks and hence lack the capacity to support other households in need. On the other hand, the impact of shocks that are idiosyncratic such as death of a household member and/or illness of household head can fully be mitigated through support received through these arrangements. Consumption per capita for households affected by floods/droughts decreases by 5.3 percent on average. In addition, these households are also more likely to consume less nutritious food items and less diverse diets. The incidence of flood/drought-related shocks decreases consumption per capita by 5.3 percent for affected households relative to those not affected.12 Flood/drought-related shocks appear to have the largest effect on household welfare compared to other shocks. The effect on other indicators of welfare such as the FCS and asset levels is equally negative. The FCS for households affected by shocks decreases by 1.933 points. Further analysis using a logit model in which the outcome variable is defined as having a low dietary diversity (defined as households who consumed less than 7 out of 13 food groups) also shows that the log odds of having low dietary diversity among households affected by flood/drought- related shocks is 0.606 higher than those not affected. Since the FCS is an index constructed from weights allocated to various food groups, it is difficult to interpret this result. However, since the weights are defined such that healthier food items such as meat, fish, and dairy products receive higher weights, higher FCS values could imply consumption of nutrient-rich food items as well as a more diverse diet. Thus, a negative effect of floods/droughts on the FCS may imply that affected households resort to consuming less nutritious food items and less diverse diets. This is in line with the observed decrease in consumption per capita for households affected by floods/droughts. Due to the incidence of such shocks, households lower their consumption per capita expenditure by consuming less nutritious food items and less diverse diets. Apart from consumption, floods/droughts deplete household assets and increase their likelihood of being poor. The asset index for households affected by floods/droughts decreases by 4.91 points relative to those not affected by shocks. Similarly, as with the FCS, the asset index is constructed by combining weights on assets owned by the households and the quantity they own such that higher weights are allocated to assets such as motor vehicles, motor cycles, televisions, and so on. As a result, it is difficult to 12Further comparison of this on food and non-food consumption indicates that the effect is larger on the latter. Food consumption per capita decreases by 4.3 percent whereas non-food consumption per capita decreases by 5.1 percent for households affected by flood/drought-related shocks. Page 31 of 49 interpret the estimated coefficient in the specification with the asset index as the outcome. However, since higher values imply wealthier households, a decrease in the asset index will imply a decrease in households’ stock on assets.13 Further analysis using households’ poverty status shows that floods/droughts increase households’ likelihood of being poor—floods/droughts increase the probability of affected households being poor by 0.5714 relative to those not affected The reduction in households’ assets and increased likelihood of being poor for households affected by flood/drought show the vulnerability of these households. The loss of assets could make it harder for these households to escape poverty, thereby trapping them in a cycle of intergenerational poverty. In terms of covariate shocks, increase in the number of conflicts has a negative effect on household welfare, particularly consumption per capita and poverty status. All else being equal, a 10 percent increase in conflict events reduces household consumption per capita by 4 percent on average. Similarly, the odds of being poor increases with conflict events—a 10 percent increase in conflict events increases the log odds of being poor by 36.7. Unlike idiosyncratic shocks such as floods/droughts, an increase in conflict events has a positive effect on the FCS and dietary diversity. The effect on household assets is however statistically insignificant. Rainfall in 2014 relative to historical levels had a positive effect on household welfare. All else being equal, consumption per capita for households in states with a 10 percent deviation of rainfall from historical averages, increases by 5.5 percent. The effect on indicators of dietary diversity is mixed- a similar change decreases household FCS and increases the logs of households having low dietary diversity. The positive effect of a positive deviation of rainfall from historical averages is possibly driven by the fact that rainfall levels in 2014 were significantly better than the previous three years, but lower than the 2007 rainfall levels which caused floods. Given that majority of households rely on rain-fed agriculture as a source of livelihood, improvements in rainfall without causing floods improves their welfare.These results highlight that both idiosyncratic and covariate shocks affect the welfare of Sudanese households. Distinguishing between individual idiosyncratic and covariate shocks as well as various indicators of household welfare further reveals the shocks with the largest impact on the welfare of Sudanese households. In terms of idiosyncratic shocks, floods/droughts have the largest effect on household welfare, particularly household consumption, assets, and poverty status. The reduction in consumption per capita occurs in both food and non-food consumption. Reduction in food consumption is likely to occur in the form of consuming less nutritious food items and less diverse diets as indicated in the effect of shocks on the FCS and dietary diversity. The reduction in overall consumption per capita is likely to be driven by further reduction in non-food items. The effect on assets and poverty status, which illustrates the increased likelihood of households affected by flood/drought to remain poor in addition to depletion of assets, further reiterates the need for effective social protection programs. Therefore, it is imperative to consider these insights in designing and implementing such programs to maximize impact. 