SRI LANKA POVERTY UPDATE Background report to Sri Lanka Poverty Assessment SRI LANKA POVERTY UPDATE Background report to Sri Lanka Poverty Assessment © 2021 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. 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. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Cover design: Wojciech Wolocznik, Cambridge, United Kingdom Interior design and typesetting: Piotr Ruczyński, London, United Kingdom Contents Acknowledgments   7 Abbreviations   7 Executive summary   8 1. Introduction     10 2. Recent trends in poverty reduction and shared prosperity    13 Progress in reducing poverty and inequality and in sharing prosperity   14 Progress in nonmonetary indicators of welfare   19 Characteristics of the poor    25 3. Drivers of poverty reduction    31 Main drivers of poverty reduction: Improvements in nonfarm labor income    32 Recent trends in the agriculture sector: Slowing of agricultural income growth with reversal of favorable price trends    35 Recent trends in the nonagriculture sector: Growth in services underpinned by tourism sector   40 Trends in nonlabor income: Localized impacts of Samurdhi and remittances    43 4. Key messages and priorities for poverty reduction and shared prosperity    50 Appendix Number of Samurdhi beneficiaries by allowance category and district, 2014 and 2015   53 Bibliography   54 Boxes Box 1 Sri Lanka’s National Poverty Line   14 Box 2 Geographic sector classifications in Sri Lanka    17 Box 3 The Shapley-Shorrocks decomposition of changes in poverty   35 Box 4 Spotlight on the coconut sector   38 Figures Figure 1 $3.20 poverty rate in Sri Lanka vs. peer countries   14 Figure B1.1 Consumption behavior for food and nonfood expenditures, 2002 vs. 2016   15 Figure 2 Poverty reduction in Sri Lanka, 2009/10 – 2016   16 Figure 3 Trend in $3.20 poverty rates by geographic sector, 2009 – 16   16 Figure 4 Share of population in the bottom 40 percent of the national per capita consumption distribution, by sector   16 Figure 5 $3.20 poverty rates in 2012/13 by district   18 Figure 6 $3.20 poverty rates in 2016 by district   18 Figure 7 Number of $3.20 poor by district   18 Figure 8  Change in per capita consumption between 2009/10 and 2012/13   18 Figure 9 Change in per capita consumption between 2012/13 and 2016   18 Figure 11 Per capita consumption growth of the bottom 40 percent and total population, circa 2012 – 17   19 Figure 10 Per capita consumption growth, bottom 40 percent vs. total population, 2009/10–2013 and 2012/13–2016   19 Figure 12 Access to electricity among poor vs. nonpoor, 2012/13 and 2016    20 Figure 13 Treatment of waste among poor and nonpoor   21 Figure 14 Ownership of household assets, 2012/13 and 2016   23 Figure 15 Ownership of vehicles, 2012/13 and 2016   23 Figure 16 Educational attainment by poverty status and sector   25 Figure 17 Size of and reasons for debt   27 Figure 18 Household debt, 2012/13 and 2016   27 Figure 19 Distance to education and health care facilities in 2016 (minutes)   29 Figure 20 Real GDP growth, 2002 – 18   32 Figure 21 Contribution of growth vs. redistribution to poverty reduction between 2012/13 and 2016, nationally and by sector   32 Figure 22 Share of the employed in agriculture, industry, and services, 2012/13 and 2016   33 Figure 23 Decomposition of growth in per capita value added   33 Figure 24 Decomposition in growth per capita value added, Sri Lanka vs. peer countries   33 Figure 25 Changes in poverty due to demographics and income sources   34 Figure 26 Output and prices for tea (left) and rubber (right), 2009 – 16   36 Figure 27 Minimum wage growth in agriculture, industry, and services, 2009 – 12 and 2012 – 16   37 Figure 28 Coconut production and exports, 2006 – 17   38 Figure 29 FOB price for processed coconut products, 2006 – 17   38 Figure 30 Minimum wage rates in the estate sector, 2010–16   39 Figure 31 People affected by droughts (left), floods (center), and landslides (right), 1980 – 2016   40 Figure 32 Net job creation by sector (in thousands), 2013 – 16   40 Figure 33 Tourist arrivals and gross tourist receipts, 2008 – 17   41 Figure 34 Foreign guest nights in graded accommodation establishments, by region, 2012 – 17   41 Figure 35 Employment in the tourism industry, 2008 – 17   41 Figure 36 Airbnb offerings by district, April 2016 – July 2017   42 Figure 37 Samurdhi budget and beneficiaries, 2011 – 16   44 Figure 38 Coverage of the poor by Samurdhi, 2012/13 and 2016   44 Figure 39 Contribution of demographic and socioeconomic factors to poverty reduction, by province   45 Figure 40 Coverage (percent, left) and monthly benefit levels (Rs, right) of social protection programs   45 Figure 41 Male and female labor force participation, by age   47 Figure 42 Number of migrants, by gender and route of employment, 1997 – 17   47 Figure 43 Number of migrants by skill level, 1994 – 2017   48 Figure 44 Number of departing migrants by province of origin, for 2009, 2012, and 2016 compared to migrants as share of provincial population in 2016   48 Tables Table 1 Share of population with access to water in 2016, nationally and by province (percent)   20 Table 2 Share of households with adequate sanitation, nationally and by province (percent)   21 Table 3 Housing conditions nationally and by province   22 Table 4 Demographic profile of the poor vs. nonpoor   25 Table 5  Labor market characteristics of the poor and nonpoor, 2012/13 and 2016   26 Table 6 Early childhood mortality rates    29 Table 7 Distribution of poor by household head’s employment status   36 Table 8 Paddy statistics, 2009 – 16   37 Table 9 Living arrangements of the elderly   46 SRI LANKA POVERTY UPDATE 7 Acknowledgments This report was drafted by a World Bank team led by Yeon Soo Kim (Senior Economist, Poverty and Eq- uity Global Practice) and comprising Anna Luisa Paffhausen (Economist) and Chinthani Sooriyamudali (Consultant). The team benefited from guidance provided by Idah Pswarayi-Riddihough (former Coun- try Director, Nepal, Sri Lanka and Maldives), Valerie Layrol (former Senior Operations Officer), and Benu Bidani (former Practice Manager, Poverty and Equity Global Practice). Comments from peer re- viewers Samuel Freije-Rodriguez (Lead Economist, Poverty and Equity Global Practice) and Kishan Abeygunawardana (Senior Economist, Macroeconomics, Trade and Investment Global Practice) are gratefully acknowledged. The team would like to thank the Government of Sri Lanka for its support and the Department of Census and Statistics (DCS) for sharing its data. Abbreviations DCS Department of Census and Statistics DHS Demographic and Health Survey FBR Family Background Report FOB free on board GDP gross domestic product HIES Household Income and Expenditure Survey PPP purchasing power parity WHO World Health Organization All dollar amounts are US dollars unless otherwise indicated. SRI LANKA POVERTY UPDATE 8 Executive summary Sri Lanka has made strong progress in reducing poverty and sharing prosperity among the less well-off in recent years. The poverty rate using the World Bank’s $3.20 poverty line (in 2011 purchasing power parity) declined from 16.2 percent in 2012/13 to 11 percent in 2016, a reduction that compares favorably to regional peers. Extreme poverty is almost eliminated. Gains were also made in nonmonetary meas- ures of welfare, including access to basic services, housing conditions, and asset ownership. Growth was inclusive but less pro-poor: per capita consumption growth of the bottom 40 percent of the consump- tion distribution accelerated to an annualized 4.2 percent but was still below the population average of 4.7 percent between 2012/13 and 2016. A significant increase in labor income, particularly from nonfarm sectors, is the major factor behind pro- gress. The economy is steadily transitioning toward industry and services, and sectors such as construc- tion and trade led job creation in recent years. Wage growth has also been strong. The expansion in ser- vices was underpinned by a booming tourism sector, as tourist arrivals quadrupled between 2009 and 2016. Real gross domestic product (GDP) growth was mainly driven by gains in labor productivity, though most of the productivity growth came from increases in within-sector productivity rather than from reallocation effects. This implies that most of the labor movement occurred from agriculture toward other sectors with low productivity. Agriculture did not contribute to poverty reduction as it had in the previous decade because stagnating productivity and a reversal of favorable prices slowed agricultural income growth. Meanwhile, the role of Samurdhi and other social assistance programs for poverty reduction was modest. Despite a significant expansion of the program in 2015, the targeting performance of Samurdhi is rela- tively weak, with only about 40 percent of the $3.20 poor receiving assistance. This means that a large share of these public resources is directed to the nonpoor, limiting the program’s effectiveness in reduc- ing poverty. Given Sri Lanka’s aging demographic profile, large informal sector, and limited availabil- ity of income support for the elderly, old-age poverty is expected to become more relevant over time. Migration has been an important survival strategy for some poor households, but this could change in response to institutional and structural changes that have affected the number and composition of migrants in recent years, and in response to the COVID-19 crisis. SRI LANKA POVERTY UPDATE Executive summary 9 The report concludes by pointing to a few key areas that are crucial for sustaining poverty reduction and shared prosperity: • The creation of more and better jobs is a key priority. Accelerating growth and poverty reduc- tion needs to rely on productivity and employment growth that can support broad-based income growth. Sustaining productivity growth will be important, especially given that progress has slowed in recent years. Accelerating labor reallocation — for example through reducing barriers to internal migration — would also help, as would creating a policy environment that promotes com- petitiveness and job creation. Impediments to transitioning to nonfarm activities need to be bet- ter understood. • Measures to increase agricultural productivity will help reduce poverty. Agricultural households have a higher poverty rate due to low productivity and low earnings in the sector. Support to in- crease paddy productivity and help farmers shift to higher-value crops can help improve rural live- lihoods and reduce poverty. • Stronger safety nets are needed to protect the poor and vulnerable. Social protection programs could be better targeted to further reduce poverty. Only a small share of estate households is cov- ered, even though the poverty rate is much higher in this sector. Benefit levels remain inadequate. Efforts to build better targeting and delivery systems and strengthen graduation programs can go a long way toward supporting the poor and most vulnerable. • Narrowing the gap in access to basic services will help achieve equal opportunities for all. Sri Lanka has made great progress in closing the gaps in access to services, but large challenges re- main in some areas, such as access to water supply. Spatial disparities are high — between urban and rural areas, and between the Western Province and the rest of the country — and contribute to inequality. While Sri Lanka has historically excelled in human capital outcomes, the gap between the poor and nonpoor is wide, and tertiary educational attainment is very low. Efforts are needed to improve conditions in the estate sector, where poverty continues to be much higher, and edu- cational and nutritional outcomes much lower, than in the rest of the country. 1  Introduction SRI LANKA POVERTY UPDATE 1 Introduction 11 Sri Lanka has made strong progress in reducing poverty and sharing prosperity among the less well- off in recent years. Strong post-conflict recovery and growth helped establish Sri Lanka as a solid mid- dle-income country. Progress is reflected in the continued pace of poverty reduction: the poverty rate (using the $3.20 poverty line) declined from 16.2 percent in 2012/13 to 11 percent in 2016. A significant increase in labor income, particularly from nonfarm sectors, is the major factor behind progress. Gains were also made in access to basic services, housing conditions, and asset ownership. Large improvements were observed in electricity coverage, which is now almost universal. Unless noted otherwise, the report defines poverty according to the World Bank’s international poverty line at $3.20 per person per day (in 2011 purchasing power parity, PPP), which is the relevant line for lower-middle-income countries accord- ing to the World Bank’s classification. While Sri Lanka has historically excelled in human development outcomes and continues to make improvements, large gaps remain in some areas, such as access to higher levels of education and access to water supply and sanitation. Rural households experienced fast progress in monetary and nonmone- tary welfare. However, regional disparities are high, and progress was particularly slow among the estate sector population, which continues to be one of the most marginalized groups as measured by poverty, human development outcomes, and access to services and economic opportunities. The aim of this report is to provide an update on progress in poverty and welfare in Sri Lanka through 2016. 1 Results from extensive analysis drawing on the Household Income and Expenditure Survey (HIES) 2016 and other secondary sources are presented. 2 The 2016 HIES is the latest available household sur- vey to date; while new HIES data were collected in 2019, the results have not been made public yet. The report documents areas where important and meaningful progress occurred, analyzes the driving forces behind such positive developments, and highlights remaining challenges where further improvements are needed. COVID-19 has exerted a significant impact on the economy and its people. Gross domestic product (GDP) is estimated to have contracted by 3.6 percent in 2020, despite swift measures implemented by the gov- ernment to contain the outbreak. Widespread jobs and earnings losses led to a significant increase in poverty, from an estimated 9.2 percent in 2019 to 11.7 percent in 2020, based on the $3.20 poverty line. The impacts of COVID-19 on livelihoods and poverty are described in more detail in a separate report (World Bank, forthcoming). 1. The last such update was published in 2016, and the analyses used data from 2012/13; see Newhouse, Suarez-Becerra, and Doan (2016). 2. The HIES is conducted by the Department of Census and Statistics (DCS); annual surveys can be found on the DCS website at http://www.statistics.gov.lk/page.asp?page=Income%20and%20Expenditure. SRI LANKA POVERTY UPDATE 1 Introduction 12 The rest of this report is structured as follows. Section 2 provides in-depth documentation of progress in poverty reduction and shared prosperity. Section 3 investigates the drivers behind changes in labor and nonlabor income that are associated with the observed changes in poverty. Section 4 summarizes the key messages and priorities for policy action. 2 Recent trends in poverty reduction and shared prosperity SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 14 Progress in reducing poverty and inequality and in sharing prosperity Poverty declined strongly between 2012/13 and FIGURE 1 $3.20 poverty rate in Sri Lanka 2016, continuing progress from previous years. 3 vs. peer countries 60 Poverty headcount ratio at $3.20 a day (2011 PPP) Poverty estimates based on the World Bank’s pover- Bangladesh ty line for lower-middle-income countries (at $3.20 50 per person per day in 2011 PPP) dropped from 16.2 percent in 2012/13 to 11 percent in 2016. 