97639 An Update on Poverty and Inequality in Albania: 9 Stylized Facts May 2015 Poverty Global Practice World Bank 1 EXECUTIVE SUMMARY This note presents 9 stylized facts that emerge related to the evolution of poverty between 2008 and 2012 in Albania. Findings from this note, using the latest Living Standard Measurement Study (2012 LSMS), show that the sharp downward poverty trend observed in the first part of the 2000s halted with the global economic crisis (Stylized fact 1). From a period of fast economic growth up to 2008, growth slowed down, labor markets weakened and remittances inflows declined sharply in Albania, resulting in a stagnant national poverty rate. Between 2008 and 2012, poverty increased in urban areas, narrowing poverty differences between urban and rural areas (Stylized Fact 2). Poverty increased in most regions particularly the Coastal region (Stylized fact 3). Not surprisingly, poverty is higher among households with low-educated members and a larger number of members. In the period under study, poverty increased particularly among household with older workers and youth, and the secondary-educated (Stylized Fact 4). Although poverty at the national level remained largely unchanged, this masks significant movements in and out of poverty that point to vulnerability of the non-poor in Albania (Stylized fact 5). In fact, the observed decline in inequality in the period 2008 to 2012 is driven by stronger consumption decline of those in the highest quintile (Stylized fact 6). The analysis suggests that lower economic growth was the main driver of the poverty trend (Stylized fact 7). In particular, weakened labor markets are behind poverty changes (Stylized fact 8), mainly through job loss, while pensions and social assistance partly offset the labor market shock (Stylized fact 9). Worryingly, labor markets have further deteriorated beyond 2012, and Albania faces multiple structural labor market challenges, with the low quality of jobs being a particularly salient issue.1 However, knowledge gaps remain in terms of understanding the constraints to more and better jobs in Albania. As growth and employment pick up in Albania, it is critical to further explore the constraints to more and better jobs so as to boost the capacity of households to generate market income and resume the downward poverty trend. Improving the ability to monitor poverty more frequently is also key for Albania. 1See accompanying labor market note (World Bank, 2015) documenting a large share of low-skilled jobs, many as unpaid family workers. 2 Table of Contents EXECUTIVE SUMMARY ............................................................................................................. 2 ACKNOWLEDGEMENTS ........................................................................................................... 4 INTRODUCTION ........................................................................................................................ 5 THE EVOLUTION OF POVERTY AND INEQUALITY IN ALBANIA .................................... 8 STYLIZED FACT 1: THE DECREASING POVERTY TREND OF THE 2000S IN ALBANIA HALTED FROM 2008 TO 2012 ............................................................................................................................................................................. 8 STYLIZED FACT 2: POVERTY INCREASED IN URBAN AREAS, NARROWING URBAN-RURAL POVERTY DIFFERENCES ............................................................................................................................................................... 9 STYLIZED FACT 3: POVERTY INCREASED IN MOST REGIONS, PARTICULARLY IN THE COASTAL REGION WHERE PREVIOUS GAINS WERE LARGELY ERODED .......................................................................... 11 STYLIZED FACT 4: POVERTY INCREASED PARTICULARLY AMONG HOUSEHOLDS WITH OLDER WORKERS AND YOUTH, AND THE SECONDARY-EDUCATED ............................................................................ 14 STYLIZED FACT 5: SMALL OBSERVED POVERTY CHANGES MASK LARGE MOVEMENTS IN AND OUT OF POVERTY........................................................................................................................................................................... 17 STYLIZED FACT 6: INEQUALITY DECLINED, FROM LARGER CONSUMPTION FALL AT THE TOP 18 UNDERSTANDING THE TRENDS ......................................................................................... 21 STYLIZED FACT 7: WEAK ECONOMIC GROWTH DROVE POVERTY INCREASES ..................................... 21 STYLIZED FACT 8: WEAKENED LABOR MARKETS AS A CHANNEL FOR POVERTY INCREASES .... 23 STYLIZED FACT 9: SOCIAL PROTECTION, PARTICULARLY PENSIONS, CONTRIBUTED TO MITIGATING THE LABOR MARKET SHOCK; REMITTANCES FLOWS SHRANK .......................................... 28 CONCLUSION ........................................................................................................................... 29 ANNEX FIGURES ...................................................................................................................... 30 ANNEX TABLES ........................................................................................................................ 35 3 ACKNOWLEDGEMENTS The note was prepared by María E. Dávalos (TTL and Senior Economist, GPVDR), César Cancho (Economist, GPVDR) and Lidia Ceriani (consultant, GPVDR) of the World Bank. The note is part of the poverty monitoring work under the Programmatic Poverty Assessment for the Western Balkans (P145041). The team is grateful for insightful comments from Carlos Rodriguez (Senior Economist, GPVDR). 4 INTRODUCTION Albania was a fast growing economy in the 2000s. Per capita gross domestic product (GDP) grew at an average annual rate of 6 percent in real terms, and Albania moved from being one of the poorest countries in Europe in the early 1990s, to reaching middle-income status in 2008 (Figure 1). The construction and services sectors were the main drivers of economic and employment growth in 2000–2008 (accounting for nearly three-quarters of growth), and significant reallocation of resources took place from rural agriculture to construction and services. Figure 1: Real GDP per capita, 2000–08, percent 6% $70,000 5% $60,000 $50,000 4% $40,000 3% $30,000 2% $20,000 1% $10,000 0% $0 Norway France Hungary Portugal Germany Luxembourg Montenegro Turkey Czech Republic Switzerland Belgium Ireland Bosnia and Herzegovina Poland Estonia Albania Greece Finland Lithuania Denmark Slovak Republic Italy Cyprus Spain Austria Serbia Latvia United Kingdom Iceland Sweden Netherlands Croatia Slovenia Moldova Romania Macedonia, FYR Bulgaria Real GDP per capita in PPP terms Real GDP per capita growth Sources: Albania Systematic Country Diagnostic, with data from World Bank and INSTAT. The global economic crisis slowed down Albania’s economic growth. After the global economic crisis hit, with significant impacts across the Western Balkans (Figure 2), economic growth slowed down significantly and persistently in Albania, linked to the Eurozone crisis and particularly to the economic performance of Greece and Italy. The region suffered another hit in 2012 (“double-dip” recession) as it struggled toward recovery and had negative impacts from climatic shocks; economic growth in Albania decelerated further. Inflation remained moderate over the period before and after the global economic crisis (below 3 percent on average from 2002 to 2012 2 ), but the labor market suffered during the economic downturn and adjusted mainly through job loss instead of wages. Today, employment rates in Albania remain low compared to the region, but the highest among the Western Balkans (Figure 3). 2 World Development Indicators, our elaboration on variable fp_cpi_totl, Consumer Price Index. 