3.2.2 Heterogeneity of the Impact of Shocks on Household Welfare The previous subsection discussed the average effect of shocks among Sudanese households. In this subsection, we focus on the heterogeneity of such effects across regions in Sudan as well as across 13To further aid in the interpretation, the weights are defined such that expensive assets such as television, air conditioner, and so on are weighted with a value of 4, whereas assets such as radio, fan, and so on are weighted with a value of 2. On the basis of this construction, a decrease of 5 points may represent a somewhat substantial decrease in household assets. 14 This is arrived at by converting the log odd ratios from the logit estimation to their probability equivalents. Page 32 of 49 households based on household livelihoods. In addition, we also examine the heterogeneity in the effect of shocks across households based on the intensity of using coping strategies, help, and loans received to pin down the effectiveness of these responses in cushioning the impact of shocks. The objective of this exercise is to draw insights on the vulnerability and resilience of Sudanese households by identifying vulnerable households from the regions and livelihoods most affected by shocks and the extent to which the use of coping strategies enhances resilience by mitigating the impact of shocks. Table 4: Estimating the Heterogeneity of the Impact of Shocks on Household Consumption Per Capita (1) (2) (3) (4) (5) Variables Region Household Use of Coping Borrowed Received Livelihood Strategy Help Household affected by shock = 1 0.0207 0.0252* −0.0717*** −0.0258** −0.00738 (0.0158) (0.0150) (0.0246) (0.0131) (0.0118) Interaction of shocks and Regiona Darfur −0.0369* (0.0206) Eastern 0.0272 (0.0227) Khartoum −0.0975*** (0.0282) Kordofan −0.00332 (0.0208) Northern −0.0480* (0.0250) Interaction of shocks and household livelihoodb Animal husbandry −0.00418 (0.0448) Wages and salaries −0.0423** (0.0171) Owned business enterprise −0.0698*** (0.0212) Property income 0.0950** (0.0441) Remittances −0.0374 (0.0394) Pension −0.0461 (0.0762) Others −0.0649** (0.0327) Interaction of shocks and household coping strategyc Middle 0.119*** (0.0351) High 0.194** (0.0825) Highest 0.196*** (0.0356) Any shock (=1) x borrowed 0.0377*** money (=1) Page 33 of 49 (1) (2) (3) (4) (5) Variables Region Household Use of Coping Borrowed Received Livelihood Strategy Help (0.0141) Any shock (=1) x help received −0.00822 (=1) (0.0178) Animal husbandry 0.0968*** 0.107*** 0.0986*** 0.101*** 0.101*** (0.0225) (0.0322) (0.0227) (0.0227) (0.0227) Wages and salaries 0.0208** 0.0433*** 0.0197* 0.0211** 0.0208** (0.0102) (0.0133) (0.0104) (0.0104) (0.0104) Owned business enterprise 0.0415*** 0.0832*** 0.0483*** 0.0495*** 0.0489*** (0.0113) (0.0156) (0.0114) (0.0114) (0.0114) Property income 0.0452** 0.0180 0.0446** 0.0458** 0.0455** (0.0222) (0.0260) (0.0220) (0.0219) (0.0219) Remittances 0.125*** 0.147*** 0.125*** 0.126*** 0.126*** (0.0222) (0.0293) (0.0221) (0.0221) (0.0222) Pension 0.0439 0.0720 0.0455 0.0492 0.0480 (0.0424) (0.0581) (0.0416) (0.0418) (0.0418) Others -0.0216 0.00604 -0.0278 -0.0274 -0.0264 (0.0170) (0.0229) (0.0171) (0.0172) (0.0172) Number of shocks reported by 0.0327*** 0.0332*** 0.0379*** 0.0349*** 0.0362*** household (0.00787) (0.00797) (0.00791) (0.00797) (0.00797) Household CSI −0.00337 −0.00515 −0.00524 −0.00534 (0.00341) (0.00347) (0.00349) (0.00348) Household size −0.0899*** −0.0916*** −0.0918*** −0.0918*** −0.0917*** (0.00213) (0.00214) (0.00214) (0.00214) (0.00214) Household dependency ratio - # -0.0377*** -0.0376*** -0.0378*** −0.0376*** −0.0376*** of (0-14+>64)/15−64 (0.00374) (0.00371) (0.00371) (0.00371) (0.00372) Household head age −0.00393** −0.00392** −0.00386** −0.00380** −0.00384** (0.00153) (0.00153) (0.00153) (0.00153) (0.00153) Household head age2 5.25e-05*** 5.09e-05*** 5.02e-05*** 4.98e-05*** 5.01e-05*** (1.49e-05) (1.50e-05) (1.49e-05) (1.49e-05) (1.50e-05) Male-headed household −0.00549 −0.00698 −0.00625 −0.00665 −0.00624 (0.0118) (0.0118) (0.0118) (0.0118) (0.0118) Rural area (=1) −0.0927*** −0.0993*** -0.0986*** −0.0991*** −0.0989*** (0.00914) (0.00912) (0.00910) (0.00911) (0.00913) Household head education level Not stated −0.0412*** −0.0318*** −0.0313*** −0.0319*** −0.0313*** (0.0108) (0.0107) (0.0107) (0.0107) (0.0108) Primary education 0.0425*** 0.0441*** 0.0444*** 0.0440*** 0.0444*** (0.0125) (0.0125) (0.0126) (0.0125) (0.0125) Secondary education 0.0711*** 0.0745*** 0.0728*** 0.0724*** 0.0730*** (0.0127) (0.0126) (0.0126) (0.0127) (0.0127) Post-secondary education 0.196*** 0.203*** 0.202*** 0.202*** 0.203*** (0.0201) (0.0201) (0.0201) (0.0202) (0.0202) Khalwa −0.0179 −0.00644 −0.00603 −0.00684 −0.00632 (0.0143) (0.0143) (0.0144) (0.0144) (0.0144) Household received help (=1) −0.0136 −0.0160* −0.0159* −0.0153* (0.00908) (0.00924) (0.00923) (0.00923) Page 34 of 49 (1) (2) (3) (4) (5) Variables Region Household Use of Coping Borrowed Received Livelihood Strategy Help Household borrowed money 0.00207 −0.00221 −0.00200 −0.00234 (=1) (0.00722) (0.00728) (0.00729) (0.00729) Extremely poor (=1) −0.463*** −0.451*** −0.451*** −0.451*** −0.451*** (0.00835) (0.00845) (0.00842) (0.00844) (0.00844) Food insecure (=1) −0.271*** −0.257*** −0.257*** −0.259*** −0.258*** (0.0120) (0.0121) (0.0121) (0.0121) (0.0121) Quintile of household asset index Poor 0.0510*** 