4 This was 40 Pakistan a further decline from 2009/10, when the pover- 30 Indonesia ty rate was 19.9 percent. Sri Lanka’s poverty rate Philippines at $3.20 per day compares relatively favorably to 20 regional peers in South and Southeast Asia for Bhutan Sri Lanka 10 which poverty estimates were available around Vietnam 2016 (figure 1). Moreover, extreme poverty has been Thailand 0 almost eliminated, with only 0.9 percent of the 0 1,000 2,000 3,000 4,000 5,000 6,000 population living on less than $1.90 per person GDP per capita (constant 2010 US$) per day (figure 2, panel a). This is a remarkable Sources: World Development Indicators database; World Bank PovcalNet, http://iresearch.worldbank.org/PovcalNet/home.aspx. Note: PPP = purchasing achievement. 5 power parity. BOX 1 Sri Lanka’s National Poverty Line Sri Lanka’s current national poverty line was established in 2002, benchmarked on consumption patterns that prevailed almost two decades ago. The national poverty line was set at Rs 1,423 per person per month in 2002 and has been inflated using the Colombo Consumer Price Index (CCPI) since then. The poverty line was derived based on the cost of basic needs method. Using this poverty line, the poverty headcount rate in Sri Lanka fell from 22.7 percent in 2002 to 4.1 percent in 2016. During this time, Sri Lanka saw strong economic growth, with GDP per capita doubling between 2002 and 2016. Sri Lanka has the oldest poverty line in the South Asia region. The steady fall in the poverty headcount rate since 2002 reflects broad-based progress in welfare. In line with improved living standards, the share of expenditure on food has fallen, from 44.5 percent in 2002 to 34.8 percent in 2016. The consumption behaviors of the reference group are increasingly more consistent with a higher 3. The years 2009/10, 2012/13, and 2016 referenced in this report correspond to the years the HIES data were collected. 4. Starting with the HIES 2009/10, the entire Eastern Province was covered; but Mannar, Kilinochchi, and Mullaitivu in the Northern Province continued to be excluded in the same survey year. Not until the 2012/13 survey did HIES cover- age extend to all 25 districts and become truly nationally representative. Because the Northern Province accounts for only a small fraction of Sri Lanka’s population, the exclusion of the three districts in 2009/10 does not have a significant effect on national-level estimates. 5. Using Sri Lanka’s national poverty line, the poverty headcount rate fell from 8.9 percent in 2009/10 to 6.7 percent in 2012/13, and then further to 4.1 percent in 2016. However, the official national poverty line was estimated in 2002 and is now too low to represent an acceptable minimum standard of living (see box 1 for details). For this reason, the analysis in this report is made in reference to the World Bank’s $3.20 poverty line unless specified otherwise. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 15 standard of living, greater dietary diversity, and urban living: FIGURE B1.1 Consumption behavior for food and they include a decrease in cereals consumption, an increase in nonfood expenditures, 2002 vs. 2016 consumption of meat and fish and prepared foods, and greater spending on nonfood consumption such as clothing and trans- Cereals Selected food port (see figure below). Most countries respond to welfare pro- expenditures gress over time by periodically revising their poverty line, both to Condiments reflect changing consumption patterns as living standards rise Meat & fish and to redefine the basic standard of living for the poor. Prepared food The current low poverty line does not adequately represent a minimum acceptable standard of living for the poor, and a Clothing, footwear Selected nonfood expenditures revision of the national poverty line is needed. The poverty Education line in 2016 was Rs 4,166 per person per month, or Rs 137 per person per day. To put this number in perspective, a family of Personal care four with one breadwinner who makes only Rs 20,000 per month Transport is unlikely to be considered poor according to the national pov- erty line. By holding the benchmark food basket unchanged 0 200 400 600 800 since 2002, the line has become less relevant as a way of iden- Monthly per capita expenditures tifying the poor, understanding their well-being, and designing (2016 Rs) pro-poor welfare programs. The HIES 2019 provides an oppor- 2002 2016 tunity to revise the poverty line. Source: Based on analysis prepared by David Newhouse, Nandini Krishnan, and Ani Rudra Silwal using HIES 2002, 2016. The depth of poverty also decreased, with improvements among the poorest of the poor. The pover- ty headcount rate is the most commonly used measure of poverty; but it does not account for the inten- sity of poverty, as it counts only the share of people whose expenditure falls below the poverty line. To complement the headcount index, the poverty gap index and the poverty severity index are also exam- ined. The poverty gap, which measures the average shortfall of the total population from the poverty line and is expressed as a percentage of the poverty line, was 2.1 percent in 2016. This means that con- sumption levels were on average 2.1 percent short of the poverty line (figure 2, panel b). Doing this cal- culation for just the poor population shows that their consumption fell short by 19 percent of the $3.20 poverty line in 2016. This is less than the shortfall in 2009/10 and 2012/13 — around 21 percent in both years (figure 2, panel c), and implies that progress was driven not only by poor people moving out of pov- erty, but also by an improvement in the depth of poverty, as the consumption level of the poor on aver- age moved closer to the poverty line. The poverty severity index measures the average squared short- fall of the total population from the poverty line and hence puts more weight on the poorest individu- als. This index also continued to decline, suggesting that reductions in poverty were accompanied by improvements among the poorest of the poor (figure 2, panel d). Poverty reduction was particularly strong in rural areas. The $3.20 poverty rate in rural areas declined from 17.6 percent to 11.5 percent between 2012/13 and 2016. However, poverty remains high in the estate sector and stood at 25.4 percent in 2016, a relatively small decline from 28 percent in 2009/10. Poverty in urban areas, already low, decreased further, from 12.5 percent in 2009/10 to 5.2 percent in 2016, with a slower pace in more recent years (figure 3). This slowing likely reflects the difficulties of reducing poverty SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 16 FIGURE 2 Poverty reduction in Sri Lanka, 2009/10 – 2016 a. Poverty headcount rate (%) b. Poverty gap (%) 20 5 16 4 Percent Percent 12 3 8 2 4 1 0 0 2009/10 2012/13 2016 2009/10 2012/13 2016 c. Poverty gap (%, poor population only) d. Poverty severity (%) 24 1.6 20 1.2 16 Percent Percent 12 0.8 8 0.4 4 0 0 2009/10 2012/13 2016 2009/10 2012/13 2016 $1.90 a day $3.20 a day National poverty line Source: World Bank staff estimates based on HIES 2009/10, 2012/13, and 2016. Note: The poverty rate for 2009/10 does not include the Mannar, Kilinochchi, and Mullaitivu districts in the Northern Province, as they were not surveyed in the 2009/10 HIES. These districts account for only a small share of Sri Lanka’s overall population and do not have a significant effect on national-level estimates. that remains deeply entrenched. In the estate sector, 70 percent of the population belongs to the bot- tom 40 percent of the per capita consumption distribution, which is significantly higher than the cor- responding share in the rural sector (41 percent) and urban sector (26 percent) (figure 4). FIGURE 3 Trend in $3.20 poverty rates by geographic FIGURE 4 Share of population in the bottom sector, 2009 – 16 40 percent of the national per capita consumption 30 distribution, by sector 25 20 Estate Percent 15 10 Rural 5 Urban 0 2009/10 2012/13 2016 0 10 20 30 40 50 60 70 80 Estate Rural Urban Percent Source: World Banks staff estimates based on HIES 2009/10, 2012/13, and 2016. Source: World Bank staff estimates based on HIES 2016. While disparities between geographic sectors appear large, an outdated sector classification makes it difficult to get an accurate understanding of the true extent and nature of urban and rural poverty. Poverty remains concentrated in rural areas, which is not surprising given that almost 80 percent of the population resides in rural areas according to official sector classifications: in fact, of the poor, 82 per- cent were rural residents, with the rest living in the urban sector (8 percent) and the estate sector (10 percent). These numbers at first glance suggest that there is little poverty in urban areas. However, the SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 17 current statistical definition of urban, rural, and estate sectors relies on administrative boundaries that were established a long time ago and underestimates the extent of urbanization in Sri Lanka. This lim- its our understanding of poverty in urban and peri-urban areas, which have expanded significantly over the past decades (see box 2). UN-Habitat (2018) shows that urban and semi-urban areas in Sri Lanka have extended beyond the official administrative boundaries. This situation raises the possibility that urban poverty is significantly understated and that the urban poor are being missed in pro-poor policies. BOX 2 Geographic sector classifications in Sri Lanka Poverty was reduced almost everywhere but the On record, Sri Lanka’s share of urban population is be- pace of progress varied significantly across the low 20 percent and among the lowest in the South Asia country. The $3.20 poverty rate declined in almost region. According to the HIES 2016, which uses the official statistical classification for urban, rural, and estate sectors, all districts between 2012/13 and 2016. The improve- only 17.4 percent of Sri Lankans live in urban areas, while the ment was particularly steep in districts where the majority of the population (78.1 percent) lives in rural are- as. By this estimate, Sri Lanka is one of the most rural soci- poverty rate was initially high — above 30 percent — in eties in South Asia and in the world. But Sri Lanka also has 2012/13, such as Mannar, Mullaitivu, Batticaloa, and sustained high GDP growth and a per capita GDP that is the second highest in the South Asia region. Sri Lanka’s low ur- Moneragala. Significant improvements were also banization rate thus appears puzzling, given that economic observed in the Uva, Southern, and North Central growth is usually accompanied by rural to urban migration. Provinces. Meanwhile, the North Western Province In fact, the urbanization rate in Sri Lanka is likely under- estimated because the official classification of sectors re- and Western Province (which includes Colombo) lies on administrative boundaries. That is, municipal coun- recorded the lowest poverty rates in 2016. Overall, cils, urban councils, and town councils are defined as urban areas; plantations with 20 acres or more and 10 or more the number of districts with poverty rates above 30 resident laborers are defined as part of the estate sector; percent declined, from five districts in 2012/13 (fig- and all other areas are classified as rural. Such an approach misses urbanization that occurs outside administrative ar- ure 5) to just two districts in 2016 (Kilinochchi and eas that would otherwise be classified as urban. Indeed, an Mullaitivu, both in the Northern Province) (figure 6). analysis of satellite data shows urban and semi-urban are- as extending beyond the official administrative boundaries that determine urban areas in Sri Lanka. While the official While the Northern and Eastern Provinces have estimate of the population of the nine provincial capitals the highest share of population living in pover- was 1.5 million in 2017, satellite imagery put the estimate at 7.4 million. Analysis of historical satellite data also reveals ty, the largest numbers of poor people live in and that urban population expanded at an average annual rate around the predominantly rural and agricultur- of 6.4 percent over the period 1995–2017. Recent studies put Sri Lanka’s urbanization rate at between 35 percent and al Highlands. District-level poverty rates meas- 45 percent (UN-Habitat 2018). While there is not a single ured at $3.20 per day are high in the Northern and standard for the classification of geographical areas, a re- vision of the definition to better reflect the actual functions Eastern Provinces, but the absolute number of poor of towns would improve understanding of population set- in these provinces is low because they are sparse- tlement and migration patterns as well as local and region- al development issues. ly populated. In comparison, the districts with the Sources: World Bank staff estimates based on 2016 HIES data (for largest number of poor are Ratnapura, Kandy, and the proportion of the urban and rural population); International Labour Organization, “Inventory of Official National-Level Statistical Badulla, which together account for about a quar- Definitions for Rural/Urban Areas,” https://www.ilo.org/wcmsp5/ groups/public/---dgreports/---stat/documents/genericdocument/ ter of all poor (figure 7). This geographic pattern of wcms_389373.pdf; UN-Habitat 2018. poverty is similar using the national poverty line. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 18 FIGURE 5 $3.20 poverty rates FIGURE 6 $3.20 poverty rates FIGURE 7 Number of $3.20 poor in 2012/13 by district in 2016 by district by district Poverty rate Poverty rate Number 52.6 37 218,431 4.7 3 6,093 Source: World Bank staff estimates based on Source: World Bank staff estimates based on Source: World Bank staff estimates based on HIES 2012/13. HIES 2016. HIES 2016. Following a marked increase in inequality between 2009/10 and 2012/13, inequality increased slightly more through 2016. After a significant increase — from 36.1 to 38.7 — between 2009/10 and 2012/13, the Gini index of inequality rose slightly further to 39.3 in 2016. Between 2009/10 and 2012/13, consumption grew relatively slowly among households in the bottom of the distribution and much faster among those in the top (figure 8). In comparison, overall consumption growth accelerated to an average of 18 per- cent between 2012/13 and 2016 and was also more balanced across the distribution. Increases were still strongest for the top 10 percent (figure 9). FIGURE 8  Change in per capita consumption between FIGURE 9 Change in per capita consumption between 2009/10 and 2012/13 2012/13 and 2016 32 32 28 28 24 24 20 20 Growth rate in mean Percent Percent 16 16 12 Growth rate in mean 12 8 8 4 4 0 0 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percentiles of per capita household consumption Percentiles of per capita household consumption Source: World Bank staff estimates based on HIES 2009/10 and 2012/13. Source: World Bank staff estimates based on HIES 2012/13 and 2016. Note: Figure shows a Growth Incidence Curve, i.e., the growth rate in per capita Note: Figure shows a Growth Incidence Curve, i.e., the growth rate in per capita consumption across percentiles of the same distribution. Calculations for consumption across percentiles of the same distribution. 2009/10 do not include the Mannar, Kilinochchi, and Mullaithivu districts in the Northern Province, as they were not surveyed. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 19 Growth was inclusive but less pro-poor. Per cap- FIGURE 10 Per capita consumption growth, ita consumption growth of the bottom 40 per- bottom 40 percent vs. total population, 2009/10–2013 and 2012/13–2016 cent recorded an annualized 4.2 percent but was still below the population average of 4.7 percent between 2012/13 and 2016. This growth represents 2009/10 – 2012/13 an acceleration relative to the previous period of 2009/10 – 2012/13, when per capita consumption of 2012/13 – 2016 the bottom 40 percent increased by only 1.5 per- cent, compared to a 3.8 percent increase for the total population over the same period (figure 10). 0 1 2 3 4 5 Percent Sri Lanka’s recent performance compares well to Bottom 40 Total Population that of other countries for which data on shared Source: Global Database of Shared Prosperity, 2019, https://datacatalog. prosperity were available (figure 11). worldbank.org/dataset/global-database-shared-prosperity. FIGURE 11 Per capita consumption growth of the bottom 40 percent and total population, circa 2012 – 17 12 10 8 6 Percent 4 2 0 -2 -4 -6 Romania Chinah Malaysia Estonia Latvia North Macedonia Lithuania Poland Namibia Croatia Tunisia Panama Dominican Republic Philippines Indonesia Chile Vietnam Hungary Montenegro Eswatini Sri Lanka El Salvador Serbia Czech Republic Kyrgyz Republic Malawi Pakistan Mauritius Sierra Leone Georgia Togo Albania Peru Kosovo Colombia Slovenia Turkey Uruguay Moldova Bolivia Bhutan Paraguay Thailand Costa Rica Bangladesh Honduras Armenia Ecuador Mongolia Belarus Botswana Ethiopia Rwanda Kazakhstan Iran, Islamic Rep. Tanzania Ghana Brazil Russian Federation Slovak Republic Ukraine Zambia West Bank and Gaza Argentinae Uganda Egypt, Arab Rep. Zimbabwe Benin Bottom 40 Total Population Source: Global Database of Shared Prosperity, 2019, https://datacatalog.worldbank.org/dataset/global-database-shared-prosperity. Note: Orange line indicates data for Sri Lanka. Progress in nonmonetary indicators of welfare Access to electricity is now almost universal, thanks to significant progress among the poor. Electricity coverage surpassed 99 percent in 2016, up from 97 percent in 2012/13 (figure 12). Access improved sig- nificantly among the poor, from 78 percent to 93 percent. The largest increases in access to electricity were seen in households in the Northern Province (increase of 26 percentage points), Eastern Province (12 percentage points), Uva Province (11 percentage points), and North Central Province (10 percentage points). However, gaps remain the largest in the Northern and Eastern Provinces, with electricity cov- erage rates of 93 percent and 94 percent in 2016, respectively. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 20 FIGURE 12 Access to electricity among poor vs. Broader and significant challenges remain on nonpoor, 2012/13 and 2016 water- and sanitation-related issues. Only about 100 35 percent of households had access to tap water as 90 of 2016. Across provinces, access to tap water was 80 70 lowest in the North Western Province (9 percent) 60 and the Northern Province (11 percent). The most Percent 50 common source of drinking water was wells — about 40 30 43 percent of households got their water from wells 20 located either within or outside the premises. 10 Households in the Northern Province were signifi- 0 Total Nonpoor Poor cantly more likely to drink water from a well. About 2012/13 2016 20 percent got drinking water from other sources Source: World Bank staff calculation using HIES 2012/13 and 2016. (e.g., tube well, river, rain water, etc.). Nationally, about 6 percent of the population lacked sufficient water to drink and 9 percent lacked water for bathing and washing (table 1). Absolute water shortages remain a challenge for some households: almost 11 percent of the poor did not have enough water to drink in the previous year, and about 14 percent of them did not have enough water for bathing and washing. TABLE 1 Share of population with access to water in 2016, nationally and by province (percent) National WP CP SP NP EP NWP NCP UP Sab Source of drinking water Well: Protected well within premises 32.4 38.7 14.9 35.7 36.1 33.9 45.4 14.6 20.0 33.1 Well: Protected well outside premises 10.1 3.9 6.9 7.2 29.2 11.4 19.1 14.4 9.2 11.3 Well: Unprotected well 2.2 1.0 1.3 4.0 3.2 1.6 3.1 2.1 1.7 3.8 Main tap line: Tap inside home 28.3 47.3 33.3 30.9 1.5 28.1 5.2 19.2 23.4 14.0 Main tap line: Tap within unit/ premises 5.1 1.7 4.5 9.3 2.8 12.8 3.1 8.6 7.3 4.0 Main tap line: Tap outside premises 1.6 1.1 1.1 1.6 6.3 2.0 0.8 2.5 1.3 1.0 Other: Project in village 7.1 3.0 14.1 7.1 1.3 3.1 5.7 6.8 16.9 11.6 Other: Tube well 3.3 2.7 2.0 1.1 11.2 5.1 8.3 1.3 1.5 0.7 Other 10.1 0.7 21.8 3.1 8.3 2.2 9.3 30.5 18.8 20.5 Enough water to drink during last year 93.5 97.8 91.6 94.5 95.4 94.5 89.3 89.0 91.3 89.9 Enough water to bath & wash during last year 90.8 96.1 87.2 91.0 96.9 93.2 88.6 87.3 80.3 85.8 Source: World Bank staff calculation using HIES 2016. Note: WP = Western Province; CP = Central Province; SP = Southern Province; NP = Northern Province; EP = Eastern Province; NCP = North Central Province; NWP = North Western Province; UP = Uva Province; Sab = Sabaragamuwa Province. Almost all Sri Lankans have access to adequate sanitation, but few are connected to a drainage sys- tem. Nearly all of the population — 98 percent — had access to a sealed toilet in 2016, and the share was only slightly lower among the poor, at 96 percent. However, less than 4 percent of the population have SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 21 a toilet that is connected to a drainage system, while 94 percent have their toilet connected to a pit or tank. Moreover, less than half of the population, and about a third of the poor, have a toilet inside the house. The Northern and North Central Provinces lag significantly in all these indicators (table 2). TABLE 2 Share of households with adequate sanitation, nationally and by province (percent)   National WP CP SP NP EP NWP NCP UP Sab Availability of toilet Within unit: Exclusively for household 46.7 70.5 38.1 36.8 22.8 55.0 40.5 26.8 34.3 35.7 Within unit: Shared with another household 1.2 1.4 1.2 0.8 0.6 1.3 1.4 1.1 1.2 1.1 Outside unit: Exclusively for household 46.2 23.9 54.0 57.1 67.2 37.3 51.2 63.7 59.1 57.4 Outside unit: Shared with another household 3.8 2.6 4.9 3.5 5.3 3.4 4.2 6.3 3.7 4.1 Other: No toilet for housing unit; toilet is 1.3 0.6 1.4 1.6 2.9 1.3 1.8 1.7 1.3 1.4 shared with another unit Other: Public toilet 0.5 1.0 0.4 0.1 0.5 0.1 0.3 0.0 0.3 0.3 Other: Not using toilets 0.3 0.0 0.1 0.1 0.7 1.6 0.7 0.5 0.1 0.1 Type of toilet Water seal connected to pit/tank 94.2 92.0 94.7 98.1 92.7 90.7 97.3 90.9 93.7 97.5 Water seal connected to drainage system/ 3.6 6.1 2.9 0.9 5.1 6.7 1.5 1.9 3.7 1.2 piped sewer No water seal 1.3 1.5 1.2 0.4 0.2 2.1 0.9 5.0 1.2 0.5 Direct pit 0.7 0.1 1.2 0.6 1.7 0.3 0.3 2.2 1.4 0.7 Other 0.1 0.3 0.0 0.0 0.2 0.2 0.0 0.1 0.0 0.0 Source: World Bank staff calculation using HIES 2016. Note: WP = Western Province; CP = Central Province; SP = Southern Province; NP = Northern Province; EP = Eastern Province; NCP = North Central Province; NWP = North Western Province; UP = Uva Province; Sab = Sabaragamuwa Province. Waste management remains a significant issue, FIGURE 13 Treatment of waste among poor as almost 80 percent of Sri Lankans burn or and nonpoor dump their garbage. Nationally, only 22 percent 50 of households have their trash collected by a gar- 40 bage truck. Most often, garbage is burned (44 per- 30 Percent cent) or dumped within the premises (28 percent). The situation is much worse among the poor (fig- 20 ure 13). Service coverage is significantly better in 10 urban areas, where 81 percent of households are 0 serviced by garbage trucks, compared to only 11 per- Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor cent and 2 percent respectively among the rural and Collected by garbage Burned Dumped within estate sector populations. The burning or dump- truck premises ing of waste can have a detrimental impact on the 2012/13 2016 environment and on population health. Source: World Bank staff calculation using HIES 2016. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 22 Housing conditions improved but are still lagging in the Northern and Eastern Provinces. In the Northern Province, over 3 percent of the population still lives in a slum/shanty, though this is an improve- ment from the 10 percent share in 2012/13. Overcrowding is much more significant in the Northern and Eastern Provinces; the share of households living in dwellings where two adults have the equivalent of less than one room is 51 percent in the Northern Province and 43 percent in the Eastern Province, com- pared to the national average of 32 percent. Electricity is the primary source of lighting — nationwide cov- erage is almost 98 percent — but 7 percent of households in the Northern Province still rely on kerosene. Firewood remains the predominant cooking fuel (69 percent) almost everywhere except the Western Province. Less than 30 percent of all households use gas (table 3). TABLE 3 Housing conditions nationally and by province National WP CP SP NP EP NWP NCP UP Sab Type of structure (% population) Single house 93.6 93.2 84.0 97.6 96.0 96.9 98.1 99.6 88.5 92.6 Slum/shanty 0.7 0.8 0.1 0.2 3.3 1.6 0.6 0.1 0.2 0.4 Other 5.7 6.0 15.9 2.2 0.7 1.5 1.2 0.3 11.3 7.0 Number of bedrooms 2.5 2.5 2.6 2.6 2.1 2.1 2.7 2.6 2.6 2.5 a Overcrowding measure  (% population) < 0.5 32.3 34.2 33.5 26.7 51.2 43.3 21.7 23.4 29.1 33.3 Between 0.5 and 1 53.4 53.7 49.6 59.2 39.8 48.3 58.7 57.4 54.7 51.8 Between 1 and 1.5 14.3 12.0 16.9 14.1 9.0 8.4 19.5 19.2 16.2 14.9 Principal type of lighting (% population) Kerosene 2.1 0.7 2.2 0.9 6.9 5.0 2.4 1.7 2.9 2.6 Electricity 97.5 99.1 97.4 99.0 92.9 94.1 97.2 98.2 96.6 97.3 Solar energy 0.3 0.2 0.4 0.1 0.2 0.9 0.4 0.1 0.4 0.1 Generator/battery 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Other 0.0 0.1 0.0 0.0 0.0 0.1 0.0 0.0 0.0 0.0 Principal type of cooking fuel (% population) Firewood 68.8 39.5 75.0 77.0 80.8 68.4 83.8 87.1 89.8 86.2 Gas 29.5 57.4 23.9 22.3 16.1 29.0 14.9 12.6 9.9 13.4 Kerosene 1.2 2.4 0.8 0.3 2.8 1.6 0.8 0.0 0.1 0.1 Electricity 0.1 0.2 0.0 0.0 0.1 0.4 0.1 0.1 0.0 0.0 Saw dust/paddy husk 0.1 0.0 0.0 0.0 0.0 0.5 0.0 0.0 0.0 0.0 Other 0.3 0.5 0.2 0.3 0.1 0.1 0.4 0.1 0.1 0.2 Source: World Bank staff calculation using HIES 2016. Note: WP = Western Province; CP = Central Province; SP = Southern Province; NP = Northern Province; EP = Eastern Province; NCP = North Central Province; NWP = North Western Province; UP = Uva Province; Sab = Sabaragamuwa Province. a. Overcrowding is defined as number of bedrooms divided by an adult-equivalent household size (adult = 1 , child = 0.5). SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 23 Gains in asset ownership, consistent with a rise in living standards, were widely observed across the dis- tribution. Ownership rates for major durable assets, such as refrigerators, washing machines, mobile phones, and televisions, increased between 2012/13 and 2016. Among the poor, ownership of TVs and mobile phones saw large increases, 9 percentage points and 12 percentage points, respectively, narrow- ing the gap between the poor and nonpoor (figure 14). Meanwhile, ownership rates for washing machines improved mainly among the better-off. There was a small increase in the ownership of transport vehicles, with a shift toward motorized modes of transport. In 2016, 40 percent of the population owned a motor- cycle or motor scooter, making these vehicles more popular than bicycles (figure 15). The level of land own- ership is high and increased between 2012/13 and 2016. Most of the owned land was occupied by a housing unit, consistent with the high rate of homeownership in Sri Lanka — 82.9 percent as of 2012 (DCS 2012a). FIGURE 14 Ownership of household assets, 2012/13 and 2016 100 90 80 70 60 Percent 50 40 30 20 10 0 Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor Refrigerator Cooker Washing machine Mobile phone Television 2012/13 2016 Source: World Bank staff estimates based on HIES 2012/13 and 2016. FIGURE 15 Ownership of vehicles, 2012/13 and 2016 50 40 30 Percent 20 10 0 Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor Total Poor Nonpoor Bicycles Motor cycles / scooters Three-wheelers Motor cars / vans 2012/13 2016 Source: World Bank staff calculation using HIES 2012/13 and 2016. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 24 Low ownership of labor-saving household durables and low access to water and other household infrastructure increase the opportunity cost of women’s time and could constrain their labor force participation. Low female labor force participation, at around 35 percent, remains a salient feature of Sri Lanka’s labor market. There has been little progress over time, despite rising educational attain- ment among women. Previous research from other countries suggests that the acquisition of labor-sav- ing household durables helps increase married women’s labor force participation by significantly reduc- ing their time spent on household chores. In Sri Lanka, only 22 percent of households own washing machines, and the share drops to below 10 percent among both the poor and bottom 40 percent. Those households with no access to tap water within the premises likely need to spend time gathering water for cooking, bathing, and laundry. Almost half of Sri Lankans don’t have a refrigerator in the house- hold, implying that women likely spend more time cooking than they would otherwise, as meals can- not be stored for long periods. According to an evaluation of an Asian Development Bank project in Sri Lanka (Third Water Supply and Sanitation Project), 82 percent of women reported that improvements in the water supply made collecting water easier, and more than half increased their incomes because they were able to substitute the time spent collecting water with income-generating activities (ADB 2011). Moreover, the use of firewood for cooking purposes places a burden on women’s time. The heating of cookstoves relies predominantly on the use of biomass fuel in the form of firewood, especially among the less well-off: only about 20 percent of the poor report owning a gas, kerosene, or electric cooker in 2016 (figure 14). Biomass fuel is disproportionately used by poorer households, as richer households are much more likely to use LPG (liquified petroleum gas) (Nandasena, Wickremasinghe, and Sathiakumar 2012). 6 The demands on women’s time posed by caring for children and the elderly and performing vari- ous household chores are likely to act as significant barriers to women’s participation in the labor market. 6. Additionally, the use of biomass fuel for cooking is detrimental to health. Smoke from the burning of firewood gener- ates harmful substances such as particulates, carbon monoxide, nitrogen oxides, and other carcinogens. The high reliance on wood-burning cooking stoves thus exposes household members to indoor air pollution, which has potentially harmful health effects such as respiratory illnesses. The exposure is greatest for those who spend the most time at home—women, children, and the elderly, the latter a rapidly growing population group. Previous studies have found that Sri Lanka faces significant morbidity and mortality risk from indoor air pollution due to the widespread use of biomass fuel: the World Health Organization (WHO) estimates that 4,300 deaths were attributable to indoor air pollution in Sri Lanka in 2004 (cited in Elledge et al. 2012). Elledge et al. (2010) cites 2009 WHO data calling respiratory diseases the fifth most important cause of neonatal mortality, and also notes that risks of stillbirths increase with exposure to biomass smoke. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 25 Characteristics of the poor Poor households are larger, younger, and less educated Poor households are larger in size and have higher dependency ratios than others (table 4). They also have higher dependency ratios, at 0.9 compared to an average of 0.7 for nonpoor households, which means that poor households have a higher number of children or elderly who need to be supported by each working-age member. 7 The heads of poor households are slightly younger and less likely to be TABLE 4 Demographic profile of the poor vs. nonpoor married than heads of nonpoor households. There Total Nonpoor Poor are no significant differences in the gender of the Household size 4.5 4.3 5.4 household head by poverty status. Dependency ratio 0.7 0.7 0.9 Age of head 34 34.3 31.1 The gap in educational outcomes between the poor Female head (%) 53 53 52 and the nonpoor remains. Sri Lanka has histor- Head is married (%) 48 49 43 ically excelled in educational outcomes, and they continued to improve between 2009/10 and 2016. Source: World Bank staff estimates based on HIES 2012/13 and 2016. Among the cohort ages 20 – 29 years old, the share with primary education or less decreased from 53 FIGURE 16 Educational attainment by poverty status percent to 45 percent during this period, while an and sector 100 increasing share passed the O-level and A-level 3 3 3 4 5 5 3 90 11 12 8 11 examinations. However, the share with a bache- 15 80 15 lor’s degree and above remained low in 2016, at 4 16 70 18 percent. The enrollment rate in early childhood 53 60 56 education among three- to five-year-olds was about Percent 56 50 49 percent nationally in 2016, at roughly the same 50 49 40 46 level as in 2012. While enrollment rates are high 30 in the Western, Northern, and Eastern Provinces 20 28 26 23 (between 56 and 63 percent), they are considerably 10 18 17 14 10 lower in the Uva and Sabaragamuwa Provinces, at 0 3 3 7 2 5 National Nonpoor Poor Urban  Rural  Estate 41 percent and 34 percent, respectively. Moreover, there is a large gap in educational achievement No schooling Less than primary between the poor and nonpoor (figure 16). Among Primary completed O-level passed working-age adults (ages 15 and above), nearly A-level passed Bachelor’s and above 90 percent of the poor have completed primary Source: World Bank staff calculation using HIES 2016. 7. Dependency ratios are calculated as the number of household members who are dependents divided by the number of working-age adults. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 26 education or less, compared to less than 70 percent of the nonpoor; about 8 percent of the poor and 16 per- cent of the nonpoor have O-level certificates. The estate sector lags significantly behind, with almost 40 percent having less than primary education and only 8 percent having O-level certification or higher. A greater share of incomes for the poor is from nonfarm sources, while indebtedness has decreased Poor households have proportionately fewer working adults, who are also more likely to be working in the low-paying agricultural sector. Working-age adults in poor households are slightly less likely than those in nonpoor households to be occupied in income-earning activities. Those who are working are more likely to be engaged in wage work and less likely to be self-employed. Moreover, the working poor are significantly more likely to be engaged in agriculture, though the importance of the sector as a source of income is declining over time (table 5). TABLE 5  Labor market characteristics of the poor and nonpoor, 2012/13 and 2016   2012/13 2016   National Nonpoor Poor National Nonpoor Poor Percentage of working adults in household 48.9 49.2 47.0 51.1 51.4 47.9 Percentage of adults in wage work 63.6 62.5 69.0 62.0 61.2 68.7 Percentage of adults in self-employment 36.5 37.6 31.0 38.0 38.8 31.3 Percentage of adults working in agriculture 25.6 22.8 39.5 23.0 21.4 35.5 Percentage of adults working in nonagriculture 74.4 77.2 60.5 77.0 78.6 64.5 Source: World Bank staff estimates based on HIES 2012/13 and 2016. Overall indebtedness decreased, including for the poor, and debt levels dropped. In 2016, 53 percent of poor and 65 percent of nonpoor households held some debt. Indebtedness slightly declined compared to 2012/13. The average ratio of outstanding debt to household annual consumption for poor households was 72 percent in 2016, equivalent to 8.6 months of household consumption. The corresponding figure in 2012/13 was 101 percent (figure 17, panel a). 8 Sources of financing have also become more formal. The poor are less likely to have debt from formal than from informal sources. Overall, the sources of debt shifted toward more formal sources and away from con- sumption-driven and expensive credit suppliers such as pawn shops and retail outlets. The share of households 8. The trends that emerge in the data may be contrary to popular perception, as poor households, especially those in former conflict areas, are perceived to be trapped in a debt cycle. More work is needed to better understand the situation of highly indebted poor households. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 27 with pawned assets halved between 2012/13 and 2016, from 42 to 21 percent, with decreases among both the poor and nonpoor. Borrowing from retail outlets also decreased, from 16 to 11 percent of households (fig- ure 18). However, indebtedness to retail outlets is about twice as common for the poor as for the nonpoor, who are most likely to borrow from banks or finance companies (considered more formal sources of credit). The poor borrow mainly to finance economic activities. Borrowing to build or repair a house is the second most common reason cited by the poor (27 percent) and the most common reason cited by the nonpoor (32 percent). The poor are much less likely than the nonpoor to finance the purchase of land, a house, or a vehicle by borrowing (figure 17, panel b). FIGURE 17 Size of and reasons for debt a. Debt as share of annual consumption (%) b. Reasons for debt 140 Economic activity 120 Construction or repair of building/house 100 Purchase of land/house/vehicle 80 Percent Purchase of domestic 60 equipment Settlement of loan 40 20 Other 0 0 5 10 15 20 25 30 35 National Nonpoor Poor Percent 2012/13 2016 National Nonpoor Poor Source: World Bank staff estimates based on HIES 2012/13 and 2016. Note: Reasons for debt relate only to debt contracted from banks, finance and leasing companies, employers, and money lenders. FIGURE 18 Household debt, 2012/13 and 2016 80 70 60 50 Percent 40 30 20 10 0 National Nonpoor Poor National Nonpoor Poor National Nonpoor Poor National Nonpoor Poor Household has debts Debts with bank Debts with retail outlet Pawned assets (land, house, jewelery, etc.) 2012 2016 Source: World Bank staff estimates based on HIES 2012/13 and 2016. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 28 Progress was slow in the estate sector, with households characterized by high poverty, lower human capital outcomes, and gaps in access to basic services Estate sector residents remain one of the most marginalized group in Sri Lanka, with about a quar- ter classified as poor. 9 Estate sector poverty rates fell gradually from 29.8 percent in 2009/10 to 28 per- cent in 2012/13 and 25.4 percent in 2016 (figure 3). This latest poverty estimate is more than double the national poverty rate. Slow progress in the estate sector is also reflected in the low ownership of assets — land ownership is only 37 percent of households in the estate sector, less than half the national average; only 20 percent own a refrigerator and only 3 percent own a washing machine. Human devel- opment outcomes also lag significantly behind: the estate population continues to be plagued by chron- ic malnutrition, poor housing conditions, and lack of access to basic services, including water and san- itation, health services, and education. Housing conditions remain poor in the estate sector, and overcrowding is severe. Some 62 percent of the estate population lived in row houses or line rooms as of 2016, with no improvement from 2012/13. 10 In stark contrast, 94 percent of Sri Lankans live in single houses. Moreover, estate residents live in severe- ly overcrowded conditions, with 71 percent having half a room or less per adult equivalent household member. Among the urban and rural population, this proportion is much lower, at 41 percent and 28 percent, respectively. In line with national trends, access to electricity increased strongly, from 85 per- cent in 2012/13 to 95 percent in 2016. But there were no improvements in access to tap water, and half the estate population gets their drinking water from rivers, tanks, and streams. Contamination of water supplies at source is high, and almost all estate households surveyed in World Bank (2017) reported that they treated the water prior to drinking by boiling, chlorinating, and filtering. Only 21 percent of estate households had toilets within their home, compared to 72 percent in urban and 44 percent in rural areas. Accessibility and quality of services are worse in the estate sector. It takes longer to reach schools (par- ticularly secondary schools), hospitals, and maternity facilities in the estate sector than it does in other 9. The estate sector consists mainly of tea or rubber plantations that rely on resident workers and are managed or owned by the state, regional plantation companies, and individuals or families. The Sri Lanka Department of Census and Statistics defines the estate sector as including land that is over 20 acres, with 10 or more resident workers. (Given that most coconut estates employ only a small resident labor force, they usually do not fall into the estate sector category.) These plantations were created during the British colonial period as self-sufficient enclaves. Labor was imported from South India, and workers were confined within the plantation structure, resulting in “residential labor.” Housing, health care, and education were pro- vided by the estate management. There was thus very little socioeconomic integration with the rest of the country, a situation that continues today, although health and education services are now the responsibility of the state. According to the 2011 Census of Housing and Population (DCS 2011), approximately 0.9 million people, or 4.4 percent of the Sri Lanka population, live on estates. Most of these plantation workers reside in the Central, Uva, and Sabaragamuwa Provinces. 10. Row houses or line rooms are small barrack-style accommodations originally constructed by the British in the early 19th century that continue to serve as the living quarters for estate workers today. Most line rooms are small and windowless and have minimal basic facilities. The dilapidated conditions further contribute to unhealthy and marginalized living conditions of estate workers. SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 29 sectors (figure 19). Among women in the estate sec- FIGURE 19 Distance to education and health care tor who have given birth, nearly all (98 percent) facilities in 2016 (minutes) received prenatal care from a skilled provider for 50 their most recent delivery, and delivery in a health 40 facility is almost universal. However, these women Travel time (minutes) are less likely to receive prenatal care from health 30 facilities (51 percent) than women in urban and 20 rural areas (69 percent and 66 percent, respectively). They are instead more likely to use medical health 10 officers and public midwives. 11 This difference could 0 result in differences in the quality of care received Preschool Primary Secondary Hospital Maternity Maternity and ultimately health outcomes. School School home clinic Urban Rural Estate Under-five mortality declined significantly in the Source: World Bank staff calculation using HIES 2o16. past decade, but children in the estate sector still suffer from poor health outcomes. Recent results TABLE 6 Early childhood mortality rates from the 2016 Demographic and Health Survey (DHS) Post-neonatal show that under-five mortality rates decreased by Under-five mortality mortality mortality mortality mortality Neonatal more than half in the estate sector, from 33 deaths Infant Child per 1,000 live births reported in the 2006/07 DHS to 15 deaths per 1,000 live births in the subsequent Urban 7 3 10 2 11 10-year period (table 6). While these improvements Rural 7 3 10 1 12 are laudable, this rate is still notably higher than that Estate 8 5 13 2 15 in the rest of the country. Higher under-five mor- Source: DCS and Ministry of Health, Nutrition and Indigenous Medicine 2017, tality is mostly driven by disparities in infant mor- table 8.2. Note: Childhood mortality rates are given for the 10-year period preceding tality, which is affected by risk factors such as the the 2016 Demographic and Health Survey. Child mortality and under-five quality of prenatal care, complications during preg- mortality measure mortality between ages one and four, and ages zero to five, respectively. All rates are expressed per 1,000 live births, except for child nancy and at birth, and low birth weight (defined as mortality, which is expressed per 1,000 children surviving to 12 months of age. less than 2.5 kg), among others. The incidence of low birth weight declined in the estate sector over the last decade but remained high in 2016, at 25 percent. The corresponding figures in urban and rural areas were 16 percent or less. Research has associated low birth weight with lower cognitive development and adverse short-term as well as long-term health implications. Undernutrition among women and children under five remains a concern in Sri Lanka, but the prob- lem is most severe in the estate sector. Despite improvements in the last 10 years, as of 2016 32 per- cent of children under the age of five in the estate sector were stunted — about double the share in other 11. Data are from the 2016 Demographic and Health Survey (DCS and Ministry of Health, Nutrition and Indigenous Medicine 2017). SRI LANKA POVERTY UPDATE 2 Recent trends in poverty reduction and shared prosperity 30 areas — and 9 percent were severely stunted. The estate sector also has a much higher prevalence of under- weight, with 30 percent of children under five underweight for their age, compared to 21 percent in the urban sector and 16 percent in the rural sector (DCS and Ministry of Health, Nutrition and Indigenous Medicine 2017).  1 2 Anemia is prevalent, affecting 53.1 percent of children ages 6 – 23 months and 30.4 percent of women ages 24 – 59 years (World Bank 2017). Improvements have been marginal since 2006. Further, 22 percent of estate sector women are considered thin, defined as a body mass index (BMI) below 18.5. Among ever married women, 15 percent are at risk of poor birth outcomes and delivery complica- tions because of short stature; this risk is three times higher than in the urban sector (DCS and Ministry of Health, Nutrition and Indigenous Medicine 2017).  1 3 12. Infant and young child feeding practices are an important determinant of malnutrition. Adherence to best practices on initiation and duration of breastfeeding is high across all sectors; especially noteworthy is the share of estate sector chil- dren receiving colostrum, up from 70 percent in 2006/07 to 97 percent in 2016. However, households in the estate sector can- not easily afford or access a variety of nutritious food, may lack information on the timing and importance of complementary food, and may be influenced by traditional beliefs in determining children’s diet (World Bank 2017). For example, the con- sumption of all protein sources, such as milk, meat, fish, poultry, eggs, legumes, and cheese or yogurt, is lower than in urban and rural areas (DCS and Ministry of Health, Nutrition and Indigenous Medicine 2017). 