ARMENIA POVERTY ASSESSMENT REPORT N0. 101680-AM HOW THE CRISIS CHANGED THE PACE OF POVERTY REDUCTION AND SHARED PROSPERITY Poverty Global Practice Europe and Central Asia June 2015 Acknowledgments This Poverty Assessment was prepared as part of the FY 15 South Caucasus Programmatic Poverty Assessment Technical Assistance (P151474). The Armenia Poverty Assessment team is comprised of Nistha Sinha (Task Team Leader), Moritz Meyer, and Sasun Tsirunyan. The report benefitted from discussions with Gohar Gyulumyan, Aleksan Hovhannisyan, Tigran Kostanyan and Donato De Rosa. Satik Nairian provided logistics and processing support. Armanda Çarçani formatted the report. The report was written under the guidance of Henry G. Kerali, Laura Bailey, Carolina Sanchez- Paramo, and Rashmi Shankar. The peer reviewers were Nandini Krishnan and Pedro Rodriguez. The team gratefully acknowledges all the comments and feedback received from the National Statistical Service of the Republic of Armenia (NSSRA), and Ministry of Labor and Social Issues. 2 Contents ACKNOWLEDGMENTS ......................................................................................................... 2 ACRONYMS .......................................................................................................................... 4 1. INTRODUCTION ............................................................................................................... 5 2. RECENT DEVELOPMENTS ............................................................................................... 8 Growth in Household Consumption ......................................................................................................... 8 Inequality ................................................................................................................................................ 10 Poverty Reduction in International Perspective ...................................................................................... 10 Poverty Reduction from the National Perspective .................................................................................. 12 3. UNDERSTANDING THE PERFORMANCE ON POVERTY REDUCTION ............................ 17 The Domestic Labor Market ................................................................................................................... 19 Remittances from Workers Abroad ........................................................................................................ 23 Pensions and Social Transfers ................................................................................................................ 24 4. CONCLUSION ................................................................................................................. 25 REFERENCES ..................................................................................................................... 26 3 Acronyms ADB Asian Development Bank ECA Europe and Central Asia ECAPOV Database of harmonized consumption data, Word Bank ECATSD Europe and Central Asia Team for Statistical Development FBP Family Benefit Program FDI Foreign Direct Investment GDP Gross Domestic Product ILCS Integrated Living Conditions Survey MPI Multidimensional Poverty Index NSSRA National Statistical Service of the Republic of Armenia PER Public Expenditure Review RCB Russian Central Bank SSPA Social Snapshot and Poverty in Armenia WDI World Development Indicators 4 1. Introduction 1.1. This report examines Armenia’s experience in reducing poverty and raising the welfare of the least well-off in the country in the years since 2009. What households spend on consumption is an indicator of their welfare. As the economy recovered from crisis, the least well-off enjoyed some growth in consumption spending, but not as much as in the years up to 2009. Moreover, growth has become less pro-poor in relative terms because the less well-off enjoyed lower growth in consumption than the better- off. As a result, although consumption did translate into a reduction in poverty, inequality is now higher than before 2009. In 2013, 32 percent of Armenia’s population lived below the national poverty line – a poverty rate higher than in pre-crisis years but down from the high of 35.8 percent in 2010. In fact, between 2012 and 2013, poverty reduction seems to have stalled. This report looks at the micro and macro aspects of Armenia’s poverty reduction experience to (a) describe the key features of post-crisis poverty, inequality, and consumption growth; (b) examine the drivers of poverty reduction in this period; and (c) explore reasons why future growth might not be as pro-poor as in the past. 1.2. In Armenia growth of the economy slowly resumed after the recession of 2009 but it has not yet reached the levels seen before the crisis (Figure 1). Up to 2009 Armenia’s real GDP had been growing at an average rate of 11.8 percent. The economic crisis broke this pattern, and in 2009 GDP went down by 14.1 percent. After a fiscal stimulus helped jump-start economic growth, real GDP growth rose to 7.2 percent in 2012 before slowing to 3.5 percent in 2013 and 3.4 percent in 2014. The trend was similar for growth in per capita GDP, which averaged 4.3 percent for 2010–13. Growth in this period was characterized by shrinking construction output and positive growth in services and agriculture. Growth in services was driven mainly by trade, information and communication, finance, insurance, food service, and recreation. Agricultural growth touched 8 percent in 2013, and although adverse weather hit the sector hard that year growth was stimulated by state assistance programs, such as free seed and subsidies to farmers for fuel and fertilizer (ADB 2014). The small contribution from industrial growth (excluding construction) was due to mining, quarrying, and the production of pharmaceuticals, food, and beverages. The government’s new export-oriented industrial development strategy is expected to push up the contribution of industry to GDP. In 2014 growth was dragged down by construction, mining, and energy, but it was also affected by the economic contraction in Russia, lower remittances and foreign direct investment (FDI) inflows, and reduced export earnings. These factors are expected to keep growth low in the next few years as well. Figure 1. Real GDP Growth Rates, 2000–15, Percent 13.2 14.0 13.9 13.2 13.7 9.6 10.5 5.9 6.9 7.2 4.7 3.5 3.4 3.0 2.2 2.7 0.8 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 -14.1 Source: World Bank, World Development Indicators (data accessed on May 25th 2015). Note: Growth rates are based on year-to-year changes in real GDP (at constant market prices). 1.3. In the last 10 years, the structure of GDP shifted away from industry toward services (Figure 2). In Soviet days, Armenia’s economy was highly industrialized, and industry has been 5 dominant in recent years as well. In 2000 the share of agricultural value-added in GDP was 26 percent and by 2005 it was down to 20 percent; in 2013 agricultural value-added was 22 percent. Meanwhile, the share of industry (mining, manufacturing, construction, electricity, water, and gas) grew substantially over the decade, rising from 39 percent in 2000 to 45 percent in 2006. Since the crisis, the share of industry has declined and the share of services in GDP has gone up notably. In 2013, services accounted for 47 percent of GDP, industry for 31 percent, and agriculture for 22 percent. These structural shifts are visible in the changing sources of growth; depending on how employment patterns respond to these shifts, they may have implications for poverty reduction and shared prosperity. Figure 2. Sectoral Share of GDP, 2000-13, Percent 50 45 40 35 30 25 20 15 10 5 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Source: World Bank, World Development Indicators (data accessed on May 25 th 2015). 