THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA Background note to Sri Lanka Poverty Assessment THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA Background note 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 Acknowledgements   6 Abbreviations   6 Executive summary   7 1. Context   8 2. The Sri Lanka Labor Market and the COVID-19 Shock    12 Low earnings and lack of safety net, partly due to high informality, lead to high levels of vulnerability    13 The COVID-19 crisis led to widespread job losses   14 3. The Distributional Impact of COVID-19    18 Earnings losses led to a significant increase in poverty   19 Livelihood support programs helped absorb the shock, but better targeting could have enhanced their effectiveness to reduce poverty   20 The new poor are more urban, but the crisis did not fundamentally change the nature of poverty   22 Inequalities will likely widen, with potential consequences for the long term   25 4. Conclusion    27 Annex 1 Summary of Relief Measures Used in Simulation   30 Bibliography   31 Figures Figure 1 Daily total (left) and newly confirmed (right) COVID-19 cases in Sri Lanka   9 Figure 2 Real GDP growth in Sri Lanka (%)   10 Figure 3 Increase in poverty from hypothetical consumption shocks   14 Figure 4 Share of jobs lost by sector   16 Figure 5 Share of displaced workers by employment status   16 Figure 6 Share of jobs lost by province   17 Figure 7 Share of displaced workers by education   17 Figure 8 Average income loss across the income distribution   19 Figure 9 Poverty impact of COVID-19 crisis   20 Figure 10 Distribution of poor and nonpoor people by province   22 Figure 11 Number of poor by province, pre- and post-COVID-19   23 Figure 12 $3.20 poverty rate by province   23 Figure 13 Map of the distribution of new poor   23 Figure 14 Distribution of working-age population (age 15+) by sector of employment   24 Figure 15  $3.20 poverty rates by household head’s employment sector   24 Figure 16 Gini index in Sri Lanka and peer countries (pre-COVID-19)   25 Tables Table 1 Household characteristics by poor and nonpoor groups   25 THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 6 Acknowledgements This report was prepared as a background note to "Sri Lanka Poverty Assessment. Accelerating Economic Trans- formation", and was written by Yeon Soo Kim (Senior Economist, World Bank) and Tiloka de Silva (Consultant and Senior Lecturer, University of Moratuwa). The work was carried out under the overall guidance of Faris H. Hadad-Zervos (Country Director for Sri Lanka, Nepal and Maldives), Zoubida Allaoua (Regional Director, South Asia), Chiyo Kanda (Country Manager, Sri Lanka and Maldives), Tae Hyun Lee (Lead Country Economist), and Andrew Dabalen (Practice Manager, Poverty and Equity). The team is grateful for comments and feedback from Ambar Narayan (Lead Economist, World Bank), Nistha Sinha (Senior Economist), Thomas Walker (Senior Econo- mist, World Bank), and Shalika Subasinghe (Senior Consultant, World Bank). Any remaining errors are the respon- sibility of the authors. Abbreviations BAU business-as-usual GCE General Certificate of Education GDP gross domestic product HIES Household Income and Expenditure Survey PPP purchasing power parity THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 7 Executive summary The COVID-19 crisis has dealt a significant shock to Sri Lanka’s economy and people. This note exam- ines the expected impact on poverty and inequality amid widespread job and earnings losses. While poverty was relatively low in Sri Lanka prior to the pandemic, pre-existing vulnerabilities were high, partly owing to high levels of informality. Many workers do not have access to employment protection or other job-related social protection benefits, making them vulnerable during times of economic cri- sis. Simulation-based results suggest that the crisis increased the international $3.20 poverty rate from 9.2 percent in 2019 to 11.7 percent in 2020; this change translates into over 500,000 new poor people. Livelihoods support programs and various relief measures implemented by the government over the course of the pandemic are expected to have mitigated the labor market shock. Inequality is expected to increase in the short run because of the unequal distribution of the shock. Moreover, reduced social mobility — as a consequence of widening disparities in access to education for example — could increase inequality in the long term. Policy measures could aim to strike a balance between those that support a resilient recovery and those that aim to include the most vulnerable in the recovery process. Achieving this balance will help reverse the impact of the pandemic and mitigate its consequences for inequality. Shifting toward a more adaptive social protection system would allow much needed support to be scaled up quickly and effectively in times of crisis. 1. Context THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 1. Context 9 Sri Lanka confirmed its first case of COVID-19 on January 27, 2020, but the first large outbreak did not occur until mid-March in the same year. The government reacted swiftly by closing the airport and halting all inbound and outbound travel. The initial outbreak led to an island-wide lockdown that lasted for a few weeks, from March 20 to April 16; thereafter the restrictions were gradually relaxed. The lock- down was completely lifted after June 28. While in effect, the first national lockdown entailed near-to- tal restrictions to movement. The situation appeared to be relatively contained until October, when the breakout of an interconnected cluster starting from an apparel industry factory in Minuwangoda, Gampaha, occurred. This outbreak, and another one originating from the Peliyagoda fish market, led to an exponential increase in cases. Localized lockdowns were reintroduced starting on October 4, 2020, primarily in high-risk areas in the highly urbanized and populous Western, Central, and North-Western Provinces. The number of newly confirmed cases started to fall significantly in late February, after sus- taining an increase amid a second wave. Figure 1 shows the evolution of daily total and newly confirmed COVID-19 cases in Sri Lanka. FIGURE 1 Daily total (left) and newly confirmed (right) COVID-19 cases in Sri Lanka 1,000 90,000 New confirmed cases and monthly average 900 80,000 800 70,000 Total confirmed casesy 700 60,000 600 50,000 500 40,000 400 30,000 300 200 20,000 100 10,000 0 0 03-10-2020 03-24-2020 04-07-2020 04-21-2020 05-05-2020 05-19-2020 06-02-2020 06-16-2020 06-30-2020 07-14-2020 07-28-2020 08-11-2020 08-25-2020 09-08-2020 09-22-2020 10-06-2020 10-20-2020 11-03-2020 11-17-2020 12-01-2020 12-15-2020 12-29-2020 01-12-2021 01-26-2021 02-09-2021 02-23-2021 New confirmed