regional patterns of human and child deprivation in … and child...regional patterns of human and...

SPECIAL ARTICLE

september 29, 2012 vol xlviI no 39 EPW Economic & Political Weekly42

Regional Patterns of Human and Child Deprivation in India

Jean Drèze, Reetika Khera

This paper takes a look at regional patterns of human

and child deprivation in India, based on district-level

data. It presents and compares two simple summary

indices of living conditions at the district level: a standard

“human development index” and a variant of it focusing

specifically on children.

Thanks are due to Tsering Wangyal and Akansha Batra for helping with the maps, and to Debraj Bagchi and Aashish Gupta for research assistance. We are also grateful to the Office of the Registrar General for unpublished data on age-specific literacy rates.

Jean Drèze ( [email protected]) is with Allahabad University and Reetika Khera ([email protected]) is with IIT Delhi.

Introduction

More than 20 years have passed since the Human Deve-lopment Index (HDI) was devised, in the context of the first Human Development Report (UNDP 1990).

The initial index was an informal aggregate of three specific indicators: life expectancy, adult literacy and per capita income. There is, clearly, an element of arbitrariness in the selection of these indicators, as well as in the aggregation formula. Neverthe-less, the HDI served some useful communication purposes: it produced a rough “ranking” of countries based on human development indicators that do capture important aspects of the quality of life, and this (along with other key messages of the Human Development Reports) helped to bring human development closer to the centre of attention in development economics and public policy.

Since then, many variants of the initial HDI have been devised, and the computation of HDI-type indexes (including creative offshoots such as the “Happy Planet Index”) has become an absorbing research activity. The limitations of these indexes are sometimes overlooked in this rush for uni-dimensional quantitative indicators of social progress.1 Some of them, however, have served useful purposes. In India, for instance, the computation of state-specific HDIs and related indicators (including the recent “multidimensional poverty index”) has contributed to a better understanding of regional disparities in human development.2

This paper takes a closer look at regional patterns of human and child deprivation in India, based on district-level data. We present and compare two simple summary indexes of living conditions at the district level: a standard “human development index”, and a variant of it focusing specifically on children. The component indicators are also discussed.

Unless stated otherwise, all the indicators presented in this paper refer to rural and urban areas combined. Due to data limitations, the north-eastern region (including Assam) is omitted from district-level analyses, as is the state of Jammu and Kashmir. For the same reason, the reference period (unless otherwise specified) is 2001, or the closest subsequent year for which the relevant data are available. Hopefully, similar work on more recent data, including the 2011 Census, will soon be possible.3

1 A District-Specific HDI

We begin by presenting a simple HDI for Indian districts, subject to the qualifications mentioned earlier. This index is based on three indicators, similar to those that initially made up the UNDP’s

SPECIAL ARTICLE

Economic & Political Weekly EPW september 29, 2012 vol xlviI no 39 43

Human Development Index: adult female literacy, child mortality, and the proportion of households with a “low standard of living”. Literacy rates are available from the Census of India 2001. Estimates of district-specific child mortality rates (more precisely “Q5”, the probability of dying before age five) for 2001, also based on census data, are available from the Population Foun-dation of India (Rajan et al 2008). The third indicator is avail-able from the second District Level Household Survey (DLHS-2), conducted in 2002-04 (International Institute for Population Sciences 2006a).4 There are significant margins of error in the individual district estimates for the last two indicators, and the main focus here is on broad regional patterns rather than on specific districts. Table 1 presents state averages of these indicators, subtracted from unity if appropriate, so that higher values indicate better outcomes (e g, the second indicator actually measures child survival instead of child mortality).

Each indicator has been “normalised” using the same formula as in the initial international HDI, as follows:

Xi ≡ (xi – xmin)/(xmax – xmin)

where Xi is the normalised indicator for district i, xi is the corresponding pre-normalisation figure, and xmax and xmin are the maximum and minimum values of the same indicator across all districts. The normalised indicator takes value 0 for the “bottom” district, 1 for the “top” district, and varies between 0 and 1 for other districts. Essentially, it tells us where a particular district stands, between the “top” and “bottom” districts (in terms of the concerned indicator), on a linear scale. For instance, a value of 0.5 means that the district is “half way” between the top and bottom districts.