1.4. Personal remittances as a share of GDP have risen since the crisis (Figure 3). Measured as a share of GDP, such flows to Armenia are among the highest in the world. In 2013, remittances were 21 percent of GDP, up from a low of 16 percent just before the crisis hit the economy. Lagged Russian GDP has the largest impact on remittances flows to Armenia because so many of the remitters work in Russia (IMF 2012). Remittances are important for the Armenian economy because they drive private consumption and thus raise aggregate demand. Some researchers argue that large remittances inflows can also adversely affect the labor market, on both the demand side, by encouraging growth of non-tradable rather than export-oriented sectors, and the supply side, by discouraging work. According to the Russian Central Bank (RCB), in the last quarter of 2014 remittances to Armenia (in U.S. dollars) fell by 31 percent because of Russia’s economic downturn and the depreciation of the ruble against the U.S. dollar (World Bank 2015). 1.5. The resumption of growth after the deep recession in 2009 is credited to fiscal stimulus; as the economy recovered, fiscal policy then switched from stimulus to consolidation while protecting and extending pro-poor spending. Armenia increased social spending in response to the crisis; in particular pensions and transfers for targeted social assistance (the Family Benefit Program, FBP) went up. Pensions consist of contributory old-age pensions (linked to years of service and the largest category in terms of number of pensioners), pensions for disability, and privileged and survivor pensions (Table 1). Between 2008 and 2013 the basic monthly pension doubled in nominal terms from AMD 6,800 to AMD 13,000—an increase in pension income that has important implications for poverty reduction and shared prosperity. When Armenia’s spending and tax policies were analyzed using 2011 data (applying the 6 Commitment to Equity methodology of Lustig et al. (2013); see World Bank (2014a), it was found that Armenia’s fiscal activities led to significant redistribution as measured by a reduction in the Gini coefficient. However, the effect is mostly driven by contributory pensions, which are not targeted but have the largest budget of all social expenditures. The analysis also calculated the extent to which fiscal spending and revenues reduced poverty. The main finding is that even though transfers are reasonably well-targeted, taxes, especially indirect taxes, fall on poorer households and offset the poverty-reducing effects of public spending. In fact in Armenia, fiscal activities cause a significant amount of downward as well as upward movement among the poor or near-poor—much more so than in Latin American countries where similar analyses have been completed. Figure 3. Remittance Flows as Share of GDP, 2000-13, Percent 1.6. 25 21 20 19 19 18 18 18 18 16 17 15 12 10 5 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Source: World Bank, World Development Indicators (data accessed on May 25 th 2015). Table 1: Pensions in Armenia, 2008–13 2008 2009 2010 2011 2012 2013 Total number of pensioners 469,747 467,555 465,084 454,488 452,505 453,917 Average monthly pension, AMD 22,556 26,056 28,647 28,701 31,248 30,962 Basic pension, AMD 6,800 8,000 10,500 1,0500 13,000 13,000 Value of one year of service, AMD 450 CPI, average yearly 109.0 103.4 108.2 107.7 102.6 105.8 Source: Armenia Pension Study (Anusic 2015). 1.7. This Poverty Assessment examines the recent trends in poverty reduction and shared prosperity and complements the analysis of poverty and social indicators conducted by the National Statistical Service of the Republic of Armenia (NSSRA) each year. The 2014 Social Snapshot and Poverty in Armenia (NSSRA 2014), an annual assessment published by the NSSRA, analyzes trends in consumption poverty and other social and economic indicators with 2008 as the reference year. The SSPA report analyzes in detail consumption poverty indicators and their trends and the profile of poverty in 7 terms of vulnerable groups, such as rural households and children. The 2014 SSPA also reported how Armenia performed on a pilot multi-dimensional poverty index developed and first analyzed by the World Bank (2014b). Using household data from the Integrated Living Conditions Survey (ILCS), this report seeks to add to the analysis in the SSPA and extend the analysis in the FY14 Poverty Assessment report by reviewing the nature of post-crisis poverty reduction and exploring reasons why growth has not been as pro-poor as in the past (sections 2 and 3). The analysis identifies three factors that shaped the pattern of poverty reduction and the pro-poor nature of growth in 2010–13: the structure of the domestic labor market, remittances from migrant workers abroad, and social transfers, especially pension income. The report concludes with discussion of what the findings imply for future poverty reduction (section 4). 2. Recent Developments Growth in Household Consumption 2.1. The resumption of aggregate growth after the crisis translated into growth of household consumption spending, with consumption growing faster for the better-off than the less well-off, unlike pre-crisis consumption growth (Figure 4A and Figure 4B). Between 2004 and 2007, when growth was averaging 13 percent, poverty was halved, from 55 percent to 26 percent, average consumption grew at 6.9 percent, and the entire distribution experiencing a growth in consumption (Figure 4A). The bottom 20 experienced slightly lower consumption growth rates at about 5.5 percent per year but in general growth seems to have been pro-poor. Between 2010 and 2013, the growth of consumption spending averaged 4.07 percent, not far off average per capita GDP growth of about 4.3 percent. However, the incidence of growth was uneven across the welfare distribution (Figure 4B). Between 2010 and 2013 although everyone enjoyed some growth in consumption, but for the better-off it was above the average. Figure 4. Growth of Consumption Expenditures A. Pre-Crisis: 2004-07 B. Post-crisis: 2010-13 9 9 8 8 7 7 6 6 5 5 4 4 3 3 2 2 1 1 0 20 40 60 80 100 0 20 40 60 80 100 Percentiles Percentiles Median spline Growth rate in mean Median spline Growth rate in mean Growth rate at median Growth rate at median Source: ILCS national data. Note: All calculations are based on the national consumption aggregate (2009 methodology). 