Child Poverty in Kyrgyzstan: Analysis of the 2008 Household Budget Survey

Transcription

1 Child Poverty in Kyrgyzstan: Analysis of the 2008 Household Budget Survey A Report for UNICEF Yekaterina Chzhen January 2010 Working Paper No. EC 2410

2 Acknowledgements Many thanks to Professor Jonathan Bradshaw for his helpful comments and suggestions. Author s contact details Department of Social Policy and Social Work, University of York. Heslington York YO10 5DD, UK. yc535@york.ac.uk. 1

3 1. Introduction This paper assesses the incidence and patterns of child poverty in Kyrgyzstan based on the 2008 Household Budget Survey (HBS). Since no single measure captures child poverty sufficiently well, this paper analyses consumption-based poverty, material deprivation and housing deprivation as well as the overlaps between these measures. A child under 18 is treated as the unit of analysis. 1 Children in Kyrgyzstan are among the groups most vulnerable to poverty. Child poverty rates are higher for children than for the population as a whole, particularly when the total poverty line is used. Children in larger families and younger children are at a particularly high risk of poverty. Thirty-six per cent of all children are classed as poor using the total poverty line, while seven per cent of children live in families whose consumption falls below the extreme (food) poverty line. Children are more likely to be poor, everything else held equal, if they have two or more siblings, if the youngest child in the household is under 6 years old, if they live in families headed by persons without secondary education, if they live in female headed households and if they live in rural areas. There are also substantial regional differences in child poverty rates. Poor children are significantly more likely to live in materially deprived households and to live in poor housing conditions, such as overcrowding and lacking important housing amenities. This paper analyses the poverty profile and living conditions of children in Kyrgyzstan. It presents the consumption-based poverty rates for children and the households characteristics that are associated with a higher risk of poverty (Section 2); the material deprivation rates, based on household ownership/lack of durable goods (Section 3); the housing deprivation rates, based on household ownership/lack of housing amenities and the number of rooms per person (Section 4); the overlaps in various deprivation measures and children s household characteristics associated with an increased likelihood of deprivation on each of these measures (Section 5); and the role of social security benefits, such as old age pensions and targeted social assistance, in alleviating child poverty (Section 6). 1 The household level HBS dataset was used and a child weight was constructed as the product of the household weight and the number of children under 18 in the household to approximate the population of children in Kyrgyzstan. The population weight is the product of the household weight and the number of household members. All estimates are weighted using the derived child weights, with the exception of population and household poverty rates in Table 1. 2

4 2. Consumption-based child poverty This paper uses consumption per capita as a basis for poverty measurement. Total monthly household consumption includes expenditure on food and relevant non-food items, as well as the value of food produced for own consumption. Two poverty lines are used: the extreme (food) poverty line of 11,710 som per year per capita and the total poverty line of 18,310 som per year per capita. Seven per cent of children live below the extreme poverty line and 36 per cent below the total poverty line, based on the HBS 2008 data. The poverty rates are four per cent and 23 per cent for all households and six per cent and 29 per cent for all individuals, respectively (Table 1). Thus, children are at a higher risk of total poverty than the population as a whole. Table 1 Poverty rates at different thresholds (%) Threshold All individuals All households Extreme (food) poverty line (11,710 som per year per) Total poverty line (18,310 som per year per capita) All children under Source: Author s estimates from HBS2008 data. Population weights are used to estimate the poverty rates for all individuals, household weights for all households and child weights for all children. Average poverty rates mask substantial variation in exposure to poverty and adverse living conditions by household characteristics. Table 2a presents the results of a descriptive analysis of child poverty based on the HBS 2008 data using the extreme and total poverty lines. Extreme poverty rates, total poverty rates, average poverty gaps 2 calculated using total poverty threshold, composition of total child poverty and composition of all children are tabulated by the relevant demographic household characteristics. Child poverty rates vary significantly with the number of children in the household, the age group of the youngest child, number of working age and pension age adults in the household, as well as the characteristics of the household head, such as gender, the highest level of education and marital status. Unfortunately there is no information on employment status of household members in the HBS. There is also significant variation by region and whether the area is rural or 2 The poverty gap shows how far a particular group is from the poverty line, on average. It is calculated as the poverty line minus the total consumption per adult equivalent divided by the poverty line, for those below the poverty line only. 3

