Report No. 63156- GE Georgia Demographic Change Implications for Social Programs and Poverty July 29, 2011 Human Development Sector Unit South Caucasus Country Department Europe and Central Asia Region Document of the World Bank Acknowledgements This note was prepared by Owen Smith (Senior Economist, ECSH1), with contributions from Victoria Levin (Economist, ECSH4). Sergiy Biletsky (Social Protection Specialist, HDNSP) carried out the pension simulations. The team is grateful to Giorgi Tsuladze (Ilia State University) for his clarifications of Georgia’s demographic data. ii CURRENCY AND EQUIVALENT UNITS Currency Unit = Lari US$0.6= 1.00 GEL (As of July 29, 2011) FISCAL YEAR January 1 – December 31 Acronyms and Abbreviations EBRD European Bank for Reconstruction and Development ECA Europe and Central Asia ECD Early Childhood Development GDP Gross Domestic Product GEL Georgian lari HBS Household Budget Survey HUES Health Utilization and Expenditure Survey ILO International Labor Organization MIP Medical Insurance Program OOP Out-of-pocket PROST Pension Reform Option Simulation Toolkit SSA Social Services Agency TSA Targeted Social Assistance UN United Nations UNFPA United Nations Population Fund UNICEF United Nations Children’s Fund WMS Welfare Monitoring Survey Regional Vice President: Philippe H. Le Houerou Country Director: Asad Alam Sector Director: Mamta Murthi Sector Manager: Jesko Hentschel Task Team Leader: Lire Ersado iii South Caucasus Programmatic Poverty Assessment Note #6 iv Table of Contents EXECUTIVE SUMMARY ...................................................................................................................................... VI I. INTRODUCTION ..............................................................................................................................................1 II. OVERVIEW OF DEMOGRAPHIC TRENDS ................................................................................................ 3 III. AGE PROFILE OF POVERTY AND SOCIAL SPENDING .........................................................................6 IV. POLICY ISSUES ............................................................................................................................................8 4.1. LABOR FORCE PARTICIPATION ...................................................................................................................8 4.2. LABOR FORCE PRODUCTIVITY .................................................................................................................. 12 4.3. COST PRESSURES OF THE PENSION SYSTEM AND OLD-AGE POVERTY ...................................................... 14 4.4. COST PRESSURES IN THE HEALTH SECTOR AND ACCESS TO CARE AMONG THE POOR............................ 16 V. SUMMARY AND POTENTIAL NEXT STEPS ............................................................................................ 19 VI. BIBLIOGRAPHY......................................................................................................................................... 23 List of Figures Figure 1: Demographic trends and key policy issues in Georgia ................................................... 2 Figure 2: Population of Georgia, 2010-2050 .................................................................................. 3 Figure 3: Population Trends, 1950-2050 ........................................................................................ 3 Figure 4: Evolution of population age groups in Georgia, 1990-2050 ........................................... 5 Figure 5: Projected dependency ratios in Georgia, 1990-2050....................................................... 5 Figure 6: Poverty by age groups in Georgia, 2009 ......................................................................... 6 Figure 7: Age distribution of social spending in Georgia, 2010 ..................................................... 7 Figure 8: Labor force participation by age and sex in Georgia, 2010 ............................................ 8 Figure 9: Labor force participation projections under alternative scenarios, 2008-2020 ............... 9 Figure 10: Labor force participation in Georgia by quintile, 2009 ............................................... 10 Figure 11: Impact of elderly females in households on labor force participation of 18-49 age group ..................................................................................................................................... 11 Figure 12: Educational achievement in Georgia by quintile, 2009 .............................................. 12 Figure 13: Pension cost simulations, 2008-2025 .......................................................................... 15 Figure 14: Aging and health spending in Georgia under alternative scenarios, 2010-2030 ......... 17 Figure 15: Financial barriers to health care in Georgia by quintile, 2010 .................................... 18 List of Tables Table 1 Linking demographic changes with social policies and the poor ..................................... 1 Table 2: Labor Force Participation in Georgia and the Region ...................................................... 8 Table A1: Regression results ........................................................................................................ 21 v Executive Summary This note provides an overview of demographic changes and their policy implications in Georgia, with particular reference to the poor. Georgia’s population is expected to decline between 2010 and 2050, and this trend will be accompanied by a growing elderly cohort and a rising total dependency ratio. The note emphasizes four interrelated policy topics. These are labor force participation, labor force productivity, and potential cost pressures arising from the pension system and from the health sector. In each area, special attention is given to the linkages between these issues, social spending programs, and opportunities for targeting the poor. The note does not present specific policy recommendations, but instead outlines broad areas where future analytical work might be undertaken to arrive at more precise policy options. Georgia’s poverty headcount is lowest among the 60+ age group and highest among the under-18 population. As is common, social spending in Georgia is more heavily concentrated among older age groups to provide income support to the elderly. Per capita social transfers to the 60+ age group are about 7 to 10 times higher than to the under-50 age group. The 0-5 age group receives a lower level of social spending per capita than school-age children or adults. The total number of workers in Georgia will fall over the next decade, but it is possible to sustain growth by mitigating this trend through policies that promote increased labor force participation. By international standards, participation rates are relatively high in Georgia—but they are somewhat lower for women compared to neighboring countries. They are also lower for persons who live in the same household as an elderly person (especially if they are chronically ill). This may be due to working-age household members remaining out of the work force due to care-giving responsibilities. There is also some evidence that pension receipt – especially by women – may reduce the pressure on younger household members to find work. Potential policy measures to address work force participation include: (i) equalizing the pension-eligibility age for women and men at 65; and (ii) closely monitoring the impact of social transfer programs – including pensions – on labor supply (particularly because increases in monthly benefit levels are