2.2. The World Bank has recently adopted an indicator to directly capture the distribution of growth; the shared prosperity indicator is defined as the growth in average consumption of the bottom 40 percent of the distribution. Since Armenia’s poverty rate is about 30 percent, the bottom 40 mostly consists of the poor. Figure 5 shows Armenia’s performance on the shared prosperity indicator, which is calculated using consumption spending harmonized for international comparisons. The patterns 8 align with the growth incidence curve in Figure 4. Between 2007 and 2013, average consumption grew for all Armenians by 1.7 percent a year, but by just 1.14 percent for the bottom 40 percent. Before the crisis, the bottom 40 percent had enjoyed annual growth in mean consumption of 4.29 percent, above the national average of 3.50 percent. The 2009 contraction of the economy hit the bottom 40 percent hard: their consumption was reduced by 7.46 percent a year—a contraction worse than that of the economy as a whole. Between 2010 and 2013, mean consumption of the bottom 40 percent grew at 2.11 percent, slightly below the 3 percent growth enjoyed by the population as a whole, and in 2012 and 2013 growth in mean consumption of the bottom 40 was essentially zero, even as average growth in consumption was 2.15 percent. It appears from the shared prosperity indicator that in the post-crisis recovery, the least well- off Armenians are not benefiting as much as others. Figure 5. Trend in Shared Prosperity: Annual Growth in Average per Capita Consumption Spending, Percent 5 3.50 4.29 3.00 1.70 2.11 2.15 1.14 -0.34 0 2007 - 2013 2007 - 2009 2009 - 2010 2010 - 2013 2012 - 2013 -5 -5.48 -7.46 -10 all bottom 40 Source: ECATSD (Europe and Central Asia Teas for Statistical Development) calculations using ECAPOV (World Bank database of harmonized consumption data, accessed on May 25th 2015). Note: Consumption aggregate drawn from ILCS data and harmonized for international comparison. Welfare aggregate includes food and nonfood consumption, durables and health expenditure harmonized. For more information on the micro data, see http://ecadataportal/. Figure 6. Shared Prosperity, Europe and Central Asia, circa 2008–13, Percent 10% Bottom 40% Total Population 8% 6% 4% 2% 0% -2% -4% Source: ECATSD (Europe and Central Asia Teas for Statistical Development) calculations using ECAPOV (World Bank 9 database of harmonized consumption data, accessed on May 25th 2015). Note: The calculations use the growth rate of consumption (income) for the bottom 40 of the welfare distribution and for the total population, 2005 purchasing power parity. Welfare aggregate includes food and nonfood consumption, durables and health expenditure harmonized. For more information on the micro data, see http://ecadataportal/. 2.3. The comparison with countries in Europe and Central Asia (Figure 6) puts Armenia’s shared prosperity performance in perspective: households in Armenia had almost the lowest consumption growth for the bottom 40. Figure 6 compares Armenia’s performance on shared prosperity between 2008 and 2013 with that of select countries in Europe and Central Asia. Even though the time period differs slightly across countries, the numbers suggest that in Armenia, unlike other countries in the region, neither the bottom 40 nor the total distribution saw any major increases in their consumption. Inequality 2.4. Since the crisis inequality has worsened. Unequal distribution of consumption in a population can deter the transmission of growth to poverty reduction. The Gini coefficient, the most commonly reported measure of inequality, ranges from 0 (complete equality) to 100 (complete inequality), although for consumption spending it is typically in the range of 30 to 50. In Armenia, the Gini coefficient for consumption spending went up from 24.2 in 2008 to 26.9 in 2012 and 27 in 2013, indicating that inequality had widened since the crisis, which was consistent with the growth pattern. Despite this increase in inequality, the Gini coefficient for consumption is lower than was typical internationally.1 Income tends to be more unequally distributed than consumption and this is the case in Armenia as well: in 2013 the Gini for income was 37.2, up from 33.9 in 2008. Poverty Reduction in International Perspective 2.5. Although Armenia’s performance on poverty reduction is noteworthy, other countries in the region have done even better (Figure 7). To facilitate cross-country comparison of poverty, the World Bank computes harmonized consumption aggregates using the latest available household survey data.2 These measures are based on a methodology that differs from the national poverty calculation methodology. Poverty is calculated by comparing these harmonized aggregates against regional poverty lines of US$1.25, US$2.50, and US$5 per person per day (adjusted using 2005 purchasing power parity). Comparisons over time show that poverty in Armenia fell steeply between 2001 and 2008; poverty at US$2.50 per person per day fell from 67 percent in 2001 to 26 percent in 2008, though it increased after the crisis. Cross-country comparisons show that in 2013, 30 percent of Armenia’s population lived on less than US$ 2.50 per person per day, a poverty rate lower than in Georgia and Kyrgyz Republic but higher than in Kosovo and Moldova. 1 One reason for the low estimated Gini for consumption spending may be that better off-households chose not to participate in the survey. As part of the fiscal incidence analysis undertaken for World Bank’s 2014 Armenia Public Expenditure Review (2014a), an exercise was undertaken to simulate the Gini (and poverty rates) if “selection -out� by richer households were taken into account. The simulation found that the Gini measure increases significantly when top incomes are included. 2 The differences between the national and ECA-harmonized poverty rates lie in differences in the consumption aggregates used and application of adult equivalence in the national measure compared to per capita in the ECA- harmonized measure. 10 Figure 7. Poverty in Europe and Central Asia 50 42.5 41.4 40 30.2 30 27.4 20 10 6.0 6.7 4.5 4.0 0 Georgia Kyrgyz Armenia Kosovo Moldova Albania Turkey (2012) Romania (2012) Republic (2013) (2011) (2011) (2012) (2012) (2012) Source: ECATSD (Europe and Central Asia Teas for Statistical Development) calculations using ECAPOV (World Bank database of harmonized consumption data, accessed on May 25th 2015). Note: Poverty is measured at the international poverty line of US$ 2.50 per person per day, 2005 purchasing power parity. Welfare aggregate includes food and nonfood consumption, durables and health expenditure harmonized. For more information on the micro data, see http://ecadataportal/. 2.6. Armenia’s comparative performance in reducing poverty has deteriorated. In a cumulative distribution of poverty changes for different countries, poverty reduction at US$2.50 per person per day in Armenia was in the past the best in the world: its performance in reducing poverty was better than roughly 98 percent of all poverty changes observed. Recently, its performance in reducing poverty has fallen to the 73rd percentile internationally. Although pre-crisis Armenia managed to reduce poverty significantly, from 2009 on the share of households living on less than US$2.50 was relatively constant (Figure 8). Between 2001 and 2008, the poverty headcount measured at the regional poverty line of US$2.50 dropped from 67.1 percent in 2001 to 26.0 percent in 2008; since 2009, poverty in Armenia has remained above 30 percent. Poverty projections for 2015 and 2016 suggest that due to sluggish economic recovery and increasing uncertainty, poverty is likely to stay close to 30 percent. Figure 8. International Poverty Trend through 2016, USD PPP 2.5 per day, Percent 80 70 60 50 40 30 20 10 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Source: ECATSD (Europe and Central Asia Teas for Statistical Development) calculations using ECAPOV (World Bank database of harmonized consumption data, accessed on May 25th 2015). Note: Poverty projections for 2014–16 build on the assumption that the elasticity between GDP growth and poverty reduction was 0.24 percent for the $2.50 day line. Welfare aggregate includes food and nonfood consumption, durables and health expenditure harmonized. For more information on the micro data, see http://ecadataportal/. 