0.0433*** 0.0435*** 0.0432*** 0.0434*** (0.0107) (0.0107) (0.0107) (0.0107) (0.0107) Middle 0.0911*** 0.0789*** 0.0792*** 0.0799*** 0.0799*** (0.0110) (0.0111) (0.0111) (0.0111) (0.0111) Rich 0.139*** 0.129*** 0.129*** 0.129*** 0.129*** (0.0115) (0.0115) (0.0115) (0.0115) (0.0115) Richest 0.279*** 0.273*** 0.275*** 0.274*** 0.275*** (0.0137) (0.0138) (0.0138) (0.0138) (0.0138) Quintile of household livestock units Middle −0.0643*** −0.0558*** −0.0554*** −0.0548*** −0.0550*** (0.0104) (0.0103) (0.0103) (0.0103) (0.0103) Rich −0.0521*** −0.0424*** −0.0440*** −0.0432*** −0.0433*** (0.0100) (0.0101) (0.0101) (0.0100) (0.0100) Richest −0.0204* −0.0191* −0.0192* −0.0181 −0.0185* (0.0111) (0.0111) (0.0111) (0.0111) (0.0111) Diversity of income activities −0.00198 −0.00359 −0.00310 −0.00315 −0.00338 (0.00732) (0.00738) (0.00736) (0.00734) (0.00735) Log (number of conflicts in −0.0141*** −0.437*** −0.431*** −0.443*** −0.445*** 2014) (0.00375) (0.0505) (0.0504) (0.0502) (0.0503) Log (deviation of 2014 rainfall 0.00259 0.577*** 0.570*** 0.588*** 0.590*** from historical avg) (0.00766) (0.0723) (0.0722) (0.0720) (0.0721) Regions Darfur −0.0344* (0.0194) Eastern −0.0150 (0.0141) Khartoum 0.0926*** (0.0218) Kordofan −0.0240 (0.0161) Northern 0.00182 (0.0221) Quintile of household coping strategy Middle −0.0276 (0.0251) High −0.150* (0.0794) Page 35 of 49 (1) (2) (3) (4) (5) Variables Region Household Use of Coping Borrowed Received Livelihood Strategy Help Highest −0.169*** (0.0263) Household borrowed money −0.0194* (=1) (0.00997) Household received help (=1) −0.0113 (0.0140) Constant 9.372*** 8.310*** 8.337*** 8.308*** 8.299*** (0.0576) (0.145) (0.144) (0.144) (0.144) State fixed effects — Yes Yes Yes Observations 11,596 11,596 11,596 11,596 11,596 R-squared 0.678 0.683 0.683 0.683 0.682 Adjusted R-squared 0.676 0.682 0.682 0.681 0.681 Notes: a. Base category for regions is central; b. Base category for household livelihood is crop farming; c. Base category for quintile of household CSI is low. In each specification, we use a binary indicator for the incidence of shocks (=1 if household reported at least one shock, and zero otherwise). This variable is correlated to region (a categorical variable indicating the six regions of Sudan) in column 1; household livelihood in column 2; four categories indicating the intensity in the use of coping strategy (constructed from the household CSI) in column 3; and binary variables indicating households’ receipt of credit and help in columns 4 and 5, respectively. We focus on household consumption per capita as our main indicator of household welfare. Each specification is estimated using ordinary least squares. Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. Households in the Central15 region affected by shocks are least affected when compared to the impact on households in Khartoum, Northern, and Darfur regions. Consumption per capita for households affected by at least one shock in Khartoum is on average 9.6 percent lower than consumption per capita for affected households in the Central region.16 In the Northern and Darfur regions, consumption per capita of households exposed to at least one shock is 4.9 percent and 3.7 percent lower, respectively, than in the Central region. Although these results illustrate the spatial heterogeneity in the impact of shocks across Sudan, differences in prevalence of shocks across states, region-specific characteristics such as the types of shocks most prevalent, and the common household livelihoods are likely to drive these results. For instance, as shown in Figure 5, prevalence of shocks in the Central region is fairly low compared to other regions. Also, in the Khartoum and Northern regions, floods/droughts (which as discussed earlier have the most negative impact on household welfare) were the most commonly reported shock in 2014/15 (see Figure 6b and Figure 6d). In addition, wages and salaries are the common livelihoods in these regions based on the 2014/15 data and as discussed below are among the livelihoods most affected by shocks. In the Darfur region, on the other hand, the high impact of shocks is possible driven by the high prevalence of shocks (around 60 percent of households reported at least one shock in 2014/15 as shown in Figure 5) and the dominance of crop farming as the main livelihood. 15 The Central region consists of Al-Gezira, White Nile, Sennar, and Blue Nile states. The Darfur region consists of the North, West, South, Central, and East Darfur states. The Eastern region consists of the Red Sea, Kassala, and Al-Gadarif states. The Khartoum region consists of the state of Khartoum. The Kordofan region consists of the North, South, and West Kordofan states. The Northern region consists of the Northern and River Nile States. 16 Although not presented here in the interest of brevity, the impact of shocks is larger on urban households than on rural households. Consumption per capita for rural households affected by a shock is on average 3.4 percent higher than urban households also affected by a shock. Page 36 of 49 In terms of household livelihoods, exposure to idiosyncratic shocks lowers consumption per capita for wage/salary earners and business owners relative to crop farmers. Consumption per capita for property owners exposed to a shock on the other hand increases relative to crop farmers. Apart from property owners, exposure to shocks lowers consumption per capita for all household livelihoods relative to crop farmers. The magnitude of the decrease can be used to identify households most likely to be affected by shocks. Consumption per capita for business owners and wage/salaried households affected by shocks is, respectively, 7 percent and 4 percent lower than that of crop farmers affected by a shock. This result further highlights the vulnerability of households in the Khartoum, Northern, and