The HDI presented here is an unweighted average of the three normalised indicators. The district-specific HDIs are dis-played in Map 1 (p 44), where districts have been grouped into “quintiles” (each quintile includes one-fifth of all districts), with darker shades indicating lower HDI values.

2 The Geography of Deprivation

Elementary as this exercise may be, it draws attention to some important aspects of the geography of human deprivation in India. For instance, looking at the darkest districts, Map 1 high-lights several regions that are known for high levels of human deprivation; these include southern Chhattisgarh (the former Bastar region), southern Orissa, northern and eastern Madhya Pradesh, the north-eastern part of Bihar, and the southern and western tips of Rajasthan. Less well understood perhaps is the fact that many districts of Uttar Pradesh, stretching across the

entire state, also belong to this group. In fact, the whole “terai” belt stretching across Uttar Pradesh and Bihar along the Nepal border emerges as a problem area. On the other side, all districts of Kerala and Himachal Pradesh belong to the top quintile of the HDI scale; this is in line with Kerala’s long-stand-ing lead in matters of social devel-opment, and Himachal Pradesh’s more recent achievements in this field.5 Other states with high levels of human development in all dis-tricts include Tamil Nadu in the south, and Punjab and Haryana in the north. While these and a few other states have relatively uniform HDI levels across districts, there are major inter-district variations in most states. Maharashtra, for in-stance, includes some districts from the top HDI quintile, but also some in all other categories including the bottom quintile (e g, Gadchiroli, bordering Chhattisgarh).

Table 1: State Averages of Component Indicators (in %)State ComponentsofHDIIndex ComponentsofABCIndexa

Literacy:Adult StandardofLiving: ChildSurvival: Literacy:Proportion Immunisation: Nutrition: FemaleLiteracy Proportionof Probabilityof ofLiterateGirlsin Proportionof Proportionof Rate,2001 Householdsabove Survivingto theAgeGroupof ChildrenFully ChildrenWhoAre ‘LowSLI’,2002-04 AgeFive,2001 10-14Years,2001 Immunised, NotUnderweight, 2002-04 2002-04

Andhra Pradesh 42.5 61.5 952 80.7 62.7 57.7

Assam 50.4 43.7 922 72.7 17.2 67.8

Bihar 28.2 33.7 938 51.2 23.0 45.4

Chhattisgarh 43.6 36.5 925 82.4 58.1 52.6

Gujarat 52.2 65.4 938 81.8 54.0 54.0

Haryana 47.6 80.7 956 88.1 59.2 64.4

Himachal Pradesh 60.8 74.8 934 95.9 79.3 63.6

Jharkhand 32.5 34.2 945 61.3 26.6 47.8

Karnataka 50.0 55.4 939 84.3 71.3 55.2

Kerala 86.2 83.9 980 98.7 78.6 64.2

Madhya Pradesh 42.6 44.4 898 77.5 30.2 44.6

Maharashtra 60.8 58.9 942 93.5 70.9 52.3

Orissa 45.1 37.6 908 74.2 53.6 57.2

Punjab 57.9 88.2 955 88.2 72.9 60.0

Rajasthan 35.6 54.7 912 70.1 24.7 41.9

Tamil Nadu 59.3 67.0 950 93.2 91.4 61.7

Uttarakhand 51.9 62.5 930 88.1 44.3 47.4

Uttar Pradesh 34.3 45.8 906 67.7 26.4 44.7

West Bengal 54.8 48.2 934 80.5 50.4 55.1

All India 47.8 57.7 941 77.0 45.8 50.8a Other than "child survival", already included under HDI. Immunisation figures refer to the age group of 12-23 months, and undernutrition figures to children aged below six years.Sources: Census of India 2001 for literacy rates; Rajan et al (2008) for child survival; second District Level Household Survey (International Institute for Population Sciences 2006a) for other indicators.