11 Poverty Reduction from the National Perspective 2.7. As growth resumed after the 2009 recession, the share of those living below the national poverty line declined but poverty is still far from the low pre-crisis rates (Figure 9). Between 2004 and 2008, Armenia had halved poverty, but after the crisis, in 2013, 32 percent of the population lived below the national poverty line, defined in terms of consumption expenditure of AMD 39,193 per adult equivalent per month. In 2010 the incidence of poverty peaked at 35.8 percent. As the economy recovered after 2009, poverty at first increased in both level and depth and has declined only moderately in recent years. In 2010, the poverty gap –measuring how far poor people are from the poverty line - was also high at 8.1 percent of the poverty line. The squared poverty gap, which measures the severity of poverty among the poor, declined from a high of 2.5 percent in 2010 to 1.7 percent in 2013. Extreme poverty, those living below the food poverty line of AMD 22,993 per adult equivalent per month, was 2.7 percent in 2013, down from a high of 3.7 percent in 2011. Though the level, depth, and severity of poverty have declined in 2010, it is worrying that from 2012 to 2013 there were almost no gains in poverty reduction or its depth and severity. Figure 9. Poverty in Armenia, 2004–13, Percent 60 53.5 50 40.1 40 35.8 35.0 34.1 32.4 32.0 30.2 30 26.4 27.6 20 10 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Poverty Poverty gap Squared poverty gap Source: ILCS national data. Note: All calculations are based on the upper national poverty line (2009 methodology). 2.8. As the pace of poverty reduction has slowed post-crisis, analysis of economic mobility shows some people moving into and out of poverty between 2010 and 2013 (Table 2). Using synthetic panel methodology (Dang and Lanjouw 2013) to explore movements in and out of poverty over the period, it appears that about 41.5 percent of households changed poverty status between 2010 and 2013: 16.5 percent were non-poor in 2010 but poor in 2013 and 25 percent of the poor in 2010 moved out of poverty by 2013. About 10 percent of the population remained persistently poor over the four-year period. A pilot multidimensional poverty index developed for Armenia (and analyzed in World Bank 2014b and NSSRA 2014) captures the complexity of poverty and also sheds light on subgroups in the population that are more likely to experience persistent or chronic poverty (Box 1). 12 Table 2. Poverty Dynamics in Armenia, 2010 and 2013 Poor in 2010 Non-poor in 2010 Poor in 2013 9.6 16.5 Non-poor in 2013 25.0 48.9 Source: ILCS data 2010 and 2013. Note. Only households whose heads were aged 25–55 in 2010. Box 1. Multidimensional Poverty in Armenia The analysis of multidimensional poverty complements analysis of consumption poverty and Figure B1: Share of population that is multidimensionally offers insights into the complexity, depth and a and consumption poor, 2013 persistence of poverty. The pilot multidimensional poverty index (MPI) analyzed 59.12 and reported in World Bank (2014b) and NSSRA (2014) measures deprivations in terms of education, health, employment, housing, and extreme poverty. The MPI is a composite index that shows the number of people who are suffering deprivations in more than one-third of the indicators. A comparison of poverty using the MPI and consumption poverty (see Battiston, Diego et 24.37 al. (2013)) can be informative in understanding 8.89 groups that are more likely to face persistent and chronic poverty. Figure B1 demonstrates that a 7.62 notable proportion of the population in Armenia is CONS non poor both multi-dimensionally poor and living below the national poverty line. In 2013 this group CONS poor comprised 7.62 percent of the population. A better MPI poor understanding of the circumstances and MPI non poor persistence of poverty can enable policy makers to target interventions to the deprivations these households experience. A second group of households is multi-dimensionally poor but lives above the upper national poverty line (8.89 percent in 2013). The deprivations make them more vulnerable to falling into poverty if their household experiences a shock. A third group of households does not experience multiple deprivations but still lives below the upper national poverty line (24.37 percent). Unlike the first and second groups, these households have adequate education, health, and housing, but often work in low-productivity jobs. a Further details, see NSSRA (2013), pp. 59 and 60. The UNECE working paper 15/2015 summarizes methodology and preliminary findings from a pilot multidimensional poverty index for Armenia (Martirosova, et al. 2015) 2.9. Urban areas outside the capital city and four of the regions have poverty rates that exceed the national average (Figures 10 and 11). Economic activity tends to be concentrated in the capital, Yerevan, which in 2010 accounted for more than half of national GDP and about one-third of the population (Eiweida and Kostanyan, 2014). Urban areas outside Yerevan experience higher poverty rates than rural areas or Yerevan (Figure 10). After the recession in 2009, poverty rose in all three areas, but rose the most in percentage point terms in cities outside Yerevan; these cities also experienced the largest decline in poverty between 2010 and 2013. In 2013, 39 percent of residents in other urban areas were poor, compared to 31.7 percent of rural residents and 25.5 percent of those living in Yerevan. Across 13 economic regions (marzes) poverty declined post-crisis everywhere except Tavush marz (Figure 11). In 2013, regional poverty rates ranged from 21.0 percent in Vayots Dzor (with 1.2 percent of the poor and 1.9 percent of population) to 45.9 percent in Shirak (with 12.1 percent of poor and 8.4 percent of the population). In Sjunik marz, which is the center of mining, which employs a large share of the residents, the poverty rate in 2013 was among the lowest in the country at 25 percent. Four regions, Lori, Kotayk, Gegharkunik, and Shirak, have nearly 46 percent of the poor, though they account for only 36 percent of the total population. The dominant economic activity (as measured by sectoral share in regional output) differs across these four regions: agriculture is dominant in Gegharkunik and Shirak and industry (manufacturing and mining) in Lori and Kotayk. These regional differences in poverty rates likely reflect a range of factors, such as types of economic activity, connectivity and availability of basic infrastructure, and topography. However, working at the marz level it is difficult to identify underlying patterns at play and sub-marz variations would be important to analyze. The poverty map currently being prepared will facilitate a better understanding on these regional patterns of poverty. Figure 10. Poverty Rates in Yerevan, other Urban Areas, and Rural Areas, 2009–13, Percent 50 45 45.4 43.6 41.5 40.3 40 39.3 35 34.9 36.0 34.5 32.1 31.7 30 26.7 27.1 27.5 25 25.6 25.5 20 15 2009 2010 2011 2012 2013 Yerevan other urban areas rural areas Source: ILCS national data. Note: All calculations are based on the upper national poverty line (2009 methodology). 