5 urban. The mean (total) poverty gap is 24 per cent, which is the average percentage by which the consumption of those classed as poor would have to be increased in order to reach the total poverty line. Children in larger families are more likely to be poor. Children with two or more siblings are at the highest risk of poverty using both extreme and total poverty thresholds. Thus, 46 per cent of children in families with three or more children under 18 are poor, compared with 36 per cent of all children, using the total poverty line. At the same time, ten per cent of children with two or more siblings live in households below the extreme poverty line, compared with four per cent of children with one sibling and three per cent of sole children. More than one-half of all children (56 per cent) live in households with three or more children under 18. Younger children are more likely to be poor. Children in families where the youngest child is five years old or younger are at the highest risk of poverty. Thus, 43 per cent of children in such families are poor, compared with 18 per cent of children in families where the youngest child is years old. Similarly, ten per cent of children in families with the youngest child aged 5 or younger are extremely poor, compared with three per cent of children in families where the youngest child is at least 6 years old. More than one-half of all children (59 per cent) live in families where the youngest child is under 6 years old, however. Children in households with four or more working age adults (19-59) are the most likely to be extremely poor. Thus, 13 per cent of children in households with four or more adults are extremely poor, compared with just two per cent of children in families with no adults or one adult only. However, a different pattern emerges when the total poverty line is used: children in families with two or three adults are the most likely to be poor. Having one pension age adult (60 and over) in the household reduces the risk of extreme poverty but increases the risk of total poverty. Thus, five per cent of children with one pensioner in the household are below the extreme poverty line, compared with 15 per cent of children with two or more pensioners and eight per cent of children with no pensioners. At the same time, 39 per cent of children with one pensioner in the household are poor, compared with 35 per cent of children with no pensioners or two or more pensioners. This suggests that pension income may play a role in lifting households with children above the extreme poverty line, but not above the total poverty line. Since the consumption measure is per capita, having more than on pensioner may reduce the value of having extra pension income. 4

6 Children in female headed households are more likely to be below the extreme poverty line, but less likely to fall below the total poverty line. Thus, nine per cent of children in female headed households are extremely poor, compared with seven per cent of children in male headed households. While 36 per cent of children in male headed households fall below the total poverty line, 35 per cent of those in female headed households are poor. However, less than one-third (28 per cent) of all children live in female-headed households. Marital status of the household head is an important predictor of child poverty. Children in households with a single (never married), widowed or divorced head are more likely to be extremely poor (8 per cent) than those in households with married or cohabiting heads (7 per cent). However, they are less likely to fall below the total poverty line. The vast majority of children (79 per cent) live in households headed by married or cohabiting persons. Living in a household with a more educated head reduces the risk of poverty. Children living in households where the household head has not completed secondary education (35 per cent), has secondary education only (40 per cent) or vocational education only (39 per cent) are substantially more likely to be poor than those in households where the head is a university graduate or has incomplete university education (22 per cent), using the total poverty threshold. Children in households where the head has no secondary education are the most likely to live in extreme poverty (11 per cent). The majority of all children (70 per cent) live in households where the head has secondary or further education. Children in rural areas are about twice as likely to be poor, irrespective of the poverty line. While four per cent of urban children fall below the extreme poverty line, nine per cent of rural children are similarly poor. At the same time, a quarter (24 per cent) of urban children lives below the total poverty line, compared with 41 per cent of rural children. It is a worrying finding, since two-thirds of children (68 per cent) live in rural areas. 5

7 Table 2a Poverty rates, gaps and composition by type of household Child poverty rate (extreme) (1) Child poverty rate (total) Average (total) poverty gap (3) Poverty composition (total) Composition of all children (2) (4) (5) Number of children under 19 One 2.5* 13.2* Two 4.4* 27.5* Three or more 10.2* 45.7* Age of the youngest child * 43.2* * 28.7* * 18.3* Number of adults (aged 19-59) None/one 1.8* 26.7* Two 7.1* 39.5* Three 9.2* 38.3* Four or more 13.4* 26.2* Number of pension age household members (aged 60+) None 7.5* 35.3* One 4.7* 38.7* Two or more 14.8* 34.7* [29.9] Gender of head of household Male 6.9* 36.0* Female 8.7* 35.3* Marital status of head Never married/widowed/divorced 8.0* 34.8* Married/cohabiting 7.3* 36.1* Highest level of education of household head Below secondary 11.3* 35.4* Secondary 9.9* 40.0* Vocational / further 5.5* 38.6* Incomplete higher/ higher 2.5* 21.6* Area Urban 3.9* 24.4* Rural 9.1* 41.1* All (Unweighted N= 3,057) [ ] weighted proportions are based on fewer than 50 unweighted cases. Child weights are used. Statistical significance: *=p<0.001 (separate cross-tabulations with chi-square tests). Child poverty rates vary substantially across seven regions of the Republic of Kyrgyzstan and Bishkek city. Table 2b shows a descriptive 6