5 Figure 2: Western Balkans growth rates crisis (per capita GDP growth rates) 10 8 6 4 Percent 2 0 -2 -4 -6 -8 European Union Albania Bosnia and Kosovo Macedonia, FYR Montenegro Serbia Herzegovina 2008 2009 2010 2011 2012 2013 Source: WB staff elaboration on World Development Indicators. Note: GDP per capita in constant 2005 US$. Figure 3: Employment rate in Europe and Central Asia, 2013 80 70 60 50 Percent 40 30 20 10 0 Source: World Development Indicators. Note: Latest data available (2013), for employment to population ratio 15+ This note presents 9 stylized facts that emerge related to the evolution of poverty and inequality over the 2002 to 2012 period, especially from 2008 to 2012, as well as some insights into the drivers of poverty changes. It complements existing work on monitoring and understanding shared prosperity – the second corporate goal of the World Bank along with reducing poverty - in Albania and the Western Balkans.3 Albania poverty estimates are based on the Living Standards Measurement Survey (LSMS). The LSMS is conducted by the Albania Statistics Office (INSTAT) with donor funding and has been carried out in 2002, 2005, 2008 and recently in 2012. The 2012 round provides an opportunity to update poverty estimates and 3World Bank (2014), “First Insights into Promoting Shared Prosperity in South East Europe”. South East Europe Regular Economic Report No.5, Special Topic. 6 analysis since 2008, capturing the effects on poverty of the global economic crisis and lower economic growth in the country that in the first part of the decade.4 Findings from this note show that downward poverty trends in the first part of the 2000s halted in Albania between 2008 and 2012. Increases took place particularly in urban areas, narrowing poverty differences between urban and rural areas. Poverty changes were driven mostly by changes in mean consumption growth. This analysis suggests that lower economic growth and particularly weakened labor markets were drivers of poverty increases, while pensions and social assistance partly offset the labor market shock. Specifically, decomposing household income growth and poverty changes by different sources of income suggests that labor markets were a key channel for poverty changes during the shock, mainly from job loss. In addition, pensions increased considerably over the period partly mitigating the crisis effects on poverty, together with social assistance income. The stylized facts behind these findings are the following:  Stylized fact 1: the decreasing poverty trend of the 2000s in Albania reversed from 2008 to 2012  Stylized fact 2: poverty increased in urban areas, narrowing urban-rural poverty differences  Stylized fact 3: poverty increased in most regions, particularly in the coastal region where previous gains were largely eroded  Stylized fact 4: poverty increased particularly among older workers and youth, and the secondary- educated  Stylized fact 5: small observed poverty changes mask large movements in and out of poverty  Stylized fact 6: inequality declined, from larger consumption fall at the top  Stylized fact 7: weak economic growth drove poverty increases  Stylized fact 8: weakened labor markets as a channel for poverty increases  Stylized fact 9: social protection, particularly pensions, contributed to mitigating the labor market shock; remittances flows shrank The rest of the note presents trends and evidence supporting each of these using the LSMS data from 2002, 2005, 2008 and 2012, with a stronger focus on the latest round. Findings presented here are complemented with two accompanying notes on labor markets in Albania using the Labor Force Survey, and on gender and access to economic opportunities using the LSMS. 4Official Poverty estimates for Albania are produced by the Albania Institute of Statistics (INSTAT) using data from the Living Standards Measurement Survey (LSMS). Poverty is based on total per-capita household consumption, regionally deflated. The poverty line has been estimated following the cost of basic needs methodology. The food poverty line or extreme poverty line was set at 3,047 LEK per month, whereas the poverty line has been set at 4,891 LEK in 2002 prices. This note focuses on the poverty line, given that poverty at the extreme poverty line is very low at 2.2 percent in 2012 (going from 5 percent in 2002, to 3.5 percent in 2005, to 1.2 percent in 2008). The poverty line in 2012 prices is 6,407.21Lek per month, which corresponds to about 3.12 US$ in 2005 PPP. 7 THE EVOLUTION OF POVERTY AND INEQUALITY IN ALBANIA STYLIZED FACT 1: THE DECREASING POVERTY TREND OF THE 2000S IN ALBANIA HALTED FROM 2008 TO 2012 The decreasing poverty trend observed from 2002 to 2008 in Albania halted between 2008 and 2012. Following a substantial decrease in poverty from 25.4 percent to 12.5 percent up to the global economic crisis (2002-2008), poverty slightly increased between 2008 and 2012 from 12.5 percent to 14.3 percent, although increases are not statistically significant.5 In addition, the average shortfall from the poverty line became deeper (poverty gap) and inequality among the poor (squared poverty gap) increased (Figure 4). The slight increase in poverty in the recent period is robust to any choice of poverty line: as shown in Figure A1, regardless of the threshold, poverty was higher in 2002 than in 2005, and higher in 2012 than in 2008. Figure 4: Poverty headcount rate, poverty gap and poverty severity (percent), 2002-2012 30.0 25.4 25.0 20.0 18.5 Percent 14.3 15.0 12.4 10.0 5.7 4.0 5.0 2.3 2.9 1.9 1.3 0.7 1.0 0.0 Headcount Poverty Gap Squared Poverty Gap 2002 2005 2008 2012 Source: WB staff analysis on LSMS data. Note: Poverty line = 6,407.21Lek per month at 2012 prices. Using a regionally harmonized welfare aggregate and poverty line, Albania is one of the poorest countries in the region. Using the regional moderate poverty line of $5.00 US$, Albania has a high poverty rate at 47 percent, compared to the regional average of 22 percent (Figure 5). The national poverty line lies between the regional poverty lines of 2.50 and 5.00 US$ 2005 PPP per day (Figure 6). 5 At 5 percent level. 8 Figure 5: Poverty headcount rate, international poverty line at 5$PPP per day, latest regionally harmonized data between 2011 and 2013 90 83 78 80 73 70 60 47 50 40 40 36 31 30 21 17 15 20 11 9 8 8 7 6 5 5 10 3 2 1 1 0 0 Lithuania Moldova Turkey Latvia Belarus Croatia Georgia Montenegro Romania Romania Bulgaria Russian Federation Kyrgyz Republic Armenia Albania Hungary Estonia Poland Poland Ukraine Slovenia Czech Republic Slovak Republic Source: WB staff analysis on ECAPOV data Figure 6: Poverty headcount rate, different poverty lines 70.0 62.3 60.0 49.1 50.5 50.0 43.9 40.0 Percent 30.0 25.4 18.5 20.0 12.0 12.4 14.3 8.5 6.4 10.0 5.2 0.0 2.50$PPP poverty line National poverty line 5.00$PPP poverty line (3.12$PPP) 2002 2005 2008 2012 Source: WB staff analysis on LSMS data. STYLIZED FACT 2: POVERTY INCREASED IN URBAN AREAS, NARROWING URBAN-RURAL POVERTY DIFFERENCES Poverty increases were more pronounced in urban areas, narrowing urban-rural poverty differences. The overall poverty increase observed in Albania from 2008 to 2012 was driven by a statistically significant increase of more than three percentage points of urban poverty6 which, coupled with the large gains in 6 Significant at the 10 percent level. 