Monthly average Total confirmed (RHS) Source: World Bank staff illustration using data from Our World in Data, “Sri Lanka: Coronavirus Pandemic Country Profile,” https://ourworldindata.org/coronavirus/ country/sri-lanka?country=~LKA The pandemic has dealt a significant shock to the Sri Lanka economy. Sri Lanka’s economy grew at an average 5.3 percent per year following the end of the civil war in 2009. Prior to the onset of the pandemic, Sri Lanka’s economy was projected to grow at 3.3 percent in 2020 (World Bank 2020a) — this pre-COVID-19 scenario will henceforth be referred to as the business-as-usual (BAU) scenario. However, the COVID-19 pandemic has brought about a severe global economic crisis, which has also affected the Sri Lanka econ- omy: according to the latest published national accounts data, gross domestic product (GDP) contracted by 3.6 percent in 2020 (figure 2). THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 1. Context 10 FIGURE 2 Real GDP growth in Sri Lanka (%) The severity of the pandemic’s economic shock 6 is confirmed by evidence from firms and house- holds, though data are still relatively limited. The 4 Department of Labour conducted an online survey 2 in April 2020 at the height of the first national lock- Percent down and found that more than half of business- 0 es had closed their operations and that almost two -2 thirds of employees were out of work. The impact -4 was severe across most industries, except agricul- 2015 2016 2017 2018 2019 2020 ture (Department of Labour 2020). A second-phase Real GDP growth (%) BAU scenario Projections survey is being undertaken as of March 2021. An Sources: Department of Census and Statistics. World Bank 2020a (for 2020 BAU projection). International Finance Corporation phone survey Note: GDP = gross domestic product; BAU = business as usual. conducted in June/July 2020 focused on formal small and medium enterprises and found wide- spread business impacts, including significant decreases in demand, difficulties meeting operational expenses and accessing financial services, and some layoffs (IFC 2020). While not representative of the enterprise sector, these data help explain the impact of the crisis on firms’ activities and on employment at different points in time. Labour Force Survey data through the second quarter of 2020 show a slight uptick in the unemployment rate and a small downward trend in employment. 1 The full extent of the crisis may become apparent only with subsequent rounds of data. The crisis threatens to reverse significant welfare gains from recent years, though a gradual recovery could return the country to a path of poverty reduction. Prior to the pandemic, growth was inclusive and poverty reduction strong; the poverty rate at $3.20 per day (in 2011 purchasing power parity terms) declined from 16.2 percent in 2012/13 to 11 percent in 2016, according to the latest available survey data. The $3.20 poverty line is the World Bank’s international poverty line applicable to lower middle-income countries, such as Sri Lanka. While a more recent household survey was conducted in 2019, 2 the data have not been released and would still only provide information on the prevailing situation prior to the pandemic. Adequate and timely data during crises are often not available or become available only with delays. Thus, it can be difficult to understand the impact of shocks, especially when they are wide-rang- ing and evolve quickly over time. Importantly, this information gap could lead to delays in the imple- mentation of interventions that seek to mitigate the welfare shock on households. 1. Unemployed slightly increased from 4.7 percent in Q1 2019 to 5.7 percent in Q1 2020, and from 4.9 percent in Q2 2019 to 5.4 percent in Q2 2020. Total employment fell in the first half of the year from about 8.2 million in 2019 to 8 million in 2020. 2. This is the Household Income and Expenditure Survey 2019, conducted by the Department of Census and Statistics. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 1. Context 11 The rest of this note is dedicated to an assessment of the short-term distributional impact of the COVID-19 pandemic. Microsimulations can help fill the data void by modeling the expected impact of a shock on the full income distribution through assumptions on income channels. The particular approach is capa- ble of simulating adjustments in employment and different types of earnings across the entire income distribution while having distinctly lower data requirements than, for example, Computable General Equilibrium (CGE) models. The primary data source for the simulation is the official Household Income and Expenditure Survey (HIES) 2016, which is conducted by the Department of Census and Statistics and also provides the data used for generating poverty estimates. The survey is representative at the nation- al and district levels and gathers information on demographic and socioeconomic characteristics of individuals and households as well as detailed information on income sources and consumption. Some preliminary findings from a World Bank rapid phone survey conducted in late 2020 are also referenced. 2. The Sri Lanka Labor Market and the COVID-19 Shock THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 2. The Sri Lanka Labor Market and the COVID-19 Shock 13 Low earnings and lack of safety net, partly due to high informality, lead to high levels of vulnerability Sri Lanka has an economically active population of around 8.6 million, of whom roughly 8.2 mil- lion were employed in 2019. About a quarter of workers are engaged in agriculture, another 28 percent in industries, and the remainder in services. By employment status, the majority are employees; 15 per- cent are employed in the public sector and 43 percent employed in the private sector. About a third are own-account workers, with many of them in agriculture. While poverty levels are relatively low, vulnerability is high, because the informal nature and the low productivity of many jobs lead to low earnings. Poverty reduction has been impressive in recent years, owing mainly to labor market improvements. Labor was reallocated from agriculture to industry and services, and wages grew strongly in the last decade. However, a large share of workers continues to be engaged in the low-productivity agriculture and services sectors, where income levels and the quality of jobs are also relatively low. Moreover, high levels of informality, at about 70 percent, and widespread precarious employment arrangements suggest a high risk of job displacement or earnings losses in the event of adverse shocks (World Bank 2020b). No formal unemployment insurance scheme is in place to protect workers during spells of joblessness, and low earnings lead to a higher