2.10. To better study the spatial correlates of poverty, and to inform geographic targeting of development projects to communities within Armenia’s 11 economic regions, the World Bank and the NSSRA are drafting a new poverty map. Several government initiatives currently use a vulnerability index calculated at the lowest administrative level (communities) to target interventions (for example, micro investments financed by the Armenia Social Investment Fund). Poverty mapping can also be used for geographic targeting because it helps in estimating consumption poverty at administrative units below the marz level. Because the ILCS sampling design does not allow for accurately measuring consumption poverty at the community level, poverty mapping combines information from households in the ILCS survey and from the population census to estimate poverty at the sub-marz level. The last poverty map for Armenia was created in 2007 using household survey data from 2004 and the population census of 2001 (World Bank 2007). The map now being prepared uses the results of the 2011, 2012, and 2013 ILCSs and will link to the 2011 census. 14 Figure 11. Regional Patterns of Poverty across Armenia, 2010 and 2013, Percent Shirak 45.9 Kotayk 42.5 Lori 38.6 Gegharkunik 35.8 Ararat 32.4 Armavir 31.3 Tavush 27.7 Yerevan 25.5 Sjunik 25.2 2013 2010 Aragatsotn 22.7 Vayots Dzor 21.0 37.1 15 20 25 30 35 40 45 50 Source: ILCS national data for 2010 and 2013. Note: The figure reports point estimates of poverty for each region. Because of the ILCS sample size and design, regional poverty rates cannot be measured precisely. Table 3.4 in the SSPA reports the 95 percent confidence intervals for regional poverty rates (NSSRA 2014). Table 3. Armenia: Poverty Profile, 2010 and 2013 Probability of Being Probability of Being Poor, 2010 Poor, 2013 Female (base: male) 0.060 0.083 Age of HH head: base: age 16 – 39 (0.019)*** (0.019)*** 40 to 59 0.076 0.054 50 to 64 0.046 0.027 (0.029)*** (0.033) 65 to 74 0.063 0.017 (0.027) (0.030) 75 and over 0.006 0.057 (0.032)** (0.035) (0.035) (0.037) Education of HH head:- base: primary education General secondary -0.037 0.004 (0.037) (0.046) Special secondary -0.122 -0.049 (0.040)*** (0.048) Tertiary -0.225 -0.136 Household size 0.047 0.053 (0.041)*** (0.049)*** Young -age dependency rate 0.070 0.056 (0.005)*** (0.006)*** (0.049) (0.054) Employment rate -0.199 -0.197 (0.028)*** (0.027)*** Location of residence: base: Yerevan, the capital Other urban areas outside Yerevan 0.156 0.108 (0.019)*** (0.021)*** Rural areas -0.020 -0.052 (0.022) (0.023)** Constant 0.157 0.088 (0.049)*** (0.058) Pseudo R2 0.12 0.11 N 7,852 5,165 ** p<0.05; *** p<0.01 Note: Marginal effects from a Probit model estimation using ILCS national data for 2010 and 2013. Employment rate defined as the share of workers among household members older than 15. Young-age dependency rate is defined as the share of children younger than 15 years in household size. Robust standard errors are in parentheses. 15 2.11. The pace of poverty reduction has slowed, and the profile of poverty has not changed much. As in past years, in 2013the poor were more likely to (1) live in urban areas outside the capital; (2) have larger households with more children; (3) have less education; and (4) live in households headed by women; and (5) they are less likely to be employed and more likely to be out of the labor force or unemployed. For example, in 2013 29 percent of the poor lived in households headed by women, compared to only 24 percent of the non-poor. Average household size (number of members living in the household) among the poor is 5.21, compared to 4.23 for the non-poor. Educational attainment among those 15 and older is generally high, with most Armenians having secondary (basic or special) education, but the share with tertiary education varies by poverty status: 14 percent of the poor compared to 26 percent of the non-poor. The average number of children is also higher in poor households. Regression results (Table 3) confirm that even after controlling for the socioeconomic characteristics of households, those headed by women are significantly more likely to be poor; and the higher the number of employed family members, the lower the probability that the household falls below the poverty line. As expected, households headed by an individual with tertiary education are significantly less likely to be poor. The poverty profile changed in some respects between 2010 and 2013. One difference is that households with heads that have had special secondary education (technical and vocational education and training) were less likely to be poor in 2010 but not in 2013. 2.12. The structure of household income differs in poor and non-poor households (Figure 12). Between 2010 and 2013, the sources of income did not change for either type. Large aggregate remittance inflows are visible in the 10 percent share of remittances in the average income of the non-poor (Figure 12). For the poor, the share of remittances in income is a much lower 5 percent. This poor –non-poor pattern reflects both the poverty-reducing impact of remittances and the skill level of emigrants from poor and non-poor households. Pensions, which went up starting in 2009, account for a large share of income for both poor and non-poor households. Labor earnings (wage or self-employment) are the largest source of income for both the poor and the non-poor. The fact that for the poor labor earnings still account for nearly half of all income suggests that they are mainly employed in sectors (such as agriculture and certain services) that do not pay enough to help people escape poverty. The report discusses next how these macro-micro patterns likely came together to reduce poverty in a less pro-poor growth environment. Figure 12. Income Sources for Poor and Non-poor, 2010 and 2013, Percent 100% 6% 6% 8% 8% 90% 6% 5% 9% 10% 80% 4% 10% 3% 9% 70% 22% 20% 60% 26% 27% 50% 40% 30% 56% 57% 49% 49% 20% 10% 0% non poor poor non poor poor 2010 2010 2013 2013 wage income pensions SA family benefit unemployment benefit remittances agriculture else Source: ILCS national data. Note: All calculations are based on the upper national poverty line (2009 methodology). 16 3. Understanding the Performance on Poverty Reduction 3.1. Decomposing the 3.8 percentage point reduction in national poverty into growth and distributional components confirms that although consumption growth contributed substantially to poverty reduction, it was counterbalanced by a change in the distribution of consumption spending (Figure 13). Growth favored the better-off. Although between 2010 and 2013, consumption growth by itself would have brought poverty down considerably (a simulated 20 percentage point reduction), a change in consumption distribution favoring the better-off counterbalanced the growth effect, dampening the net reduction in poverty to only 3.8 percentage points. Figure 13. Effects of Consumption Growth and Distributional Changes on Poverty 20 16.51 positive values indicate gain in poverty) Percentage point change in poverty rate (negative values indicate reduction and 15 10 5 0 Total change in poverty rate Change in poverty due to Change in poverty rate due to -5 consumption growth change in distribution of -3.79 consumption -10 -15 -20 -20.3 -25 Source: ILC national data 2010 and 2013. Note: Using methodology from Datt and Ravallion 1991. Total change in poverty rate (head count rate, FGT0) is broken down into (a) the change in poverty due to an increase in mean consumption expenditure for a given distribution; b) the change in poverty due to a change in the distribution of consumption expenditure for a given level of mean expenditure. 