9 poverty reduction in rural areas from 2005 to 2008, reduced significantly the urban-rural poverty divide compared to 2005 (Figure 7). Poverty rates in 2012 were higher than in 2005 (13.6 percent as opposed to 11.2 percent), reversing a slight gain for urban areas in the second half of the 2000s. In line with the national trend, the severity and depth of poverty increased in both urban and rural areas, but more so in the former (Table A1). In recent years (2008-2012), the urban population grew from 50 percent to 54 percent of the total population, and the concentration of poor became nearly evenly distributed between urban and rural areas given a large increase in the poor in urban areas (Figure 8). Figure 7: Poverty headcount urban and rural areas Figure 8: Distribution of the poor urban and rural areas, 2008-2012, percent 35.0 70 29.6 59.7 30.0 60 24.2 51.1 48.9 25.0 50 19.5 40.3 20.0 Percent 40 Percent 14.6 15.3 15.0 13.6 30 11.2 10.1 10.0 20 5.0 10 0.0 0 Urban Rural Urban Rural 2002 2005 2008 2012 2008 2012 Source: WB staff analysis on LSMS data. Note: Poverty line = 6,407.21Lek per month at 2012 prices. In line with poverty changes, household expenditures fell with a particularly pronounced drop in urban areas. The two percent increase in overall poverty between 2012 and 2008 is associated to a 5 percent mean reduction in total per capita expenditure. This contraction in consumption was higher in urban areas (-9 percent) than rural areas (-3 percent) (Figure 9). The second quintile of the population was the most affected (Table A2). Figure 9: Mean per capita expenditures, 2002-2012 10 15,000 14,000 Lek per capita per month 13,000 12,000 11,000 10,000 9,000 8,000 2002 2005 2008 2012 Total Rural Urban Source: WB staff analysis on LSMS data. Note: Poverty line = 6,407.21Lek per month at 2012 prices. STYLIZED FACT 3: POVERTY INCREASED IN MOST REGIONS, PARTICULARLY IN THE COASTAL REGION WHERE PREVIOUS GAINS WERE LARGELY ERODED Poverty increased within all regions except mountain areas, which experienced a considerable decline in poverty. In Tirana and in the Coastal regions not only did poverty increase in 2012 with respect to 2008, but it reached higher levels than in 2005. In 2012 the incidence of poverty was higher among individuals living in Coastal region, where 17.6 percent of the population was living below the poverty line (see Figure 10 and Table A3). The mountain areas experienced a sharp decline in poverty (from 27 percent in 2008 to 15 percent in 2012), despite a quite stable population share (from 8.3 percent in 2008 to 9.2 percent in 2012); it remains, however, the second poorest region in Albania. More insights are nevertheless needed to understand the drivers of poverty changes in this region. 11 Figure 10:Poverty Headcount, by region, 2002, 2005, 2008, and 2012 50 45 40 35 30 Percent 25 20 15 10 5 0 2002 2005 2008 2012 Coastal Central Mountain Tirana Source: WB staff analysis on LSMS data. Note: Poverty line = 6,407.21Lek per month at 2012 prices. The concentration of poor people has increased in the Coastal region. The Coastal region has accounted for 31 percent of the total population over the decade (2002 to 2012), while the relative contribution to poverty increased going from 25 percent in 2002, to 27 percent in 2005, to 32 percent in 2008 and 38 percent in 2012. In 2012, therefore, the Coastal region accounts for a larger share of poor individuals than the Central region, which is the most populous area with 42 percent of total population (Figure 11). Figure 11: Evolution of population share and relative contribution to poverty, by regions Population share Relative contribution to poverty 100% 100% 11.35% 7.97% 5.59% 12.69% 12.93% 16.17% 90% 18.21% 18.47% 90% 15.68% 11.75% 11.35% 20.62% 80% 80% 17.86% 9.82% 8.30% 9.19% 70% 70% 60% 60% 46.01% 44.76% 36.31% 42.83% 41.55% 51.39% 50% 50% 46.35% 37.14% 40% 40% 30% 30% 20% 20% 37.69% 30.88% 31.20% 30.66% 30.79% 27.34% 32.07% 10% 25.06% 10% 0% 0% 2002 2005 2008 2012 2002 2005 2008 2012 Coastal Central Mountain Tirana Coastal Central Mountain Tirana Source: WB staff analysis on LSMS data. Note: Poverty line = 6,407.21Lek per month at 2012 prices. 12 Figure 12:Incidence of poverty, by prefectures, 2012 Further disaggregating the data by prefectures also shows a heterogeneous poverty picture across the country. The LSMS 2012 data allow for a further decomposition of poverty by the 12 counties or prefectures (quarku or prefektura) in which the country is administered. Kukes is the poorest county, with 22 percent of population living below the poverty line followed by Fier and Lezhe both with 18 percent of poor individuals, while Elbasan and Gjirokaster are the least poor, with 11 percent of population living below the poverty line (see Figure 12and Table A4). The average poverty shorfall is highest in Lezhe (4.3 percent of the poverty line), while in Shkoder there is the highest concentration of poor at the bottom of the distribution (the squared poverty gap is the highest). A further decomposition by strata allows picturing the different prefectures by urban and rural settlements. In seven of the twelve prefectures (namely Berat, Fier, Gjirokaster, Korce, Kukes, Shkoder, Tirane) the incidence of poverty is higher in rural than in urban areas (see Figure A2). Urban poverty ranges from 5.6 percent in Kukes to 17.5 percent in Lezhe. Rural poverty ranges from 9.7 percent in Elbasan to 29.5 percent in Kukes. Overall, household expenditures declined particularly sharply in the Coastal and Source: WB staff analysis on LSMS 2012 data. Tirana regions (-7 and -13 percent, respectively). Looking at Note: prefectures are coloured according to their poverty poverty by prefectures and urban/rural, weighted by the total headcount level – from lower poverty (lighter blue) to higher poverty (darker blue). population in the country, data show (Figure 13, panel a.) that poor individuals are mainly concentrated in urban Tirana (19 percent), rural Fier (10 percent), and urban Durres (8 percent). Figure 13, panel b. shows instead the distribution of poor individuals as share or urban and rural population. Urban poor are mainly found in Tirana (38 percent) and Durres (16 percent); while rural poor live mainly in rural Fier (21 percent) and rural Tirana (16 percent). 13 Figure 13: Share of population and incidence of poverty by prefectures and rural/urban, 2012 a. As a share of total population b. As a share of urban/rural population Note: Each bar shows individuals and poor individuals living in each Note: Left-hand bars show individuals and poor individuals as share prefecture as a share of total population. The red (green) bars of total urban population. Right-hand bars show individuals and correspond to poor individuals living in urban (rural) areas. poor individuals as share of total rural population. The red (green) bars correspond to poor individuals living in urban (rural) areas. Source: WB staff analysis on LSMS 2012 data. STYLIZED FACT 4: POVERTY INCREASED PARTICULARLY AMONG HOUSEHOLDS WITH OLDER WORKERS AND YOUTH, AND THE SECONDARY-EDUCATED Poor households seem to have a higher share of youth and older people. The 15 percent increase in poverty (or 2 percentage points) has hit in particular households that concentrate mature working-age adults (age 50 to 64) - who bore a 73 percent increase in poverty with respect to 2008 - and youth (age 15-24) where the incidence of poverty increased by 24 percent (Table 1). Table 1: Poverty headcount, by age groups, percent Age groups 2002 2005 2008 2012 % change 2012-2008 0-14 32.8 25.1 19.8 21.1 6.8 15-24 25.7 18.3 12.0 14.8 23.5 25-49 24.1 18.0 13.1 14.4 10.0 50-64 15.3 12.0 5.4 9.3 73.0 65+ 20.1 13.4 8.5 9.6 13.1 14 Total 25.4 18.5 12.4 14.3 15.9 Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data. Nevertheless, the poor among younger age groups accounts for a higher share of the total poor. This is partly explained by a higher share of children and youth in the total population, observed in the age and gender population pyramid. Interestingly, the population pyramids evidence a missing population group between the ages of 30 and 40 years, likely linked to migration (Figure 14): around 2.8 percent of the Albanian population