planned). It is also worth exploring the option of using SSA centers to better link the disadvantaged with jobs. Enhancing labor force productivity should also be a priority, in order to promote growth within the context of a shrinking workforce. Measures to strengthen the education sector would lie at the heart of this agenda. In addition to improving schools, it is worth exploring opportunities for life-long learning that enables workers to enhance their skills. Perhaps most importantly, serious consideration should be given to better reaching younger age groups with social spending—in particular, through the implementation of early childhood development programs to help promote equality of opportunity. International evidence increasingly suggests that rates of return on human capital investments are highest at this stage. Specific policy options could be designed that incorporate the experiences of other countries that have successfully instituted early childhood development as a foundation for their human development strategies. vi Neither the pension system nor the health system faces serious fiscal sustainability issues in their configurations as of 2011—but over time both will need to strengthen their safety net function for the poor. Adjustments to the pension system (e.g., monthly benefit levels and indexation options) and the health financing system (e.g., insurance coverage and benefit package design) will need to strike a balance between reducing old-age poverty and maintaining long-term fiscal sustainability in view of the growing percentage of elderly persons in Georgia. In both cases, targeted rather than universal policies and programs may help balance these objectives. vii I. Introduction 1. Georgia is currently experiencing significant demographic changes, which have far- reaching policy implications. Shifts in a country’s demographic profile happen only gradually, and thus do not often figure prominently in day-to-day policy deliberations. However, an awareness of these trends and their ramifications is essential to inform decision-making on a wide range of issues. This policy note: (i) provides an overview of key demographic changes unfolding in Georgia; (ii) explores their implications for working-age and elderly populations; and (iii) highlights linkages with social spending and the poor. It does not present specific policy recommendations, but instead outlines broad areas where future analytical work could help identify more tailored policy options. 2. The major driver of Georgia’s demographic shift is an aging population. Long- standing low fertility rates and a gradually increasing life expectancy have resulted in rising dependency ratios, with fewer people of working age expected to support a growing number of elderly persons. The two major implications of an aging population are a smaller share of the population is of working age – which raises important issues for labor force participation and productivity – and a larger share of the population is made up of older age groups – which poses an obvious challenge for the fiscal sustainability of pension and health programs. These issues are depicted in Figure 1. Each of these topics raises specific issues for the poverty-reduction agenda, as indicated in Table 1. 3. An aging population is a reality that cannot be avoided in Georgia—however, significant policy space exists for mitigating its potential negative effect on Georgia’s overall development trajectory, including for the poor. Policy levers are available to promote labor market participation and productivity to counter rising dependency ratios; education systems can be strengthened to support these efforts; and measures can be taken to ensure that pension and health spending is sustainable. These issues are expanded on below. For each potential policy measure, the implications for addressing poverty and vulnerability are addressed. A more detailed discussion of the impact of aging across the Europe and Central Asia region is contained in the 2007 World Bank publication From Red to Gray. Table 1 Linking demographic changes with social policies and the poor Demographic trend Key policy challenges Linkages with social policies Shrinking working-age Promoting labor force x Work disincentives of transfer programs population participation x Dependent care options (e.g., long-term care) x Linking the unemployed with jobs x Raising retirement ages Raising labor force x Early childhood education productivity x Lifelong learning x Active labor market programs Rising elderly Containing pension cost x Choosing benefit levels, retirement age population pressures x Adequate replacement rates, social pensions Containing health system cost x Enabling access to care by lowering out-of- pressures pocket payments x Service delivery reforms 1 4. The note is structured as follows. The next section summarizes the key demographic trends underway in Georgia. Section 3 presents a profile of poverty and social spending by age. Section 4 analyzes the four policy issues listed in column 2 of Table 1. In each of these sections, the poverty reduction angle is emphasized. The final section provides a summary of the key messages, with an emphasis on policy priorities going forward. Figure 1: Demographic trends and key policy issues in Georgia Demographic Trends 2010 and Potential Effects 2050 100+ 100+ Men Women Men Women 90-94 90-94 More elderly: 80-84 ↑ Pension spending 80-84 ↑ Health spending 70-74 70-74 60-64 60-64 Age 50-54 Age 50-54 40-44 40-44 30-34 Fewer workers: 30-34 ↓ Labor force 20-24 ↓ Output 20-24 10-14 10-14 0-4 0-4 180 140 100 60 20 20 60 100 140 180 180 140 100 60 20 20 60 100 140 180 Population ('000s) Population ('000s) Source: UN and WB staff 2 II. Overview of Demographic Trends 5. Georgia is undergoing a period of significant demographic change. In 2010, it had a population of about 4.4 million; by 2050, this is expected to shrink, although by how much is uncertain (see Figure 2).1 Very recent developments, including the emergence of positive net migration and an uptick in fertility, will result in a higher population if sustained. Under the ‘medium’ scenario, Georgia’s population will be about the same in 2050 as it was in 1950, in contrast to Armenia and Azerbaijan, which are expected to experience population increases of about 2.5 and 3.5 times, respectively, over the same period. In this respect, Georgia parallels more closely the better-off countries in Eastern Europe (see Figure 3; ‘Eastern Europe’ includes Belarus, Bulgaria, Czech Republic, Hungary, Moldova, Poland, Romania, Russia, Slovakia and Ukraine). Figure 2: Population of Georgia, 2010-2050 Figure 3: Population Trends, 1950-2050 400 5.0 Armenia Azerbaijan 4.5 350 Georgia Eastern Europe 4.0 Total population (1950=100) 300 Population (millions) 3.5 250 3.0 Low 200 2.5 Medium 2.0 150 High 1.5 100 1.0 50 0.5 0.0 0 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Source: UN 2010 2015 2020 2025 2030 2035 2040 2045 2050 Source: UN 6. Georgia’s declining population is largely due to the historical trend of its fertility rate, which fell earlier and further than the rate in neighboring countries in the South Caucasus. By 1950, the average number of children per woman in Georgia had fallen to 3, significantly lower than in Armenia (4.5) and Azerbaijan (5.5). The impact continues to be felt, even as all three converged to a similar fertility level (just under 2 children per woman) by 2010. An additional factor has been external migration, particularly of the Russian population, during the 1990s. The Georgian population has been more stable. A further noteworthy trend is the emergence of skewed sex ratios (the number of male per female births); this currently stands at 1.11 in Georgia, as compared to 1.06 in Eastern Europe and 1.05 in the west. The long-term implications of this trend are unclear.2 7. An additional cause of Georgia’s declining population size is that low fertility rates have not been significantly offset by longevity gains. Between 1970 and 2010, life expectancy in Georgia increased from about 68 to 72. This increase is similar to that of Georgia’s neighbors, 1 This technical note uses the UN’s ‘medium scenario’ in Figure 2. A key reference on the trends described here is the “Demographic Yearbook of Georgia” (2009), by UNFPA and Ilia State University. 