3.2. What was driving poverty reduction between 2010 and 2013? A decomposition analysis that links household consumption expenditures to their sources shows that labor income (employment and earnings), pensions, agricultural incomes, and remittances were the main drivers of poverty reduction between 2010 and 2013 (Figure 14).3 Post-crisis, domestic labor market conditions helped bring poverty down. Growth in the share of adult household members who were employed and in their earnings (including self-employment income) was the leading source of poverty reduction. Growth in remittances from Armenians who emigrated, most to Russia, also helped to reduce poverty. Among non-earned sources of income, higher pensions were also important. Changes in other sources of non-earned income, such as unemployment benefits and social assistance transfers (targeted and non-targeted), did not do much to reduce poverty reduction in this period. Demographic change between 2010 and 2013 led to a small increase in the share of household members of non-working ages, which did not contribute to poverty reduction. 3 Based on micro decomposition techniques. See, for example, Inchauste et al. 2014. 17 Figure 14. Consumption Poverty Reduction between 2010 and 2013 Decomposed dependency rate 0.4 -2.5 employment rate -1.8 wage income -1.8 pension income unemployment benefits 0.5 income from family and SA 0.9 -1.3 remittances -0.9 income from agriculture other income 0.2 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 contribution to total poverty reduction (2010 to 2013: -3.8 percentage points) Source: ILCS national data 2010 and 2013. Note: Based on the micro decomposition of poverty change, negative numbers indicate that the factor helped to reduce poverty, positive numbers that the factor had the opposite effect. The effect of the consumption-to-income ratio is not shown; this ratio can be considered to represent household savings as well as measurement error; in the micro decomposition for the figure, when all other sources of income were kept constant, the ratio had the effect of raising poverty. 3.3. To understand why growth post-recovery favored the better-off, drivers of growth in consumption spending across the distribution were compared; pension income and remittances stand out for their effect on consumption by the better-off (Figure 15). Figure 4.B shows that the better-off (top 60 percent of the distribution) experienced higher growth in consumption than did the less- well-off (bottom 40). Figure 15 shows that the drivers of consumption growth differ across the distribution. For consumption growth, growth in the share of household members employed mattered more for the bottom 40 than the top 60. Labor earnings and agricultural income had a similar effect. Growth in remittances (mainly from Russia) and growth in pension income (driven by the government’s decision to raise basic pensions) mattered more for consumption growth for the top 60 percent. For the poorest 40 percent, consumption grew because both earnings and the share of household members employed grew. For this less-well-off group, interestingly, income from the FBP, which is poverty- targeted social assistance, declined post-crisis; this likely reflects a reduction in the scope of the program between 2010 and 2013. Thus, differential consumption growth and erosion of pro-poor growth (in a relative sense) seem to be rooted in growth in pension income and remittances and a reduction in poverty- targeted social assistance. 18 Figure 15: Contributors to Consumption Growth, Bottom 40 and Top 60, between 2010 and 2013 dependency rate employment rate wage income pension income unemployment benefits bottom 40 income from family and SA top 60 remittances income from agriculture other income -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3 3.5 contributors to consumption growth acros the distribution (2010 to 2013) Source: ILCS national data 2010 and 2013. Note: Based on micro decomposition of income growth, positive numbers indicate that the factor contributed to income growth, negative numbers that it had the opposite effect. Effect of the consumption-to-income ratio is not shown. 3.4. Bringing the macro context and the micro analysis together is the next step in understanding poverty reduction in Armenia. Outcomes in the domestic labor market, remittances from workers abroad, and social transfers (mainly pensions) have been identified as important factors underpinning poverty reduction performance after the 2009 crisis and the extent to which growth was or was not pro-poor or not. The Domestic Labor Market 3.5. The labor market showed some improvements between 2010 and 2013 (Table 4). The labor force participation rate went up from 61.0 to 63.4 percent, mainly because more women (up from 52 to 56 percent) and urban residents (56 to 58 percent) were participating. Employment rates also rose for all workers (from 49.6 percent to 53.2 percent), and the unemployment rate fell from 19 to 16.2 percent. In 2013 urban unemployment was still high at 23.4 percent, but the female unemployment rate at 18 percent was higher than the men’s 14.4 percent. It is not clear what drove the increase in women’s labor force participation and employment. It may be that growth in the services sector is pulling women into the labor market. 3.6. Despite overall low employment and high unemployment, wages have gone up in Armenia. The employment rate suggests that the economy underutilizes its labor resources, yet despite the slack in the labor market, in some sectors real wages have been growing (see below), underpinned by an increase in the minimum wage. Rising wages are reflected in the role of earnings in reducing poverty (Figure 14). The minimum wage increased from AMD 30,000 per month (US$62; AMD 22,912 in real 2005 AMD) to AMD 45,000 per month (US$93; AMD 29,424 in real 2005 AMD). Correspondingly, the average nominal gross wage rose from AMD 102,652 (US$212; AMD 78,401 in real 2005 AMD) to AMD 146,524 (US$303.78; AMD 95,809 in 2005 AMD). These work out to an annual increase in real terms between 2010 and 2013 of 8 percent for the minimum wage and 6 percent for the average gross wage. 19 Table 4. Labor Market Indicators, 2010 and 2013, Percent Labor Force Participation Employment Rate Unemployment Rate 2010 2013 Rate 2010 2013 2010 2013 Total 61.2 63.4 49.6 53.2 19.0 16.2 Male 72.3 72.8 59.9 62.3 17.0 14.4 Female 52.2 55.9 41.1 45.8 21.2 18.1 Urban 55.9 58.3 40.4 44.6 27.8 23.4 Rural 71.1 72.5 66.8 68.1 6.1 6.0 Source: SSPA 2014 for the years 2010 and 2013. Note: Population aged 15-75 years. 3.7. The sectoral composition of employment is well aligned with the rise of the services sector as the main contributor to GDP, and the non-poor have a higher presence in services, which enables them to benefit from the new sectoral growth structure (Figure 16). Among the employed in 2013, services (46.8 percent) accounted for the largest share, followed by agriculture (36.1 percent), and then industry (17 percent). Services also account for the largest share of employment for the non-poor, so that they benefit from this dominant source of GDP growth. The share of the poor in agriculture employment has declined as jobs in services have grown. The poor have a higher share of industry employment than do the non-poor. The recent adoption of an export-oriented government development strategy for industry could affect its share of employment. Agriculture, which is benefitting from public investment but contributes only 20 percent to GDP, is no longer the largest employer of the poor but still employs 40 percent of the poor and 36 percent of the non-poor. Figure 16. Sectoral Composition of Employment, Percent 46.8 47.6 48.9 44.2 44.4 38.3 39.9 41.0 36.1 36.1 35.8 34.8 17.5 17.1 19.5 19.2 16.7 16.4 2010 2013 2010 2013 2010 2013 All Poor Nonpoor Agriculture Industry Services Source: ILCS and SSPA data for the years 2010 and 2013. Note: Working-age population (aged 15–75). 