8 analysis of child poverty across the regions. The differences by region are significant at p<0.001 using both extreme and total thresholds. Extreme child poverty rates range from the low of two per cent in Bishkek to the high of 19 per cent in Issyk-Kol. Total child poverty rates range from 13 per cent in Bishkek to 55 per cent in Talas. However, children in Issyk-Kol live in families furthest from the total poverty line, with the poverty gap of 28 per cent. The largest number of poor children live in Osh (32 per cent), although on average this region has middling poverty rates. Table 2b Poverty rates, gaps and composition by regions Child poverty rate (extreme) (1) Child poverty rate (total) (2) Average (total) poverty gap (3) Poverty composition Composition of all children (4) (5) Issyk-Kol Zhalal-Abad Naryn Batken Osh Talas Chui Bishkek [18.8] All (Unweighted N= 3,057) Child weights are used. [ ] weighted proportions are based on fewer than 50 unweighted cases. Table 3 shows the estimated odds of being in poverty for each of the household characteristics analysed above, holding other characteristics constant. The results confirm the findings from the descriptive analyses above. Children with two or more siblings and children in families where the youngest child is under 6 are the most likely to be poor, everything else held equal. Thus, sole children are only 20 per cent as likely to be poor as children with two or more siblings, using either poverty line. Children in families with four or more adults are the most likely to fall below the extreme poverty line, while children with three co-resident adults are the most likely to fall below the total poverty line. Children with two or more pensioners in the household are the most likely to fall below the extreme poverty line, while those with no coresident pensioners are the most likely to fall below the total poverty line. Characteristics of the household head are important predictors of child poverty. Children in households with a female head are 4.4 times more likely to be extremely poor and 81 per cent more likely to be poor than children in 7

9 male headed households. Interestingly, children in households with married heads are somewhat more likely to be poor than those in households with single, divorced or widowed heads, everything else held equal. Children in households with university educated adults are only 31 per cent and 59 per cent as likely to be poor using the extreme and total poverty lines, respectively, as children in households where the head has secondary education only. Regional differences persist after controlling for household characteristics. Rural children are 67 per cent more likely to fall below the extreme poverty line and 31 per cent more likely to fall below the total poverty line, everything else held equal. Children in the region of Issyk-Kol are more than seven times as likely to fall below the extreme poverty line and two and a half times as likely to fall below the total poverty line as children in Osh. Children in the capital and in the region of Chui are the least likely to be poor, irrespective of the poverty line used. Table 3 Odds of being consumption poor Child poverty rate (extreme) Child poverty rate (total) Number of children under 18 (ref: three or more) One 0.21* 0.22* Two 0.31* 0.53* Age of the youngest child (ref: 0-5) * 0.55* * 0.55* Number of adults (ref: two) None/one 0.11* 0.44* Three 1.54* 1.15* Four or more 3.00* 0.48* Number of pensioners 60+ (ref: none) One 0.19* 0.92* Two or more 1.44* 0.83* Female head of household 4.41* 1.81* Married/cohabiting 1.49* 1.07* Highest level of education of household head (ref: secondary) Below secondary 1.50* 0.94* Vocational / further 0.54* 1.16* Incomplete higher/higher 0.31* 0.59* Rural area 1.67* 1.31* Region (ref: Osh) Issyk-Kol 7.35* 2.47* Zhalal-Abad 2.04* 1.26* Naryn 2.85* 1.40* 8