2 See Mesle et al. (2005), who attribute this to sex-selective abortions which are prevalent among third births. A skewed sex ratio has been proposed as a cause of China’s high savings rate (Wei and Zhang, 2009). 3 but it is much lower than the increase in most countries in East Asia and Latin America.3 The future trend in life expectancy in Georgia is quite uncertain, and will depend on investments by government and households both inside and outside the health sector. Improved progress on health outcomes may slow down the decline in the size of the overall population only a small degree—and it would also imply a larger percentage of elderly persons. A total population of 3.3 million in 2050, as projected in Figure 2, is predicated on the assumption that life expectancy will increase from 72 to 78. 8. Low fertility rates and gradual increases in longevity are the main factors underlying Georgia’s most important demographic trend: an aging population. The median age is projected to increase from 31.2 in 1990, to 37.6 in 2010, and to 45.8 in 2050. Figure 4 shows the predicted evolution of four age groups between 1990 and 2050, with respect to both the number of persons and the share of the population. The population share of the under-15 age group is expected to fall from 25 percent to 15 percent, and the working age population will fall from 66 to 60 percent. In contrast, the population shares of the over-65 and particularly the over- 80 age groups will rise significantly. 9. The combination of a shrinking workforce and an expanding elderly population will clearly have wide-ranging policy implications. After several years of a declining dependency ratio,4 the year 2010 marked a turning point in Georgia, with the total dependency ratio expected to increase from 90 dependents per 100 persons of working age in 2010 to nearly 140 dependents per 100 persons by 2050. This will be driven largely by a rising old-age dependency ratio, because the child dependency ratio is expected to remain relatively stable (see Figure 5). 10. In sum, Georgia is facing major demographic changes, including a declining overall population, an aging population structure, and rising dependency ratios . The note will now turn to an overview of poverty and social spending by age group, before exploring a range of policy issues that arise in that context. 3 Rajaratnam et al. (2010), “Worldwide mortality in men and women a ged 15-59 years from 1970 to 2010: A systematic analysis”. Lancet: 375: 1704-1720. 4 The total dependency ratio is the ratio of the sum of the population aged 0-14 and that aged 65+ to the population aged 15-64. The child dependency ratio is the ratio of the population aged 0-14 to the population aged 15-64. The old-age dependency ratio is the ratio of the population aged 65+ to the population aged 15- 64. All ratios are presented as the number of dependents per 100 persons of working age. 4 Figure 4: Evolution of population age groups in Georgia, 1990-2050 Figure 5: Projected dependency ratios in Georgia, 1990-2050 140 Dependents/ 100 persons of working age 120 100 80 60 40 20 0 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 Child dependency ratio Old-age dependency ratio Total dependency ratio Source: UN 5 III. Age Profile of Poverty and Social Spending 11. Poverty rates in Georgia vary by age group, with the young significantly more affected than the old (see Figure 6). While the overall prevalence of poverty is 25.7 percent, the poverty headcount among the under-18 group is 29 percent, compared to 22.2 percent among the over-60 population. The working age population is in between. Moreover, the prevalence of child poverty is even higher in multi-children families: in households with three or more children, the poverty headcount is 35.2 percent. Figure 6: Poverty by age groups in Georgia, 2009 40% 100% 90% 17% 22% 21% 20% 35% 24% 29.0% 80% 30% 26.3% Poverty headcount 25.9% 24.7% 70% 25% 22.2% 60% Share 20% 58% 55% 50% 56% 57% 61% Age 60+ 15% 40% Age 18-59 30% 10% Age < 18 20% 5% 25% 23% 24% 10% 19% 18% 0% 0% 0 to 17 18 to 29 30 to 49 50 to 59 60 and up 1 2 3 4 5 Age group Quintile Source: WMS Source: WMS 12. Social spending in Georgia is mostly received by older age groups, and does not offset the over-representation of younger cohorts among the (post-transfer) poor. Figure 7 depicts the results of an exercise to disaggregate all social spending in Georgia according to the age of recipient. The results are based on 2010 administrative data from the Social Services Agency (SSA), in tandem with an analysis of recent household surveys (UNICEF’s Welfare Monitoring Survey and the Health Utilization and Expenditure Survey). Data on municipal spending, including kindergarten programs, is not available. Private household spending is not shown, as the focus here is on public policy. The top panel shows per capita spending by age group according to the age of the intended recipients. The bottom panel assumes full intra- household sharing of social transfers. For example, in the case of a pensioner who receives GEL 80 per month and who lives with two middle-aged adults and one child, GEL 20 would be “allocated” to each of the four household members. Although the second method entails a “redistribution” of social spending to younger age groups, even in this scenario the major share of social spending – particularly non-education expenditures – is received by older age groups. The difference is particularly marked in the case of social protection spending: older age groups receive an average of nearly GEL 1200 per capita annually (GEL 100 per month), while the highest average per capita social benefit for any age group under 50 is GEL 120 (GEL 10 per month). This broad pattern is likely to be found in most countries due to the common predominance of pension spending, but the contrast across age groups may vary. 6 13. In sum, social spending is more concentrated among the elderly, leaving younger cohorts somewhat over-represented among the poor. The next section will explore various policy issues that arise in relation to demographic pressures, and their relevance for social programs. Figure 7: Age distribution of social spending in Georgia, 2010 (a) Transfer receipts accrue only to the intended beneficiary 1,400 1,200 Annual spending per capital (GEL) 1,000 Education Child programs 800 Health IDPs State compensation 600 TSA Political pension 400 Survivor's pension Disability pension Old age pension 200 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Age group Source: SSA, WMS, HUES (b) Transfer receipts are shared equally within households 1,200 1,000 Annual spending per capita (GEL) 800 Education Child programs Health IDPs 600 State compensation TSA Political pension 400 Survivor's pension Disabled pension Old age pension 200 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Age group Source: SSA, WMS, HUES 7 IV. Policy Issues 4.1. Labor force participation 14. The shrinking working-age population poses an obvious challenge to sustaining and enhancing labor force output and growth in Georgia. The government of Georgia may consider policy options aimed at creating higher labor force participation rates, including stronger attachment to the formal labor market (especially among the poor), and possibly immigration. It may also monitor the potential for transfer programs (including pensions) to discourage labor market participation. This section summarizes the challenges and potential policy responses. 