was estimated to be an international migrant in 2010.7 Figure 14: Age-gender pyramids, share of total population and share of poor (Percent), 2012 90-95 80-85 70-75 60-65 Age in years 50-55 40-45 30-35 20-25 10-15 0-5 6 4 2 0 2 4 6 Share in total population, % Poor females Poor males Females Males Note: The left(right)-hand side shows the share of males(female) and poor males(poor females) in each age-braket as shares of the total population. Specifically, the distance from the middle to each side in blue and pink denotes the total population share in that age group for males and females. The distance in light gray and purply is the proportion of poor people in that age group of the total number of poor, again for each gender. Average levels of education increased during the 2000s in Albania. During the period 2002-2012 average educational attainment improved: the share of individuals aged 25 or more who attained tertiary education increased by 75 percent over the 2000s, from 8 percent in 2002 to 14 percent in 2012. At the same time, there are 4 percentage points fewer individuals with at most incomplete primary education, and general secondary education became more important with respect to vocational types of secondary schools (see Figure 15). 7 2010 data from the World Development Indicators for the international migrant stock (% of population). 15 Figure 15: Share of population (age 25+) by highest level of education attained 8.4% 8.6% 11.5% 14.0% 15.8% 15.7% 18.4% 27.7% 17.8% 18.8% 13.9% 6.1% 56.1% 56.8% 55.6% 51.8% 2.0% 0.1% 0.6% 0.5% 2002 2005 2008 2012 None or less than 4 Incomplete (5-12) Specialized Secondary General Secondary Tertiary Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data Not surprisingly, education levels are negatively related to poverty. The incidence of poverty is highest in households in which the head have none or incomplete primary education at 18 percent (Figure 16 and Table A6). Conversely, poverty levels are lowest for individuals headed by someone who attained tertiary education at 2 percent in 2012. The increase in poverty in the years 2008-2012 has affected all individuals, regardless of the education level of the household head. Nevertheless, individuals with secondary education seem to have been particularly hit. On average, 66 percent of poor individuals live in households where the head has no education of incomplete primary education, as opposed to 45 percent of individuals in the top 60 percent of the distribution. On the other hand, only 2 percent of poor individuals live in households where the head has tertiary education, as opposed to 16 percent of individuals in the top three quintiles (Figure 16, right panel). Figure 16: Poverty and Education level Poverty Headcount by level of education of the Distribution of individuals by level of education of household head the household head 20.0 17.2 18.4 70.0 60.0 15.0 11.5 50.0 Percent Percent 9.5 40.0 10.0 8.1 5.6 30.0 5.0 1.9 2.2 20.0 10.0 0.0 0.0 None or General Specialized Tertiary None or General Specialized Tertiary Incomplete Secondary Secondary Incomplete Secondary Secondary Primary Primary 2008 2012 Poor Bottom 40 percent Top 60 percent Source: WB staff analysis on LSMS 2008 and 2012 data. Poor individuals live in larger households. On average, poor individuals live in households made of 6 individuals, as opposed to just 3 individuals for households in the top 60 percent. Individuals in the bottom 16 40 percent live in households about as numerous as the poor’ (see Figure 17). Only a minority of households are female-headed in Albania, so profiling on this characteristics might be less relevant. As Figure 18 shows, at most 10 percent of individuals in the top three quintiles and in urban settlements live in female-headed household. For poor individuals, in rural settlements 7 percent live in female-headed households, and only 5 percent of those living in urban settlements. Figure 18: Distribution of individuals in Female- Figure 17: Average Household size headed households 6.0 5.7 11.0 9.9 5.5 5.4 9.5 5.5 10.0 9.0 5.0 9.0 Average HH size 4.5 5.1 5.2 8.0 7.8 5.0 7.6 7.3 Percent 7.0 4.0 6.0 3.5 5.0 3.0 3.6 7.2 3.3 4.0 6.0 2.5 3.2 4.7 3.0 2.0 2.0 Overall Urban Rural Overall Urban Rural Poor Bottom 40 percent Top 60 percent Poor Bottom 40 percent Top 60 percent Source: WB staff analysis on LSMS 2012 data. STYLIZED FACT 5: SMALL OBSERVED POVERTY CHANGES MASK LARGE MOVEMENTS IN AND OUT OF POVERTY Even if overall poverty did not increase significantly over the period, economic mobility analysis suggest high churning. Using the synthetic panel methodology 8 to explore movements in and out of poverty over the period, results show that around 15 percent of households, both in 2005-2008 and in 2008- 2012, changed socio-economic status (falling into poverty or escaping poverty) (Figure 19). However, while around 7 percent of households escaped poverty in the recent period, around 8 percent of non-poor households fell into poverty (in net resulting in very small poverty changes). In 2005 and 2008, although movements out of poverty more than offset movements into poverty – thus resulting in poverty decreases overall – vulnerability was still present with 10 percent of poor households escaping poverty, while around five percent of the non-poor fell into poverty. Table 2 shows the full transition matrices of movements in and out of poverty in the two periods estimated using the abovementioned method, given the lack of panel data for Albania. 8The synthetic panel methodology builds on an imputation methodology to construct synthetic panel data with predicted consumption using two different rounds of cross sections. The approach relies on time- invariant individual and household characteristics, and makes assumptions on the error structure. For more details see Dang et al. (2014) on synthetic panels using point estimates. 17 Figure 19: Economic mobility –share of households moving in and out of poverty, and net escapes from poverty (in percent) 12% 10% 8% 6% 4% 2% 0% -2% -4% -6% -8% -10% 2005-2008 2008-2012 Escaped poverty Fell into poverty Net escapes from poverty Source: WB staff analysis on LSMS 2005, 2008, 2012 data. Notes: Methodology based on Dang et al. (2014) on synthetic panels using point estimates. Table 2: Mobility in Albania: Transition in and out of poverty (2005-2008 and 2008-2012) Destination (2008) Poor Non-poor Total Poor 6.9 4.7 11.6 Origin (2005) Non-poor 10.1 78.4 88.5 Total 17 83.1 100 Destination (2012) Poor Non-poor Total Poor 4.8 6.9 11.7 Origin (2008) Non-poor 7.7 80.6 88.3 Total 12.5 87.5 100 STYLIZED FACT 6: INEQUALITY DECLINED, FROM LARGER CONSUMPTION FALL AT THE TOP Between 2008 and 2012 inequality declined. The Gini index went from 28.2 in 2008 to 26.9 in 2012 (Figure 20), as consumption expenditures of those at the top of the distribution declined more sharply than for others (Table A2). Other measures of inequality shows the same path of inequality overtime. Tirana remains the region with higher inequality, with a Gini index of 28.4, even if inequality fell more sharply in this region. Inequality – measured with the Theil index – is almost fully explained by inequality within regions. 18 Figure 20: Inequality in per-capita expenditure distribution, total and by region 35 Gini Index 29.6 30 28.2 28.2 35 32.6 26.9 30 28.4 25 27.1 27.2 25.7 26.2 24.8 25 22.4 20 14.4 20 15 12.9 13.0 13.4 11.8 12.1 12.2 11.1 15 10 10 5 3.6 3.8 3.5 3.3 5 0 0 Gini Theil p90/p10 Atkinson (e=1) Coastal Central Mountain Tirana 2002 2005 2008 2012 2008 2012 Source: WB staff analysis on LSMS data. The decline in inequality resulted from a larger decline in consumption for the well-off, compared to the rest of the population. The incidence of growth across the consumption distribution was different pre and post crisis, and in urban and rural areas. Pre-crisis, as Figure 21 shows, the growth rate of the bottom quintiles was higher than the growth rate of the rest of the distribution; in other words, growth benefitted disproportionately more the poor and less well-off resulting in a good performance in shared prosperity. In the period 2008 to 2012, which covers the global economic crisis, consumption growth is negative for all groups and more so for those in the highest quintile (Figure 22 and Table A2), which is behind the falling inequality. 