Darfur regions where wage/salaries and crop farming are common livelihoods. Households whose main livelihood is property income appear to be least affected by shocks. Consumption per capita of those affected by shocks is 9.5 percent higher than crop farmers affected by shocks. Therefore, the impact of idiosyncratic shocks is higher on wage/salary earners and business owners than on crop farmers. This is perhaps because crop farmers can be insured from fluctuations in consumption in the event of a shock by consuming farm harvests. Property owners, on the other hand, appear to be least affected when compared to the impact on crop farmers. The possibility that property income (unlike agricultural income, business revenue, and wages/salaries) is less likely to be volatile is what insures households whose main livelihood is property income from the negative effect of shocks. Increase in use of coping strategies and access to credit appears to effectively mitigate the impact of shocks on household welfare. More intense use of coping strategies significantly lowers the impact of shocks on household welfare. Consumption per capita for households in the top 40 percent of the CSI distribution and affected by at least one shock is 19 percent higher than for households equally affected by shocks but in the bottom 40 percent of the same distribution.17 Similarly, households who were affected by shocks but were able to borrow money are able to increase their consumption per capita by 3.8 percent relative to those affected by shocks but did not borrow money. However, support provided by friends/family/government/NGOs does not significantly improve the welfare of households affected by shocks; the interaction of incidence of shocks and receipt of support is statistically and economically insignificant. Therefore, households who do not have the capacity (whether in the form of assets or savings) or those who are credit constrained are likely to be less resilient to shocks. Exposure to shocks for these households may significantly affect household welfare. There appears to be high returns on education in Sudan. Across all specifications, consumption per capita is nearly 4 percent, 7 percent and 19 percent higher for households whose heads have some primary, secondary and post-secondary education respectively than households whose heads have no education. These differences in welfare across households driven by education levels of household heads shows that the importance of human capital accumulation. Higher levels of education and the resulting increase in access to economic opportunities can increase welfare and strengthen the resilience of households to shocks. Thus, while existing social protection programs may be inadequate, these results illustrate the effectiveness of human capital accumulation in strengthening the resilience of households to shocks. Therefore, ongoing reforms to restore economic stability and economic growth can complement existing 17Further analysis focusing on the effectiveness of using savings and selling assets in response to shocks indicates that consumption per capita for households using these coping strategies is 26 percent higher than for households using other or no coping mechanisms. Page 37 of 49 social protection programs to lower vulnerability and build resilience by increasing labor market demands and job opportunities. While these results indicate that the use of coping mechanisms (such as savings or assets) appears to cushion the impact of shocks in the short run, the long-run consequences of depleting stock of assets and savings may affect households’ ability to escape poverty, particularly when shocks are persistent or households live in communities frequently affected by shocks. It is for this reason that there is a need for effective social protection programs on which households affected by shocks can rely instead of using their assets and savings or resorting to extreme responses that may significantly affect their welfare. Page 38 of 49 4. Coping with and Resilience to Shocks The previous section shows that shocks have negative effects on the welfare of Sudanese households. In the absence of formal insurance, credit, and/or effective social protection programs, households often rely on various strategies to cope with shocks. These include ex ante strategies to manage the risk of exposure to shocks (such as crop choices in the case of agricultural households) as well as ex post strategies used to mitigate the impact of shocks when they occur (such as selling assets, dissaving, increasing labor supply, reducing consumption, removing children from school, and so on). Households also rely on other informal strategies such as help from family members, friends, and so on. The second set of results presented in this paper showed that coping strategies used by Sudanese households in response to shocks mitigate the impact of shocks. However, the type and extent to which these strategies are used can provide insights into the resilience of households to shocks. For instance, the use of severe coping strategies such as selling assets, using savings, reducing expenses, and so on (which as discussed earlier are common in Sudan) may be able to mitigate the impact of shocks in the short term—but often at the expense of weaker resilience to future shocks. For poor farming households who are vulnerable to frequent agriculture and weather-related shocks, this relationship between the use of coping strategies and resilience to shocks may have significant effects on their ability to escape poverty. In this section, we describe the coping strategies used by Sudanese households in response to shocks. We draw insights from the use of these strategies to highlight their implications on the resilience of Sudanese households to shocks (particularly the most vulnerable) and the need for social protection reforms. 