13. Short stature is an indicator of past nutritional deficiencies and a risk factor for poor birth outcomes and delivery compli- cations. The cut-off point at which mothers are considered at risk because of short stature is usually 145 cm. 3  Drivers of poverty reduction SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 32 Main drivers of poverty reduction: Improvements in nonfarm labor income Strong post-conflict recovery and growth helped establish Sri Lanka as a solid middle-income coun- try. Real GDP grew at an average of 5.7 percent between 2002 and 2009. Following the cessation of the civil war, annual GDP growth accelerated and reached an average of 8.5 percent between 2010 and 2012, after which it slowed down significantly (figure 20). GDP growth was relatively high during the period of FIGURE 20 Real GDP growth, 2002 – 18 2012 – 16 — the focus of the analysis in this report — at 10 5.4 percent annually. Structural transformation 8 continued at a gradual pace, with economic activ- 6 Percent ities shifting from agriculture to industry and ser- 4 vices. The expansion in the services industries was 2 mainly driven by low-value-added activities such 0 as transport and trade. 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: World Development Indicators database. Poverty reduction in recent years was primari- Note: Dotted line shows average over years indicated. ly driven by an increase in average consumption, rather than a redistributive effect. To understand FIGURE 21 Contribution of growth vs. redistribution the driving factors behind the progress in pover- to poverty reduction between 2012/13 and 2016, nationally and by sector ty reduction, a simple decomposition is performed 4 to quantify the relative contribution of growth 2 and redistribution to changes in poverty. 14 The 0 results from this exercise, shown in figure 21, sug- -2 Percent gest that strong growth in real consumption helped -4 the poor climb out of poverty. The redistributive -6 effect had a poverty-increasing impact, meaning -8 that if consumption had not grown in real terms, -10 National Urban Rural Estate poverty would have actually increased. Alternatively, Growth Redistribution if consumption growth had been distributed more evenly across the distribution, poverty would have Source: World Bank staff calculation using HIES 2012/13 and 2016. Note: The bars show the contribution (in percentage points) of the growth and declined much faster. redistribution components to the change in poverty. 14. Specifically, changes in poverty are decomposed into a balanced growth component and a redistributive component, by fix- ing either relative inequalities or the relative poverty line. The growth effect is estimated as the change in poverty that is due to a variation in mean, holding inequality constant. The redistribution effect is estimated as the change in poverty that is due to a variation in the Lorenz curve, holding the mean constant. For details, see Datt and Ravallion (1992). SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 33 Labor market outcomes improved, as jobs were FIGURE 22 Share of the employed in agriculture, created and wages grew strongly. Employment industry, and services, 2012/13 and 2016 100 grew by over 100,000 between 2012 and 2016, and of those employed, an increasing share was work- 80 ing in industry and services. Agriculture account- 60 Percent ed for 24.8 percent of the employed in 2016, down 40 from 28 percent in 2012/13. The share working in industry increased from 26.1 to 27.3 percent and 20 the share in services from 45.9 to 47.9 percent (fig- 0 ure 22). In real terms, earnings grew by an annual- 2012/13 2016 ized 7 percent between 2013 and 2017.  1 5 Agriculture Industry Services Source: World Banks staff calculation using HIES 2012/13 and 2016. Gains in labor productivity spearheaded the growth in real GDP per capita. Real GDP per cap- FIGURE 23 Decomposition of growth in per capita value added ita grew at an annual rate of 5.2 percent between 2002–18, Total = 5.2% 2002 and 2018. A large share of the growth was 2012–18, Total = 3.7% due to increases in labor productivity, which led to 2002–18, Total = 6.1% better jobs and higher wages. However, productiv- -1 0 1 2 3 4 5 6 7 ity growth has fallen in recent years, after enjoy- Percentage yearly contribution to growth ing a boost between 2002 and 2012. The impact of Within-sector productivity Static reallocation Dynamic reallocation Employment rate demographic change was negative, reflecting the Participation rate Demographic change decline in the share of the working-age popula- Source: Based on World Bank Job Structure Tool and data from World tion (figure 23). Development Indicators. FIGURE 24 Decomposition in growth per capita value Most of the productivity growth came from added, Sri Lanka vs. peer countries increases in within-sector productivity, particu- Malaysia larly in services, and much less from realloca- Philippines tion effects. Labor productivity rises when work- Bangladesh ers move from low- to high-productivity sectors or Vietnam Thailand when productivity levels within sectors improve. In Sri Lanka Sri Lanka, reallocation yielded limited productiv- -1 0 1 2 3 4 5 6 ity gains, as most of the movement occurred from Percentage point annual change agriculture toward sectors with low productivi- Within-sector productivity Static reallocation ty, such as trade. Instead, most of the productivity Dynamic reallocation Employment rate Participation rate Demographic change growth was due to improvements in within-sector Source: Based on World Bank Job Structure Tool and data from World productivity. In peer countries such as Vietnam and Development Indicators. 15. Estimates are from DCS Labour Force Surveys, 2013 – 17. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 34 Bangladesh, productivity growth was comparably strong, but a significant share was due to reallocation, and these countries also benefited from favorable demographic trends (figure 24). Reflecting improvements in labor productivity, poverty reduction between 2012/13 and 2016 is pre- dominantly associated with higher labor income in nonfarm sectors. A further decomposition anal- ysis that measures the relative importance of changes in demographics, employment, public transfers, and remittances for poverty reduction highlights the importance of nonfarm labor income and indicates that agriculture did not contribute to poverty reduction as it had in the previous decade (see box 3 for details on the methodology). 16 During that time, a boost in farm earnings was mainly driven by high- er food and commodity prices, rather than productivity growth. The increase in nonfarm labor income results from a combination of a larger share of adults working outside the agricultural sector and those adults obtaining higher labor earnings. This combination accounts for two-thirds of poverty reduc- tion between 2012/13 and 2016 (figure 25). Increases in Samurdhi income explain about 22 percent, and increases in the number of adults engaged in economic activities contributed another 20 percent to the overall change in poverty. The consumption-to-income ratio decreased and had a positive association with poverty reduction, i.e., households had a tendency to spend proportionately less of their income, and if the ratio had stayed the same, poverty would have decreased even more. The next sections fur- ther unpack the trends at the sectoral level to better understand what led to improved earnings in the nonagricultural sectors and the decline in the agricultural sector. FIGURE 25 Changes in poverty due to demographics and income sources Consumption-to-income ratio Percent adults in household Percent adults working in household Percent working in agriculture Agricultural income Percent working in nonagriculture Nonagricultural income Samurdhi income Remittances income Other nonlabor income In-kind income -80 -60 -40 -20 0 20 40 60 80 Percent Source: World Bank staff calculation using HIES 2012/13 and 2016. Note: The figure shows percentage contribution of different demographic factors and income sources to poverty reduction. A negative contribution means that the component helped reduce poverty. The sum of all bars amounts to 100 percent. 16. Previous analysis shows that poverty reduction between 2002 and 2012/13 was driven by growth in labor income, attrib- uted to an increase in returns to nonfarm wage workers and higher returns to self-employed farm workers (Ceriani, Inchauste, and Olivieri 2015). SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 35 BOX 3 The Shapley-Shorrocks decomposition of changes in poverty Researchers and policy makers are often interested in knowing what role is played in poverty reduction by changes in demographics, employment, public transfers, and remittances. Recognizing that poverty is a function of several components — total household per capita consumption (itself dependent on the number of household members), consumption-to-income ratio, and household income (see equation (1)) — one can decompose changes in poverty into changes in each of these components. (1) where θ is the consumption-to income ratio, n is the number of household members, na is the number of adults in the household, no is the number of occupied (employed) adults, and yiL and yiNL are labor income and nonlabor income of individual i, respectively. By changing each component, one at a time, until a complete change from the base period to the period of interest is achieved, one can simulate counterfactual welfare distributions and associated poverty rates. These are then used to assess the contribution of changes in each factor to changes in poverty. For instance, by changing labor income to its value in the period of interest while keep- ing all other factors at their base values, and then comparing poverty rates in the actual and the hypothetical scenarios, the con- tribution of changes in labor income to poverty reduction can be determined. The value of the contribution of any factor, however, depends on the sequence in which each factor is changed. In the case of three components, labor income could, for instance, be changed first, as in the example just given, or second, or third, providing different results for the contribution of labor income to pov- erty reduction. To overcome this issue of path dependency, the Shapley value is calculated; this value is the average contribution of a factor as the average over all possible ways of changing it. This method was originally developed by Shapley (1953) in the context of game theory to provide a solution to a cooperative game with transferable utility, where the objective is to share fairly the output of a technology jointly owned and operated by a fixed group of agents. Shorrocks (2013) proposed the use of the Shapley value as a general framework for decomposition of distributional questions. There are some important caveats to this method of decomposing changes in poverty. First, the decomposition cannot identify whether changes in endowments of the population (e.g., higher educational levels or increases in productive assets) or changes in returns to these endowments underlie changes in poverty. Second, the counterfactual distributions used in the decomposition result from a ceteris paribus exercise in which one component at a time is modified, keeping all other components constant and not taking into account general equilibrium effects. Source: Based on Azevedo, Nguyen, and Sanfelice 2012; Azevedo et al. 2013; Inchauste et al. 2014. Recent trends in the agriculture sector: Slowing of agricultural income growth with reversal of favorable price trends Agricultural households are more likely than others to be poor. Among households headed by some- one performing self-employed work in agriculture, 11.9 percent are poor. Among households headed by someone performing agricultural wage work, the poverty rate is significantly higher, estimated at 21.5 percent. If the household head is working outside of the agricultural sector, the poverty rate is less than 9 percent. For comparison, the overall $3.20 poverty rate is 11 percent. However, agricultural households do not account for the majority of the poor. Households whose head is working in agriculture account for about 27.5 percent of all the poor, whereas those whose head works SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 36 in nonagriculture account for 43.2 percent of the TABLE 7 Distribution of poor by household poor. Households whose head is inactive account head’s employment status for the remaining 29.4 percent (table 7). Households of this type as share Household head is . . . of all poor (%) Falling output and prices in key export prod- Self-employed in agriculture 14.5 ucts and a stagnating paddy sector contributed Wage worker in agriculture 13.0 to a slowdown in the growth of the agricultur- Nonagricultural worker 42.7 al sector in recent years. The period 2012 – 16 saw Other nonagricultural worker 0.5 decreases in both output and prices of tea and rub- Out of labor force 29.4 ber, both key export products, compared to the years before. Tea is Sri Lanka’s most important Source: World Bank staff calculation using HIES 2016. export commodity by value. Export quantity fell by almost 10 percent between 2012 and 2016, and the value of tea exports started to decline after years of sustained growth, though the fall in export value was buffered by the rupee depreciation. Nearly all — 99 percent — of tea produced is used for exports: in 2016, of 293 million kg of tea produced, 289 million kg was exported. Profit margins, proxied using a ratio of auction price over production cost, did not improve between 2012 and 2016 (figure 26, left). Rubber saw a steep decline over this period, both in production and export volume, accompanied by an equally sharp drop in prices (figure 26, right). These trends mark a stark contrast to the trends between 2009 and 2012, when both production and profitability increased. The value added of paddy to GDP declined amid stagnating productivity and volatile production, despite an increase in the total area sown (table 8). FIGURE 26 Output and prices for tea (left) and rubber (right), 2009 – 16 340 1.4 180 4.5 330 160 4.0 1.2 140 3.5 320 1.0 120 3.0 Price/cost ratio Price/cost ratio 310 Million kg Million kg 0.8 100 2.5 300 0.6 80 2.0 290 60 1.5 0.4 280 40 1.0 270 0.2 20 0.5 260 0 0 0 2009 2010 2011 2012 2013 2014 2015 2016 2009 2010 2011 2012 2013 2014 2015 2016 Export volume Export volume Non-export production Auction price/cost of production FOB price/cost of production Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. Note: FOB = free on board. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 37 TABLE 8 Paddy statistics, 2009 – 16 2009 2010 2011 2012 2013 2014 2015 2016 Area sown (thousand ha) 978 1,065 1,223 1,067 1,227 964 1,254 1,114 Irrigated area (thousand ha) 730 809 940 824 958 717 984 904 Rain-fed area (thousand ha) 248 257 278 242 269 248 269 237 Production (thousand MT) 3,652 4,301 3894 3,846 4,621 3,381 4,819 4,420 Yield per ha (kg) 4,337 4,527 3,970 4,353 4,329 4,264 4,429 4,372 Price (Rs/bushel) Guaranteed price 605 605 605 605 699 699 939 793 Open market price 676 598 635 618 637 843 787 772 Value added (million Rs) 41,179 48,377 44,325 43,596 52,084 43,386 76,256 52,649 Percentage of GDP 1.7 1.8 1.5 1.4 1.6 1.2 0.9 0.6 Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” 2018. Note: ha = hectare; MT = metric ton. Reflecting these trends, there was limited improvement in agricultural earnings. Minimum wages in the agricultural sector increased by 89 percent during 2009 – 12 (in nominal terms); however, the increase was a modest 7 percent between 2012 and 2016 (figure 27). This was significantly lower than the wage growth in industry and commerce (44 percent) and services (25 percent) during the same period. FIGURE 27 Minimum wage growth in agriculture, industry, and services, 2009 – 12 and 2012 – 16 The coconut sector experienced remarkable 100 growth, contributing to improvements in rural 80 incomes. In recent years, coconut rose to become 60 Percent the second most important food commodity by 40 export value as output grew substantially. Export 20 values doubled between 2012 and 2016 after record- ing a 43 percent increase in the 2009 – 12 period (fig- 0 Agriculture Industry and Services commerce ure 28). The timing coincides with the rise in glob- 2009–2012 2012–2016 al demand, reflecting the expanding popularity of coconut water, coconut oil, and other processed Source: Department of Labour 2016. Note: Figure shows the growth rate of the minimum wages index in wages board coconut products (Hexa Research 2019). Celebrity sectors. endorsements and the marketing of coconut as a “superfood” boosted demand from health-conscious consumers around the world. Between 2012 and 2016 alone, prices of major coconut products rose between 25 percent and 53 percent (figure 29), leading to better margins for producers. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 38 FIGURE 28 Coconut production and exports, FIGURE 29 FOB price for processed coconut products, 2006 – 17 2006 – 17 70 3.5 800 60 3.0 700 50 2.5 600 FOB price (Rs) Nuts, billions Rs, billions 500 40 2.0 400 30 1.5 300 20 1.0 200 10 0.5 100 0 0 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Export value, coconut non-kernel products Coconut oil Export value, three major coconut products Desiccated coconut Coconut production Copra Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. Note: FOB = free on board. Copra is dried coconut meat. While Sri Lanka’s coconut farmers may have benefited from rising global demand, coconut produc- tion requires large investments, and the longer production cycle involves risks. Domestic consump- tion of fresh coconuts has been relatively stable in the past decade, at around 1.7 – 1.9 billion nuts annu- ally. This amounts to about 80 coconuts per Sri Lankan per year. Coconut is an essential part of the native cuisine, and domestic demand will likely remain stable. However, global demand can be unpre- dictable, and thus sustainability needs to be carefully managed. Box 4 provides a snapshot of the coco- nut industry in Sri Lanka. BOX 4 Spotlight on the coconut sector Coconut has evolved from a primary food crop to an industrial crop in recent years, as recognition of its health and nutritional benefits boosted global demand. Health-conscious customers, looking for an alternative to dairy and vegetable-based oil, have turned to products such as coconut to fill the gap. The diet fad and celebrity endorsements have helped boost demand for a variety of coconut products across the world. This global trend has created opportunities for Sri Lanka, which is the world’s fifth largest coconut producer. Coconut and coconut by-products make up 12 percent of all agricultural produce in Sri Lanka. Sri Lanka is best known for desiccated coconut and coconut oil; it is a market leader in desiccated coconut, which accounted for 18 percent of total export income in 2016. Export income from coconut and coconut-based products quadrupled between 2006 and 2016. Coconut production and processing is concentrated in the districts of Kurunegala, Puttalam, and Gampaha, which form the famous “coconut triangle.” These districts in the North Western Province cover about 66 percent of total coconut acreage. Demand for coconut-based products from outside of Sri Lanka has led to a growing number of industries in coconut manufac- turing. Fresh coconuts have always been abundant in Sri Lanka but they are rarely used for purposes other than traditional cook- ing or hair care. Sri Lanka coconuts are less sweet than the varieties found in other Southeast Asian countries, but the oil is better suited for producing butter and spreads. Innovations such as those by the Coconut Research Institute, which created a variety of king coconut that retains the taste of coconut water for a shelf life of six months, have also helped exports. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 39 Expanded production, improved quality, and increased investments are needed to further benefit from this global trend, which can help increase agricultural incomes. Sri Lanka produced about 3 billion nuts in 2016, but nearly two-thirds of production went toward domestic consumption. Domestic demand tends to be inelastic, as coconut is an essential part of the traditional cuisine. The challenge will be to increase production above and beyond the stable demands of the domestic market. Securing cultivable land is key, and because coconut land tends to be flat, there is competition for suitable land from housing and real estate. The area within the coconut triangle, situated to the north and east of Colombo, is valuable real estate and therefore creates pressure for develop- ment. Unlike tea and rubber production, coconut production is mostly carried out by smallholder farmers, who face difficulties in raising the large investments required for the long coconut production cycle (five or more years) and for processing coconuts into higher-value-added products. Sources: Based on Coconut Cooperative Blog 2017; Coconut Research Institute n.d.; Financial Times 2015; Roar Media 2016. In the estate sector, poverty reduction was helped by increases in earnings, supported by the introduc- tion of a budgetary relief allowance, among other things. The main source of income for estate residents is wage employment on plantations. Wage payments are governed by collective agreements between plan- tation firms and trade unions that stipulate four components of minimum wages: a minimum dai- FIGURE 30 Minimum wage rates in the estate ly wage, a daily attendance incentive, a fixed-price sector, 2010–16 share supplement, and a newly introduced budget- 450 ary relief allowance (figure 30). The minimum dai- 400 ly wage of plantation workers was raised substan- 350 300 tially in 2009. The daily attendance wage is paid 250 to workers whose attendance amounts to at least Rs 200 75 percent of the number of days of work offered 150 per month, excluding statutory holidays. However, 100 despite this increase, wages remain very low, and 50 are often below Rs 20,000 per month. 0 2010 2011 2012 2013 2014 2015 2016 Finally, natural disasters have been occurring with Minimum daily rate of wages Daily attendance incentive Daily price share supplement Budgetary relief allowance increased intensity, introducing further volatil- ity and constraining growth in the agricultural Source: Department of Labour 2016. sector. Floods, droughts, and landslides have been occurring with greater intensity over the last decade (figure 31). The prolonged drought event that start- ed in 2016 and lasted through a good part of 2017 was arguably the worst drought in 40 years, affecting almost 4 million people. The full impact of this event is likely not reflected in the latest survey because data collection took place in 2016 only. Widespread crop failure was reported during this time, which led to large income drops. The prices of staple food increased, and food security deteriorated signifi- cantly. This led households to resort to negative coping strategies, such as the sale of assets (WFP 2017). Better management of climate-related risks and enhanced resilience will be needed to reduce the impact of disasters on livelihoods. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 40 FIGURE 31 People affected by droughts (left), floods (center), and landslides (right), 1980 – 2016 3.5 1.4 140 3.0 1.2 120 People affected (thousands) People affected (millions) People affected (millions) 2.5 1.0 100 2.0 0.8 80 1.5 0.6 60 1.0 0.4 40 0.5 0.2 20 0.0 0 0.0 0 0 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 Source: DesInventar database, https://www.desinventar.net/. Recent trends in the nonagriculture sector: Growth in services underpinned by tourism sector Recent economic growth boosted demand for labor, with many new jobs created in construction, followed by trade, manufacturing, and transportation. The construction sector added almost 80,000 jobs and led job creation between 2013 and 2016. Trade and manufacturing each added around 55,000 jobs, and the trans- port sector grew by 48,000 workers. Accommodation and food services added almost 32,000 workers (fig- FIGURE 32 Net job creation by sector (in thousands), ure 32). Most of these sectors employ a dispropor- 2013 – 16 Construction tionately large share of low-skilled workers, which Trade helped increase incomes of poorer households. Manufacturing Transportation Household activities The tourism sector exhibited remarkable growth Education Public administration in the last decade and contributed to the expan- Accommodation & food services sion of the services sector. Sri Lanka enjoyed an Financial & insurance activities Other service activities impressive growth in the number of tourist arrivals, Human health & social work especially after 2012. The number of visitors sur- Administrative & support service Information & communication passed 2 million in 2016, and gross tourism earn- Other ings amounted to $4 billion in 2017 (figure 33). The Professional, scientific & technical Mining & quarrying sector is thus the third highest foreign exchange Agriculture earner, after remittances and textiles/garments. -180 -160 -40 -20 0 20 40 60 80 The total contribution of tourism to GDP was esti- Number of jobs (thousands) mated to be around 11.6 percent in 2017 (WTTC 2018). Source: DCS, Labour Force Survey Annual Reports, 2013 – 16. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 41 The rapid growth in the sector was facilitated by infrastructure investments that significantly reduced travel time. In particular, the completion of the Southern Expressway halved the travel time between Colombo and the southern coastal area. Tourist activities are increasingly concentrated in and around Colombo and the Southern Province: Colombo remains the main port of entry, and the towns along the southern coast are known for their beautiful beaches and idyllic scenery. Tourism activities also spread more broadly across the country, such as to the High Country and near the ancient cities. The number of overall guest nights in the Northern and Eastern Provinces is very low, accounting for less than 5 per- cent of the total (figure 34). FIGURE 33 Tourist arrivals and gross tourist receipts, FIGURE 34 Foreign guest nights in graded 2008 – 17 accommodation establishments, by region, 2012 – 17 2.5 5 4.5 Number of foreign guest nights Number of tourist arrivals (millions) Gross tourist receipts (billion $) 4.0 2.0 4 3.5 3.0 (millions) 1.5 3 2.5 2.0 1.0 2 1.5 1.0 0.5 0.5 1 0 2012 2013 2014 2015 2016 2017 0 0 Colombo City Greater Colombo 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 South coast East coast Total tourist arrivals High country Ancient cities Gross tourist receipts Northern region Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. The tourism industry has potential to accelerate FIGURE 35 Employment in the tourism industry, poverty reduction, as it makes intensive use of 2008 – 17 450 low-skilled workers, requires relatively low invest- Number employed in tourism (thousads) 400 ment, and can particularly benefit the rural popu- 350 lation. As Sri Lanka’s popularity as a tourist desti- 300 nation grows, the tourism sector is increasingly an 250 important contributor to job creation. The estimat- ed sum of direct and indirect employment in the 200 tourism industry increased from about 162,000 in 150 2012 to over 400,000 in 2019 (figure 35). The sector 100 has large job creation potential: it can provide flex- 50 ible, part-time jobs to rural and urban populations, 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 and thus could particularly benefit women and help increase Sri Lanka’s persistently low female labor Direct Employment Estimated Indirect Employment force participation. As a labor-intensive industry, Source: Central Bank, “Economic and Social Statistics of Sri Lanka,” various years. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 42 tourism provides a range of different employment opportunities across the skills spectrum, including the less-skilled and unskilled segments of the labor force, such as youth. In addition, tourism-related activities could help households complement their primary source of live- lihood, especially in rural areas. Tourism tends to thrive in places that have a warm climate, rich cul- tural heritage, inspiring landscapes, and abundant biodiversity — factors that are particularly appar- FIGURE 36 Airbnb offerings by district, April ent in Sri Lanka’s rural areas. Tourism has a long 2016 – July 2017 13 and diversified supply chain, as it includes many Ampara different output activities and inputs. Spending by 12 Anuradhapura Badulla tourists can benefit a wide range of sectors such as Batticaloa small-scale agriculture, handicrafts, transport, and 11 other services, all of which can directly contrib- Colombo 10 ute to poverty reduction. The role of the “sharing economy” has also become more prominent in Sri 9 Lanka, as seen from the rising number of Airbnb 8 Number of units (thousads) rental units available in key tourist destinations Galle (figure 36) and the wide availability of ride-shar- 7 ing services such as Uber and Pickme. 