19 Figure 21: Growth-incidence curve (2002-2008) Figure 22: Growth-incidence curve (2008-2012) a. Overall Population (2002-2008) a. Overall Population (2008-2012) Total (years 2006 and 2008) Total (years 2008 and 2012) 6 0 Growth incidence 95% confidence bounds Growth incidence 95% confidence bounds Growth in mean Growth at median Growth in mean Growth at median Mean growth rate Mean growth rate -1 5 Annual growth rate % Annual growth rate % -2 4 -3 3 -4 2 1 10 20 30 40 50 60 70 80 90 100 1 10 20 30 40 50 60 70 80 90 100 Expenditure percentiles Expenditure percentiles b. Urban (2002-2008) b. Urban (2008-2012) Total (years 2002 and 2008) Total (years 2008 and 2012) 0 8 Growth incidence 95% confidence bounds Growth in mean Growth at median Mean growth rate -2 Annual growth rate % Annual growth rate % 6 -4 -6 4 Growth incidence 95% confidence bounds Growth in mean Growth at median Mean growth rate -8 1 10 20 30 40 50 60 70 80 90 100 Expenditure percentiles 2 1 10 20 30 40 50 60 70 80 90 100 Expenditure percentiles c. Rural (2002-2008) c. Rural (2008-2012) Total (years 2002 and 2008) Total (years 2008 and 2012) 5 1 0 4 Annual growth rate % Annual growth rate % -1 3 -2 2 -3 Growth incidence 95% confidence bounds Growth incidence 95% confidence bounds Growth in mean Growth at median Growth in mean Growth at median -4 Mean growth rate Mean growth rate 1 1 10 20 30 40 50 60 70 80 90 100 1 10 20 30 40 50 60 70 80 90 100 Expenditure percentiles Expenditure percentiles Source: World Bank Staff analysis over LSMS 2002, 2008 and 2012. 20 UNDERSTANDING THE TRENDS STYLIZED FACT 7: WEAK ECONOMIC GROWTH DROVE POVERTY INCREASES Economic growth decelerated significantly following the global economic crisis. The GDP per capita growth rate suddenly dropped from 7.5 percent in 2008 to 3.4 percent in one year, and in 2013 it showed a merely 1.4 percent growth of the economy (Figure 23). Household consumption from National Accounts shows an even more serious deceleration, consistent with the negative consumption growth rates captured in the LSMS. Figure 23: Annual percentage growth rate of GDP per capita and consumption (national accounts), 2002- 2014 12.0 9.4 10.0 7.5 8.0 5.9 5.9 5.7 5.5 5.0 6.0 4.3 3.7 Percent 2.9 3.4 4.0 2.5 1.6 1.4 1.3 2.0 0.0 -2.9 -2.0 -4.0 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 GDP growth HH consumption growth Source: WB staff analysis on World Development Indicators. Note: GDP and Household final consumption expenditure in constant local currency (indicators ny_gdp_pcap_kn and ne.con.petc.kn) In the period under study, the construction sector was the most affected by the economic downturn. While growth rates for other sectors were modest but positive, the construction sector shrank by around 9 percent annually from 2008 to 2012. The share of the sector in GDP fell from 16 percent to 10 percent over the period (Figure 24). These sectoral patterns are consistent with an increase in urban poverty – as construction is usually concentrated in urban areas and is a labour-intensive sector. In 2009, 10 percent of the total employed in Albania (mostly males) worked in the construction sector. This represented around 114,882 jobs. By 2011, this number had gone down to 94,488 workers. The large majority of workers are in the agriculture sector (around 45 percent pre crisis, 2008) – mostly concentrated in rural areas – which experienced modest but positive growth rates over the period.9 9 Data estimated using Labor Force Surveys. 21 Figure 24: GDP by sector: sectoral share and annualized sectoral growth rate, 2008-2012 30% 25% 20% 15% 10% 5% 0% Agriculture, Mining and Manufacturing Construction Trade, Hotels Transport Post and Other Services -5% Hunting, Quarrying and Restaurants Communications Forestry & -10% Fishing -15% GDP share 2008 GDP share 2012 Annualized sectoral growth rate Source: Own calculations with statistics from INSTAT Changes in mean consumption growth, not changes in inequality, drove poverty increases between 2008 and 2012. Decomposing poverty changes into those attributed to growth and redistribution shows that the shift in mean growth was poverty-enhancing (pushing poverty up by 3.95 percentage points), and were not fully offset by changes in the shape of the distribution. The growth effect was stronger in urban areas than in rural areas (Figure 25). In the first part of the 2000s, however, both changes in mean growth and redistribution contributed to the downward trend in poverty, mainly driven by growth. When compared to what happened in the previous period (2002-2008), data (Figure 25) shows that the thirteen percentage points reduction in poverty can be attributed to growth primarily (-11 percentage points, 84 percent of the change), while a relatively smaller but poverty-reducing share to redistribution (-2 percentage points, 16 percent of the change). This pattern is similar in both rural and urban areas, although the distribution effect was relatively stronger for urban areas (21 percent of the change compared to 14 percent in rural areas). 22 Figure 25: Growth-redistribution decomposition of poverty changes (Shapley Approach), percentage points 2002-2008 2008-2012 Urban Rural Total Urban Rural Total 2 6 0 5 -2 4 Percentage Points Percentage Points -4 3 5.4 -6 2 3.95 -8 1 2.02 -10 0 -12 -1 -1.9 -1.4 -1.99 -14 -2 -16 -3 Growth Redistribution Growth Redistribution Source: WB staff analysis on LSMS 2002, 2008, 2012 data. STYLIZED FACT 8: WEAKENED LABOR MARKETS AS A CHANNEL FOR POVERTY INCREASES Labor markets weakened during the global economic crisis of 2009. With the slowdown in economic growth, employment rates declined, both from higher inactivity and unemployment rates.10 The results here focus on labor market outcomes derived from the LSMS. Some data issues emerge when measuring labor market activity with the LSMS, particularly in 2012, which could introduce bias to the results. Nevertheless, the overall shock to labor markets is validated with LFS data. 11 The share of households with an unemployed or inactive (non-retired) head increased in Albania between 2008 and 2012. Particularly, the share of heads in unemployment rose from around 3 to 12 percent over the period (Figure 26); similarly, a large share of household heads became inactive (from 7 to around 14 percent). This is partly due to a sharp decline in people working as own-account and unpaid family workers during the economic downturn. In tandem with a deterioration in labor market outcomes of household heads, the share of the total population (15 years and older) in unemployment and inactivity increased from 4 percent to 12 percent, and from 17 percent to 28 percent, respectively (see Table A7). 10 See Albania labor market note based on the LFS, produced by the Poverty Global Practice. 11 Some potential data challenges related to changes in survey fieldwork need to be highlighted. The 2012 LSMS was carried out in the Fall of 2012, while previous rounds carried out the data collection in the Spring. The impact of this change on key indicators is not fully established. The extent to which it affects the consumption aggregate and poverty rates cannot be determined; however, under the assumption of consumption smoothing, it could be relatively less than other more seasonal indicators such as labor market outcomes (including labor income). Is this is the case, some sharp changes in labor market outcomes observed between 2008 and 2012 could be partly explained by fieldwork changes rather than true trends in the economy. Nevertheless, the overall shock to labor markets is validated with LFS data in the Albania labor market note (2015) produced by the Poverty Global Practice. 