4.1 Household Coping Strategies Informal systems of support between households is a common coping strategy in Sudan, particularly among households affected by death of a household member or loss of livestock. Help received from other households is the most common coping strategy among households in Sudan. For instance, nearly 60 percent of households affected by death of a household member in 2009 and more than 40 percent of households affected by similar shocks in 2014/15 received help from other households (see Error! Reference source not found.). Similarly, 41 percent and 36 percent of households who lost livestock in 2009 and 2014/15, respectively, relied on help received from other households. Livestock ownership is a common buffer against shocks in most developing countries, particularly farming households who use them on their farms to draw farm implements, as a means of transport, as a source of food, and as liquid assets. As such, loss of livestock results in a significant depletion in households’ capacity to absorb the effect of shocks. In Sudan, these households resort to relying on informal systems of support to cope with such losses. Households affected by fire and robbery/assault also rely on these informal coping strategies. The idiosyncratic nature of these shocks makes it possible for affected households to rely on help from other households. Selling/renting assets, using savings, and increasing labor supply are common coping strategies for households affected by floods/droughts. The most common coping strategy for households affected by floods/droughts is selling assets or using savings or renting out farms. In both 2009 and 2014/15, more than 30 percent of households affected by floods/droughts resorted to selling assets, dissaving, or renting out their farms. Another common coping strategy for households affected by droughts/floods is increase Page 39 of 49 in labor supply (see Error! Reference source not found.). The covariate nature of shocks such as floods/droughts makes it difficult for affected households to rely on other coping strategies such as help from other households. Therefore, the common use of assets or savings and increasing labor supply in response to floods/droughts is in line with empirical evidence. Households affected by severe illness/accidents often cope by selling assets or using savings and borrowing money. Like households affected by floods/droughts, the use of assets or savings is a common coping strategy for households affected by severe illness/accidents. More than 40 percent of households affected by these shocks in 2009 and more than 30 percent in 2014/15 reported that they resorted to selling their assets or using their savings to cope with the impact of the shock. Another common coping strategy for these households is borrowing money—almost 20 percent of households affected by illness of a household member reported that they coped with the shock by borrowing money (see Error! Reference source not found.). Therefore, in addition to informal systems of support between households, other informal arrangements such as informal credit are equally likely to play a significant role in mitigating the impact of shocks in Sudanese communities. Given that support between households in times of need is a common coping strategy in Sudan, the design and implementation of social protection programs must identify and incorporate the roles of informal institutions to effectively lower vulnerability. Similarly, given the common use of assets and savings, the state of asset markets and the extent of financial inclusion, particularly for vulnerable households, should be considered. Page 40 of 49 Figure 11: Shocks in Sudan: Risk-Coping Strategies, 2009 and 2014/15 (a) 2009 70% 60% 50% 40% 30% 20% 10% 0% Sell Assets/Use Savings/Rent Farm Increase Labor Supply Borrowed Money Received Help Reduced Expenses or Consumption Kids Migrate or removed from School Other (c) 2014/15 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Sell Assets/Use Savings/Rent Farm Increase Labor Supply Borrowed Money Received Help Reduced Expenses or Consumption Kids Migrate or removed from School Other Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Page 41 of 49 In addition to coping strategies, households also rely on help provided by NGOs, the government, and family/friends to mitigate the impact of shocks. We examine the different types of help provided to households affected by shocks. The poorest households in Sudan are the highest receivers of food aid, government benefits, and help from NGOs and Zakat centers, whereas the richest households receive most help from non-family members. Households in the bottom 20 percent of the consumption per capita distribution receive most help in the form of food aid. In Error! Reference source not found.a and Error! Reference source not found.b, it is can be observed that 26 percent and 32 percent of the poorest households received food aid in 2009 and 2014/15, respectively. This compares to 6 percent and 5 percent of the richest households in the same periods. A similar trend is also observed in the provision of government benefits and support from NGOs and Zakat Centers—the poorest households are the largest beneficiaries. While this shows that Sudan’s current SSN programs are poverty responsive, integrating these programs with DRM strategies to ensure shock-responsive is necessary to strengthen resilience. On the other hand, help from non-household members is commonly provided to richer households—63 percent and 57 percent of the richest households received help from non-family members in 2009 and 2014/15, respectively, compared to 27 percent and 19 percent to the poorest households in the same periods. For poor households, access to support from governments, NGOs, and other institutions can minimize