risk of poverty and allow workers little room to accumulate savings that they can resort to during times of crisis. However, formal workers are not immune to shocks, either, as the COVID-19 crisis has made clear. For example, the export-oriented apparel industry, which employs about half a million workers, experienced significant challenges due to low global demand and a shortage of raw materials at the early stages of the pandemic, and it was severely affected by movement restrictions. Sri Lanka has a highly protective labor law regime, but it does not include provisions to deal with exceptional situations such as the COVID-19 pandemic, making employment protection challenging in the current circumstances (Department of Labour 2020; World Bank 2020b). In a context of high vulnerability, large shocks can quickly lead to significant deterioration in live- lihoods and welfare. Simple simulations using pre-pandemic data from 2016 illustrate that a relative- ly small shock could bring about notable deteriorations in poverty. For example, a hypothetical 10 per- cent decline in consumption levels could increase the $3.20 poverty rate from a baseline of 11 percent to 15.5 percent. A 20 percent or 30 percent decline further increases the poverty rate to 21.8 percent or 29.7 percent, respectively. The shocks yield similarly significant increases in poverty when simulations use the national poverty line (figure 3). This finding suggests that there are many households that are near-poor and clustered just above the poverty line, making them highly vulnerable to shocks such as those experienced as a result of the COVID-19 pandemic. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 2. The Sri Lanka Labor Market and the COVID-19 Shock 14 FIGURE 3 Increase in poverty from hypothetical The COVID-19-induced crisis was transmitted to consumption shocks the labor market through the mobility restric- 32 tions that broadly slowed down economic activ- 28 ities, though supply- and demand-side disrup- tions also played a role. Prevailing social-distanc- 24 ing measures and various mobility restrictions, 20 including strict lockdowns, had a direct impact on Percent 16 almost all sectors of the economy, leading to wide- 12 spread disruptions in economic activities and sub- 8 sequent earnings losses. Sectoral GDP data show that industries have been affected most severe- 4 ly but there are large variations across subsectors. 0 Construction and textile manufacturing suffered $1.90 poverty rate $3.20 poverty rate National poverty rate 2016 baseline 20% shock the largest impact as they are sensitive to demand 10% shock 30% shock shocks and the work requires physical presence. In Source: World Bank staff estimation using HIES 2016. the services sector, the overall impact was small but the aggregate impact masks heterogeneity across subsectors — for example, large output losses were experienced in transport, food and accommodation, and personal services. Extended travel restrictions shut down most of the tourism industry, except for some domestic tourism. 3 Meanwhile agricultural production was largely undisrupted, partly due to gov- ernment efforts to ramp up domestic production and promote import substitution, but there were chal- lenges with transport and logistics. The fishery sector also suffered a significant shock. Weak external demand has likely impacted export-oriented subsectors and their prevailing wages. The COVID-19 crisis led to widespread job losses Deteriorations in labor market outcomes directly impacted households’ economic wellbeing. As noted above, Sri Lanka made remarkable progress in poverty reduction in recent years, largely underpinned by improvements in labor market outcomes. Labor reallocation and growth in nonfarm incomes have been the main drivers of poverty reduction in recent years (World Bank 2021). However, the COVID-19- induced crisis has hit construction, transport, manufacturing, and food accommodation particularly hard — the very sectors that created a large number of jobs in recent years, supporting the movement of labor out of agriculture and in turn reducing poverty. Since the COVID-19 outbreak, many workers have been compelled to work or conduct their businesses from home but some jobs lend themselves better 3. Following the closure of airports in March to stem the outbreak, tourism earnings have been effectively zero for the rest of the year in 2020. Sri Lanka started welcoming tourists again in January 2021. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 2. The Sri Lanka Labor Market and the COVID-19 Shock 15 than others to being conducted remotely; jobs that require physical proximity are more economically vulnerable given social distancing policies. In Sri Lanka, as in many other developing countries, compar- atively fewer jobs are expected to be amenable to a work-from-home environment (Hatayama, Viollaz, and Winkler 2020). In addition to labor market deteriorations, declines in remittances could further raise poverty, although thus far these have been resilient. Sri Lanka has been a migrant-sending country for a long time, and remittances are an important source of income for some poor households, especially in some parts of the country where outmigration has increased in recent years (World Bank 2021). As the economic situa- tion in migrant destination countries worsened and an increasing number of migrants returned, remit- tances had been expected to fall significantly. However, such expectations did not materialize, though the trend will need continuous monitoring as the global economic crisis continues to unfold. In order to assess the distributional impact of the COVID-19 crisis, we adopt a methodological approach that has low information demands and can be deployed relatively quickly. The model follows an approach that uses macroeconomic projections and overlays them on behavioral models built with microdata to generate estimates of the full income distribution at the individual and household level. These predic- tions are then further converted to generate poverty projections. The specific steps involved are as fol- lows: (i) macroeconomic projections of output at the sectoral level are translated into employment pro- jections via historical estimates of GDP-employment elasticities; (ii) the latest household survey data are used to build behavioral models of employment status and earnings, as functions of individual and household characteristics; (iii) the employment projections at the macro level are replicated at the house- hold level via the behavioral models estimated in (ii); (iv) individual income and household income are adjusted based on predictions of employment shocks and earnings shocks, including remittances projec- tions; and (v) poverty estimates are derived by mapping estimates for income into consumption space. 