23 Figure 26: Labor force status of household heads in the total population, 2002-2012 100 7.26 6.76 7.02 90 14.41 80 21.29 22.44 28.70 70 25.75 6.16 3.56 60 13.91 2.87 6.82 11.68 50 36.14 3.33 40 20.52 23.08 17.92 30 20 28.20 32.77 30.94 26.63 10 0 2002 2005 2008 2012 Employed Self-Employed Unpaid Family Workers Unemployed Retired OLF Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data. Note: Figures for category “Student” are close to zero. Self-employed includes both employers (a very small share of the total) and own-account workers. For 2002, it is not possible to disentangle self-employed (employers and own-account workers) from unpaid family workers. Along with a higher share of unemployed and inactive household heads in Albania, the concentration of the poor increased dramatically in households with these categories. Of the total number of poor in the country, 19 percent are in households with an unemployed head - the highest of the decade and in contrast to around 4 percent in 2008 - and 25 percent in households with a head out of the labor force (not retired) (Table 3). In 2002, in turn, the highest concentration of the poor was among the self-employed (44 percent), dropping sharply by 2012 (17 percent). Similarly, of the poor population 15 years and older, more were concentrated among the unemployed and inactive in 2012, compared to 2008. Table 3: Distribution of the poor by labor force status of the household head (percent) 2002 2005 2008 2012 Employee 20.06 25.50 25.36 20.18 Self-Employed* 18.82 22.99 12.92 43.23 Unpaid family worker 23.04 11.37 4.51 Unemployed 9.54 6.00 3.39 18.67 Retired 16.39 16.12 23.23 19.19 Out of the labor force (excl. retired) 9.92 10.47 13.56 24.53 Total 100.0 100.0 100.0 100.0 Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data. Note: Figures for category “Student” omitted. Self-employed includes both employers (a very small share of the total) and own-account workers. For 2002, it is not possible to disentangle self-employed (employers and own-account workers) from unpaid family workers. * Employers and own-account workers. Accordingly, poverty is significantly higher in households with an unemployed or inactive head. Individuals living in household where the household head is occupied (either employed or self-employed) or retired are less likely to be poor than individuals living in households headed by unemployed or inactive 24 (Table 4). In 2012, in fact, the incidence of poverty in households with an employed or retired head was, respectively, 11 percent and 13 percent as opposed to 22 and 24 percent, in turn, for households with unemployed and inactive heads. Between 2008 and 2012 the incidence of poverty has increased by 5 percentage points among individuals living in household headed by unemployed individuals, while it decreased by 2 percentage points among individuals whose household head was self-employed. Among employees, public sector workers have significantly lower poverty rates (see Table A8). Table 4: Poverty headcount by labour force status of the household head (percent) 2002 2005 2008 2012 Employee 18.01 14.08 9.66 10.55 Self-Employed 28.37 21.04 12.89 10.64 Unemployed 39.04 30.84 14.59 21.17 Retired 22.11 15.33 11.97 12.81 Out of the labor force (excl. retired) 35.80 29.53 24.10 23.94 Total 25.41 18.51 12.44 14.35 Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data. Note: Figures for category “Student” omitted. Not surprisingly, employment played an important role in poverty changes. Although capturing income is not the focus of the LSMS and welfare is measured using consumption data, utilizing the available data on income, with caveats in mind12, could point to some insights on poverty changes. Decomposing poverty by income source shows opposite trends in the factors behind poverty reduction between 2005-2008 and 2008-2012. From 2005 to 2008 poverty reduction was led by favourable changes in labor income – from a combination of higher wages/earnings, and higher numbers of adults and employed adults in the household – and changes in pensions. From 2008 to 2012, the loss of jobs (“shared of employed”) is a key factor behind poverty trends with a poverty-increasing effect (Figure 27). Looking at growth in household income shows a similar pattern (Figure 28), with job loss across quintiles. Increases in wages/earnings were particularly relevant for income growth of those in the lowest quintiles in that period (Figure 28, panel a). Poverty decomposition results show that wages/earnings continued contributing to poverty reduction, signalling that the labor market adjusted in quantities during the crisis. 12Income is not the best proxy for welfare in a country like Albania with very high informality and a large agricultural sector. Specifically, income tends to be underreported, especially in societies with large agricultural or self-employed populations for various reasons, including people forgetting items sold, reluctance to disclose the full extent of income, or reluctance to report income earned illegally. In addition, income measures do not capture how household smooth consumption when facing a negative or positive economic shock. Finally, income does not capture the advantage of having free or subsidized access to services, such as housing, health care, education or child care. 25 Figure 27: Poverty decompositions by income source, percentage points 2005-2008 2008-2012 Other 0.1 Other -1.0 Social Assistance -0.2 Social Assistance -3.2 Pensions -2.9 Pensions -5.2 Wages/Earnings -1.8 Wages/Earnings -2.8 Share of Employed -0.2 Share of Employed 5.6 Share of Adults -2.2 Share of Adults -2.5 Ratio Ratio 1.3 10.4 Consumption/Income Consumption/Income -4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 -10.0 -5.0 0.0 5.0 10.0 15.0 Source: WB staff analysis on LSMS 2002, 2005, 2008, 2012 data. Note: Labor income is captured by wages/earnings, shared of adults (demographic effect) and share of employed. “Other” includes mainly remittances, as well as property and other income. 26 Figure 28: Decomposition of total household income growth 2005-2008 a. Income growth by source and quintile b. Average changes in income growth 0.4 Total Income 15% 0.3 Other -1% 0.2 Percent 0.1 Social Assistance 2% 0 Pensions 7% -0.1 Wages/Earnings -3% -0.2 Lowest II III IV Highest Share Employed 2% Quintile Quintile Share Adults Share Employed Wages/Earnings Share Adults 7% Pensions Social Assistance Other -15% -5% 5% 15% 25% 2008-2012 a. Income growth by source and quintile b. Average changes in income growth 0.8 0.6 Total Income 9% 0.4 Other -5% Percent 0.2 Social Assistance 6% 0 Pensions 14% -0.2 Wages/Earnings 5% -0.4 Lowest II III IV Highest Share Employed -11% Quintile Quintile Share Adults Share Employed Wages/Earnings Share Adults 0% Pensions Social Assistance Other -15% -5% 5% 15% 25% Source: WB staff analysis on LSMS 2005, 2008 and 2012 data. Note: Labor income is captured by wages/earnings, shared of adults (demographic effect) and share of employed. “Other” includes mai nly remittances, as well as property and other income. 