10 Batken 2.21* 0.50* Talas 1.82* 1.99* Chui 0.36* 0.49* Bishkek 0.83* 0.33* Constant 0.05* 0.89* Pseudo R-square Statistical significance: *=p< To summarise, having controlled for other household characteristics to eliminate any spurious associations, children are most likely to be living in poverty if There are three or more children in the household The youngest child is under six years old There are three or more working age adults in the household The household head is female The household head is married or cohabiting They live in Issyk-Kol region. However, poverty is not limited to these most vulnerable children. The majority of poor children live in households that do not appear to be at the highest risk of poverty and Live in families with two working age adults Live in families with no pension age adults Have a male head of household. 3. Material deprivation To complement the consumption-based poverty analysis, this section analyses the material deprivation of children in Kyrgyzstan. It is measured as households lack of durable assets using a simple count index and a prevalence weighted index. The following seven durable goods have been included in the analysis: colour TV, refrigerator, washing machine, stereo, mobile telephone, car or van and a vacuum cleaner. These items are chosen because at least ten per cent of all households in HBS 2008 report owning them. However, it is not clear whether the households that lack these items cannot afford them or choose not to own them. Table 4 shows the proportion of children living in households lacking each of these items and Table 5 shows the proportions of children lacking a number of these items. Poor children are substantially more likely to live in households lacking each of these durable goods than all children. Children in extremely poor households are the most likely to lack each of these items. For example, while 9

11 39 per cent of all children live in households without a colour TV, over one-half (54 per cent) of poor and two-thirds (66 per cent) of extremely poor children live in households lacking this item. More than 90 per cent of poor and extremely poor children live in households without a mobile phone, a car or a vacuum cleaner. Table 4 Durable goods lacked (%) All children Poor children (Extremely) poor children Colour TV Refrigerator Washing machine Stereo Mobile phone Car/van Vacuum cleaner There are noticeable differences in deprivation rates between all children and poor children. Poor children are more likely to live in households lacking more durable goods than children overall. Around one per cent of all children live in households not lacking any of these durable goods, compared with only 0.2 per cent of poor children (Table 5, Figure 1). None of the extremely poor children lives in the households lacking fewer than two of the studied items and less than one per cent lack only two items. Poor and extremely poor children are substantially more likely to live in families lacking all seven durable assets, 29 per cent and 37 per cent, respectively, than all children (18 per cent). To achieve a housing deprivation rate that is comparable with the total consumption child poverty rate for 2008 (36 per cent), the deprivation threshold is drawn at lacking six or seven items. This results in 34 per cent of all children, 49 per cent of poor children and 65 per cent of extremely poor children experiencing material deprivation. 10

12 Table 5 Number of durable goods lacked (%) Number of durable goods lacked All children Poor children (Extremely) poor children Figure1: Number of durable goods lacked (%) An obvious problem with this methodology is that the items included in the simple count index may not be of equal importance to the households wellbeing, but the HBS provides no information about the desirability or importance of these durable goods. Furthermore, there is no information on whether the item is lacked because the household cannot afford it or because it is not wanted. Using the prevalence weighted deprivation index helps overcome this drawback at least in part because it is based on the assumption that households are relatively more deprived if they lack an item that most other household have. For example, lacking a TV carries more weight than lacking a car because more households have a TV than a car. Each score of 1 (item lacked) is multiplied by the proportion of children in the weighted sample who live in households owning this item. The scores are then summed across all items and divided by the total number of items, i.e. seven items, for 11

13 each household. The resulting score is multiplied by 100 to create a continuous variable that ranges from 0 (not lacking any items) to 100 (lacking all items that everybody else owns). Poor children have a higher prevalence weighted deprivation score, on average. While the mean score for all children is 19.4, it is substantially higher at 23.6 and 25.7 for poor and extremely poor children, respectively (Table 6). This suggests that poor children live in households lacking more of the items that other households tend to own. Table 6 Average prevalence weighted deprivation score and deprivation rates All children Poor children (Extremely) poor children Mean Standard Deviation Housing deprivation Housing problems can have an adverse impact on children s health, safety, education and social development. The HBS 2008 includes questions about housing, such as the number and condition of amenities and rooms. Poor children often live in accommodation lacking important amenities. Children in poor households are consistently more likely to live in dwellings without each of the housing facilities analysed 3 : electricity, telephone, central heating, sewerage, mains water, bathtub or shower, and hot running water (Table 7). However, children in extremely poor households are not necessarily more likely to lack these housing amenities than all poor children, which points to the potential unreliability of the extreme poverty measure. For example, extremely poor children are less likely to live in households without electricity, a landline telephone or mains water supply than all poor children. Yet, at the same time, extremely poor children are the most likely to live in dwellings without sewerage, bathtub or shower, central heating or hot running water. More than 95 per cent of extremely poor children lack each of these four amenities. At the same time, with the exception of electricity, which is available to almost all households, at least 70 per cent of all children lack each of the remaining six amenities. 3 The amenity is either not available or not in working condition. 12

