Chapter 6 Education and Multidimensional Poverty at Household Level
6.4. The relationship between Education and Basic Needs and Capabilities
Poverty is a multidimensional phenomenon that extends well beyond its monetary aspect (Sen, 1981; 1985; Haq, 1995; World Bank, 2001; Stewart et al., 2007). In this section, basic needs and capabilities – as defined in Chapter 2 – are examined in terms of the variables shown in Table 6-8. The wide range of variables associated with the basic needs and capabilities of slum dwellers has been delineated as follows: (1) access to safe water; (2) access to sanitation, including latrines and drainage; (3) health condition; (4) access to electricity; (5) food security; (6) political rights; (7) right to housing and the condition thereof; (8) access to credit and financial services; and (9) overall development of the slum in which a household resides.
Table 6-8 Definitions of variables
Variable Definition Mean Std. Dev.
Water index 1 Availability of drinking water (hours) multiplied by household has private source = 2; household obtains drinking water from public source = 1
16.88 14.29
Water index 2 Degree of lack of sufficient drinking water (household
frequently lacks water = 1; household sometimes lacks water = 2; household never lacks water = 3) multiplied by household has private source = 2; household obtains drinking water from public source = 1
2.25 1.50
Toilet Household has private toilet = 1; household does not have private toilet = 0
0.08 0.28
Drainage Household has drainage = 1; household does not have drainage = 0
0.88 0.33
Institutional birth Proportion of household members born in any medical institution
0.11 0.20
Sickness Proportion of household members not debilitated by sickness for more than 7 days during previous 12 months
0.91 0.17
Health Proportion of household members reporting substantial improvement, improvement, or no change in health respectively in comparison to the previous 12 months
0.93 0.15
Medical insurance Household has medical insurance = 1; household does not have medical insurance = 0
0.02 0.13
Electricity No electricity = 0; household has access to neighbour's supply = 1; household has own illegal electricity connection = 2; household has own legal connection = 3
2.34 0.58
Ration card Household has ration card = 1; household has no ration card = 0
0.77 0.42
Voter ID Household has voter ID card = 1; household has no voter ID card = 0
0.80 0.40
Rented house Household rents house = 1; household does not rent house = 0 0.13 0.34
Token Household has token = 1; household does not have token = 0 0.41 0.49
Bank account Household has account at bank or post office = 1; household does not have account at bank or post office = 0
0.17 0.38
Slum development Unweighted sum of the following: paved roads (household lives in slum where all internal roads are paved = 1; not all internal roads are paved = 0) + street lighting (household lives in slum where at least one street light functions = 1; no streetlight functions = 0) + spraying (household lives in slum where vector-control spraying has been provided during previous 12 months = 1; no spraying during previous 12 months = 0) + refuse collection (household lives in slum where refuse is collected = 1; household lives in slum where refuse is not collected = 0) + electricity (household lives in slum with legal electricity connection = 1; household lives in slum with no legal electricity connection = 0) + mobile health clinic (household lives in slum where government or private mobile health clinic has been available during previous 12 months = 1; no such health clinic = 0) + decision-making body (household lives in slum where decision-making body has been organised = 1; no such body = 0)
4.01 1.71
integrated basic needs and capabilities index. Therefore, factor analysis – more specifically, maximum likelihood factor analysis – was conducted. In this process, some variables were discarded in order to avoid the manifestation of ‘Heywood cases’.60 Variables were thus combined to generate a composite index of basic needs and capabilities (abbreviated in equations as BN/C) as follows:
BN/C INDEX (𝑖) = � 𝐹𝐹𝐹 (𝑖)𝑋𝐹
𝑛 𝐹=1
Where FL is the factor loading j=1…n corresponding to the number of variables, and i represents the ith significant factor.
In the second stage, composite indices generated on the basis of the factor loading for each of the significant factors (Eigen value is more than 1) were combined using the proportion of Eigen values as weights.
) ( / ) ( ) ( / 1 i INDEX C BN i EV i EV INDEX C BN k i
∑ ∑
= = k < nWhere i ranges from 1 to k, the number of significant factors
Using varimax rotation in order to obtain statistically independent factors, the results of the analysis suggest four significant items (Table 6-9). These four factors together explain 87.9% of the variation. For factor 1 – the most dominant and that which explains 30.2% of the variation – possession of a ration card (0.92) and voter ID (0.82) have the highest loadings. Variables with moderate loadings for this factor include rented accommodation, which has a negative value (-0.58), and possession of a token61 (0.36).