27 STYLIZED FACT 9: SOCIAL PROTECTION, PARTICULARLY PENSIONS, CONTRIBUTED TO MITIGATING THE LABOR MARKET SHOCK; REMITTANCES FLOWS SHRANK Pension income increased significantly in Albania over the period. The pension system deficit as a share of GDP, including subsidies for rural pensions, increased from 2.7 percent in 2005 to 3.6 percent in 2012. This is partly linked to large increases in average pensions: administrative data shows that average urban pensions rose by 18.9 percent and average rural pensions by 61.4 percent during 2007 –2012. In line with administrative data, pension income increased significantly for households in rural areas (around 19 percent between 2008 and 2012) accounting for the majority of total income increases in those locations. Overall, pensions accounted for a higher share of total household income in 2012 compared to 2008 (Tables A9, A10 and A11). Largely pensions, and to a lesser extent social assistance, contributed to mitigating the poverty effect of the loss of jobs. Income from pensions and social assistance increased for all quintiles (Figure 21) – particularly for the poorest - and both played an important role in mitigating poverty impacts from a labor market shock (Figures 20). In relative terms, by 2012, households – poor and non-poor - relied much more on pensions and private transfers as income sources than in 2008 (Table A9 through A10) Remittances declined as the crisis hit the Eurozone, hitting some groups of the population in Albania. Remittances fell from 10.3 percent of GDP in 2008 to 6.8 percent in 2013, largely due to the close links to poorly-performing Greek and Italian economies where a large share of migrants reside. The stock of Albanian emigrants stands at 40 percent: in 2013 – very high compared to neighbouring countries. Almost 1.3 million Albanians were living outside the country, most of them in Greece (46 percent) and Italy (36 percent).13 The decline in remittances affected relatively less the poorest – hence the lower role in increasing poverty – but more so the income of the better-off. Those in the fourth quintile, who benefited significantly from remittances between 2005 and 2008, suffered the largest contraction in this source of income during the crisis period. While remittances represented around 8 percent of income for the non-poor in 2008, the share of income from remittances was around 3 percent for the poor. This share declined dramatically by 2012 (Table A10 and A11), in line with the declining trends in aggregate remittances previously mentioned. For the better-off in rural areas, remittances played a larger role both pre-crisis (increasing their income) and post crisis (reducing their income) (Figure A5 through A8). 13 World Bank (2015). Next Generation Albania: A Systematic Country Diagnostic. 28 CONCLUSION The stylized facts in this short note aimed at updating poverty trends for Albania using the 2012 LSMS and exploring initial hypothesis on trends and drivers of poverty. Since poverty numbers and analysis were not available since the 2008 LSMS round, this analysis can be useful in presenting issues and areas for further research. Findings show that poverty reduction halted in Albania in the recent period (2008-2012). Net trends were driven by a combination of lower economic growth and particularly weakened labor markets, as well as increases in social protection benefits particularly pensions. Poverty results are consistent with the story on shared prosperity for Albania: the country performed well before the global economic crisis in terms of the shared prosperity indicator (consumption of the bottom 40 percent of the distribution grew by 2.64%), but this trend did not continue during the most recent period. In fact, from 2008 to 2012, mean consumption declined by around 1.31% and 1.22% for the bottom 40 percent. Beyond the period analysed in this note, prospects for poverty reduction in the coming years are positive in Albania but risks are on the downside. Economic developments in Greece and the Eurozone could negatively impact economic growth, labor markets, and remittances inflows, adversely affecting households. Labor markets in Albania have been a key mechanism, as in most countries, driving poverty changes and overall improvements or declines in welfare. The drop in employment and increase in both inactivity and unemployment documented in this note, and their close association with poverty status, point to a critical agenda in terms of job creation. Worryingly, labor markets further deteriorated beyond 2012, and Albania faces multiple structural labor market challenges, with the low quality of jobs being a particularly salient issue.14 However, knowledge gaps remain in terms of understanding the constraints to more and better jobs in Albania. In addition, the role of social protection is critical – as previously discussed - and improvements need to be made there as well, as pointed in the recent Albania Systematic Country Diagnostic (World Bank, 2105). Encouragingly, efforts are ongoing to better target the main social safety net program and make it more compatible with work. However, other instruments that can respond to times of crisis should be re-visited to ensure they are flexible enough to do so. Given that remittances are a source of income for many households in Albania, and that remittances flows are highly vulnerable given economic struggles of sending countries, counting on tools to protect the poor and vulnerable, alongside efforts to increase their market income-generating capacity, is critical. A poverty measurement agenda also emerges. Specifically, the main constraint for monitoring poverty and overall welfare in Albania lies in the frequency of the surveys (thus far surveys have been carried out every 3 to 4 years) and availability of donor funding (used thus far to finance the household surveys). Tools for a more frequent and sustained welfare monitoring are needed. 14See accompanying labor market note (World Bank, 2015) documenting a large share of low-skilled jobs, many as unpaid family workers. 29 ANNEX FIGURES Figure A1: Poverty Incidence Curve (poverty headcount at different levels of the poverty line), by year .25 Year 2002 Year 2005 Year 2008 Year 2012 .2 2002 2005 .15 FGT(0), % 2012 .1 2008 .05 0 0 1281.442 2562.884 3844.326 5125.768 6407.21 Poverty Line, New Leks, 2012 prices 30 Figure A2: Incidence of poverty, by prefectures, urban and rural settlements, 2012 Urban Rural Source: WB staff analysis on LSMS 2012 data. Note: prefectures are coloured according to their poverty headcount level – from lower poverty (lighter color) to higher poverty (darker color). Note: Poverty line = 6,407.21 Lek per month at 2012 prices. Figure A3. Expenditures vs. income: Log-Distribution of per capita income and expenditure, 2012 .8 Expenditure Income .6 Kernel density .4 .2 0 0 5 10 15 Logarithm of per capita Income or Expenditure Source: WB staff analysis on LSMS 2008, 2012 data. 31 Figure A4: Age-gender pyramids, share of total population and share of poor (Percent) ALB2002 ALB2005 90-95 90-95 80-85 80-85 70-75 70-75 60-65 60-65 Age in years Age in years 50-55 50-55 40-45 40-45 30-35 30-35 20-25 20-25 10-15 10-15 0-5 0-5 8 6 4 2 0 2 4 6 6 4 2 0 2 4 6 Share in total population, % Share in total population, % Poor females Poor males Poor females Poor males ALB2008 90-95 80-85 70-75 60-65 Age in years 50-55 40-45 30-35 20-25 10-15 0-5 6 4 2 0 2 4 6 Share in total population, % Poor females Poor males Note: The left(right)-hand side shows the share of males(female) and poor males(poor females) in each age-braket as shares of the total population. Specifically, the distance from the middle to each side in blue and pink denotes the total population share in that age group for males and females. The distance in light gray and purply is the proportion of poor people in that age group of the total number of poor, again for each gender. 