4 Measuring the impact of the COVID-19 crisis just on employment has the potential to underestimate the broader impact, as the employment response following large external shocks has usually been relative- ly modest in the past. 5 In order to adequately capture the scope of the shock, ad hoc wage cuts are addi- tionally built into the simulation, with the magnitude of the shock varying at the sectoral level. 6 This approach allows the modeling of labor market responses at the intensive as well as the extensive margin. Results indicate that the COVID-19 crisis led to widespread losses in livelihoods, particularly in indus- try and services. Figure 4 compares the sectoral distribution of jobs before COVID-19 to jobs lost as a result of the crisis. Workers in urban areas and those engaged in industries and services were proportionately 4. For more details on the methodology, please see Olivieri et al. (2014). 5. Employment did not fall dramatically following the Easter attack in 2019 and the financial crisis in 2009. 6. These are applied to capture the impact at the intensive margin and are based on changes in sub-sectoral GDP. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 2. The Sri Lanka Labor Market and the COVID-19 Shock 16 more affected, while agriculture workers were less exposed. 7 Before the pandemic, about 27.3 percent of workers were engaged in the industry sector; this sector experienced 42.5 percent of pandemic-related job losses, implying that the labor market shock was disproportionately large compared to its employ- ment share. A disproportionate labor market shock was also felt in the services sector which before the pandemic had employed 47.9 percent of workers but accounted for nearly 60 percent of job losses. Meanwhile, agricultural activities have been permitted continuously through the pandemic, including during strict lockdowns, to minimize disruptions to food supply, and they have been at the center of renewed import substitution efforts. However, tea exports slightly declined and there were intermittent disruptions in transport and logistics. The fishery subsector suffered a significant shock and led the con- traction in the primary sector. The brunt of the crisis-induced job loss has been borne by workers in ser- vices and industries, who were less likely to be poor prior to the pandemic than workers in agriculture. FIGURE 4 Share of jobs lost by sector FIGURE 5 Share of displaced workers by employment status 70 65 70 60 65 60 55 55 50 50 45 45 40 Percent 40 Percent 35 35 30 30 25 25 20 20 15 15 10 10 5 5 0 0 Agriculture Industry Services Public sector Private sector Employer Own account Unpaid employee employee worker family worker Share of jobs before COVID Share of jobs before COVID Share of jobs lost due to COVID Share of jobs lost due to COVID Source: World Bank staff estimation using HIES 2016. Source: World Bank staff estimation using HIES 2016. Job displacement was more likely to happen in urban areas and among those with primary education or more. Job losses were concentrated among private sector employees and own account-workers (figure 5) and were disproportionately high in more urbanized areas, such as the Western Province (figure 6): the Western Province accounted for about 18.6 percent of total employment before COVID-19 but for more 7. Self-employed and unpaid family workers account for about two thirds in agriculture, more than any other sector. Job loss is less likely in this context, particularly given that activities largely continued even during strict lockdown periods. For this reason, it is assumed that employment in agriculture is unaffected. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 2. The Sri Lanka Labor Market and the COVID-19 Shock 17 than a third of all jobs lost. Further, displaced workers were on average slightly better educated than the average worker as most had completed primary education or more (figure 7). Poor households suffered disproportionately fewer job losses, likely because of the larger share of workers engaged in agriculture. FIGURE 6 Share of jobs lost by province FIGURE 7 Share of displaced workers by education 40 70 35 60 30 50 25 40 Percent Percent 20 30 15 20 10 5 10 0 0 WP CP SP NP EP NCP NWP UP Sab Less than Primary Passed O/L A/L and primary completed above Share of jobs before COVID Share of jobs before COVID Share of jobs lost due to COVID Share of jobs lost due to COVID Source: World Bank staff estimation using HIES 2016. Source: World Bank staff estimation using HIES 2016. Note: WP = Western Province; CP = Central Province; SP = Southern Province; Note: O/L = Ordinary Level; A/L= Advance Level. NP = Northern Province; EP = Eastern Province; NCP = North Central Province; NWP = North Western Province; UP = Uva Province; Sab = Sabaragamuwa Province. 3. The Distributional Impact of COVID-19 THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 19 Earnings losses led to a significant increase in poverty The labor market shock has been unequal across the income distribution. It is important to consider the wider impact on earnings across the distribution because looking just at job losses could underes- timate the impact of the shock. The need to consider earnings as well as employment is also borne out by preliminary findings from a recent World Bank COVID-19 rapid phone survey. 8 This survey found FIGURE 8 Average income loss across the income that among respondents engaged in the labor mar- distribution ket prior to the pandemic, more than half suffered 8.0 7.5 a labor market shock, primarily in the form of earn- 7.0 6.5 Loss in per capita income (percent) ings losses (reported by more than 30 percent) while 6.0 a more modest impact occurred through temporary 5.5 5.0 absence and job losses. 9 Figure 8 shows the share of 4.5 income losses across the income distribution. The 4.0 3.5 poorest experienced the largest proportionate earn- 3.0 2.5 ings shock while the smallest proportionate income 2.0 losses were suffered by the richest. The latter tend 1.5 1.0 to have formal, secure jobs and better access to 0.5 digital technology that allows them to conduct 0.0 0 10 20 30 40 50 60 70 80 90 100 wage work or business operations remotely. They Income percentile are also more likely to be working in the services Source: World Bank staff estimation using HIES 2016. sector which suffered the smallest contraction. 