the use of emergency coping strategies which may result in poverty traps. Agricultural households are among the largest beneficiaries of food aid programs and government support, whereas households relying on aid and remittances receive a lot of help from non-household members. Across household livelihoods, some variation is observed about help received. For instance, the percentage of households who received food aid is highest among agricultural households (crop farmers and households engaged in animal husbandry) (see Error! Reference source not found.c and Error! Reference source not found.d). On the other hand, households relying on remittances, pension, and aid are the largest beneficiaries of help from non-household members. For agricultural households, given their high vulnerability to shocks (particularly agricultural related shocks) and low resilience (due to their low capacity to fully mitigate the impact of shocks), receipt of help that is effective to cope with shocks can significantly lower their risk of being trapped into poverty. Page 42 of 49 Figure 12: Help Received Quintiles of Consumption Per Capita (a) 2009 (b) 2014/15 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Poorest Poor Middle Rich Richest Poorest Poor Middle Rich Richest Food Aid Received Gov't Benefits Food Aid Received Gov't Benefits Other NGO Non Hh. Members Other NGO Non Hh. Members Other Groups Other Groups Zakat Center Household Livelihoods (c) 2009 (d) 2014/15 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Food Aid Received Gov't Benefits Food Aid Received Gov't Benefits Other NGO Non Hh. Members Other NGO Non Hh. Members Other Groups Other Groups Zakat Center Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. 4.2 Resilience of Sudanese Households to Shocks Drawing insights from the previous section on coping strategies, this paper examines the resilience of Sudanese households to shocks. For this analysis, resilience to shocks is examined through the lens of Page 43 of 49 households’ ability to cushion the impact of shocks based on the severity of current coping strategies, households’ ability to replenish stock of assets and/or savings used to mitigate current shocks, and the belief that shocks can reoccur.18 As a result, households that use severe coping strategies such as selling productive assets or dissaving may be unable to mitigate the impact of repeated shocks and thus become less resilient. Based on the coping strategies discussed in the previous section, we construct a household CSI by weighting the individual strategies reported by the households based on their severity. For instance, strategies such as selling assets, dissaving, or borrowing money carry a higher weight than others such as reducing consumption, cutting on number of meals, or relocating consumption among household members. Therefore, higher values of the CSI indicate an intense use of coping strategies when faced with a shock, hence a higher likelihood of less resilience. Poor households are likely to be less resilient. The average CSI for poor households was 3.9 in 2009 compared to 3.57 for non-poor households. In 2014/15, the average for poor households was higher than for non-poor households—2.93 compared to 2.72. Similarly, households in the bottom 40 percent of the consumption per capita distribution had the highest CSI. This implies that poor Sudanese households use more severe coping strategies to mitigate the impact of shocks. In the context of resilience to shocks, the use of severe coping strategies such as selling assets may be triggered by the low capacity to use less severe coping strategies such as increasing labor supply, petty trading, and borrowing to mitigate the impact of shocks. Therefore, to improve the resilience of poor Sudanese households, access to markets and financial institutions needs to be improved to reduce the use of severe coping strategies in the event of a shock that may trap them into poverty. In terms of household livelihood, agricultural households and those relying on remittances are likely to be less resilient. Crop farmers in Sudan use the most severe coping strategies. The average CSI for crop farmers was 4.25 in 2009 and 3.16 in 2014/15. Severe coping strategies are also observed among households relying on remittances and animal husbandry. For agricultural households, assets such as farm land, seeds, and livestock are important inputs in their livelihood. Selling these assets in the event of a shock (particularly given that a majority of these households rely on rain-fed farming and variations in rainfall exposes them to significant risk) may affect their productivity and welfare. However, in the absence of formal insurance and/or credit, social protection programs, or programs to build resilience, they can only resort to selling these assets to mitigate the impact of shocks. Across states, households in the Darfur, Kordofan, and parts of the Nile regions appear to be less resilient. The use of severe coping strategies in these regions is in line with results described earlier. The main livelihood of households in these regions is agriculture related (Figure 6). Thus, the observation that agricultural households are likely to be less resilient to shocks is shown across household livelihoods and across states. In states such as Khartoum, where wages and salaried activities are the main livelihood, coping strategies used to mitigate shocks are less severe, indicating better resilience to shocks. 