14 Table 7 Housing amenities lacked or not in working order (%) Dwelling lacks All children Poor children (Extremely) poor children Electricity Landline phone Mains water supply Sewerage Bathtub or shower Central heating Hot running water Poor children are more likely to lack more of the housing amenities than all children. Although almost no children live in households not lacking any of these amenities, seven per cent of all children live in households lacking only one, compared with two per cent of poor children and 0.1 per cent of extremely poor children (Table 8; Figure 2). To achieve a housing deprivation rate that is comparable with the total consumption child poverty rate for 2008 (36 per cent), the deprivation threshold is drawn at lacking six or more amenities. This definition results in 65 per cent of all children experiencing housing deprivation. The corresponding rates for all poor and extremely poor children are noticeably higher at 70 per cent and 63 per cent, respectively. These rates are substantially higher than the estimated consumption poverty and material deprivation rates, since the majority of all children live in households lacking six essential amenities. Table 8 Number of housing amenities lacked or not in working order (%) All children Poor children (Extremely) poor children

15 Figure 2 Number of housing amenities lacked (%) Poor children are more likely to live in overcrowded accommodation. The average number of rooms per person in the primary dwelling is higher for all children (0.70) than for poor children (0.63) or extremely poor children (0.54). The differences in means between poor and non-poor children are statistically significant at p< If the threshold is drawn at 0.5 or fewer rooms per person, the overcrowding rate for all children is 37 per cent, compared with 42 per cent for all poor children and 63 per cent for extremely poor children (Table 12). Table 12 Average number of rooms per person and overcrowding rates All children Poor children (Extremely) poor children Mean (SD) 0.70 (0.34) 0.63 (0.24) 0.54 (0.24) Overcrowding rate (%) Overlaps in poverty indicators Table 14 summarises the rates of poverty based on the measures analysed in the previous sections. 14

16 Table 14 Percentage of children poor by each indicator (2008) % children Consumption poor (extreme poverty line) 7.4 Consumption poor (total poverty line) 35.8 Materially deprived (based on durable goods lacked) 34.4 Housing deprived (based on amenities lacked) 64.6 Living in overcrowded accommodation 37.3 Excluding the extreme poverty measure, 86 per cent of children are poor on at least one of the four indicators: total consumption poverty, material deprivation, housing deprivation, and overcrowding. More than one-half (56 per cent) of all children are deprived on at least two and onequarter (25 per cent) are poor on at least three indicators, while seven per cent are poor on all four measures (Table 15). To achieve a composite deprivation rate that is comparable with the total consumption child poverty rate of 36 per cent, children living in households deprived on at least three out of four indicators can be defined as deprived. This composite measure is, therefore, based not only on consumption poverty, but also on material and housing deprivation as well as overcrowding. Table 15 Proportion of children poor or deprived (six indicators) % children No ways 13.6 At least one 86.4 At least two 56.1 At least three 24.6 All four ways 7.1 There is also a substantial degree of overlap between total consumption poverty and the measures of deprivation. A significantly higher proportion of poor than non-poor children are deprived on each of the studied indicators (Table 16). For example, 49 per cent of poor children are also materially deprived, compared with only 28 per cent of non-poor children. 15

17 Table 16 Overlap between total poverty and deprivation (column %) Not poor Poor Materially deprived (based on durable goods lacked) Housing deprived (based on amenities lacked) Living in overcrowded accommodation All associations are statistically significant at p< There is a considerable degree of overlap among all four indicators by household characteristics. Children with more siblings, those in families with children under 6 years old, those in families with lower educated household heads and those in rural areas are the most likely to be poor on each of the indicators: total consumption poverty, material deprivation, housing deprivation and overcrowding (Table 17a). Other household characteristics make children vulnerable to some kinds of poverty or deprivation but not to others. For example, children in households with two or three working age adults are the most likely to be consumption poor and materially deprived, but somewhat less likely to be housing deprived or live in overcrowded accommodation than other children. Similarly, children in households with one pension age adult are the most likely to be consumption poor and materially deprived, but it is children with two or more pensioners who are the most likely to be housing deprived or live in overcrowding conditions. This suggests that larger households tend to have worse housing conditions, on average. Furthermore, children in female headed households are more likely to be materially deprived but less likely to be poor on any of the other three measures than children in male headed households. At the same time, there are no significant differences by marital status of the household head with regards to material deprivation, while children in households where the head is married or cohabiting are more likely to be poor on any of the other three measures. However, some of these associations may be spurious because other important household characteristics are not controlled for. 16