10 Note: Income percentile is based on per capita household income. With jobs and earnings lost, poverty increased significantly, and over 500,000 people are estimated to have fallen into poverty due to the crisis. The $3.20 poverty rate is projected to have increased from 9.2 percent in 2019 to 11.7 percent in a post-COVID world in 2020. This more than reverses progress since 2016 when the poverty rate was 11 percent, implying a significant setback. This is also in stark contrast to the no-COVID (BAU) scenario under which poverty would have declined to 8.2 percent in 2020 — part of a counterfactual world had the COVID-19 crisis not happened. Projections using the national poverty line suggest similar patterns of welfare losses. Extreme poverty (as measured by the $1.90 a day poverty 8. The World Bank conducted a rapid phone survey across eight South Asian countries. In Sri Lanka, the survey was imple- mented between September and December 2020, and primarily aimed to understand changes in the labor market among dif- ferent groups. Additional questions were included on households’ ability to meet basic needs, safety nets, and coping mecha- nisms. Full survey results with more detailed analysis will become available in the coming months. 9. The simulation incorporates changes in jobs and earnings losses only, whereas temporary absence from jobs is not consid- ered, for example. 10. Estimates of sectoral GDP growth rates in 2020 are as follows: agriculture ( – 2 .4 percent), industry ( – 6.9 percent) and ser- vices ( – 1.5 percent). Source: Department of Census and Statistics. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 20 line) doubled from estimated 2019 levels (figure FIGURE 9 Poverty impact of COVID-19 crisis 9). The poverty gap, which measures the distance 14 to the poverty line, is estimated to have increased 12 10 from 17.9 percent in 2019 to 20 percent in 2020. This Percent 8 implies not only that there are more poor people, 6 4 but also that the poor have fallen deeper into pov- 2 erty. 11 The number of poor (using the $3.20 poverty 0 2016 2017 2018 2019 2020 2021 line) is projected to increase from about 1.95 million $1.90 poverty rate $3.20 poverty rate in 2019 to 2.5 million in 2020. The poverty rate will National poverty rate gradually start declining again in 2021 but it will Source: World Bank staff estimation using HIES 2016. take a few years for it to return to pre-crisis levels. Note: Estimates for 2017 onward are based on simulations. Livelihood support programs helped absorb the shock, but better targeting could have enhanced their effectiveness to reduce poverty To mitigate the impact of the pandemic on the poor and vulnerable, the government implement- ed several livelihood support programs early in the crisis.  1 2 The government initiated several mitiga- tion measures, mainly through existing welfare schemes such as Samurdhi, elderly allowance, disabili- ty allowance, and the chronic kidney disease (CKD) allowance. The first such programs were implement- ed across all 25 districts during the first lockdown period, in April and May, and included extending the allowances to wait-listed and newly identified families and individuals, as well as making one-off top-up payments to existing beneficiaries under the Samurdhi program and elderly allowance program. In addi- tion, for low-income families not covered under the Samurdhi program, livelihood support was provid- ed where one or more members had lost their livelihood as a result of the pandemic. Following the sec- ond wave in October and November, low-income families who were quarantined or in lockdown areas also received relief, either in cash or in kind.  1 3 11. The results represent annualized COVID-19 impacts even though the shock was not distributed evenly throughout the year, and they could thus mask sharp reductions in income in some months for affected households. 12. The annex 1 summarizes these programs and others implemented as the crisis continued. 13. Some municipal councils conducted their own efforts to distribute relief packages to those in distress; however, more detailed information on these local initiatives is difficult to collect and these are therefore not part of the modeling exer- cise. See for example Daily News, “SLAF Together with the CMC Distribute Dry Rations to Low Income Families,” April 1, 2020, https://www.dailynews.lk/2020/04/01/local/215590/slaf-together-cmc-distribute-dry-rations-low-income-families; or The Morning, “Colombo Municipal Council: 100,000 Families to Receive Dry Rations,” November 26, 2020, http://www.themorn- ing.lk/colombo-municipal-council-100000-families-to-receive-dry-rations/. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 21 A large budget was expended on these mitigation efforts which covered a large proportion of the pop- ulation, especially in areas hit hard by the COVID-19 outbreak. Particularly in the Western Province and areas where the lockdown was in effect for extended periods of time, large shares of the population were covered. More than 4.9 million families received a payment of Rs 10,000, administered through the Department of Samurdhi Development, during the first lockdown period, in addition to the indi- vidual allowances for elderly or disabled individuals or chronic kidney disease patients; around 1.4 mil- lion families received relief payments of (or in-kind transfers equivalent to) Rs 5,000 during the second wave. Given Sri Lanka’s population size of around 21 million, these programs are expected to have had very wide coverage. An estimated Rs 105.1 billion have been incurred for the livelihood support pro- gram, with another Rs 47.7 billion spent on the Rs 5,000 allowance during the months of April and May. These expenses are in addition to the regular livelihood programs such as Samurdhi and fertilizer sub- sidy programs (Treasury 2020). The wide coverage of these mitigation efforts also stood out in a comparison against South Asian peer countries. Nearly half (about 44 percent) of respondents in the World Bank rapid phone survey confirmed that they had received new assistance since March 2020 — a far higher share than is reported in other countries in the region (e.g., only 12 percent in Bangladesh reported receiving new assistance). Additionally, public sector training and employment programs were launched. In 2020, the government launched Program for Placement of Unemployed Graduates, under which 49,450 graduates were provid- ed with employment starting from September. It also announced a program for public sector employ- ment of 100,000 individuals from low-income families, with the intention of providing selected candi- dates with formal vocational training before permanent employment. The first batch of approximate- ly 35,000 trainees have completed their leadership/soft skills development programs conducted by the National Youth Corp and have commenced a vocational training program at the National Apprentice and Industrial Training Authority. After completion of the course, the trainees are expected to receive National Vocational Qualification level 3 