Figure 1: Proportion of Underweight Children (NFHS-2 vs DLHS-2)

Prop

ortio

n (%

) of u

nder

wei

ght c

hild

ren,

NFH

S-2,

199

8-99

Proportion (%) of underweight children, DLHS-2, 2002-04

20

30

40

50

60

70

20 30 40 50 60 70

SPECIAL ARTICLE


Broader regional patterns also emerge. Specifically, low-HDI districts are heavily concentrated in one massive, largely con-tiguous “lump” stretching across the large north Indian states (Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh, Orissa, Rajasthan and Uttar Pradesh). All the bottom-quintile districts are part of this lump, and so are most of those in the second quintile (from the bottom), with the main exception of a few contiguous districts in northern Karnataka (Bellary, Gulbarga, Raichur and Koppal).6 Further, there are very few high-HDI districts in this area (e g, virtually none from the top quintile of the HDI scale). The main exceptions are districts located around major cities such as Kanpur in Uttar Pradesh or Bhopal and Indore in Madhya Pradesh.

Outside this low-HDI area (hereafter the “northern heart-land”), most districts fall into two contiguous zones. There is, first, a contiguous zone in the south, which includes the four southern states (Andhra Pradesh, Karnataka, Kerala and Tamil Nadu), and most of Gujarat and Maharashtra. This zone is virtually free of low-HDI districts, except for the patch of contiguous districts mentioned earlier (around northern Karnataka). Within this region, the most advanced area is Kerala, where every district belongs to the top quintile of the HDI scale. The second contiguous set of relatively high-HDI districts is located in the north-west, stretching across most of Punjab, Haryana and Himachal Pradesh. There are also a few high-HDI districts in West Bengal.

Another interesting aspect of Map 1 is that it suggests an important role for state policies. Within a state, human deve-lopment levels tend to be relatively similar among contiguous districts. However, contiguous districts located in different states often have very different levels of human development. In other words, there are few discrete “jumps” (in levels of human development) across district boundaries within a state, but there are discrete jumps across state boundaries, e g, between Maharashtra and Madhya Pradesh, or between Andhra Pradesh and Orissa, or between Uttar Pradesh and Uttarakhand.7 In some cases, this may reflect the fact that state boundaries coincide with some sort of natural or environmental barrier. State policies, however, are also likely to play a role.

3 A Child Development Index

We now extend this enquiry by presenting a summary index of the well-being of children. This is a normalised variant of the “Achievements of Babies and Children” (ABC) index, intro-duced in the Focus On Children Under Six (FOCUS) report.8 Briefly, this modified ABC index is constructed in the same manner as the HDI index, using the following four indicators: probability of surviving until age five; proportion of children fully immunised in the age group of 12-23 months; proportion of children aged 12-35 months who are not underweight; and female literacy rate in the 10-14 age group. Each indicator captures an important aspect of the well-being of children: survival until age five, full immunisation, adequate nourish-ment, and literacy, respectively. The last indicator focuses on girls specifically, to impart some gender sensitivity to the in-dex.9 For state averages of these indicators, see Table 1.

The ABC index is presented in Map 2, in a format similar to Map 1, based on a grouping of districts into five even “quintiles”. The broad patterns are much the same as with the HDI (Map 1): the northern heartland is the big “problem region”, flanked by areas of higher child development in the north and south, with a patch of relatively low child development within the southern region, around northern Karnataka. The similarity between Map 1 and Map 2, in terms of broad contours as well as in some

Map 1: Human Development Index (2001)

Human Development Index

Top quintile (.63-1)

Second quintile (.51-.62)

Third quintile (.38-.50)

Fourth quintile (.29-.37)

Bottom quintile (0-.28)

Missing values

Source: See text.