18 Table 17a Poverty and deprivation rates by household characteristics Child poverty rate (total) Material deprivation Housing deprivation Overcrowding Three out of four ways Number of children under 19 One 13.2* 19.8* 56.6* 11.6* 3.9* Two 27.5* 32.9* 52.0* 33.4* 18.1* Three or more 45.7* 39.4* 73.3* 46.9* 33.2* Age of the youngest child * 35.2* 64.0* 42.2* 29.4* * 37.6* 69.3* 33.7* 18.2* * 25.9* 59.0* 24.7* 16.6* Number of adults (aged 19-59) None/one 26.7* 40.9* 54.2* 23.4* 21.8* Two 39.5* 33.9* 67.5* 37.3* 26.5* Three 38.3* 33.8* 61.5* 35.2* 21.7* Four or more 26.2* 28.6* 68.3* 58.5* 22.6* Number of pension age household members (aged 60+) None 35.3* 34.4* 63.5* 36.2* 23.1* One 38.7* 35.1* 71.3* 36.7* 31.9* Two or more 34.7* 30.0* 63.0* 63.0* 27.4* Gender of head of household Male 36.0* 33.8* 69.0* 39.2* 25.3* Female 35.3* 35.8* 53.4* 32.5* 22.7* Marital status of head Never married/widowed/divorced 34.8* * 35.1* 23.8* Married/cohabiting 36.1* * 37.9* 24.8* Highest level of education of household head Below secondary 35.4* 40.5* 73.2* 49.1* 35.6* Secondary 40.0* 47.8* 74.9* 33.6* 34.7* Vocational / further 38.6* 27.2* 62.4* 35.9* 18.1* Incomplete higher/ higher 21.6* 11.5* 39.3* 40.4* 5.6* Area Urban 24.4* 19.8* 33.4* 43.2* 12.7* Rural 41.1* 41.3* 79.6* 34.5* 30.2* All (Unweighted N= 3,057) * 34.4* 64.6* 37.3* Statistical significance: *=p<0.001 (separate cross-tabulations with chi-square tests). There are also significant regional differences in poverty and deprivation rates. Children in Talas are the most likely to be consumption poor, while children in Batken, Osh and Talas are the most likely to be materially deprived and housing deprived (Table 17b). Children in Bishkek are the least likely to be consumption poor and housing deprived, but they are the most likely to live 17

19 in overcrowded accommodation. Children in Bishkek and in the Chui region are the least likely to be materially deprived. Table 17b Poverty and deprivation rates by region Child poverty rate (total) Three out of four ways Material deprivation Housing deprivation Overcrowding Issyk-Kol Zhalal-Abad Naryn Batken Osh Talas Chui Bishkek All (Unweighted N= 3,057) All chi-square tests are statistically significant at p< Child weights are used. Table 18 shows the estimated odds of being poor on each of the deprivation indicators and on the composite measure of poverty/deprivation for each of the household characteristics, holding other characteristics constant. It is important to control for a number of factors to eliminate spurious associations. The household and regional characteristics included in the separate logistic models reported in Table 18 do a better job explaining the variation in household deprivation rates based on the number of amenities lacked 4 than predicting the odds of being deprived on each of the other indicators or the odds of being deprived on the composite deprivation index (being deprived on three out of four items). The number of children and age of the youngest in the household are crucial predictor of deprivation. Confirming the results of descriptive analyses in Table 17a, children with two or more siblings are the most likely to be consumption poor, housing deprived, to live in overcrowded accommodation or to be poor on the composite indicator. However, children with just one sibling are the most likely to be materially deprived, everything else held equal. Children in families where the youngest children is under 6 years old are the most likely to be consumption poor or to live in overcrowded accommodation, but children in families where the youngest child is 6-11 years old are the most likely to be materially deprived and housing deprived, 4 R-square=