3.8. The composition of non-agricultural employment differs between the poor and the non-poor (Figure 17A and 17B). In 2013 manufacturing, wholesale and retail trade, and education/health/social services each accounted for 10 percent of total employment. As expected, the poor have little or no presence in sectors associated with high skills, such as financial intermediation and public administration (6 percent in 2013). Construction continues to employ a notable share of the poor (7 percent) despite the 20 reduction in its output in recent years. For the non-poor, public administration, education/health/social services, manufacturing, and wholesale and retail trade each account for about 10 percent of employment share. There were also interesting changes between 2010 and 2013, when among the poor the share of employment in manufacturing, wholesale and retail trade, and education/health services increased. Among the non-poor the share of public administration and manufacturing increased over this period. Consistent with the steep decline in construction output, the share of construction sector employment declined for both poor and non-poor. Figure 17. Detailed Sectoral Composition of Employment, Percent A. Poor workers B. Non-poor workers Agriculture, hunting, forestry and 44.4 Agriculture, hunting, forestry and 35.8 fishing 39.9 fishing 34.8 Mining and quarrying 0.9 Mining and quarrying 1.0 0.5 0.9 Manufacturing 6.7 Manufacturing 5.4 9.3 8.0 Electricity, gas and water supply 3.1 Electricity, gas and water supply 3.7 2.1 2.3 Construction 8.8 Construction 6.6 7.2 5.2 Wholesale and retail trade; repair of 7.9 Wholesale and retail trade; repair of 9.9 motor vehicles, motorcycles 10.7 motor vehicles, motorcycles 9.7 Тransportations and warehouse 6.1 Тransportations and warehouse 5.9 economy 5.7 economy 6.5 Accommodation and food service 1.8 Accommodation and food service 1.9 activities, Private HHs activities 2.4 activities, Private HHs activities 1.9 Financial services, insurance 0.6 Financial services, insurance 1.6 activities and information 0.8 activities and information 1.9 Public administration, support 7.4 Public administration, support 9.4 services and technical activities 6.2 services and technical activities 10.8 Education, human health and social 9.6 Education, human health and social 15.2 work activities 11.3 work activities 14.0 Arts, entertainment and other 2.7 Arts, entertainment and other service 3.7 service activities 4.0 activities 4.1 poor 2010 poor 2013 non-poor 2010 non-poor 2013 Source: ILCS and SSPA data for the years 2010 and 2013. Note: Working-age population (aged 15–75). 3.9. Between 2010 and 2013, real hourly wages grew in most sectors (Figure 18).4 The number of workers employed in agriculture declined by 2.7 percent a year between 2010 and 2013, and the shedding of workers appears to be pushing up hourly wages (2013 was also a good year of agriculture output). Meanwhile, employment in construction also plunged, by 8 percent a year, while wages grew by 1 percent a year. In 2007, at the peak of construction- driven growth, the sector employed 91,000 workers; by 2013 the number was down to 66,000. For industry (mainly manufacturing) and trade and tourism, real wages tracked growth in employment. For manufacturing, this pattern could reflect the 2012 government decision to emphasize certain export- oriented industries. The transport and communication and financial services sectors experienced some growth in employment and real wages. For this analysis public administration and education/ health/social services are bundled together; this is the only group that experienced a reduction in the growth of real wages. The residual sector consisting of arts and entertainment, private households, and a few other sectors saw dramatic growth in employment but a reduction in real wages. 4 This analysis uses the slightly different categorization of sectors (merging some subsectors) than the NSSRA uses. 21 22 Figure 18. Growth in Real Wages and Employment by Sector, 2010 to 2013, Percent 15% 12.7% employment growth real wage growth (median) 10% 7.2% 5% 3.0% 2.6% 1.8% 2.4% 1.0% 0.9% 1.0% 0.3% 0.6% 0.4% 0% Other Construction Administration and Finance and Real Agriiculture Industry Transportation and Trade and Tourism Communication -2.7% -2.5% -5% Services Estate Public -5.0% -10% -8.3% Source: ILCS data for 2010 and 2013. NSS sectoral classifications, wages deflated by the consumer price index used by ECA TSD (Europe and Central Asia Team for Statistical Development). Note: Median hourly wages computed using data on earnings and hours worked. Information on real wage growth in the agricultural sector may be biased due to lack of information on hours worked. Industry excludes construction. 3.10. Despite the positive trends in the domestic labor market, some concerns are emerging. The share in total paid employment in non-agriculture sectors has remained unchanged; almost all the rise in the share of non-agricultural employment comes from the increase in those who are self-employed from 5.5 to 7 percent of all employment (65,000 in 2010; 77,900 in 2013). For urban job seekers who face an unemployment rate of 23 percent (almost 1 in 4), in 2013 the average job search had lasted 21 months (NSSRA 2014) on average—23.7 months for women and 16 for men. Long job searches and more workers self-employed suggest problems of skills mismatches and low net job creation. The latter is certainly a factor. The NSSRA Statistical Yearbook reported that in 2013, employers planned to hire 1,514 workers by year-end, and there were 38 applications for every vacancy advertised. About 63 percent of the unemployed have work experience; most had been laid off due to job reduction and company liquidation or bankruptcy or termination of temporary work. Skills mismatches look likely given that 22 percent of the job seekers with prior work experience had tertiary education and another 26 percent had specialized secondary education. Remittances from Workers Abroad 3.11. Poverty reduction and the incidence of growth in Armenia are also shaped by international labor markets through remittances. Between 2010 and 2013, nearly 8 percent of Armenian households had working-age members who had migrated to areas within or outside the country (NSSRA 2014). Sixty percent of male and 25 percent of female migrants had moved in search of work. Russia is the destination of choice for almost 90 percent of those who leave to seek jobs and many are employed in construction. While most migrants, like most other Armenians, have secondary education, a notable share of migrants, about 30 percent of those employed in industry or retail trade, have tertiary education (Figure 19). Remittances go to 10 percent of poor Armenian households and 16 percent of non-poor households (see Figure 12). Some of the differential in the coverage of remittances is driven by the skills of the migrants who remit as well as the poverty-reducing effect of remittances. The current high inflows of remittances, while good for aggregate demand and for families, may also be influencing the labor market in two ways that are detrimental to the prospects of inclusive growth based on high employment: (1) remittances could discourage those of working-age from seeking employment, and (2) exchange rate appreciation due to remittance inflows can make exports less profitable, pushing jobs into non-tradable sectors. Karapetyan 23 and Harutyunyan (2013) found evidence for such patterns in Armenia that merit further examination for their implications for future poverty reduction. Figure 19. Migrant Education by Sector of Employment, 2013, Percent Retail