Central Administration for Statistics and World Bank Snapshot of Poverty and Labor Market Outcomes in Lebanon based on Household Budget Survey 2011/20121 May 25, 2016 Version 2 1 This brief is a result of a one year collaboration between CAS and the World Bank team. The authors of the report are Najwa Yaacoub and Mayssaa Daher from CAS, and Dean Jolliffe and Aziz Atamanov from the World Bank. This brief is based on analysis of the 2011-12 household budget survey (HBS) implemented by Central Administration for Statistics (CAS) with technical assistance from the World Bank. The survey was conducted during the period of September 2011 to November 2012, and was stratified across nine regions. The sample was designed to cover 4,805 households, but due to high non-response, it only includes 2,476 participating households. Poverty numbers presented in this note are not comparable with poverty estimates for other years due to differences in the instruments, fieldwork implementation and to some extent sample design; and also due to differences in the methodology for constructing welfare aggregate and the poverty line.2 All regional estimates in this report should be viewed with caution given concerns about significant levels of nonresponse and relatively small sample sizes within regions. CAS and the World Bank are working together to improve the quality of future surveys. According to the most recent estimates obtained by the Central Administration for Statistics and the World Bank, the average level of consumption per person in 2011-12 was about 7,788 thousand LBP per year in Lebanon. The data indicate that in 2011-12, 27 percent of the population were poor. This implies about one million people had levels of consumption that were below the annual poverty line, which was set at the level of 4,729 thousand LBP per capita per year. The Gini coefficient of per capita consumption, which measures level of inequality, was estimated to be 0.32. The poorest 20 percent of the population were found to consume about 8 percent of total consumption, while the top 20 percent of the population consumed about 40 percent. Confidence intervals for the regional poverty estimates are wide and overlap with each other, indicating that many of the observed regional differences in poverty are not statistically significant, but some comparisons are. In particular, poverty in Beirut is significantly lower than the average in the country, and poverty rates in the poorest North Lebanon and Bekaa regions are higher than in Beirut and Mount Lebanon (Figure 1). Geographically, the highest poverty count is observed in Mount Lebanon, North Lebanon and Bekaa which account for about 77 percent of all poor people in the country (Figure 2). The regions with the lowest counts of poor people are Beirut and Nabatieh: about 58-59 thousand people in each. Figure 1. Poverty headcount rates by regions, 95% Figure 2. Number of poor people by governorates confidence intervals 400,000 342,815 50 350,000 headcount, 95% confidence 40 287,950 36 38 300,000 poor population 30 31 25 27 250,000 20 22 internval 16 10 200,000 169,711 0 150,000 117,723 100,000 59,024 57,581 50,000 0 Beirut Mount North Bekaa South Nabatieh Lebanon Lebanon poverty rate lower bound upper bound Source: Authors’ calculation based on HBS 2011/12. Notes: Population weights were adjusted for non-response. Around seven percent of households in the HBS have non-Lebanese heads, but are Lebanese residents. The poverty rate for this group of the population is very high: 55 percent compared to 26 percent for the population from households headed by Lebanese citizens. Households with a non-Lebanese head are more likely to be found at the bottom of the distribution. Fourteen percent of households in the bottom quintile (i.e. poorest one-fifth of the population) are headed by non-Lebanese resident, while among the richest quintile this share is only about 3 percent. 2 Detailed methodology on non-response correction and poverty measurement is presented in joint CAS-World Bank publications (available at CAS website). 2|Page The data indicate that in 2011/12, 45 percent of individuals 15 years of age and above were in the labor market and employed. Fifty-one percent of the population in this age range were inactive (i.e. not in the labor force). The unemployment rate, defined as the ratio of unemployed to total active population in the labor force, was about 9 percent. 3 Largest sectors of employment are in education, health and public administration, trade and services (Figure 3). There are substantial differences between employment rates among men and women which widens after age 34 (Figure 4). Participation in the labor market is closely correlated with education level. Thus, among those with some tertiary education, the share of employed is 61 percent. Among the population with less than primary education, the employment rate is about half the rate at 31 percent. Figure 3. Employment by economic sectors in %, 15+ Figure 4. Employment rates by gender and age groups, % Financial Agriculture, Mining and 100 90 92 Education, intermediati 4 manufacturi 77 health and on, 3 ng, 12 80 67 % of population, 15+ public 58 administrati Construction 60 on, 21 ,7 48 32 40 27 36 30 25 20 16 Trade, 22 6 3 0 Transport 15-19 20-24 25-34 35-49 50-64 65+ Total Services, 24 and communicati age group on, 6 men women Source: Authors’ calculation based on HBS 2011/12. There is no clear pattern in the relationship between employment and poverty at the individual level. Employment increases somewhat across quintiles, but drops for the richest quintile. If only heads of household are analyzed, employment is much higher for the poor heads than the rich. For example, share of households with employed heads is 72 percent in the bottom quintile versus 51 percent among the richest households from the top quintile. There can be several reasons for this, including the possibility that the poor simply cannot afford to be without any sort of work. In contrast to the rate of employment, the sector in which one works is closely associated with economic wellbeing (Figure 5). For example, poverty rates in agriculture and construction are particularly high, while poverty rates for those working in the public sector or in services are much lower. In addition to sector of employment, the contractual type of employment is also associated with higher poverty (Figure 6). Poverty rates among weekly or daily employees are much higher than the rates for those who are monthly paid employees. 