15. Overall, the labor force participation rate in Georgia compares quite favorably with other countries in the region. Among persons in the 15-64 age group, it is currently about 74 percent for men and 55 percent for women, or about 64 percent overall (see Table 2). These rates are higher than the rates in Europe, and slightly lower than those in Armenia and Russia, primarily due to lower participation rates by women. The ILO projects small increases over the next ten years. Figure 8 shows labor force participation by age group, with the familiar inverted U-shape: the youngest and oldest age groups are least likely to be in the labor market. Addressing youth unemployment is a key related challenge. Table 2: Labor Force Participation in Georgia and the Region Georgia: Labor force participation rates, ECA: Labor force participation rates, 1990-2020 2009 Men Women Total Men Women Total 1990 78.4 59.9 68.5 EU-27 66.4 51.8 58.9 1995 72.3 56.3 63.7 EU-10 64.4 50.1 56.9 2000 74.0 54.7 63.6 Russia 73.6 62.9 67.8 2005 73.3 55.4 63.7 Armenia 74.6 59.6 66.3 2010 74.1 55.2 63.9 Azerbaijan 66.8 59.5 63.0 2015 75.7 56.5 65.3 Source: ILO 2020 75.7 56.9 65.6 Source: ILO Figure 8: Labor force participation by age and sex in Georgia, 2010 100 90 80 70 60 50 % Total 40 Male 30 Female 20 10 0 [15-19] [20-24] [25-29] [30-34] [35-39] [40-44] [45-49] [50-54] [55-59] [60-64] 65+ Source: ILO 8 16. Some basic simulations provide an indication of the potential evolution of labor force participation in Georgia under alternative scenarios. Figure 9 shows a baseline ILO scenario for labor force participation, under which the number of workers declines modestly from about 2.27 million to 2.15 million between 2008 and 2020. This decline could be partly offset if the participation rate was held constant at 2008 levels, or increased by 2 percent across all age-gender groups, or increased by 6 percent in the 40-59 age group. However, in all cases there would still be a decline. 17. Another option for sustaining higher participation rates would be to equalize the pension age eligibility for men and women. At present it is 65 for men and 60 for women, despite the fact that life expectancy for women is more than five years longer. Household budget survey (HBS) data indicate that about 72 percent of women aged 55 to 59 are economically active, while only 56 percent of women 60 to 64 remain so. Raising the retirement age is a difficult measure for any government to adopt, but it is a step that could contribute to efforts to offset the declining working-age population. Figure 9: Labor force participation projections under alternative scenarios, 2008-2020 2,300 2,250 Number of workers (000s) 2,200 2,150 ILO baseline projections LFPR constant at 2008 levels 2,100 LFPR increases by 2% 2,050 LFPR increases among middle ages (40-59) by 6% 2,000 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Source: ILO and WB calculations 18. Labor force participation is obviously a major issue for the poor. Figure 10 shows the labor force status of workers (15-59 age group) in Georgia by quintile. The poorest and richest quintiles are almost equally likely to be economically active, but the rich are more likely to have a job and far more likely to earn regular wages. Self-employment and informal work are more common among the poor. One way of addressing this issue would be through the implementation of initiatives to better link the unemployed with job openings. The Social Services Agency (SSA) is already well-advanced in the establishment of “one-stop shop” service centers for administering social programs, including the targeted social assistance (TSA) program aimed at the extreme poor. These centers could also be used as information exchanges to better assist the poor with their job search efforts. 9 Figure 10: Labor force participation in Georgia by quintile, 2009 80% 70% 60% 50% 40% % economically active 30% % having a job 20% % wage employed 10% 0% 1 2 3 4 5 Source: HBS Quintile 19. It is also important to be cognizant of the potential impact of social transfer programs on labor force participation. For example, Georgia’s TSA program for vulnerable families is one of the most generous such programs in the region. It has a cash benefit of GEL 30 per month for the household head, and GEL 24 per month for additional household members. This amounts to an average of about 35 percent of recipients’ total household consumption. In principle this might serve as a disincentive for finding a (formal) job or working longer hours. However, while an appropriate data-set to rigorously assess this proposition is not available, a simple comparison of labor market participation by those who receive the TSA (i.e., with a proxy means test score below 57,000) with those who do not get TSA but do receive medical insurance for the poor (i.e., those with a score between 57,001 and 70,000) does not detect any significant difference. However, a full analysis will have to await more complete data. 20. Certain features of Georgia’s pension system may also have an impact on labor market participation. First, the old-age pension of GEL 80 per month is funded out of general tax revenues and is based strictly on age-eligibility criteria (65 for men and 60 for women), with only a small top-up based on work history. Thus, it is effectively a cash benefit like the TSA, but with different eligibility requirements. Second, about 16 percent of Georgians are old-age pensioners. However, due to a high prevalence of multi-generational families, about 53 percent of all households include at least one pensioner. To the extent that pension receipts are shared within a household, there could be an impact on the working-age population. Third, the pension budget is about four times larger than the TSA budget, with a correspondingly larger potential to affect household behaviors. The average replacement rate (benefit as a share of average wage) of the old age pension is about 15 percent,. However, as discussed below, this may rise markedly in the near future. 21. There is some evidence that pension receipt has an effect on labor market participation. To explore this issue, a model was estimated using HBS data. The probability of the working-age household members being economically active (or having a job) is predicted on the basis of standard controls and a series of dummy variables for the presence of ten different age/gender groups of household members aged 50 and over.5 The results suggest a significant 5 The approach follows the methodology used by Bertrand et al. (2003) with reference to the similar pension system in South Africa during the 1990s. Controls include gender, a quartic in age, educational achievement, 10 negative impact of about 4 percentage points on the probability of working-age household members being economically active following the achievement of pension eligibility by an older household member, but only if the older person is female (that is, after they turn 60). Figure 11 shows this impact for the 18 to 49 year old working population. This is consistent with previous work on a similar pension system in South Africa. It is suggestive of intra-household reallocation of pension receipts, particularly by an elderly woman in the family. This result does not hold if the analysis is restricted to the poorest quintile (the effect is clearest in the middle quintile). 