18Both rounds of data indicate that agriculture- and weather-related shocks were common in Sudan in both 2009 and 2014/15. Frequent occurrence of these shocks may weaken resilience of households—particularly agricultural households who often rely on selling their assets to cope with shocks. Page 44 of 49 Figure 13: Household CSI (a) Poor versus Non-Poor (b) Quintile of Consumption Per Capita 4.50 4.50 3.90 3.88 3.96 4.00 4.00 3.71 3.67 3.57 3.42 3.50 3.50 2.93 2.94 2.91 3.00 2.72 3.00 2.79 2.70 2.62 2.50 2.50 2.00 2.00 1.50 1.50 1.00 1.00 0.50 0.50 0.00 0.00 Poorest Poor Middle Rich Richest Non-Poor Poor 2009 2014/15 2009 2014/15 (c) Household Livelihood (d) States 4.50 6.00 4.00 5.00 3.50 3.00 4.00 2.50 2.00 3.00 1.50 2.00 1.00 0.50 1.00 0.00 0.00 Central Darfur Kassala Khartoum Al-Gadarif White Nile Blue Nile Western Darfur River Nile Al-Gezira Northern West Kordufan Red Sea Sinnar Northern Darfur East Darfur Northern Kordofan Southern Darfur Southern Kordofan 2009 2014/15 2009 2014/15 Source: Authors’ calculation using 2009 NBHS and 2014/15 NHBPS data. Page 45 of 49 5. Conclusion and Policy Implications In this paper, we discussed the incidence of shocks in Sudan by combing two household survey datasets collected in 2009 and 2014/15. By combining both rounds of data, we were able to highlight the main shocks affecting Sudanese households, the coping strategies available to and used by these households as the heterogeneity in the incidence of shocks across space (rural/urban, states, and so on), and household attributes such as livelihoods and poverty status. Drawing insights from these descriptions (such as the type and intensity of coping strategies used, the category of households mostly affected by shocks), we profile households who are likely to be resilient or vulnerable to shocks. In the second part of the paper, we used the latest round of the data to estimate the impact of shocks on the welfare of Sudanese households. Since the impact of shocks is likely to vary across specific types of shocks and across indicators of household welfare, we consider this multidimensionality in our analysis. We also consider the heterogeneity in the impact of shocks across households distinguished by space, livelihood, use of coping mechanisms, and access to credit and informal systems of support. The results indicate that poor agricultural rural households are most exposed to idiosyncratic shocks in general and agricultural-related shocks such as crop or livestock loss. Most of these households rely on assistance provided by family/friends, NGOs, or the government as well as on selling their assets or dissaving to cope with shocks. The fact that these strategies may be insufficient to fully mitigate the impact of shocks and enable households to escape poverty shows the need for social protection programs to reduce vulnerability and build resilience. In terms of the impact of shocks, weather-related shocks such as floods/droughts have the largest negative impact on household welfare. On average, households affected by floods/droughts experience a 5.3 percent decrease in consumption per capita relative to those not affected by shocks. They also experience a significant decrease in their assets and are more likely to be poor and have a low dietary diversity. The effect of shocks on consumption per capita is particularly more pronounced among households in Khartoum, Northern, and Darfur regions of Sudan (where flood/drought-related shocks are commonly reported) and business owners and wage/salary-dependent households. The large negative effect of floods/droughts is possibly driven by the fact that Sudanese households rely on help provided by family members, friends, NGOs, or the government and the sale of assets or dissaving to mitigate the impact of shocks. Since floods/droughts are likely to be covariate in nature and likely to result in the destruction assets, the mere exposure to these shocks equally reduces the capacity of the affected households to mitigate the impact. Thus, although the use of these coping mechanisms can reduce the impact of shocks on household welfare, it is likely to be temporal and insufficient. Furthermore, the long- term consequences of these strategies such as the possibility of poverty traps show the need for social protection programs. The results from the analysis have policy implications. First, ongoing reforms to restore macroeconomic stability and spur growth must be reinforced to strengthen the resilience of Sudanese households. These reforms must continue to tackle emerging challenges such as rising prices and volatile exchange rates; increase economic opportunities such as jobs; and ensure the provision of social services such as education, skills and capacity building. For agricultural households who face significant exposure to rainfall related shocks, these reforms will strengthen their resilience by reducing income fluctuations and facilitating the diversification of economic activities. They Page 46 of 49 will also complement existing social protection programs to reduce vulnerability of households to shocks and reduce incidences of poverty traps. Secondly, although expansion of existing social protection programs to increase coverage may be constrained by the current macroeconomic environment and government’s limited resources, the integration of these programs into DRM strategies is necessary to strengthen the resilience of Sudanese households by increasing their responsiveness to shocks. Well-targeted social safety net programs tailored with effective disaster management can help mitigate the impact of shocks on the poor and vulnerable groups- particularly in countries like Sudan where weather-related shocks have larger negative effects on household welfare. Existing social protection programs appear to be poverty-responsive and are insufficient. Coping strategies used by households to complement these programs in the event of a shock such as selling assets, using savings and increasing labor supply are not only weak but may also increase the likelihood of poverty traps. By integrating social protection programs with DRM strategies, their effectiveness can be improved through raising awareness on resilience building and