3 A person is considered unemployed if he or she did not work during the past 7 days and was not temporarily absent. In addition, he or she should be available and actively search for work. 3|Page Figure 5. Poverty rates by sectors of employment, 15+ Figure 6. Poverty rates by type of employment %, 15+ 50% 50% 47% 38% 40% 40% 33% 29% poverty rate, % 25% 40% 30% poverty rate, % 20% 16% 17% 28% 30% 7% 10% 19% 0% 20% 14% manufacturing communication Construction Agriculture intermediation Services Trade Education, health Transport and Mining and and public… Financial 10% 0% employer monthly own account weekly, daily paid worker paid employee employee Source: Authors’ calculation based on HBS 2011/12. Education is also a strong predictor of poverty. People with some university education have significantly lower poverty rates that average. Nine percent of people with some university education are poor, while about one third of the people with primary schooling or less are poor (Figure 7). Enrollment rates drops dramatically for children aged 7-21 from the poorest quintile of the population after reaching age of 13. For example, enrollment for the poorest quintile drops from 92 percent to 72 percent between age 13 and 15 (Figure 8). The drop in enrollment is more pronounced for men and non-Lebanese population. For example, overall enrollment rate among Lebanese children ages 7-21 is about 83 percent compared to 43 percent among the non-Lebanese reference group. Figure 7. Poverty rates by level of education in %, Figure 8. School enrollment in % by age and welfare 22+ years 40% 32% 34% 30% % of population, 22+ 24% 20% 16% 15% 9% 10% 0% Source: Authors’ calculation based on HBS 2011/12. Access to health services depends on availability of health insurance. The proportion of the population covered by health insurance is far from universal in Lebanon with the poor being particularly less protected. For instance, only 28 percent of household members are insured in the bottom quintile of the distribution compared to 67 percent among the wealthiest quintile. Similarly, only 19 percent of households in the bottom quintile have all members insured compared to 52 percent of households in the top quintile. 4|Page There is a substantial gap between poor and non-poor households in Lebanon in terms of access to such assets as flat TVs, air conditioners, microwaves, computers, washing machines and cars (Figure 9). The poorest households have lower access to basic infrastructure as well. For example, households from the bottom quintile are more likely to have unprotected sources of drinking water, and to use open sewage sources for drainage, both of which can cause health problems (Figure 10). Figure 9. Assets ownership by poverty status, % Figure 10. Source of drinking water and sewage by quintiles, % 100% 10% 10% bottom 20 top 80 80% poor non poor % of population 8% 60% % of population 6% 40% 6% 4% 20% 4% 2% 0% 2% 0% unprotected sources of open sewage or no drinking water drainage system Source: Authors’ calculation based on HBS 2011/12. Note: unprotected sources include unprotected wells, spring, rainwater collectors, surface water and carts with small tanks. Access to mobile lines is relatively high in Lebanon. Around 85 percent of households had at least one family member with access to mobile line (Figure 11). This share does not change much across the distribution. However, when examining the total number of adults with mobile lines, the picture changes (Figure 12). On average, one adult from two in the poorest quintile has access to a mobile phone, while among the richest quintile two of every three adults have access to a mobile line. In contrast to cell phones, household-level access to the internet is quite low. Only 28 percent of households report having at least one member with access to internet. This ratio is much smaller for the poorest households. In particular, 11 percent of households report access to internet versus 41 percent among the richest quintile (Figure 11). Figure 11. Household access to mobile phone line and Figure 12. Access to mobile line by quintiles among Internet by quintiles in % population 15+, % 100 85 84 89 85 84 85 80 72 70 67 65 80 61 % of households share of adults, 15+ 60 53 60 41 35 40 24 28 40 18 20 11 20 0 bottom 2 3 4 top Total 0 consumption per capita quintiles bottom 2 3 4 top Total any member has mobile phone line consumption per capita quintiles any member has internet home Source: Authors’ calculation based on HBS 2011/12. The findings in this brief are drawn from HBS 2011/2012. This microdata was collected by Central Administration of Statistics www.cas.gov.lb and financed by the World Bank. Detailed methodology on non-response correction and poverty measurement is provided in two joint technical papers prepared by CAS and the World Bank: CAS and World Bank (2015). Accounting for unit non-response in the Lebanon Household Budget Survey (HBS) 2011/2012. Washington: DC [www.cas.gov.lb] 5|Page CAS and World Bank (2015). “Measuring Poverty in Lebanon using 2011 Household Budget Survey. Washington: DC [ www.cas.gov.lb] 6|Page