22. Irrespective of the impact of the pension system itself, the presence of a chronically ill older person in the household may also affect labor market participation. Survey data provide tentative evidence that this may be an issue. For example, a chronically ill older person in the household is associated with a lower probability of middle-aged members having a job (by about 2 percentage points), even after controlling for other factors. An impact of similar magnitude is observed in the case of households with children under 5 years of age. These patterns are probably due to a perceived need to stay at home to take care of dependent family members. About 38 percent of the over-50 population in Georgia reports having a chronic illness. One policy implication is that better health and/or long-term care services for the elderly (or child care for the young), could have a positive impact on labor force participation by working-age family members who are otherwise pre-occupied with their care-giving roles. (Long-term care can also improve system-wide efficiency by reducing hospitalization). Satisfaction with health status in Georgia is very low (Gallup 2007), and reliance on out-of- pocket payments for health care is very high. Addressing these issues could help tackle the challenge of a shrinking labor force. Figure 11: Impact of elderly females in households on labor force participation of 18-49 age group 0% Pension % change in probability of being eligibility -1% economically active -2% -3% -4% -5% 50-54 55-59 60-64 65-69 70+ Presence of women 50+ in household Source: WB staff based on HBS N.B.: Blue markers indicate statistical significance. There is also a statistically significant difference between coefficients for elderly women aged 55-59 and 60-64. rural/urban location, family size, hous ehold income net of that individual’s own earnings (if any), and the presence of under-5 and school-age children in the household. See Annex for further details. 11 23. In brief, a shrinking working-age population in Georgia highlights the importance of policy instruments that could boost labor market participation. These options include: (i) raising the female retirement age to 65; (ii) using SSA centers to better link the poor with job opportunities; (iii) monitoring potential labor disincentives embedded in cash transfer programs; and (iv) ensuring adequate health, long-term care, and day-care services to reduce dependent care-giving responsibilities of working-age household members. 4.2. Labor force productivity 24. An additional response to the shrinking working-age population would be to implement policies aimed at improving the productivity of workers, and thus promoting economic growth more broadly. Education reforms would be front and center in this agenda, but it would also be important to consider programs aimed at persons outside of the traditional school ages. Georgia’s strong performance in international comparisons of investment climate should also have positive long-term impacts on labor productivity, by promoting foreign direct investment and technology transfers. 25. Enhancing educational opportunities for the disadvantaged may be seen as the key priority for raising the future labor productivity of today’s poorer households. Figure 12 shows the expected correlation between educational achievement and socioeconomic status. In the bottom two quintiles in 2009, more than twice as many people had not gone beyond secondary school as compared to those who had. Providing opportunities for the poor to stay in school longer can help equalize the playing field. Moreover, this agenda does not need to be limited to those of school age: international experience suggests that programs for life-long learning through training opportunities for the working-age population can be very effective at raising labor productivity. Again, SSA centers could be a good forum for linking the disadvantaged with information about training options. Figure 12: Educational achievement in Georgia by quintile, 2009 80% 70% 60% 50% % stopped at 40% secondary 30% % more than 20% secondary 10% 0% 1 2 3 4 5 Source: HBS Quintile 26. According to the age profile of social expenditures shown earlier in Figure 7, the 0 to 5 age group receives less spending per capita than either school-age children or adults. Persons in the 0-5 age group can receive cash benefits under the targeted social assistance 12 program (TSA) if their household is eligible and has applied, but there are few other social programs that are likely to reach this group. There is a small level of spending for programs in support of orphanages (including health insurance) and adoptions, and a new food voucher for 0 to 3 year olds, but otherwise there are no pre-school or day-care programs and no specific child benefits. (Municipal spending for kindergartens was not included in the analysis due to data constraints, and private spending is not reflected as the focus here is on the public policy angle). 27. The relatively low level of social spending directed to the 0 to 5 age group stands in contrast to growing international evidence that successful early childhood development programs offer potentially the highest rates of return on human capital investments. An emerging body of literature has identified the strong linkages between early-childhood experiences and later life outcomes, including educational achievement, employment, and health outcomes.6 This stage is particularly important because it corresponds to the period when brain development is at its most rapid, and thus it is crucial for laying the foundation for skills development (both cognitive and behavioral) and physical health that are so important for success later in life. Indeed, as a general rule, “the earlier the better” appears to hold true with regard to the impact of programs aimed at skills development. Rates of return for the early- childhood period are estimated to be significantly higher than for the primary school age period, which in turn is better than for programs aimed at youth or young adults. 28. Social programs targeting the adult population can have a positive impact on “equality of outcomes”, particularly among those who can no longer work, social programs targeting the youngest can make a key contribution towards achieving “equality of opportunity.” In Georgia, as elsewhere, life chances are often strongly influenced by circumstances beyond the control of the individual. For example, there is a spatial dimension: poverty rates are significantly higher in rural areas and in particular regions (e.g., Mtskheta- Mtianeti). There is also a strong inter-generational channel: poverty headcounts are about 2.5 times higher in households in which the mother has not achieved a secondary education, a factor beyond the control of her children. The notion that poverty is caused by much more than just individual effort is well-supported by the Georgian population: only 8.5 percent attributed the existence of poverty to “laziness and lack of willpower”, whereas 59 percent pointed to “injustice in society” or being “unlucky”.7 29. In sum, poverty-reduction policies that seek to enhance equality of opportunity, particularly among the young, merit serious consideration. This would entail exploring options for expanding social spending targeted to younger age groups, and in particular for adapting international experience with early childhood development programs to the local context. 