60
A factor solution that produces an error variance estimate of less than zero (Hair et al., 2006).
61
The tokens distributed during Prime Minister V. P. Singh’s administration (1989–90) constitute formal proof of residence in a slum (Ghertner, 2010) – although they are sometimes traded on the black market – and confer the right to reside in the city, including the right to resettlement in the event of the demolition of the slum.
For factor 2, which explains 25.3% of the variation, there are two water access-related indicators: (1) the composite index of hours of water availability, and whether a household has access to safe drinking water from its own private source; and (2) degree of difficulty in obtaining water in the previous dry season. These two indices have the highest loadings (0.99 and 0.82 respectively).
For factor 3, there are two health-related variables: (1) the proportion of household members reporting either much better health, better health, or no change in comparison to 12 months ago and (2) the proportion of household members not debilitated by illness or injury for more than 7 consecutive days in the last 12 months. These two indices have the highest loadings (0.76 and 0.75 respectively).
For factor 4, plumbed drainage at home (0.71) and the slum development index (0.62) have the highest loadings; while an electricity supply via any kind of legal or illegal connection (0.31) has a modest loading.
Table 6-9 Factor analysis results
Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Uniqueness
Water index 1 -0.0159 0.8203 -0.0033 -0.0667 -0.0205 -0.2122 0.2770 Water index 2 0.0182 0.9857 0.0025 0.0182 -0.0456 0.1027 0.0152 Toilet 0.1680 0.1490 0.0257 -0.0028 -0.0085 0.0848 0.9416 Drainage 0.2267 0.0680 -0.0135 0.7087 -0.0276 0.0992 0.4309 Sickness 0.0364 0.0675 0.7546 -0.0061 -0.0256 0.0230 0.4235 Health -0.0173 -0.0628 0.7589 0.0753 0.0172 -0.0190 0.4136 Electricity 0.1382 -0.1351 0.0532 0.3114 0.3699 0.2794 0.6479 Ration card 0.9212 0.0339 0.0091 0.1345 0.0193 -0.0413 0.1299 Voter ID 0.8191 -0.0144 -0.0006 0.0831 0.1209 0.0680 0.3027 Rented house -0.5822 0.0282 -0.0293 0.2225 -0.2126 -0.0839 0.5577 Token 0.3554 -0.1547 -0.0373 0.1055 0.5959 -0.0108 0.4820 Bank account 0.1388 0.1335 -0.0281 0.0993 0.2396 -0.0477 0.8925 Slum development 0.0504 -0.1495 0.1434 0.6221 0.2238 -0.1312 0.5002 Eigenvalue 2.1076 1.7643 1.1727 1.0923 0.6640 0.1846 - Variance explanation 0.3017 0.2526 0.1679 0.1564 0.0950 0.0264 - Cumulative variance 0.3017 0.5543 0.7222 0.8785 0.9736 1.0000 -
Table 6-10 shows the distribution of the basic needs and capabilities index. Variation within the same range of indices is apparently great. In particular, the lower the index group, the higher the variation. This implies that some households, particularly those who only achieve a low score in the index, face very limited access to basic needs and capabilities. Approximately 6.2% of households correspond to the lowest value, while 12.4% have the highest composite value. The second lowest group represents the largest percentage of sample households (41.1%), followed by the middle group of values between 5 and 7.5 (27.0%). It is notable that the correlation ratio of this index to MPCE is only 0.05. Thus, the basic needs and capabilities index clearly exhibits a different aspect of deprivation from monetary poverty.