certificates. The effectiveness of these measures in mitigating the increase in poverty is analyzed, albeit with imper- fect information. The effect of the mitigation measures is simulated under two scenarios: a “realistic” scenario and a “perfect targeting” scenario. Under the realistic scenario, the targeting of relief meas- ures is imprecise, to reflect targeting amid the urgency of the crisis and in the absence of a social safe- ty net system that could quickly identify the vulnerable population. In this simulation, beneficiaries are randomly allocated while maintaining the existing distribution of beneficiaries between districts. Simulating the support distributed to families on the Samurdhi waiting list requires making assump- tions about new beneficiaries, and information on the distribution of existing beneficiaries across con- sumption quintiles is also utilized. The “perfect targeting” scenario is simulated under an assumption of perfect targeting of poor households (other than for the simulation of the program for unemployed THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 22 graduates). This is a rather unrealistic scenario in any context, but the results help explain the possible range of impact, given the lack of information on the characteristics of the actual beneficiaries. Results suggest that the combined mitigation measures helped absorb the labor market shock and soften the impact on poverty. Estimates suggest that the $3.20 poverty could have been reduced from 11.7 percent to 10.3 percent under a more practical scenario, and even further to 9.7 percent if targeting had been perfect under the livelihood support programs. While the programs helped some poor house- holds, a large share of resources was spread across wide population groups, amounting to modest aver- age transfers per household. Going forward, a more targeted approach that allows more resources for the poor and vulnerable could prove to be more cost-effective while ensuring that the benefits are direct- ed to those most in need. The lack of safety nets and coping mechanisms is worrying: about 44 percent of households in the rapid phone survey reported that they did not have any source to help them cov- er emergency expenses. The new poor are more urban, but the crisis did not fundamentally change the nature of poverty Three different groups are distinguished to compare the characteristics of the poor and nonpoor. These are: (i) the “old poor,” the group of poor as of 2019; (ii) the “new poor,” the group that is newly poor as a result of the COVID-19-induced crisis; and (iii) the “nonpoor,” the group of households expected to remain above the poverty line after COVID-19. The new poor are more likely to live in urban FIGURE 10 Distribution of poor and nonpoor people areas than the old poor. Compared to those who by province were poor before the crisis, the new poor are slight- 30 ly more likely to be living in urban areas: about 25 12 percent of the new poor, compared to 6.1 per- 20 Percent 15 cent of the old poor, are urban dwellers. This trend 10 is also reflected in the number of the new poor: 5 the highly urbanized Western Province accounted 0 for the largest share of the new poor across prov- WP CP SP NP EP NCP NWP UP Sab inces (figure 10) and has seen the number of poor Old poor (based on 2019) New poor (after COVID-19) Non-poor (after COVID-19) residents rise by over 40 percent during the crisis Source: World Bank staff estimation using HIES 2016 (figure 11). Some neighborhoods in the Colombo, Note: WP = Western Province; CP = Central Province; SP = Southern Province; Gampaha, and Kalutara districts that are part of the NP = Northern Province; EP = Eastern Province; NCP = North-Central Province; NWP = North-Western Province; UP = Uva Province; Sab = Sabaragamuwa Western Province went through extensive periods Province. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 23 FIGURE 11 Number of poor by province, pre- and of lockdown, which led to severe disruptions in post-COVID-19 livelihood activities and therefore reduced earn- ings. While the Western Province has the lowest 2019 poverty rate across provinces, at around 4 percent 2020 post-COVID-19 in 2019, an increase to 5.7 percent is expected in 0 400 800 1,200 1,600 2,000 2,400 2,800 2020. The Western Province’s large share of the Thousands new poor is mainly owing to its large population; WP SP EP NWP Sab it has experienced a relatively small increase in the CP NP NCP UP poverty rate itself. Source: World Bank staff estimation using HIES 2016 Note: WP = Western Province; CP = Central Province; SP = Southern Province; The vast majority of the poor continue to live in NP = Northern Province; EP = Eastern Province; NCP = North-Central Province; NWP = North-Western Province; UP = Uva Province; Sab = Sabaragamuwa Province. rural areas, and the largest increase in the $3.20 poverty rate occurred in places whose rate was high even before the crisis (figure 12). The $3.20 poverty rate was estimated at around 15 percent for the Northern and Eastern Provinces before the crisis and exceeded 19 percent in both places post-COV- ID-19. In the Central Province, poverty increased from 12.6 percent to 15.9 percent, and in the Uva and Sabaragamuwa Provinces from around 16 to 19 percent. Across districts, Kandy and Ratnapura — which are highly rural and before the pandemic had the largest number of poor — a lso account for a large share of the new poor, as well as Gampaha (figure 13). While the crisis may have slightly shifted the compo- sition of the poor, it did not fundamentally change the nature of poverty in Sri Lanka, as most of the poor continue to live in rural areas. FIGURE 12 $3.20 poverty rate by province FIGURE 13 Map of the distribution of new poor 20 Number of new poor 0 – 20,000 18 20,000 – 40,000 16 40,000 – 60,000 60,000 – 80,000 14 80,000 – 100,000 12 Percent 10 8 6 4 2 0 WP CP SP NP EP NCP NWP UP Sab 2019 2020 post-COVID-19 Source: World Bank staff estimation using HIES 2016 Source: World Bank staff estimation using HIES 2016 THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 24 Poverty increased the most in sectors that accounted for much of the job and earnings losses. The new poor are disproportionately represented in industries which experienced the largest overall shock while the nonpoor are overrepresented in services which experienced the smallest shock (figure 14). For instance, poverty nearly doubled among households where the head of the household was employed in accommodation and food services and increased by nearly 50 percent for those in the construction sec- tor (figure 15). FIGURE 14 Distribution of working-age population FIGURE 15  $3.20 poverty rates by household (age 15+) by sector of employment head’s employment