30. In brief, a shrinking working-age population in Georgia highlights the importance of utilizing policy instruments that could boost labor market productivity and thus economic growth. A stronger education system would lie at the heart of this agenda, and while reforms to schools and colleges are ongoing, a relatively neglected area has been early-childhood development programs for children in the critical period from ages 0 to 5, especially programs aimed at providing greater equality of opportunity for poorer households to escape poverty. Labor market productivity can also be increased through programs for life-long learning for the working-age population that help workers keep their skills up to date. Other options include 6 See, for example, Almond and Currie (2010), “Human Capital Development before Age 5”. NBER. 7 World Bank-EBRD Life in Transition Survey, 2007 13 using SSA centers to link the poor with training opportunities, and sustaining Georgia’s excellent track record in terms of strengthening the investment climate. 4.3. Cost pressures of the pension system and old-age poverty 31. Fewer workers and more elderly will clearly have implications for the fiscal sustainability of future social spending in Georgia. The two major programs from a budgetary perspective are old-age pensions and health care. The following section presents a brief analysis of the potential costs under various alternative scenarios, and a discussion regarding the need to achieve balance with old-age safety nets. 32. The old-age pension budget accounts for nearly 4 percent of GDP, and close to two- thirds of social protection spending. All women over 60 and men over 65 receive GEL 80 per month as of mid-2011, regardless of work history. A small top-up of between GEL 2 and GEL 8 is also available, depending on work experience. In total, about 665,000 old-age pensioners received benefits in 2010. The monthly benefit has increased rapidly in recent years, from GEL 14 per month in 2003 to GEL 38 per month in 2007 to the current GEL 80. An increase to GEL 100 is planned starting in September 2011. There is no indexation of pensions. However, there are plans for further increases in the near-term, including a target of “hundred-dollar pensions” (about GEL 165 at the current exchange rate) beginning in late 2012. 33. Figure 13 shows the implications for pension spending as a share of GDP under various alternative scenarios. The first graph depicts four alternative ways of projecting the current benefit of GEL 80 forward: (i) with no indexation; (ii) with wage indexation; (iii) with price indexation; and (iv) with price indexation and retirement ages for men and women equalized at 65 by 2020 (as discussed above). In all scenarios except wage indexation, the GDP growth would more than offset the pace of a rising pensioner group and other considerations, so pension spending as a share of GDP would fall. The second graph depicts the same four alternative projections, but with the monthly benefit increased to GEL 150 by 2012 to approximate the implications of pursuing a target of “hundred-dollar pensions”, as proposed. This would imply spending 6 percent of GDP on pensions, but it would fall over the course of time in all scenarios except wage indexation. 14 Figure 13: Pension cost simulations, 2008-2025 Pension spending scenarios with GEL 80 benefit, % of GDP 10% 9% Wage indexation Price indexation 8% Price indexation, equalize M/F retirement ages Pension spending (% of GDP) 7% No indexation 6% 5% 4% 3% 2% 1% 0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 Pension spending scenarios with GEL 150 benefit, % of GDP 10% Wage indexation 9% Price indexation 8% Price indexation, equalize M/F retirement ages No indexation Pension spending (% of GDP) 7% 6% 5% 4% 3% 2% 1% 0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 Source: WB estimates using PROST 34. The simulations indicate that the Georgian pension system does not pose a serious fiscal problem in its current configuration. It should remain sustainable, provided that any increases in the monthly benefit are kept modest (e.g., an immediate increase to GEL 150 for all pensioners would pose a financing gap), and no wage indexation is adopted. The scenarios that include the equalization of the pension age eligibility for men and women at 65, which would lower required pension outlays, provides evidence of the advantages of including this measure in addition to the productivity angle discussed above. 35. However, fiscal considerations have to be balanced with the need to address old-age poverty. A policy of non-indexation coupled with no further ad-hoc monthly benefit increases in the scenarios above would result in a steadily deteriorating replacement rate (pension benefit 15 relative to average wage). At present, the replacement rate is about 15-20 percent. Thus, it would fall to an unacceptably low level in a few years without a further increase. 36. The significant poverty-reducing impact of pensions is the major reason for the lower poverty rates among the elderly depicted in Section 3. Based on 2009 data, it has been estimated that the overall poverty rate would increase by half, from 25.7 percent to 39.1 percent, in the absence of old-age pension transfers. This reflects in part the fact, as noted, that while pensioners account for about 1 out of every 6 Georgians, due to a high level of multi- generational living, just over half of all Georgian households include at least one pensioner. Moreover, the old-age pension budget accounts for nearly 4 percent of GDP, and close to two- thirds of social protection spending. 37. In brief, the current pension system is affordable going forward, but maintaining the fiscal sustainability of the system while adequately addressing poverty among the elderly will remain a challenge. Since pensions account for about one-sixth of the overall government budget, this will be a key public finance challenge in the years to come. One potential approach to strike this balance would be to provide means-tested pension benefits (social pensions) to poorer households as a top-up to the basic monthly amount. 4.4. Cost pressures in the Health sector and access to care among the poor 38. Concerns about health spending sustainability also arise frequently in the context of aging populations. Older people are typically in worse health, and therefore require more medical care. In Georgia, the 50+ age group spends on average 50 percent more on health than the 18-49 group.8 As the population ages, overall health spending should rise. The challenge will be to achieve sector objectives – such as better health and financial protection – subject to demographic constraints. 39. However, international literature suggests that population aging is typically not the major driver of health spending increases over time. While an aging population does contribute to rising health costs, it happens only gradually. Thus, it cannot explain the significant increases in health expenditures over time that are common around the world. There are more significant factors -- in particular, the adoption of new technologies and the expansion of insurance coverage. These can be viewed as “age-specific” costs – in other words, these factors can drive up the amount that will be spent on a 65-year old in five years compared to the amount that is spent on a 65-year old today. These factors may be particularly important in Georgia during the years ahead, because at present only about 25 percent of the population is covered by insurance, and the availability of technologies (e.g., to treat cardiovascular disease and cancer) is far behind Europe and the U.S. However, unlike aging-related costs, age-specific expenditures are very much amenable to policy influence: judicious expansions of insurance programs and benefit packages can help ensure that cost increases are manageable within the available budget. 