6 Gampaha Meanwhile, some of the jobs created in the trans- Hambantota 5 Jaffna port sector were the result of increased demand Kalutara Kandy for alternative transport services. Buses are the 4 Kegalle Kilinochchi main modality of mass transportation in Sri Lanka. Kurunegala 3 Mannar Public buses command about 40 percent of the mar- Matale Matara ket share in urban areas and 60 percent in rural Moneragala 2 areas. While bus fares are very affordable, service Mullaitivu Nuwara Eliya is known for being slow and inefficient, with long 1 Polonnaruwa Puttalam wait times and frequent accidents. Safety concerns Ratnapura 0 Trincomalee have been raised by female commuters. Private bus April 2016 Sept. 2016 April 2017 July 2017 services suffer from similar issues. This has led to Source: Airbnb data from http://tomslee.net. a shift toward heavier use of private and paratransit vehicles, including personal cars and three-wheel- ers, contributing to employment expansion in the transport sector. This trend has helped employ a large number of low-skilled, prime-age men; the longer-term chal- lenge will be to better invest in the skills of this workforce while at the same time improving the availability and quality of transport services. The market served by three-wheelers has been expand- ing, owing to their ready availability and flexibility. Three-wheelers have helped fill a gap in transport SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 43 services, with their numbers exceeding 1 million as of 2017, 17 but this growth has also led to congestion and traffic safety issues. A recent study showed that most three-wheeler drivers have low levels of edu- cation (IPS 2019), and they constitute part of Sri Lanka’s large informal workforce. The challenge will be to increase the skills and employability of prime-age job seekers, improve job matching, particular- ly given persistent labor shortages in many sectors (DCS 2017b), and offer wages and working conditions that can attract youth as well as middle-aged men to these jobs. Trends in nonlabor income: Localized impacts of Samurdhi and remittances Nonlabor income generally consists of transfers from public or private sources. Public sources gen- erally include payments from social protection programs, such as payments from social assistance pro- grams, elderly support, or disability payments, while private sources mainly comprise remittances from family members or relatives outside of the household. In Sri Lanka, the two most significant sources of transfers are from the flagship social assistance program, Samurdhi, and remittances received from a large contingent of migrants working abroad. Samurdhi, Sri Lanka’s largest poverty alleviation program, is an integrated welfare program that pro- vides cash transfers, food stamps, and housing assistance. The cash transfer component is a means-test- ed program that bases eligibility on self-reported income. However, the income-based cutoffs are not easy to implement, given high rates of informal employment; and much discretion is applied in prac- tice (Walker 2018). Moreover, reliance on manual registration and a weak beneficiary identification sys- tem have resulted in poor coverage and targeting. The Samurdhi program underwent a significant expansion in 2015, which led to large increases in benefit amounts. The annual program expenditure almost tripled, from about Rs 15 billion to Rs 40 billion, between 2014 and 2015 (figure 37). The increase in coverage was comparatively limited, from 42 to 45 percent of the poor according to the national poverty line (figure 38). Most resources were used to increase the benefit amounts of existing beneficiaries. In fact, not only were all the Samurdhi benefit brackets at least doubled — from Rs 210, Rs 750, Rs 1,200, and Rs 1,500 to Rs 420, Rs 1,500, Rs 2,500, and Rs 3,500 — but the share of beneficiaries receiving higher values also increased significantly. The number of beneficiaries in the lowest bracket decreased by 6 percent, in the next higher bracket by 56 percent, and in the second highest bracket by 7 percent, whereas the number in the highest bracket increased from 14,889 to 594,594. The appendix shows the number of beneficiaries by province and by benefit category. 17. Not all three-wheelers are used for the purpose of providing transport services to others. The Economic Census 2013/14 conducted a separate sample survey, which revealed that about 47.2 percent of all three-wheelers were used for passenger transport, followed by 35.9 percent used for household purposes (DCS 2017a). SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 44 Average benefit levels among beneficiaries increased from around Rs 770 to Rs 2,580, according to esti- mates from the 2016 HIES data. However, benefit amounts are still largely inadequate and are not indexed to inflation, which means that they continue to erode as prices in the economy rise. FIGURE 37 Samurdhi budget and beneficiaries, FIGURE 38 Coverage of the poor by Samurdhi, 2011 – 16 2012/13 and 2016 40 1.60 50 Number of beneficiaries (thousands) 35 1.55 40 Expenditure (million Rs) 30 1.50 30 25 Percent 20 1.45 20 15 1.40 10 10 1.35 5 0 National Poor (national Poor ($3.20 0 1.30 poverty line) poverty line) 2011 2012 2013 2014 2015 2016 2012/13 2016 Beneficiaries Expenditure Source: Staff calculation using HIES 2012/13 and 2016. Source: Department of Divineguma Development 2014, 2015, 2016. Note: Y-axis is truncated. Despite high poverty rates, only 8 percent of the estate population was covered by Samurdhi in 2016. In rural and urban areas, where poverty rates were much lower, the program covered 21 and 12 percent of the population, respectively. Moreover, the average benefit received was significantly lower in the estate sector, than in urban or rural areas. It was previously suggested that the likelihood of enrollment in the Samurdhi program is lower in the estate sector because Samurdhi officers rarely visit the estates (World Bank 2017). The targeting performance of the program remains weak, even after the program’s expansion. About 40 percent of the poor according to the $3.20 poverty line receive assistance from Samurdhi, whereas the share among nonpoor households was 16 percent. (Under the national definition of poverty, the shares are 45 percent of the poor and 18 percent of the nonpoor). Among the poorest 10 percent of households in which at least one member receives Samurdhi, average monthly benefit levels are Rs 2,423, equal to about 12 percent of monthly consumption for these households. Weak targeting of the poor leads to low effectiveness in reducing poverty and sends a large share of public resources to the nonpoor. The contribution of Samurdhi to poverty reduction was analyzed using the decomposition method presented in section 3.1, applied at the province level. In all provinces, non- farm income was the most important contributor to poverty reduction (figure 39). Samurdhi’s role was overall modest with the exception of the Northern Province, where it was the second most important factor contributing to poverty reduction. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 45 FIGURE 39 Contribution of demographic and socioeconomic factors to poverty reduction, by province 50 6 40 4 30 20 2 Percentage point change 10 0 0 -10 Percent -20 -2 -30 -4 -40 -50 -6 -60 -70 -8 -80 -90 -10 Western Central Southern Northern Eastern North North Uva Sabaragamuwa Western Central Consumption-to-income ratio Percent adults Percent adults working Percent working in agriculture Agricultural income Percent working in nonagricultural Nonagricultural income Samurdhi income Remittances Other nonlabor income In-kind income Total percentage point change in poverty rate Source: World Bank staff calculation using HIES 2012/13 and 2016. The targeting performance and benefit levels of other social protection programs remain low (figure 40). Besides Samurdhi, there is a range of other welfare programs: pensions, disability payments, elder- ly payments, disease payments, a school food program, the Triposha food program, and scholarships. FIGURE 40 Coverage (percent, left) and monthly benefit levels (Rs, right) of social protection programs Disability / relief Disability / relief payments payments Elderly payment Elderly payment Pension payment Pension payment Samurdhi Samurdhi Educational & Educational & Scholarships Scholarships School food program School food program Tuberculosis/kidney Tuberculosis/kidney diseases payment diseases payment Triposha food Triposha food program program 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 Percent Rs (thousands) National Poor Nonpoor National Poor Nonpoor Source: World Bank staff calculation using HIES 2016. Note: Benefit levels indicate amount received per month. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 46 Targeting performance is low, as most of these programs reach less than 10 percent of poor households. For example, less than 12 percent of poor households received elderly payments in 2016. A large share of public resources allocated to welfare programs continue to go to the nonpoor. Pensions are more likely to be received by rich households than by poor: only 2 percent of the poorest 10 percent receive pension benefits, compared to 18 percent of the richest 10 percent. This finding is not surprising, given that pen- sions are an entitlement mainly reserved for formal public sector workers. Benefit levels are also inade- quate for all programs except pensions, which provide average benefits of Rs 24,187 per month. Undercoverage is more than 40 percent and leakage is over 80 percent for all social assistance pro- grams. Undercoverage, defined as the percentage of the poor not receiving a transfer, was 44 percent in 2016. Leakage is defined as the receipt of transfers by nonpoor beneficiaries. For social assistance pro- grams overall, leakage is 73 percent in terms of number of beneficiaries and 66 percent in terms of trans- fer amount. The targeting differential, defined as the difference between coverage rate of the poor and the participation rate of the nonpoor, is estimated to be 16 percent. Social assistance programs have a modest impact on poverty reduction. This result is simulated by esti- mating the poverty rate in the absence of the relevant transfer, assuming that household welfare will diminish by the full value of the transfer. Compared to a baseline $3.20 poverty rate of 11 percent in 2016, eliminating all social assistance programs would result in a small increase in poverty, raising it to 13 percent. In the absence of all social protection programs (including pensions), poverty would increase to 15 percent. 18 Given Sri Lanka’s aging demographic profile, there TABLE 9 Living arrangements of the elderly are increasing concerns about income support Living situation Share (%) $3.20 Poverty rate for the elderly. The share of the elderly (individ- Alone 6.9 4.7 uals ages 65 and above) is projected to surpass 20 With spouse 15.7 7.6 percent of the population by 2045, up from 11 per- With other elderly cent in 2020. 19 Elderly poverty was 10.8 percent in 1.6 8.4 (not spouse) 2016, similar to the national average. More than 60 With children 60.7 12.3 percent of the elderly live with their children. Only With other relatives 14.9 11.6 about 7 percent live alone, and the poverty rate for With others 0.2 18.0 this group is low (table 9). There are concerns about All elderly 100.0 10.8 the lack of formal income support schemes — the coverage of elderly payments is low, and there is Source: World Bank staff calculation using HIES 2016. 18. Estimates of undercoverage, leakages, and the impact of programs on poverty measures were provided by the World Bank data compilation ASPIRE: The Atlas of Social Protection Indicators of Resilience and Equity, http://datatopics.worldbank.org/ aspire/, which is gratefully acknowledged. 19. Estimates are from United Nations population projections (median variant), https://population.un.org/wpp/Download/ Standard/Population/. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 47 otherwise little available in the way of old-age income support. The elderly benefit provides Rs 2,000 per month to individuals ages 70 and above with no income; but the benefits reach only 19.1 percent of the target population. The benefit amount is less than half of the already low national poverty line and unlikely to offer effective relief. The coverage of pensions remains low outside of the public sector, where employees are entitled to noncontributory pensions. Beyond strengthening social protection, encouraging longer working lives among the elderly can help prevent them from falling into poverty. Male labor force participation increases steeply until age 25, after which it remains at around 95 percent until it drops precipitously at around age 55. Those work- FIGURE 41 Male and female labor force participation, ing in the agriculture sector tend to remain in the by age 100 labor force longer than others. Women exit the labor force even sooner, after reaching peak par- 80 ticipation around age 45 (figure 41). 20 60 Percent 40 Remittances are an important source of income for 20 some poor households, and there have been nota- 0 ble changes in migration trends in recent years. 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 Sri Lanka has historically been a migrant-sending Female Male country, though the skills and gender composition Source: World Bank staff calculation using HIES 2016. of migrants are undergoing structural changes amid a decline in departures. Figure 42 shows annual FIGURE 42 Number of migrants, by gender and route departures for foreign employment by gender and of employment, 1997 – 17 route of employment (self-basis or through agen- 320 Number of migrants (thousands) 280 cy) between 1997 and 2017. The annual outflow of 240 migrants continued to rise through 2014, when it 200 surpassed 300,000, but has since been falling rap- 160 120 idly. Housemaids have made up a large share of all 80 migrants, but their numbers have been decreasing 40 since 2005. Generally speaking, such a decline is 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 to be expected; as Sri Lanka’s economy grows and living standards improve, there is a narrowing of Self-basis male Through agency male the expected wage gap between Sri Lanka and the Self-basis female Through agency female destination country. Source: Sri Lanka Bureau of Foreign Employment. 20. While these figures are not derived from panel data, which would offer a more accurate life-cycle perspective, the patterns of early labor market exit appear to have held over time. Calculations with earlier rounds of the HIES show similar patterns. SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 48 Several trends have emerged more recently: (1) there has been a rapid decline in overall departures since 2013, which is mainly due to a decline in the number of female migrants; (2) migrants are increas- ingly more likely to find employment opportunities on their own compared to other routes, though the trend is much stronger among male migrants; and (3) both men and women have been less likely to FIGURE 43 Number of migrants by skill level, find employment through agencies in recent years. 1994 – 2017 320 280 Number of migrants (thousands) These trends are likely the result of institution- 240 al and structural factors, the impact of which 200 will need continued monitoring. In 2013, the Sri 160 Lanka Bureau of Foreign Employment introduced 120 a Family Background Report (FBR) requirement in 80 an attempt to restrict labor migration of wom- 40 en with young children. The new regulation was 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 strictly enforced and led to an immediate drop in the number of women migrating overseas to work Professional level Skilled Housemaid as housemaids. 21 Meanwhile, the decline in the Middle level Semi-skilled Clerical & related Unskilled number of male agency-led migrants could be the Source: Sri Lanka Bureau of Foreign Employment. result of lower demand for low-skilled labor in the Middle East. Finally, migrants increasingly possess FIGURE 44 Number of departing migrants by province better skills (figure 43). of origin, for 2009, 2012, and 2016 compared to migrants as share of provincial population in 2016 80 3.5 An increasing number of migrants has been com- Migrants in population (percent) 70 Number of departing migrants 3.0 ing from the Eastern Province in recent years, 60 which helped reduce poverty among household 50 2.5 (thousands) members left behind. Most migrants are still from 40 2.0 the populous Western Province, but migrants 30 1.5 from the Eastern Province increased by 35 percent 20 1.0 between 2009 and 2016. This trend likely reflects 10 0 0 a lack of job opportunities (figure 44). Total remit- Central Eastern North Central Northern North Western Sabaragamuwa Southern Uva Western tances received as a share of GDP remained at slight- ly below 9 percent between 2012 and 2016, and had a relatively modest impact on overall pover- 2009 2012 2016 ty reduction (figure 25). However, among house- Percent of population holds in the Eastern Province, remittances were Source: Sri Lanka Bureau of Foreign Employment; population figures from DCS 2012b. 21. The policy has reportedly led to an increase in cases of human trafficking of women who tried to circumvent the regula- tion; there are concerns about the effectiveness of the policy in safeguarding the welfare of children left behind (ILO 2018). SRI LANKA POVERTY UPDATE 3 Drivers of poverty reduction 49 the second biggest factor contributing to poverty reduction (figure 39), which is consistent with the migration trend observed. The migrant-sending strategy may have been a survival strategy for some poor households, but it has not come without a cost. In-depth interviews in ILO (2018) suggest that female migrants continue to look for opportunities overseas to finance the building of a house and to meet education and health needs of their children. In households that have a male migrant, women’s vulnerabilities are intensified by the breakdown of families, which leaves women caring for their children with little to no financial support from their husbands. This trend is particularly worrying in former conflict-affected areas, where there are a lot of female-headed households. 