sector 60 Agriculture 55 Manufacturing 50 45 Construction 40 Commerce 35 Percent 30 Transport 25 Accommodation and food services 20 Activities of households 15 10 Other 5 Not working 0 Not working Agriculture Industry Services 0 5 10 15 20 25 30 35 Old poor (based on 2019) New poor (after COVID-19) Percent Non-poor (after COVID-19) 2019 Post COVID-19 Source: World Bank staff estimation using HIES 2016. Source: World Bank staff estimation using HIES 2016. The new poor are more educated and live in smaller households. The share of new poor who have com- pleted primary education, the General Certificate of Education (GCE) Ordinary Level (O/L), and the GCE Advanced Level (A/L) is higher than among the old poor, though the new poor are still significantly less educated than the nonpoor. Similarly, average household size and dependency ratios for the new poor are lower than for the old poor but higher than for the nonpoor. Overall, the characteristics of the new poor appear to position them as a middling group in between the old poor and nonpoor. The pandem- ic’s effect on per capita household consumption varies significantly across these groups, with the new poor falling into poverty as a result of a reduction in their per capita consumption by nearly a quarter; a much smaller reduction is expected among the nonpoor (4.7 percent) (table 1). THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 25 TABLE 1 Household characteristics by poor and nonpoor groups Old poor New poor Nonpoor (2019) (after COVID-19) (after COVID-19) Share of females 53.3% 53.4% 53.2% Average age 31.0 31.0 34.5 Average household size 4.8 4.5 3.7 Average dependency ratio 0.42 0.39 0.34 Share of female-headed households 25.2% 23.9% 25.9% % reduction in real per capita consumption (between 2019 and 2020 after COVID-19) 6.2% 22.4% 4.7% % of households by head’s education No schooling 9.5% 5.2% 2.8% Below primary 28.6% 23.2% 13.3% Primary completed 55.5% 63.9% 54.0% O/L passed 5.3% 6.1% 15.8% A/L and above 1.1% 1.6% 14.1% A/L passed 0.9% 1.1% 10.9% Bachelor’s and above 0.2% 0.5% 3.2% Source: World Bank staff estimation using HIES 2016. Note: O/L = Ordinary Level; A/L= Advanced Level. Inequalities will likely widen, with potential consequences for the long term The pandemic is expected to widen inequalities that will manifest through the labor market in the short-term. As seen in figure 8, earnings losses were distributed unevenly across the distribution, with the richest households experiencing much smaller reductions compared to those in the bottom 40 per- FIGURE 16 Gini index in Sri Lanka and peer countries cent of the distribution. Consistent with this, the (pre-COVID-19) Gini index, a commonly used aggregate measure Malaysia of inequality, is expected to slightly increase from Sri Lanka Bhutan 39.3 to 39.8 as of 2020. This is concerning given Thailand that Sri Lanka’s Gini index was already relatively Vietnam high in comparison to peers even before the pan- Pakistan Bangladesh demic (figure 16). Maldives 0 5 10 15 20 25 30 35 40 45 Combined with pre-existing inequalities, the Gini index potential long-term impact on inequality through Source: World Bank PovcalNet, accessed March 4, 2021. World Bank staff reduced social mobility could be significant. While calculation for Sri Lanka using HIES 2016. Note: The Gini index varies from 0 to 100, with a higher number representing this note focuses mainly on the pandemic’s impact a higher level of inequality. Year of data varies between 2015 and 2018. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 3. The Distributional Impact of COVID-19 26 on livelihoods and poverty, there are concerns about broader socioeconomic impact and the implica- tions for inequality. An area of particular concern is access to education, as schools were closed for most of the pandemic period; a reopening date has been set for April 2021. Widening disparities in education- al outcomes due to lack of access to digital technology and e-learning content could leave long-lasting scars and exacerbate inequalities between urban and rural areas and different socioeconomic groups. The potential disparities are corroborated using computer ownership as an imperfect proxy for access to digital connectivity: according to HIES 2016, only 19.6 percent of rural households and 6.0 percent of estate households had a computer. Percentages were higher among urban households, though still low at around 34.8 percent. Social mobility could be adversely affected by this situation to the extent that better education serves as a ladder out of poverty — and there is increasing evidence that that is the case in Sri Lanka, with more educated individuals being more productive and more likely to hold better jobs and earn more. 4. Conclusion THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 4. Conclusion 28 This note presented short-term projections on changes in Sri Lankans’ livelihoods and poverty that were derived from a microsimulation analysis to assess the impact of the pandemic on households’ welfare. The primary channels considered were those through the labor market, as poverty reduction in the past had mainly been driven through improvements in labor market outcomes. High levels of preexisting vulnerabilities imply that relatively small adverse shocks can lead to significant increases in poverty. In the absence of adequate and detailed data, a simulation tool that links macroeconomic pro- jections to micro-level behavioral models was deployed to capture the impact of shocks at the house- hold level. The main findings are that the COVID-19 crisis induced widespread livelihoods losses, leading to a sig- nificant increase in poverty. The sharp economic slowdown is estimated to increase the $3.20 pover- ty rate from 9.2 percent in 2019 to 11.7 percent in 2020, leading to over 500,000 new poor people. The economic impact of the pandemic is expected to be felt broadly, but particularly among those working in the industry and services sectors, among the new poor in more urbanized areas such as the Western Province, and in places with high numbers of poor people prior to the pandemic — including the Northern, Eastern, Uva and Sabaragamuwa Provinces. The new poor are more likely to be urban than rural, given the impact of the shock in the industry and services sectors, and they are slightly more educated than those who had been poor prior to the pandemic. However, while a disproportionate share of the new poor resides in urban areas, poverty in Sri Lanka remains overwhelmingly rural. Mitigation measures implemented since the onset of the pandemic helped absorb the labor market impact and soften the impact on poverty. In addition to estimates of first-order impact, we also attempt- ed to simulate how the mitigation efforts by the