lowering risk of exposure to shocks. Finally, the role of the agricultural sector as a main source of livelihood for most households particularly in rural areas means that ongoing reforms to ensure economic stability and growth must also include the agricultural sector. Given the prevalence and recurrence of weather shocks, and the associated welfare impacts on households, one way to mitigate their negative impact is to develop agricultural institutions such as extension services to build farmers’ capacity in early detection of crop pests and livestock diseases and promote the adoption of drought-resistant crops. These reforms can be complemented by informal and semi-formal institutions such as social networks to increase the diffusion of climate-smart and pest- resilient agricultural practices. With a large share of Sudanese farmers susceptible to crop and livestock losses and reduced yields because of pests, recurring droughts, and irregular rainfall, drought-resistant and high-yield-certified seeds should help mitigate the impact of such risks and smooth year-to-year production and household welfare. Farm households that are vulnerable to weather shocks such as insufficient and erratic rainfall, particularly in arid areas where there are limited alternatives for irrigation, can significantly benefit from these interventions. The vulnerability of agricultural households results from their difficulty to cope with shocks due to lack of assets and limited access to credit and insurance markets. As a result, they resort to selling their few assets to mitigate the immediate impact of shocks while compromising their expected future income. Such coping strategies are generally inefficient and result in a vicious circle of chronic poverty. Given the macroeconomic challenges the GoS currently faces, financial and technical support will be crucial in implementing these reforms. Ensuring economic stability, improving the targeting of social protection programs, and reforming the agricultural sector while facing significant budget constraints requires financial resources, technical support and humanitarian assistance. With the support of institutions such as the World Bank Group, implementing these reforms will significantly contribute towards reducing poverty and inequality by lowering vulnerability and building resilience of Sudanese households. Page 47 of 49 References Bardhan, P., and C. Udry. 1999. Development Microeconomics. New York: Oxford University Press. Central Bureau of Statistics. 2017. Dissemination of the Sudan Household Budget and Poverty Survey. Sudan Central Bureau of Statistics. Dasgupta, P. 1993. An Inquiry into Well-Being and Destitution. Oxford: Clarendon Press. Deaton, A. 1992. Understanding Consumption. Oxford: Oxford University Press. Dercon, S. 1996. “Risk, Crop Choice, and Savings: Evidence from Tanzania.” Economic Development and Cultural Change 44 (3): 485–513. ———. 2002. “Income Risk, Coping Strategies, and Safety Nets.” The World Bank Research Observer 17 (2): 141–166. ———. 2004. “Growth and Shocks: Evidence from Rural Ethiopia.” Journal of Development Economics 74: 309–329. Dercon, S., and P. Krishnan. 2000. Vulnerability, Seasonality, and Poverty in Ethiopia. Journal of Development Studies 36 (6): 25–53. Harrower, S., and J. Hoddinott. 2005. “Consumption Smoothing in the Zone Lacustre, Mali.” Journal of African Economies 14 (4): 489–519. Hoddinott, J. 2006. “Shocks and Their Consequences Across and within Households in Rural Zimbabwe.” Journal of Development Studies 42 (2): 301–321. Hoddinott, J., and B. Kinsey. 2001. “Child Growth in the Time of Drought.” Oxford Bulletin of Economics and Statistics 63: 406–436. Jacoby, H., and E. Skoufias. 1997. “Risk, Financial Markets, and Human Capital in a Developing Country.” Review of Economic Studies 64 (3): 311–335. ———. 1998. “Testing Theories of Consumption Behavior Using Information on Aggregate Shocks: Income Seasonality and Rainfall in Rural India.” American Journal of Agricultural Economics 80 (1): 1–14. Jalan, J., and M. Ravallion. 1999. “Are the Poor Less Well Insured? Evidence on Vulnerability to Income Risk in Rural China.” Journal of Development Economics 58 (1): 61–81. Kjellgren, A., C. Jones-Pauly, A. H. El-Tayeb, E. Tadesse, and A. Vermehren2014. Sudan Social Safety Net Assessment. Washington, DC: World Bank. Kochar, A. 1998. “Smoothing Consumption by Smoothing Income: Hours-of-Work Responses to Idiosyncratic Agricultural Shocks in Rural India.” The Review of Economics and Statistics 81 (1): 50–61. Morduch, J. 1995. “Income Smoothing and Consumption Smoothing.” Journal of Economic Perspectives 9 (3): 103–114. Page 48 of 49 Paxson, C. 1992. “Using Weather Variability to Estimate the Response of Savings to Transitory Income in Thailand.” American Economic Review 82 (1): 15–33. Porter, C., and R. Hill. 2015. “Shocks, Safety-nets and Vulnerability to Poverty in Ethiopia.” Unpublished Manuscript. Rosenzweig, M. R., and H. Binswanger. 1993. “Wealth, Weather Risk, and the Composition and Profitability of Agricultural Investments.” Economic Journal 103 (1): 56–78. Skoufias, E., and A. R. Quisumbing. 2005. “Consumption Insurance and Vulnerability to Poverty: A Synthesis of the Evidence from Bangladesh, Ethiopia, Mali, Mexico, and Russia.” The European Journal of Development Research 17(1): 24–58. Townsend, R. 1995. Consumption Insurance: an Evaluation of Risk-Bearing Systems in Low Income Economies. Journal of Economic Perspectives 2 :83–102. World Bank. 2011. A Poverty Profile for the Northern States of Sudan. Washington, DC: World Bank Group. ———. 2014. World Development Report- Risk and Opportunity- Managing Risk for Development. Washington, DC: World Bank Group. ———. 2018. Macro Poverty Outlook. Washington, DC: World Bank. ———.2019. Poverty and Inequality in Sudan: 2009-2014. Washington, DC: World Bank. Page 49 of 49