4 Key messages and priorities for poverty reduction and shared prosperity SRI LANKA POVERTY UPDATE 4 Key messages and priorities for poverty reduction 51 Poverty continued to decline in Sri Lanka through 2016, and prosperity was being created and shared. Between 2012/13 and 2016, Sri Lanka observed significant progress in poverty reduction. The poverty rate according to the $3.20 poverty line declined from 16.2 percent in 2012/13 to 11 percent in 2016. Moreover, the depth of poverty also fell, implying an improvement among the poorest of the poor. Sri Lanka’s pov- erty rate compares favorably with poverty rates in peer countries. Encouragingly, improvements in labor income were the main driver of the progress in poverty reduc- tion, and these continued during a period of important structural changes at the sector level. Sustained economic growth translated into better labor market outcomes in the form of higher employment and wages. In a notable shift from previous years, the contribution of agriculture to rural incomes declined, as favorable price trends reversed and productivity stagnated. During the same period, most jobs were created in transport, commerce, and construction, which helped lift the earnings of the less skilled and less well-off. Samurdhi underwent a significant expansion in 2015, but coverage did not change much, and the contribution to poverty reduction was relatively modest. Remittances played a small role com- pared to other factors, but have provided important support to some poor households. Overall, limited progress in meeting structural labor market challenges poses a concern for longer- term progress in poverty reduction and shared prosperity. Many of the new jobs are of low productivity and in sectors with high levels of informality. There was little progress in raising the labor force partic- ipation among women. Participation is almost on par with men for highly educated women (those with a tertiary education) but significantly lower at all other educational attainment levels. Given the high burden of household chores and child care, the opportunity cost of working is high, and the challenges are compounded by the difficulties of finding a decent, well-paying job. A large share of the net employ- ment growth among women came from increased employment in the public sector. The following key areas should be the focus of efforts to accelerate job creation and poverty reduction: • The creation of more and better jobs is a key priority. Accelerating growth and poverty reduc- tion needs to rely on productivity and employment growth that can support broad-based income growth. Labor productivity has been growing strongly but most of the growth came from increases in within-sector productivity and little from reallocation effects. This implies that most of the labor movement occurred from agriculture toward other sectors with low productivity. Supporting the movement of workers toward sectors with higher productivity is important to further promote productivity gains, especially given that progress has slowed in recent years. Accelerating labor reallocation — for example through reducing barriers to internal migration — would also help, as would creating a policy environment that promotes competitiveness and job creation. SRI LANKA POVERTY UPDATE 4 Key messages and priorities for poverty reduction 52 • Measures to increase agricultural productivity will help reduce poverty. Agricultural house- holds have a higher poverty rate due to low productivity and low earnings in the sector. Support to increase paddy productivity and help farmers shift to higher-value crops can help improve rural livelihoods and reduce poverty. • Stronger safety nets are needed to protect the poor and vulnerable. Social protection programs could be better targeted to further reduce poverty. Only a small share of estate households is cov- ered, even though the poverty rate is much higher in this sector. Benefit levels remain inadequate. Efforts to build better targeting and delivery systems and strengthen graduation programs can go a long way toward supporting the poor and most vulnerable. • Narrowing the gap in access to basic services will help achieve equal opportunities for all. Sri Lanka has made great progress in closing the gaps in access to services, but large challenges remain in some areas, such as access to water supply. Spatial disparities are high — between urban and rural areas, and between the Western Province and the rest of the country — a nd contribute to inequality. While Sri Lanka has historically excelled in human capital outcomes, the gap between the poor and nonpoor is wide, and tertiary educational attainment is very low. Progress was slow in the estate sector, where poverty continues to be much higher, and educational and nutritional outcomes much lower, than in the rest of the country. Finally, a revision of the official national poverty line, accompanied by a more accurate classification of geographic sectors, would help reflect the true extent of poverty and vulnerability. The current benchmark for measuring poverty was established almost two decades ago and needs to be revisited to reflect the consumption patterns of today. Similarly, the current definition of geographic sectors relies on administrative boundaries, and several recent studies have shown that official statistics likely under- estimate the actual extent of urbanization. Better data and estimates will enhance our understanding of the patterns of poverty and vulnerability, and thus help in devising better interventions. SRI LANKA POVERTY UPDATE 53 Appendix Number of Samurdhi beneficiaries by allowance category and district, 2014 and 2015   2014 2015 District  Rs 210 Rs 750 Rs 1,200 Rs 1,500 Rs 420 Rs 1,500 Rs 2,500 Rs 3,500 Colombo 1,787 48,650 0 237 14,268 214,945 9,339 17,012 Gampaha 7,731 106,481 4,273 315 74,349 414,593 22,260 45,417 Kalutara 6,241 46,033 11,74 8 115 64,063 23,711 10,995 21,804 Kandy 7,94 4 61041 21002 114 74,409 30,972 15,071 33,807 Matale 3,330 35,743 3,466 194 34,087 15,803 7,862 14,965 Nuwara Eliya 8,124 24,946 6,617 15 84,065 12,419 6,031 13,126 Galle 8,250 55,284 9,126 53 74,893 20,246 11,806 31,254 Matara 4,982 50,076 12,799 32 44,527 23,227 11,137 26,399 Hambantota 6,830 26,812 21,520 5 64,397 14,660 9,772 23,952 Jaffna 0 38,873 10,374 4,660 0 13,418 9,205 31,283 Mannar 0 2,636 7,875 2,655 0 3,020 2,307 7,839 Vavuniya 2,805 8,234 846 102 24,759 2,822 1,599 4,829 Mullaitivu 44 2,088 7,198 1,714 44 1,926 2,614 6,530 Kilinochchi 0 3,109 6,059 2,565 0 2,716 2,050 6,968 Batticaloa 17,296 62,14 3 84 2 164,502 19,764 10,269 32,677 Ampara 14,324 57,720 2,858 108 134,960 20,742 10,404 29,336 Trincomalee 8,175 28,385 2,208 600 74,330 10,218 5,899 15,602 Kurunegala 9,338 120,091 20,030 485 84,555 51,155 27,581 58,440 Puttalam 1,752 38,034 15,671 15 14,559 22,123 9,446 20,959 Anuradhapura 5,671 4 8,124 8,743 474 54,245 23,020 11,507 22,050 Polonnaruwa 4,484 21,742 3,235 112 44,304 7,691 5,781 11,241 Badulla 6,419 27,553 22,956 45 64,087 18,608 9,699 22,126 Moneragala 7,397 30,445 9,209 96 74,306 12,656 8,045 18,677 Ratnapura 12,311 56,983 42,419 143 114,504 25,974 20,698 51,572 Kegalle 8,881 37,866 21,428 33 84,382 19,214 11,869 26,729 Total 154,096 1,039,092 271,734 14,889 145,595 459,643 253,246 594,594 Source: Department of Divineguma Development 2014, 2015, 2016. SRI LANKA POVERTY UPDATE 54 Bibliography ADB (Asian Development Bank). 2011. “Validation Report. DCS (Department of Census and Statistics). 2017b. “Sri Lanka Sri Lanka: Third Water Supply and Sanitation Project.” Labour Demand Survey 2017.” http://www.statistics.gov. Manila. https://www.adb.org/sites/default/files/ lk/industry/Labour_Demand_Survey_2017_Report.pdf. evaluation-document/35540/files/in270-11.pdf. DCS (Department of Census and Statistics). Various years. Azevedo, João Pedro, Gabriela Inchauste, Sergio “Household Income and Expenditure Survey.” Olivieri, Jaime Saavedra, and Hernan Winkler. 2013. http://www.statistics.gov.lk/page.asp?page=Income%20 “Is Labor Income Responsible for Poverty Reduction? and%20Expenditure. A Decomposition Approach.” Policy Research Working DCS (Department of Census and Statistics). Various years. Paper 6414, World Bank, Washington, DC. “Sri Lanka Labour Force Survey: Annual Report.” Azevedo, Joao Pedro, Minh Cong Nguyen, and Viviane http://www.statistics.gov.lk/page.asp?page=Labour%20 Sanfelice. 2012. “Shapley Decomposition by Components Force. of a Welfare Aggregate.” MPRA Paper 85584, Munich DCS (Department of Census and Statistics) and Ministry Personal RePEc Archive. https://mpra.ub.uni-muenchen. of Health, Nutrition and Indigenous Medicine. 2017. “Sri de/85584/1/MPRA_paper_85584.pdf. Lanka Demographic and Health Survey 2016.” http:// Central Bank. Various years. Annual Report. www.statistics.gov.lk/page.asp?page=Health. https://www.cbsl.gov.lk/en/publications/ Department of Divineguma Development. 2014. economic-and-financial-reports/annual-reports. “Performance Report 2014.” http://www.samurdhi.gov.lk/ Central Bank. Various years. “Economic and Social Statistics web/images/cercular/performance%20report%20eng- of Sri Lanka.” https://www.cbsl.gov.lk/en/publications/ lish.pdf. other-publications/statistical-publications/economic-an Department of Divineguma Development. 2015. d-social-statistics-of-sri-lanka. “Performance Report 2015.” http://www.samurdhi.gov.lk/ Ceriani, Lidia, Gabriela Inchauste, and Sergio Olivieri. 2015. web/images/cercular/divineguma%20performace%20 “Understanding Poverty Reduction in Sri Lanka: Evidence report%202015%20english.pdf. from 2002 to 2012/13.” Policy Research Working Paper Department of Divineguma Development. 2016. 7446, World Bank, Washington, DC. “Performance Report 2016.” https://www.parliament.lk/ Coconut Cooperative Blog. 2017. “New Coconut Triangle uploads/documents/paperspresented/performance-repor in Sri Lanka.” January 22, 2017. t-department-of-divineguma-development-2016.pdf. https://www.thecoconutcoop.com/blog/2017/1/22/a-ne Department of Labour. 2016. “Labour Statistics.” w-coconut-triangle-in-sri-lanka. http://www.labourdept.gov.lk/images/PDF_upload/sta- Coconut Research Institute. n.d. “Strategic Plan 2016 – 2020.” tistics/stats2016.pdf. https://www.cri.gov.lk/web/images/stories/statistics/cri_ Elledge, Myles F., Sumal Nandasena, Michael J. Phillips, srategic_plan.pdf. and Vanessa E. Thornburg. 2010. “Environmental Datt, Gaurav, and Martin Ravallion. 1992. “Growth and Health Risk and the Use of Biomass Stoves in Redistribution Components of Changes in Poverty Sri Lanka.” RTI Research Brief, RTI International, Measures: A Decomposition with Applications to Brazil Research Triangle Park, NC. and India in the 1980s.” Journal of Development Economics https://www.rti.org/sites/default/files/resources/ 38, no. 2: 275 – 95. rti-publication-file-daa5543c-b021-4757-bc41- DCS (Department of Census and Statistics). 2011. “Census ec6bb736ed7e.pdf. of Population and Housing — Population Tables.” Elledge, Myles F., Michael J. Phillips, Vanessa E. http://www.statistics.gov.lk/PopHouSat/CPH2011/index. Thornburg, Kibri H. Everett, and Sumal Nandasena. 2012. php?fileName=Activities/TentativelistofPublications. “A Profile of Biomass Stove Use in Sri Lanka.” International DCS (Department of Census and Statistics). 2012a. “Census Journal of Environmental Research and Public Health 9, no. of Population and Housing — Housing Tables.” 4: 1097 – 1110. http://www.statistics.gov.lk/PopHouSat/CPH2011/Pages/ Financial Times. 2015. “Asia’s Coconuts Activities/Reports/Finalhousing.pdf. Go Global — But What about the Farmers?” DCS (Department of Census and Statistics). 2012b. “Census February 26, 2015. https://www.ft.com/content/ of Population and Housing: 2012.” Final report. a1a635d0-4951-37d8-b0e4-32bf7eaf7508. http://www.statistics.gov.lk/PopHouSat/CPH2011/Pages/ Hexa Research. 2019. “Coconut Water Market Size and Activities/Reports/FinalReport/FinalReportE.pdf. Forecast, by Form (Liquid, Powder), by Packaging DCS (Department of Census and Statistics). 2017a. (Tetra Pack, Plastic Bottle, Others), by Distribution “Economic Census 2013/14: Final Report on Informal Channel (Offline, Online), by Region, and Trend Non-Agricultural Activities.” http://www.statistics.gov. Analysis, 2015 – 2025.” https://www.hexaresearch.com/ lk/Economic/FinalReport_Informal_NonAgri.pdf. research-report/coconut-water-market. SRI LANKA POVERTY UPDATE Bibliography 55 ILO (International Labour Organization). 2018. “Sri Lankan Shapley, L. S. 1953. “A Value for n-Person Games.” Female Migrant Workers and the Family Background In Contributions to the Theory of Games. Vol. 2. Edited Report.” International Labour Organization. by H. Kuhn and A. Tucker, 307 – 17. Princeton, NJ: https://www.ilo.org/colombo/whatwedo/publications/ Princeton University Press. WCMS_632484/lang--en/index.htm. Shorrocks, A. F. 2013. “Decomposition Procedures for Inchauste, G., J. P. Azevedo, B. Essama-Nssah, S. Olivieri, Distributional Analysis: A Unified Framework Based T. V. Nguyen, J. Saavedra-Chanduvi, and H. Winkler. on the Shapley Value.” Journal of Economic Inequality 11, 2014. Understanding Changes in Poverty. Directions no. 1: 99 – 126. in Development. Washington, DC: World Bank. UN-Habitat. 2018. “State of Sri Lankan Cities IPS (Institute of Policy Studies of Sri Lanka). 2019. “Allowing 2018.” UN-Habitat. https://unhabitat.org/ Youth to Tuk-Tuk or Not Tuk-Tuk: Should Access the-state-of-sri-lankan-cities-2018-report. to Three Wheeler Market in Sri Lanka Be Regulated?” Walker, Thomas. 2018. “Policy Note: Social Protection in Sri Talking Economics blog, January 30, 2019. http://www. Lanka.” Unpublished draft. ips.lk/talkingeconomics/2019/01/30/allowing-yout WFP (World Food Programme). 2017. “Sri Lanka: Joint h-to-tuk-tuk-or-not-tuk-tuk-should-access Assessment of Drought Impact on Food Security and -to-three-wheeler-market-in-sri-lanka-be-regulated/. Livelihoods.” https://www.wfp.org/publications/ Nandasena, Sumal, Ananda Rajitha Wickremasinghe, Sri_Lanka_Joint_Drought_Assessment. and Nalini Sathiakumar. 2012. “Biomass Fuel Use for World Bank. 2017. “Multisectoral Nutrition Assessment in Sri Cooking in Sri Lanka: Analysis of Data from National Lanka’s Estate Sector.” World Bank, Washington, DC. Demographic Health Surveys.” American Journal World Bank. Forthcoming. “The COVID-19 Impact of Industrial Medicine 55, no. 12: 1122 – 28. on Livelihoods and Poverty in Sri Lanka.” Background Newhouse, David Locke, Pablo Suarez-Becerra, and Dung paper to Sri Lanka Poverty Assessment. World Bank, Thi Thuy Doan. 2016. “Sri Lanka Poverty and Welfare: Washington, DC. Recent Progress and Remaining Challenges.” World Bank, WTTC (World Travel and Tourism Council). 2018. “Travel Washington, DC. and Tourism: Economic Impact 2018: Sri Lanka.” https:// Roar Media. 2016. “Coconuts Are Trending: How Sri Lanka www.wttc.org/economic-impact/country-analysis/ Is Cashing In (Part II).” October 14, 2016. https://roar. country-reports/. media/english/life/economy/coconuts-trending-sri-lank a-cashing-part-ii/.