Table 6-10 Distribution of households in the basic needs and capabilities index
Index No. of households Percentage Coefficient of variation
≤ 2.5 25 6.19 0.1973 2.5 < index ≤ 5 166 41.09 0.1801 5 < index ≤ 7.5 109 26.98 0.1220 7.5 < index ≤ 10 54 13.37 0.0564 > 10 50 12.38 0.0571 Total 404 100 0.5740
This composite index shows that there is a non-linear correlation between average basic needs and capabilities, and different levels of education (Table 6-11). For example, the index for households with an average of more than 5 years of education (6.57) is close to that of those with an average of more than zero but less than 2.66 years (6.47) (Table 6-11-1). The index of those households in which no one has completed primary school (6.06) is slightly greater than that of the highest proportion of household members who have completed at least this level of education (6.04) (Table 6-11-4). Similarly, even if the household head or any other family member is educated to tertiary level, the basic needs and capability index is not necessarily higher than that of households whose head or most highly educated member is schooled to a level no higher than secondary (tables
Table 6-11 Distribution of basic household needs and capabilities by various education indicators
1. Ave rage le ve l of e ducation (ye ars)
Level of Education No. of households Percentage Average index Index variation coefficient
0 110 27.23 5.89 0.5267
0 < index ≤ 2.66 87 21.53 6.47 0.5514
2.66 < index ≤ 5 116 28.71 5.41 0.5844
> 5 91 22.52 6.57 0.6099
Total 404 100 6.03 0.5740
2. House hold he ad's e ducation (ye ars)
Level of Education No. of households Percentage Average index Index variation coefficient
0 205 50.74 6.07 0.5692 1-5 80 19.80 5.99 0.6295 6-8 69 17.08 6.08 0.5327 9-10 33 8.17 5.62 0.5529 11-12 8 1.98 7.37 0.6086 13+ 9 2.23 5.19 0.7067
3. Most highly e ducate d house hold me mbe r (ye ars)
Level of Education No. of households Percentage Average index Index variation coefficient
0 63 15.59 5.82 0.4700 1-5 123 30.45 5.91 0.5704 6-8 131 32.43 6.04 0.5912 9-10 60 14.85 5.97 0.5987 11-12 14 3.47 8.05 0.5521 13+ 13 3.22 6.14 0.7285
4. Proportion of house hold me mbe rs who have comple te d at le ast primary school
Level of Education No. of households Percentage Average index Index variation coefficient
0 126 31.19 6.06 0.5187
0 < index ≤ 0.015 57 14.11 6.12 0.6095
0.015 < index ≤ 0.02 52 12.87 5.83 0.6744
> 0.02 169 41.83 6.04 0.5746
5. Proportion of house hold me mbe rs who have comple te d primary school
Level of Education No. of households Percentage Average index Index variation coefficient
0 171 42.33 6.08 0.5479
0 < index ≤ 0.008 17 4.21 7.25 0.6007
0.008 < index ≤ 0.01 30 7.43 5.48 0.6942
> 0.01 186 46.04 5.96 0.5762
6. Proportion of house hold me mbe rs who have comple te d middle school
Level of Education No. of households Percentage Average index Index variation coefficient
0 279 69.06 6 0.5431
0 < index ≤ 0.008 15 3.71 7.92 0.5329
> 0.008 110 27.23 5.84 0.6489
7. Proportion of house hold me mbe rs who have comple te d se condary e ducation
Level of Education No. of households Percentage Average index Index variation coefficient
0 359 88.86 5.90 0.5688
<0 45 11.14 7.04 0.5824
8. Proportion of house hold me mbe rs who have comple te d highe r se condary school
Level of Education No. of households Percentage Average index Index variation coefficient
0 391 96.78 6.04 0.5710
<0 13 3.22 5.72 0.6906
9. Proportion of house hold me mbe rs who have comple te d te rtiary school
Level of Education No. of households Percentage Average index Index variation coefficient
0 388 96.04 6.03 0.5696
<0 16 3.96 5.87 0.6975
Notes: N = 404. For the average education level, 2.66 is the median. For the proportion of
household members who have completed at least primary school, 0.015 is the median and 0.02 the mean. For the proportion of household members who have completed primary school, 0.008 is the median and 0.01 the mean. For the proportion of household members who have completed middle school, 0.008 is the mean.
6-11-2 and 6-11-3). However, this might reflect the fact that very few heads or any other household members are highly educated.
Regressions of the composite basic needs and capabilities index on different levels of education reveal statistically insignificant t-ratios, which supports the hypothesis that there is no stable correlation between education level and this index.