20 while those in families with the youngest child aged are the most likely to be poor on the composite indicator. Numbers of working age and pension age adults in the household make an important difference to deprivation rates, even after controlling for other factors. Children with three working age adults in the household are the most likely to be consumption poor and to live in overcrowded accommodation, while those with just one adult are the most likely to be materially deprived and to be poor in three out of four ways. Children with two adults in the household are the most likely to be housing deprived. At the same time, children with no pension age adults in the household are the most likely to be consumption poor, materially and housing deprived and to be poor on the composite indicator, while those with two or more pensioners are the most likely to live in overcrowded accommodation. This suggests that pension makes a positive contribution to the living standard of the household, but having pensioners in the household results in fewer rooms per person. Characteristics of the household head are also important predictors of child poverty/deprivation. Children in female headed household are 81 per cent more likely to be consumption poor, but those in male headed households are 23 per cent more likely to be materially deprived, 32 per cent more likely to be housing deprived, 12 per cent more likely to live in overcrowded accommodation and 39 per cent more likely to be poor on the composite measure. Children with married or cohabiting household heads are more likely to be consumption poor or housing deprived, but less likely to be materially deprived, to live in overcrowded accommodation or to be poor on the composite measure. Children with university educated household heads are the least likely to be consumption poor, materially or housing deprived, or to be poor on the composite measure, but it is children with heads educated to secondary level only who are the least likely to live in overcrowded accommodation. Regional differences in child deprivation rates remain even after household characteristics are controlled for. Rural children remain at the highest risk of poverty and deprivation, regardless of the measure used. The differences are the most striking with regards to housing deprivation: rural children are 6 times more likely to lack important housing amenities than urban children. Overall, children in the Issyk-Kol region are the most likely to be consumption poor, those in the region of Osh are the most likely to be materially deprived, those in Batken are the most likely to be housing deprived, while children in Naryn and in Bishkek are the most likely to live in overcrowded accommodation, everything else held equal. Children in Naryn are also the most likely to be poor on three out of four indicators. 19

21 Table 18 Odds of being deprived Child poverty rate (total) Material deprivation Housing deprivation Overcrowding Three out of four ways Number of children under 18 (ref: three or more) One 0.22* 0.58* 0.96* 0.09* 0.09* Two 0.53* 1.28* 0.64* 0.37* 0.59* Age of the youngest child (ref: 0-5) * 1.15* 2.04* 0.75* 0.58* * 0.87* 1.40* 0.91* 1.07* Number of adults (ref: two) None/one 0.44* 1.38* 0.55* 0.56* 1.05* Three 1.15* 1.03* 0.66* 1.27* 1.00 Four or more 0.48* 0.70* 0.55* 3.56* 0.66* Number of pensioners 60+ (ref: none) One 0.92* 0.54* * 0.86* Two or more 0.83* 0.47* 0.41* 3.51* 0.74* Female head of household 1.81* 0.81* 0.76* 0.89* 0.72* Married/cohabiting 1.07* 0.90* 2.16* 0.60* 0.85* Highest level of education of household head (ref: secondary) Below secondary 0.94* 1.22* 1.23* 1.77* 1.45* Vocational / further 1.16* 0.41* 0.52* 1.14* 0.45* Incomplete higher/higher 0.59* 0.16* 0.25* 1.21* 0.14* Rural area 1.31* 2.21* 6.22* 0.55* 1.66* Region (ref: Osh) Issyk-Kol 2.47* 0.33* 0.26* 1.93* 0.88* Zhalal-Abad 1.26* 0.51* 0.29* 0.43* 0.32* Naryn 1.40* 0.50* 0.83* 2.06* 1.22* Batken 0.50* 0.95* 1.79* 0.81* 0.58* Talas 1.99* 0.84* 1.13* 0.64* 1.28* Chui 0.49* 0.06* 0.12* 1.69* 0.18* Bishkek 0.33* 0.26* 0.14* 2.05* 0.26* Constant 0.89* 1.14* 1.59* 1.93* 1.15* Pseudo R-square Statistical significance: *=p< Role of social protection benefits in poverty alleviation Old age pensions Old-age pensions make a small difference to average child poverty rates. 22 per cent of all children live in households where at least one person is reportedly in receipt of an old-age pension. Table 19 shows what difference 20