40. Several simple projections are laid out in Figure 14 that illustrate the potential impact of aging and other factors on future health care costs in Georgia. In 2010, government spending on health in Georgia was 1.8 percent of GDP. If medical spending per 8 These estimates are based on survey data that capture out-of-pocket spending, which represents about two- thirds of total health spending in Georgia. It is assumed that a similar spending-by-age differential applies to that portion of expenditures covered by government. In other countries, this ratio is usually higher than 1.5:1. 16 person rises at the same rate as GDP until 2030, health spending will actually fall to 1.7 percent of GDP, despite the impact of aging. This is because the anticipated decline in overall population noted earlier, coupled with GDP growth, will more than offset the medical expenses due to the rising share of elderly persons. If, however, age-specific factors such as technology and insurance cause medical spending per person to increase at a rate of 1 or 2 percentage points faster than GDP, government health spending would increase to 2.0 or 2.4 percent of GDP, respectively. Therefore, age-specific factors are likely to be more important drivers of health spending increases in Georgia going forward than the aging population per se. 41. With health spending posing little challenge to fiscal sustainability at present, priority should be given to improving access and lowering out-of-pocket payments for the poor. In 2010, the poor were significantly more likely to forego medical care in the case of sickness, or not buy medication when prescribed, because they could not afford it (Figure 15). A previous technical note highlighted the high reliance on out-of-pocket payments (OOP) in Georgia’s health financing system, including high drug prices relative to the EU.9 Although high OOP may help promote cost containment in the face of demographic pressures, they also act as a hindrance to broader sectoral objectives. Fortunately, a policy solution is within reach: the Medical Insurance Program for the poor has proven to have a large impact on reducing OOP. Currently about 20 percent of the population is eligible for the program, and expanding eligibility beyond this group would be an important step in improving access to health care for the poor. Figure 14: Aging and health spending in Georgia under alternative scenarios, 2010-2030 3.0% 2.4% 2.5% 2.0% 2.0% 1.8% % of GDP 1.7% 1.5% 1.0% 0.5% 0.0% 2010 - 2030 - 2030 - 2030 - Actual Impact of aging Health spending Health spending only rises 1% point rises 2% points Source: WB calculations faster than GDP faster than GDP 9 South Caucasus Programmatic Poverty Technical Note, World Bank, 2010. 17 Figure 15: Financial barriers to health care in Georgia by quintile, 2010 30% 25% % respondents 20% 1 15% 2 3 10% 4 5% 5 0% No medical care sought due No prescription purchased to financial barriers due to financial barriers Source: HUES 42. In sum, as in the case of pensions, the impact of an aging population on health costs in Georgia does not appear to threaten the system’s sustainability. Age-specific drivers, such as insurance and technology, are more likely to be the cause of spending increases—but these factors can be managed through careful policy design. Moreover, government spending in Georgia is among the lowest in the region—and in all of the scenarios presented above, health expenditures would remain well below regional averages. In addition – as noted earlier – a certain portion of additional spending might have a positive impact on labor force participation. As health spending increases, priority should be accorded to improving access to the lower socioeconomic groups not yet covered by MIP. 18 V. Summary and Potential Next Steps 43. This note aimed to provide an overview of demographic changes and their policy implications in Georgia, with particular reference to the poor. Georgia’s population is expected to decline between 2010 and 2050. This trend will be accompanied by a growing percentage of elderly persons, and a rising total dependency ratio. Meanwhile, Georgia’s poverty headcount is highest in the under-18 population and lowest in the 60+ age group. Overall social spending is mainly concentrated on older age groups. 44. The note emphasized four interrelated policy topics. These are: (i) labor force participation; (ii) labor force productivity; (iii) potential cost pressures arising from the pension system; and (iv) potential cost pressures arising from the health sector. In each area, special attention was given to the linkages between these issues, social spending, and opportunities for targeting the poor. 45. The key policy implications and a potential agenda going forward can be summarized under each of these four areas as follows: x Potential policy measures to address labor force participation include equalizing the pension-eligibility age for women and men at 65, and using SSA centers to better link the disadvantaged with job opportunities. It will also be important to closely monitor the impact of social transfer programs, including pensions, on labor supply (particularly because increases in monthly benefit levels are under consideration). Lastly, it may be worth exploring long-term care programs for the elderly that would make it easier for younger household members to join the labor market. x Measures to strengthen the education sector lie at the heart of the agenda for addressing labor force productivity. In addition to improving schools, it is worth exploring opportunities for life-long learning that enable workers to enhance their skills (again potentially using SSA centers as information brokerages that can reach the poor). Early childhood development (ECD) is crucial for improving future skills and labor force productivity. At present, public spending on ECD programs (and on the youngest age group in general) is relatively low. Leaving this issue to private household spending is unlikely to result in equality of opportunity. Specific policy options could be designed that incorporate the experiences of other countries that have successfully instituted ECD as a foundation for their human development strategies. x Cost pressures in the pension system are not a major problem in its current configuration, but over time there will be a need to strengthen the safety net function for the poor. Adjustments such as raising monthly benefit levels and indexation options will need to strike a balance between reducing old-age poverty and maintaining long-term fiscal sustainability in view of the growing share of elderly persons. Targeting increases to the poor rather than using universal approaches could help balance these objectives. x Cost pressures in the health system also do not pose a major challenge at present. However, adjustments to the health financing system (e.g., insurance coverage and benefit package design) will need to achieve a balance between improving access and reducing out-of-pocket payments on the one hand and maintaining system affordability 19 on the other. Again, targeted (e.g., through MIP) rather than universal policies and programs may help balance these objectives. 