Basic needs and capabilities Index = 6.04 – 0.01 average level of education (23.49)***(-0.07) R2 = 0.0000 Basic needs and capabilities Index = 6.10 – 0.02 household head’s education
(26.98)*** (-0.51) R2 = 0.0006 Basic needs and capabilities Index =5.73 + 0.05 highest level of education
(17.49)*** (1.07) R2 = 0.0028
Basic needs and capabilities Index = 6.17 – 6.49 proportion of household members
(28.74)*** (-1.05) who have completed
at least primary school R2 = 0.0028
Basic needs and capabilities Index = 6.21 – 6.69 primary school ratio - 6.47 middle
(26.15)*** (-0.60) (-0.41)
school ratio + 1.54 secondary school ratio + 78.88 higher secondary ratio + (-0.82) (-0.11)
64.12 tertiary education ratio R2 = 0.0064 (-0.79)
Indeed, similar education levels amongst households do not necessarily mean consistent access to various needs and capabilities, and the study generally found no strong evidence of gains in basic needs and capability index associated with educational attainment per se.
Next, the basic needs and capabilities index was divided into five groups with the lowest being assigned a value of 1 and the highest 5, and an ordered probit regression conducted. The dependent variable was the basic needs and capabilities index (groups
1–5), and in order to ensure that the assessment was compatible with analyses of other concepts of poverty, the explanatory variables were the same as those utilised in regressions of other indices in this chapter. The results are given in Table 6-12.
The index value tends to rise significantly as the proportion of household members who have been employed during the previous 12 months increases. Other variables, such as LPG, out-remittance, household size, asset index, house index, and microfinance participation, also show a positive effect on the basic needs and capabilities index, although none of the coefficients are statistically significant, unlike their significant correlation with MPCE (Table 6-6).
The difference in correlation with regard to MPCE (Table 6-6) is the male ratio (although none of the coefficients are statistically significant). The index value tends to fall as the proportion of male household members rises. The value is clearly affected when migrant males live alone or male migrant relatives live together. It is assumed that such males might not bother accessing various basic amenities or facilities in Delhi if they regard themselves as temporary residents of the city, or nobody in the household has time to investigate better living conditions.
In terms of the correlation between education and non-monetary poverty, all education variables show negative signs but none of them are statistically significant. This implies that in contrast to the linkage between education and monetary poverty, education does not have a strong positive association with basic needs and capabilities. This indicates that slum dwellers’ education does not readily lead to claims of various basic rights as citizens, and it may be the case that income is a more urgent necessity for the educated.
Table 6-12 Ordered probit estimations of basic needs and capabilities index Eq (1) Eq (2) Eq (3) Eq (4) Eq (5) Employment ratio 0.5717 ** 0.5514 ** 0.5868 ** 0.5671 ** 0.6034 ** (0.2743) (0.2787) (0.2742) (0.2742) (0.2734) LPG 0.1410 0.1149 0.1235 0.1705 0.1334 (0.1183) (0.1163) (0.1193) (0.1203) (0.1173) Out-remittance 0.0418 0.0203 0.0044 0.0858 0.0218 (0.1664) (0.1668) (0.1665) (0.1729) (0.1674) Household size 0.0377 0.0361 0.0433 0.0506 0.0478 (0.0336) (0.0340) (0.0340) (0.0350) (0.0339) Asset index 0.0042 0.0039 0.0036 0.0041 0.0038 (0.0032) (0.0031) (0.0031) (0.0032) (0.0032) House index 0.0002 0.0002 0.0002 -0.0001 0.0000 (0.0005) (0.0005) (0.0005) (0.0005) (0.0005) Male ratio -0.3930 -0.4101 -0.3963 -0.4215 -0.4232 (0.3248) (0.3225) (0.3265) (0.3281) (0.3258) Debt -0.0018 -0.0125 -0.0141 -0.0025 0.0097 (0.1633) (0.1620) (0.1615) (0.1676) (0.1647) Microfinance 0.1191 0.1285 0.1178 0.1611 0.1249 (0.2255) (0.2244) (0.2255) (0.2268) (0.2258) Average education -0.0324 (0.0209)
Household head's education -0.0148 (0.0142) 0.0101 (0.0164) -2.1489 (3.5128) -7.2012 (4.4116) -0.0345 (2.0801) -45.6863 (30.3174) -25.7722 (16.7753) -2.7450 (1.6690) Pseudo R2 0.0122 0.0112 0.0106 0.0157 0.0119 Most highly educated
Primary school ratio Middle school ratio Secondary education ratio Higher secondary school ratio Tertiary education ratio At least primary school ratio
Notes: N = 400. ***, ** and * indicate significance at 1%, 5%, and 10% respectively. Figures in
parentheses indicate robust standard errors.
6.5. Education and Subjective Wellbeing