22 pensions make to average consumption-based child poverty rates. If pensions are deducted from total yearly household expenditure, which is then divided by household size, the extreme child poverty rate would increase from seven per cent to ten per cent, while the total child poverty rate would go up from 36 per cent to 45 per cent. Thus, pension income makes some difference to average child poverty rates. Of course, this analysis, as well as the analyses below, assumes that all of the pension income is consumed by the household. Table 19 Child poverty rates with and without old-age pension income Threshold Child poverty rate With pensions Without pensions Extreme poverty line Total poverty line Old-age pension income can make a difference to whether a child is poor or not. Table 20 shows what difference old-age pensions can make to children in poor (old-age pension recipient) households. If pensions were deducted from their total household consumption, 19 per cent of children who are currently not poor based on the extreme poverty line would have been classed as poor. At the same time, 47 per cent of children who are currently not poor based on the total poverty line would have been classed as poor if pension income were deducted from their household consumption. This suggests that pension income makes an important difference to about onehalf of the children living in pension-recipient households. Table 20 Poverty rates with and without old-age pension income for those in old-age pension recipient households Below extreme poverty line (without pensions) Below total poverty line (without pensions) Lifted above extreme poverty line (with pensions) 18.5 Lifted above total poverty line (with pensions)

23 Targeted social assistance Targeted social assistance (TSA) benefit income makes a small difference to average child poverty rates. Around 15 per cent of all children live in households receiving targeted social assistance benefits. Table 21 shows what difference it makes to average child poverty rates if TSA income were deducted from total household consumption. While extreme child poverty rate would go up from seven per cent to ten per cent, total child poverty rate would go up from 36 per cent to 43 per cent. This suggests that TSA levels are too low and too few households receive the benefit for it to make a large difference to average child poverty rates. Table 21 Child poverty rates with and without TSA benefit income Threshold Child poverty rate With benefits Without benefits Extreme poverty line Total poverty line Family benefit income can make a difference to whether a child is poor or not. Table 23 shows the re-calculated poverty rates for children in TSA recipient households who are not currently poor. If TSA income were deducted from their household consumption, nine per cent of children who are currently not poor based on the extreme poverty line would have been classed as poor. At the same time, 19 per cent of children who are currently not poor based on the total poverty line would have been classed as poor if TSA income were deducted from their household consumption. This suggests that TSA income makes a difference to about one-fifth of children living in TSA recipient households. Table 22 Poverty rates with and without family benefit income for those in TSA benefit recipient households Lifted above extreme poverty line (with family benefit) Below extreme poverty line 9.1 (without family benefit) Below total poverty line (without family benefit) Lifted above total poverty line (with family benefit)

24 Conclusions Thirty-six per cent of children in Kyrgyzstan live in consumption poor households and seven per cent fall below the extreme (food) poverty line. Children are somewhat more likely to be poor than all households and the overall population. Poor children are more likely to live in households lacking important durable goods and to live in adverse housing conditions, such as the lack of essential housing amenities and overcrowding. Overall, holding other factors constant, the following household characteristics are associated with a higher risk of total consumptionbased child poverty: There are three or more children in the household The youngest child is under six years old There are three or more working age adults in the household The household head is female The household head is married or cohabiting They live in Issyk-Kol region. The following household characteristics are associated with a higher risk of being poor on three out of four poverty measures (consumption poverty, material deprivation, housing deprivation and overcrowding), everything else held equal: There are two or more children in the household The youngest child is aged or under 6 years old There is only one or no working age adults in the household There are no pension age (60 and over) adults in the household The household head is male The household head is not married/cohabiting The household head does not have secondary education They live in a rural area They live in the region of Naryn. Targeted social assistance appears to make little difference to average child poverty rates, but the introduction of a universal child benefit could help alleviate child poverty. Given that the majority of children are affected by at least one dimension of poverty or deprivation, benefits that are targeted to particularly vulnerable groups of the population may not reach all of the poor or deprived children. Universal child benefits are meant to be relatively easy to administer and could raise the living standards of all families with children. Table 23 shows the potential reduction in total child poverty for different (hypothetical) child benefit levels. This simple analysis is based on the assumption that all of the child benefit income would be spent by the household, thus entering the consumption-based child poverty estimation. A 23

25 child benefit equal to the food poverty line (11,710 som/year) per child under 18 would reduce the total child poverty rate by 70 per cent. A more modest child benefit of one-half of the food poverty line would reduce the child poverty rate by one-third. Table 23 Total child poverty rates with universal child benefit (drams per child per month) Threshold Total poverty rates CB=0 CB=Foo d poverty line/4 CB= Food poverty line /2 CB= Food poverty line CB= Food poverty line *1.25 CB= Food poverty line * Base: all children 24