46. An analytical work program to help support further work related to these policy options could include the following: x A deeper analysis of current workforce skills in Georgia, including the role of the education system and the potential for introducing active labor market policies. This would entail a comprehensive assessment of both labor supply and labor demand characteristics, the constraints associated with each, and specific measures that could help overcome those constraints. For example, the potential role of activation policies to improve labor supply, including prospects for shrinking the size of the informal sector, could be explored. x Ongoing survey work, as new data becomes available, to analyze the impact of social spending (especially TSA and pensions) on labor force decisions by the household. Household budget survey data offer an opportunity to monitor key indicators of relevance to social sector policies in Georgia on an ongoing basis. The labor module within the current questionnaire could also be reviewed for potential areas of improvement. x More detailed simulations of pension reform options and their fiscal implications. Key policy variables for the existing system include the monthly benefit, more institutionalized measures (e.g., indexation) to ensure an adequate old-age safety net over the medium to long-term, and potential differentiation of the benefit level according to different beneficiary characteristics (e.g., targeting of increases to the poor or oldest old). In addition, there is an important agenda for considering additional pension system pillars such as voluntary savings vehicles (including tax advantages) to encourage greater savings for old age among those who can afford to do so. 20 Annex 1. Georgia’s pension system and labor force participation 47. This technical annex provides further details on labor force participation as discussed in Section 4.1. The labor force participation of younger household members may be affected by the pension eligibility of older members, due to the disincentive effect of receiving a monthly transfer. Pension eligibility is strictly age-based: 60 for women and 65 for men. 48. The analysis used a probit model to predict the labor force participation of individuals aged 18 to 49 as a function of a range of household and individual characteristics. These included 10 dummy variables for the presence of older household members. The age groups were: 50-54, 55-59, 60-64, 65-69, and 70+ and over, for men and women separately. Other controls, listed in the order they appear in Table A1 below, include children under 5 in the household, school aged children, gender, a quartic for age, urban/rural, the presence of a chronically-ill household member over 50, family size, educational attainment of the potential worker, and household income exclusive of that individual’s earnings (if any). Panel HBS data over the four quarters of 2009 was used. Within-household clustering was addressed in the estimated model. Table A1: Regression results ------------------------------------------------------------------------------ | Robust aqt | dF/dx Std. Err. z P>|z| x-bar [ 95% C.I. ] ---------+-------------------------------------------------------------------- old50m~h*| -.0151399 .0099313 -1.54 0.123 .123893 -.034605 .004325 old55m~h*| -.0263205 .0121098 -2.22 0.026 .088347 -.050055 -.002586 old60m~h*| -.0400520 .0147378 -2.81 0.005 .057112 -.068938 -.011166 old65m~h*| -.0286243 .0167067 -1.76 0.079 .039772 -.061369 .004120 old70m~h*| -.0301054 .0112023 -2.75 0.006 .097018 -.052061 -.008149 old50f~h*| -.0127452 .0112740 -1.14 0.253 .106561 -.034842 .009351 old55f~h*| -.0148620 .0123191 -1.22 0.222 .09706 -.039007 .009283 old60f~h*| -.0456372 .0149717 -3.16 0.002 .071729 -.074981 -.016293 old65f~h*| -.0310280 .0154523 -2.06 0.039 .060548 -.061314 -.000742 old70f~h*| -.0376603 .0102187 -3.78 0.000 .165853 -.057689 -.017632 under5hh | -.0820275 .0061837 -13.27 0.000 .251052 -.094147 -.069908 school~h | -.0114994 .0046938 -2.45 0.014 .658501 -.020699 -.002300 gender | .2598792 .0065487 38.28 0.000 1.4782 .247044 .272714 age | .9640487 .0900621 10.73 0.000 33.5018 .787530 1.140570 agesq | -.0398050 .0043079 -9.25 0.000 1215.64 -.048248 -.031362 age3 | .0007241 .0000886 8.18 0.000 46998.2 .000551 .000898 age4 | -4.88e-06 6.62e-07 -7.37 0.000 1.9e+06 -6.2e-06 -3.6e-06 urban | .1228742 .0070924 17.62 0.000 1.45304 .108973 .136775 chroni~s*| .0146314 .0078423 1.85 0.064 .268453 -.000739 .030002 familysi | .0184373 .0034186 5.45 0.000 4.13904 .011737 .025138 educ*| .1647995 .0071264 22.52 0.000 .492377 .150832 .178767 shemos~t | -.0000642 .0000108 -5.93 0.000 372.271 -.000085 -.000043 ---------+-------------------------------------------------------------------- obs. P | .7184591 pred. P | .7551334 (at x-bar) ------------------------------------------------------------------------------ (*) dF/dx is for discrete change of dummy variable from 0 to 1 z and P>|z| correspond to the test of the underlying coefficient being 0 49. The results show a significant negative impact on labor force participation when the elderly pensioner is female, but the results are less clear when the pensioner is male. Specifically, there is a significant difference between the coefficients for females 55-59 and 60- 64, indicating a discontinuity at the pension eligibility threshold, but not for men 60-64 compared to 65-69. The results are sensitive to the exact model specification, because this 21 significant difference for female pensioners is not always present with alternative left-hand side variables (e.g., wage employment or self-employment), or sub-groups of the 18-49 year old working-age population. The analysis is intended to be suggestive of a potential issue more commonly associated with cash transfer programs for the poor, but which may nonetheless apply to Georgia’s pension system, particularly because monthly benefits are expected to rise steeply in the near future. 22 VI. Bibliography Almond, Douglas, and Janet Currie (2010). “Human Capital Development before Age Five”. NBER Working Paper 15827. Bertrand, Marianne, et al. (2003). “Public Policy and Extended Families”. World Bank Economic Review 17(1): 27-50. International Labour Organization (ILO) (2009). Economically Active Population Estimates and Projections (5th edition, revision 2009). http://laborsta.ilo.org/applv8/data/EAPEP/eapep_E.html (retrieved October 19, 2010). Meslé, France, Jacques Vallin, and Irina Badurashvili (2007). “A Sharp Increase in Sex Ratio at Birth in the Caucasus. Why? How?” In Isabelle Attané and Christophe Z. Guilmoto (eds.), Watering the Neighbour’s Garden: The Growing Demographic Female Deficit in Asia (pp. 73-88). Paris: Committee for International Cooperation in National Research in Demography. Rajaratnam, Julie Knoll, Jake R Marcus, Alison Levin-Rector, Andrew N Chalupka, Haidong Wang, Laura Dwyer, Megan Costa, Alan D Lopez, Christopher JL Murray (2010). “Worldwide mortality in men and women aged 15-59 years from 1970 to 2010: A systematic analysis.” Lancet 375 (9727): 1704-1720. UN (2008). World Population Prospects: The 2008 Revision. Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat. http://esa.un.org/unpp (retrieved October 18, 2010). UNFPA and Ilia State University (2010). “Demographic Yearbook of Georgia”. Wei, Shang-Jin. and Xiaobo Zhang (2009). “The Competitive Saving Motive: Evidence from Rising Sex Ratios and Savings Rates in China.” NBER Working Paper 15093. World Bank (2007). From Red to Gray: The “Third Transition” of Aging Populations in Eastern Europe and the former Soviet Union. Washington, DC: The World Bank. World Bank and EBRD (2007). Life in Transition Survey data. http://www.ebrd.com/pages/research/analysis/publications/transition/data.shtml 23