32 Figure A5: Decomposition of total income growth (2005-2008), Urban Settlements a. Income growth by source and quintile b. Average Changes 0.3 Total Income 2% 0.2 Other -4% 0.1 0 Social Assistance 1% -0.1 Pensions 6% -0.2 Wages/Earnings -9% -0.3 Share Employed 2% Lowest II III IV Highest Quintile Quintile Share Adults 6% Share Adults Share Employed Wages/Earnings Pensions Social Assistance Other -15% -5% 5% 15% 25% Source: WB staff analysis on LSMS 2005, 2008 data. Figure A6: Decomposition of total income growth (2005-2008), Rural Settlements a. Income growth by source and quintile b. Average Changes 0.7 0.6 Total Income 0.5 0.4 Other 10% 0.3 0.2 Social Assistance 3% 0.1 Pensions 6% 0 -0.1 Wages/Earnings 5% -0.2 -0.3 Share Employed -2% Lowest II III IV Highest Quintile Quintile Share Adults 7% Share Adults Share Employed Wages/Earnings -15% -5% 5% 15% 25% Pensions Social Assistance Other Source: WB staff analysis on LSMS 2005, 2008 and 2012 data. Note: Labor income is captured by wages/earnings, shared of adults (demographic effect) and share of employed. “Other” includes mainly remittances, as well as property and other income. 33 Figure A7. : Decomposition of total income growth (2008-2012), Urban Settlements a. Income growth by source and quintile b. Average Changes 0.6 0.5 Total Income 8% 0.4 0.3 Other -2% 0.2 0.1 Social Assistance 7% 0 -0.1 Pensions 11% -0.2 Wages/Earnings 1% -0.3 -0.4 Share Employed -8% Lowest II III IV Highest Quintile Quintile Share Adults -1% Share Adults Share Employed Wages/Earnings Pensions Social Assistance Other -15% -5% 5% 15% 25% Source: WB staff analysis on LSMS 2005, 2008 data. Figure A8: Decomposition of total income growth (2008-2012), Rural Settlements a. Income growth by source and quintile b. Average Changes 1 0.8 Total Income 5% 0.6 0.4 Other -11% 0.2 Social Assistance 6% 0 -0.2 Pensions 19% -0.4 -0.6 Wages/Earnings 12% Lowest II III IV Highest Quintile Quintile Share Employed Share Adults Share Employed Share Adults 0% Wages/Earnings Pensions Social Assistance Other -15% -5% 5% 15% 25% Source: WB staff analysis on LSMS 2005, 2008 and 2012 data. Note: Labor income is captured by wages/earnings, shared of adults (demographic effect) and share of employed. “Other” includes mainly remittances, as well as property and other income. 34 ANNEX TABLES Table A1: Poverty gap and squared poverty gap (percent), total, urban and rural, 2002-2012 Poverty Gap (P1) Squared Poverty Gap (P2) 2002 2005 2008 2012 2002 2005 2008 2012 Rural 6.59 5.31 2.64 3.02 2.14 1.75 0.71 1.01 Urban 4.47 2.33 1.98 2.85 1.60 0.80 0.61 0.90 Total 5.7 4.0 2.3 2.9 1.9 1.3 0.7 1.0 Source: WB staff analysis on LSMS data. Table A2: Mean per capita expenditure for different groups (Leks per month, 2012 prices) 2002 2005 2008 2012 Percentage change 2012-2008 Overall Population 10,218.90 11,932.24 12,748.25 11,709.51 -8.15 Type of settlement Rural 9,447.16 10,306.62 11,585.19 11,145.23 -3.80 Urban 11,297.68 14,006.83 13,935.81 12,189.69 -12.53 Regions Coastal 11,029.2 12,548.8 12,639.2 11,217.4 -11.2 Central 9,819.8 11,151.0 12,308.3 12,026.9 -2.3 Mountain 8,079.7 9,354.1 9,319.9 9,939.6 6.6 Tirana 11,845.6 15,476.4 15,528.6 12,696.5 -18.2 Quintiles poorest 4,647.41 5,135.02 5,873.66 5,552.52 -5.47 2 6,872.33 7,769.21 8,591.42 7,999.16 -6.89 3 8,824.20 10,348.73 11,053.61 10,271.29 -7.08 4 11657.89 13,551.51 14,257.5 13,444.44 -5.70 least poor 19106.11 22,869.55 23,974.75 21,290.25 -11.20 Source: WB staff analysis on LSMS 2012 data. 35 Table A3. Poverty headcount by region, 2002-2012, percent Change 2012-2008 Region 2002 2005 2008 2012 (Percentage points) Coastal 20.6 16.2 13.0 17.6 4.61 Central 25.6 21.2 10.7 12.5 1.80 Mountain 44.5 25.6 26.6 15.3 (11.31) Tirana 17.8 8.1 8.8 12.6 3.78 Total 25.4 18.5 12.4 14.3 1.97 Table A4. Poverty, different indices, by prefectures, 2012, percent Headcount Poverty Gap Squared Headcount Headcount Poverty Gap Urban Rural Berat 12.67 2.36 0.71 10.19 14.57 Diber 13.01 2.32 0.68 15.83 12.05 Durres 16.23 3.31 1.10 16.48 15.39 Elbasan 10.72 2.25 0.75 12.38 9.70 Fier 17.54 3.52 1.06 11.02 21.82 Gjirokaster 10.69 2.23 0.87 9.62 11.79 Korce 12.22 2.49 0.71 10.50 13.28 Kukes 21.79 3.71 0.91 6.55 29.50 Lezhe 17.48 4.30 1.56 17.53 17.41 Shkoder 15.69 3.69 1.59 14.95 16.27 Tirane 14.15 2.81 0.85 14.12 14.24 Vlore 11.68 2.40 0.83 12.14 10.75 Total 14.34 2.93 0.95 15.27 13.56 Table A5. Gini coefficient 2002 2005 2008 2012 Total 0.2818 0.2963 0.2823 0.2692 Region Coastal 0.2792 0.2940 0.2708 0.2720 Central 0.2687 0.2866 0.2567 0.2624 Mountain 0.2716 0.2443 0.2478 0.2236 Tirana 0.2980 0.2984 0.3262 0.2843 36 Table A6. Poverty headcount by education level, percent 2002 2005 2008 2012 Change 2012-2008 Incomplete (5-12) 27.0 20.9 13.3 15.0 1.7 General Secondary 16.7 10.3 7.7 10.9 3.2 Specialized Secondary 12.3 6.6 4.2 7.5 3.3 Tertiary 3.2 1.7 1.6 3.1 1.5 Population 25+ 20.6 14.9 9.6 11.8 2.2 Table A7. Poverty, distribution of poor and population by employment status, population 15 years and older, percent Poverty Headcount Rate Distribution of the Poor Distribution of Population 2002 2005 2008 2012 2002 2005 2008 2012 2002 2005 2008 2012 Employee 14.0 10.2 7.1 7.7 11.3 12.8 15.2 11.1 18.1 20.5 22.4 18.1 Self Employed 25.6 13.8 8.6 8.4 40.5 9.3 10.8 6.5 35.3 11.1 13.1 9.8 Unpaid Family Worker 24.3 13.0 10.5 29.4 17.4 3.3 19.7 13.9 3.9 Unemployed 33.3 20.2 21.6 18.2 11.1 5.9 9.2 16.9 7.5 4.8 4.4 11.7 Out of the labor force Retired 18.4 12.5 7.7 9.8 12.8 12.2 14.2 13.2 15.6 15.9 19.3 17.0 Student 15.2 7.9 6.2 10.6 4.0 4.8 5.8 10.1 5.8 10.0 9.8 11.9 Other OLF 25.4 23.2 16.8 17.8 20.2 25.5 27.4 38.9 17.8 17.9 17.1 27.5 Population 22.3 16.3 10.4 12.6 100 100 100 100 100 100 100 100 15+ Table A8. Poverty by type of ownership of the job place of the household head, 2012, percent Poverty Population Absolute Relative share contribution contribution Private 13.38 62.74 8.40 79.61 1.97 1.55 1.28 4.16 Public 4.79 33.11 1.59 15.03 1.23 1.50 0.41 3.66 Total employee HH 10.55 100.00 10.55 100.00 1.39 0.00 1.39 0.00 Note: Population shares of private and public employees do not add up to 100 because 4 of observations have missing information on ownership of the job place. Table A9: Share of total income by income source, by consumption deciles 2008 and 2012, percent 37 Deciles Wage Income Pension Social Assistance Private Transfers 2008 2012 2008 2012 2008 2012 2008 2012 1 70.4 42.1 16.2 38.1 9.2 16.6 2.8 1.0 2 75.7 41.2 16.4 41.5 4.9 14.5 2.1 1.3 3 75.2 48.6 13.7 32.4 3.9 14.6 2.4 1.3 4 71.0 56.4 20.4 29.7 3.6 10.6 3.6 1.6 5 76.7 57.3 16.2 31.9 3.7 7.1 2.7 2.5 6 76.4 56.2 17.4 27.2 2.4 10.4 2.6 2.5 7 57.7 59.5 14.5 24.5 2.3 10.5 8.2 2.2 8 64.3 60.8 12.2 26.4 2.0 7.7 19.5 2.9 9 62.6 63.7 13.1 26.7 3.2 4.8 15.2 3.2 10 78.2 72.5 10.6 15.1 3.4 2.9 4.8 3.7 Total 69.9 58.6 14.2 27.3 3.4 8.7 8.0 2.5 Unemployment Deciles Property Other Incomes Total Benefit 2008 2012 2008 2012 2008 2012 1 0.1 0.8 0.6 0.2 0.7 1.1 100 2 0.2 0.7 0.4 0.4 0.3 0.2 100 3 0.2 0.7 4.5 0.3 0.1 2.1 100 4 0.3 0.7 1.1 0.7 0.1 0.3 100 5 0.1 0.6 0.6 0.2 0.2 0.3 100 6 0.0 0.1 1.1 2.6 0.1 0.9 100 7 0.0 0.4 16.9 2.6 0.2 0.3 100 8 0.0 0.1 1.6 1.6 0.3 0.6 100 9 0.2 0.1 4.6 0.9 1.1 0.5 100 10 0.1 0.2 2.6 5.0 0.2 0.5 100 Total 0.1 0.4 4.1 1.9 0.4 0.6 100 Source: WB staff analysis on LSMS 2012 data. Table A10: Share of total income by income source, by poverty status, 2008, percent Wage Social Private Unemployment Other Pension Property Total Income Assistance Transfers Benefit Incomes Non 69.9 14.1 3.0 8.2 0.1 4.3 0.3 100 poor Poor 71.1 15.1 9.7 2.9 0.1 0.6 0.6 100 Total 69.9 14.2 3.4 8.0 0.1 4.1 0.4 100 Source: WB staff analysis on LSMS 2008 data. Table A11: Share of total income by income source, by poverty status, 2012, percent Wage Social Private Unemployment Other Pension Property Total Income Assistance Transfers Benefit Incomes 38 Non 60.4 26.1 8.0 2.6 0.3 2.0 0.6 100 poor Poor 40.3 39.9 16.6 1.1 1.0 0.4 0.8 100 Total 58.6 27.3 8.7 2.5 0.4 1.9 0.6 100 Source: WB staff analysis on LSMS 2012 data. 39 40