trade 7.9 39.0 22.8 30.4 Construction 7.0 63.4 18.2 11.4 Manufacturing, mining, utilities 1.5 46.4 21.5 30.7 Agriculture, forestry and fishing 26.6 43.1 23.4 6.9 general, primary and less secondary vocational higher education, tertiary Source: ILCS data 2013. Pensions and Social Transfers 3.12. In Armenia pensions and other social assistance are important social protection programs that are also significant sources of income for households. Although well-targeted, the FBP is a small social assistance program. FBP transfers account for nearly 10 percent of the income of poor households, and 60 percent of the benefit reaches the poor. Nevertheless, FBP transfers did not help reduce poverty between 2010 and 2013. Pensions are universal and individuals become eligible to receive them in case of income loss. The most common pension types are old age and disability pensions. Currently, the pensionable age is 63 for both women and men and the share of pension age-workers is about 12 percent of the population (67 percent of the population is of working age). Pensions account for 27 percent of the income for poor households and 20 percent for non-poor households. The coverage of pensions is also high across the consumption distribution (Figure 20). Almost 90 percent of the bottom quintile lives in a household that receives pensions, although coverage declines to 33 percent for the richest quintile. The old age pension may be either social or contributory (linked to years of service); the share of those receiving a contributory pension is likely higher among the better-off. Pension income is an important driver of not only poverty reduction but also the income growth of the top 60. As Table 1 shows, the government significantly raised the pension amounts in the years after the crisis, and the poverty-reducing effects are clear. However, it is important to note that pension-income growth has favored the better-off who receive contributory pensions. Figure 20. Pension Coverage by Consumption Quintile, 2013, Percent 100% 88% 80% 60% 60% 52% 36% 33% 40% 20% 0% quintile 1 quintile 2 quintile 3 quintile 4 quintile 5 Source: ILCS national data 2013. Note: Quintiles are based on pre-pension consumption. 24 4. Conclusion 4.1. Limited poverty reduction and less shared prosperity since the crisis raise questions about why growth has become less pro-poor. This note attempted to understand what may have been driving the pace of poverty reduction and shared prosperity (growth in the consumption of the bottom 40) in 2010–13. Clearly, low growth limits how much poverty can be reduced and the predicted any worsening of the macro conditions will not be helpful. In the current circumstances, the government’s decision to raise pensions and minimum wages is paying off in reduced poverty. Growing remittances, largely from Russia, also helped bring poverty down. But pensions and remittances also appear to be benefitting the better-off more than the bottom 40, making growth relatively less pro-poor. More analysis is needed to understand how and why remittances and pensions affect the less-well-off. In particular, there are concerns, and some evidence for Armenia, that remittances flows could discourage work, something that deserves further study. 4.2. There are also some positive structural changes in the economy that matter for sustainable future poverty reduction. The structure of GDP has moved away from construction-driven growth to growth led by services, though with some resurgence in manufacturing in response to the government’s export-oriented development strategy. Though agriculture only contributes a small share of GDP, its growth has been favorable in recent years. Employment patterns have shifted away from agriculture toward services and manufacturing, with services dominant in employment. These shifts exacerbate some of the challenges because the less-well-off, who have less education, will not be able to fully tap into the poverty reduction potential of services-led growth. Moreover, decomposition of the drivers of poverty reduction and consumption growth shows clearly the importance of employment—but the Armenian labor market has seen only limited growth in the employment rate. This suggests that if notable poverty reduction and gains in shared prosperity are to be achieved, policies must continue to tackle fundamental issues of how to enhance the skills of the less well-off and stimulate job creation in high-productivity sectors, like manufacturing. 4.3. Compounding long-standing structural issues in its labor market, Armenia is confronted by a number of short-term challenges. Rising energy tariffs, the impact of the Eurasian Customs Union, and the spillovers from a slowing Russian economy could all minimize the possibility for reducing poverty and enhancing shared prosperity. Indeed, as the economy slowed in 2013, poverty reduction had already stalled: there was zero change in poverty rates between 2012 and 2013. Looking ahead, the expected lower economic growth, combined with higher prices due to currency devaluation, rising energy prices, and a projected decline in remittances of up to 30 percent, will all negatively affect income and consumption growth among the poor and vulnerable. The impact could well lead to increases in poverty in 2015 and 2016. 25 References Anusic (2015): “Pensions in Armenia�, unpublished mimeo, World Bank, Washington DC. Asian Development Bank, ADB (2014): “Asian Development Outlook�, Asian Development Bank, Manila. Battiston, Diego et al. (2013): "Income and Beyond: Multidimensional Poverty in Six Latin American Countries," Social Indicators Research, vol. 112(2), pages 291-314. Dang, Hai-Anh and Peter Lanjouw (2013): “Measuring Poverty Dynamics with Synthetic Panels Based on Cross-Sections,� Policy Research Working Paper No. 6504, World Bank, Washington DC. Eiweida, Ahmed and Tigran Kostanyan(2014): “Armenia: Trends and Challenges in Regional Development�, unpublished mimeo, World Bank, Washington DC. Ghazaryan, Armine and Guillermo Tolosa (2012): “Remittances in Armenia: Dynamic Patters and Drivers�, unpublished mimeo, International Monetary Fund, Washington DC. Inchauste, Gabriela et al. (2014): “Understanding Changes in Poverty�, World Bank, Washington DC. Karapetyan, Lili and Liana Harutyunyan (2013): “The Development Side Effects of Remittances in CIS countries: The Case of Armenia�, Consortium for Applied Research on International Migration, European University Institute, Florence. Lustig et al. (2013): “Commitment to Equity: The Impact of Taxes and Social Spending in Inequality and Poverty in Argentina, Bolivia, Brazil, Mexico, Peru, and Uruguay: An Overview. CEQ working paper No. 13. Martirosova, Diana et al. (2015): “Monitoring Poverty in Armenia Using Multidimensional Poverty Indicators�, UNECE working paper No. 15/2015, Geneva. National Statistical Service of the Republic of Armenia, NSSRA (2013), “Social Snapshot and Poverty in Armenia 2013,� Yerevan. National Statistical Service of the Republic of Armenia, NSSRA (2014), “Social Snapshot and Poverty in Armenia 2014,� Yerevan. World Bank (2014a): “Armenia Public Expenditure Review�, World Bank, Washington DC. World Bank (2014b): “Poverty Assessment for Armenia, Promoting Shared Prosperity in Armenia�, World Bank, Washington DC. World Bank (2015): “Macro Poverty Outlook�, World Bank, Washington DC. World Bank (2007): “Armenia: Geographic distribution of poverty and inequality�, Report No. 41181- AM, World Bank, Washington DC. 26 World Bank: “World Development Indicators (data accessed on May 25th 2015)�, World Bank, Washington DC. 27