government may have helped buffer the adverse shock. The government measures had wide coverage among the population, especially in the Western Province. However, the average transfer amount per household was relatively modest; and while transfers helped offset some of the adverse economic impact, more targeted spending could have been more effective at reaching those most in need. The current social protection system could be improved to provide targeted support to those most in need and to reintegrate those who lost their jobs into the labor market. In the medium term, social safety nets can aim to improve targeting toward the poor and vulnerable, while adopting a system that allows support to be scaled up quickly and effectively in times of crises. In the absence of a strong social protection response, households are likely to cope with livelihoods losses by drawing down on savings, selling assets, or reducing food intake, all of which have longer-term negative implications for household wellbeing. There are some signs that households are indeed resorting to these negative coping strate- gies: in the rapid phone survey, about 80 percent of households reported that they reduced the purchase of preferred food due to budget, with a similar share reporting that they ran out of food and lacked the money to buy more. These responses highlight the lack of safety nets and adequate coping mechanisms. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 4. Conclusion 29 While this note focuses on the short-term distributional impact on livelihoods and poverty, the potential longer-term socioeconomic consequences of the pandemic need to be better understood. These include disruptions in education and health services as well as aspects of food security. The data and information gaps remain large, and it is not clear yet what shape the recovery process will take and whether the poorest and most vulnerable will be able to participate in it equally. A lot more work will be needed going forward. Looking ahead, a resilient recovery will require support to bring back jobs and livelihoods, along with efforts to mitigate long-run impacts on inequality. Sri Lanka’s economy is expected to gradually recover with a projected GDP growth rate of 3.4 percent in 2021. The share of people living on less than $3.20 per day is expected to decline accordingly to 10.9 percent. Containing the health crisis with the help of effec- tive vaccine rollouts will be an immediate priority and a prerequisite to fully resuming normal economic activities. Policy measures could aim to strike a balance between those that support a resilient recovery and those that aim to include the most vulnerable in the recovery process. The latter will require efforts to reverse the impact of the pandemic and mitigate its consequences for long-run inequality. THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 30 Annex 1 Summary of Relief Measures Used in Simulation Months implement- Payment Number Program ed (2020) (Rs) of beneficiaries Program for Placement of Unemployed Graduates (less than 35 September onwards 20,000 each month 49,450 individuals years and unemployed for a period of one year or more at the time of application) One-off payment to existing Samurdhi beneficiary families April – May 5,000 each month 1,768,600 families Payment to families on the Samurdhi waiting list April – May 5,000 each month 707,630 families Livelihood support to non-Samurdhi families affected by the April – May 5,000 each month 1,727,086 families pandemic Payment to families making appeals under programs 3 and 4 April – May 5,000 728,523 families One-off payment to existing recipients of the elderly allowance April – May 3,000 each month 416,667 individuals Payment to individuals on the waiting list and newly identified April – May 5,000 each month 209,420 individuals beneficiaries under the elderly allowance in April and 212,636 in May Payment to individuals on the waiting list and newly identified April – May 5,000 each month 52,940 individuals beneficiaries under the disability allowance in April and 51,641 in May Payment to individuals on the waiting list and newly identified April – May and 5,000 each month 19,472 individuals beneficiaries under the chronic kidney disease (CKD) allowance September onward Livelihood support provided during second wave October – November 5,000 1,343,214 families Provision of food packs for families in lockdown areas or quaran- October – November Food pack 74,961 families tined during second wave (average value of 5,000 – 10,000) Source: Ministry of Public Administration (for program 1); State Ministry of Samurdhi, Household Economy, Micro Finance, Self-Employment, Business Development, and Underutilized State Resources Development (for programs 2 – 9); Department of National Planning, Ministry of Finance (for programs 10 and 11). THE COVID-19 IMPACT ON LIVELIHOODS AND POVERTY IN SRI LANKA 31 Bibliography Ceriani, Lidia, Gabriela Inchauste, and Sergio Olivieri, 2015. Mongey, Simon, Laura Pilossoph, and Alex Weinberg. 2020. “Understanding Poverty Reduction in Sri Lanka: Evidence “Which Workers Bear the Burden of Social Distancing from 2002 to 2012/13.” Policy Research Working Paper Policies?” NBER Working Paper 27085, National Bureau 7446, World Bank, Washington, DC. of Economic Research, Cambridge, MA. Department of Census and Statistics. 2018. Household Olivieri, Sergio, Sergiy Radyakin, Stainslav Kolenikov, Income and Expenditure Survey 2016. HIES Final Michael Lokshin, Ambar Narayan, and Carolina Report. Department of Census and Statistics. Ministry Sánchez-Páramo. 2014. Simulating Distributional Impacts of National Policies and Economic Affairs. of Macro-dynamics: Theory and Practical Applications. Department of Labour. 2020. “Covid 19 & Beyond — T he Washington, DC: World Bank. Impact on the Labour Market of Sri Lanka.” Survey Treasury. 2020. “Fiscal Management Responsibility Report report of the e-survey conducted on private sector estab- 2020 – 2021.” https://www.treasury.gov.lk/documents/ lishments. May 2020. budget/2021/FMRP-Report-2020-21-(English).pdf. Hatayama, Maho, Mariana Viollaz, and Hernan Winkler. World Bank. 2020a. Global Economic Prospects, January 2020: 2020. “Jobs’ Amenability to Working from Home: Evidence Slow Growth, Policy Challenges. Washington, DC: from Skills Surveys for 53 Countries.” Policy Research World Bank. Working Paper 9241, World Bank, Washington, DC. World Bank. 2020b. “Informality, Job Quality, and Welfare IFC (International Finance Corporation). 2020. “Gendered in Sri Lanka.” World Bank, Washington, DC. Impacts of COVID-19 on Small and Medium-Sized World Bank. 2021. “Sri Lanka Poverty Update.” World Bank, Enterprises in Sri Lanka.” World Bank Group, Washington, DC. Washington, DC.