Map 2: ABC Normalised Index (2001)

Top quintile (.68-1)

Second quintile (.59-.68)

Third quintile (.49-.59)

Fourth quintile (.40-.49)

Bottom quintile (<.40)

Missing values

Source: See text.

ABC Normalised Index

SPECIAL ARTICLE


(not all) of the finer details, is not entirely surprising. But it is not a trivial pattern either, considering that differences could emerge for at least three reasons: first, the two maps focus on different aspects of human development; second, Map 1 reflects the cumulative impact of living conditions over a long period of time (including, for instance, the state of the schooling system many years ago), while Map 2 captures children’s living environment in the recent past (that is, recent relative to 2001, the reference year); third, the HDI, unlike the ABC index, includes an indicator of economic “living standard” (this is why, for instance, Punjab and Haryana fare better in terms of HDI than ABC). In spite of the potential differences, however, the summary indexes point to very similar regional patterns.

Having said this, a few distinct features of Map 2 are worth noting. First, the north-south contrast in Map 2 is particularly striking. In the southern tip of India, including not only Kerala and Tamil Nadu but also substantial parts of southern Karna-taka, almost every district belongs to the highest-ABC quintile. Conversely, in the north, there is a vast expanse of very low-ABC districts, stretching across most of Rajasthan, Uttar Pradesh, Bihar, Jharkhand and Madhya Pradesh.

Second, Chhattisgarh is now clearly demarcated from this problem region. There is some evidence that child develop-ment indicators have improved quite rapidly in Chhattisgarh in recent years.10 What Map 2 shows is that Chhattisgarh was already ahead of other parts of the so-called BIMARU states (undivided Bihar, Madhya Pradesh, Rajasthan and Uttar Pradesh) around the time of its formation in 2000. Quite likely, the gap is even wider today.

Third, in the southern region, the ABC map sharply brings out not only Kerala’s but also Tamil Nadu’s “lead” in the field of child development. Kerala’s achievements in this regard are well known, but Tamil Nadu is not far behind, and the fact that all districts of Tamil Nadu are doing well in this respect is quite remarkable. This is consistent with, and consolidates, recent evidence of rapid improvements in child nutrition, healthcare and elementary education in Tamil Nadu in the post-Independence period, plausibly linked with active social policies including free and universal provision of essential public services.11

Both the HDI and the ABC index point to a resilient north-south contrast in matters of human and child development. Strictly speaking, this is not so much a north-south contrast as a contrast between what we call here the “northern heartland” and the rest of the country. Indeed, as one moves further north from there, e g, into the extreme north-west (or for that matter north-east), human and child development indicators improve again. But with this qualification, the north-south dichotomy is quite striking.

This broad contrast is not new.12 It also relates, quite likely, to similar and widely discussed contrasts in gender relations, which tend to be more unequal and patriarchal in the north than in the south (although here again, “north-south” is actually a simplifi-cation of complex regional patterns).13 What is interesting is that the contrast shows little sign of disappearing or even diminishing. In fact, informal scrutiny of similar district-level data for 1981 (not presented here) suggests that the concentration of depriva-tion in the northern heartland has quite possibly increased

between 1981 and 2001. This is also consistent with the fact that this pattern is somewhat more pronounced in the ABC map than in the HDI map (bearing in mind that the latter reflects the cu-mulative impact of living conditions over a long period of time).

4 Component Indicators

The composite indexes presented so far help to highlight broad regional patterns of human and child deprivation. For instance, the heavy concentration of deprivation in the northern heart-land, though fairly well known, emerges with special clarity in this approach. Having said this, much information is lost in the aggregation exercise, and the composite indexes should not substitute for continued attention to the component indicators.

To illustrate the issue, Maps 3 to 6 (p 46) present the compo-nents of the ABC index, in the usual format (with districts grouped in quintiles, and darker shades denoting worse indi-cators). There are broad similarities between the four maps, including the familiar north-south pattern, and some of these maps, notably the child immunisation map, look much like the ABC map. However, there are also significant differences. Com-pare, for instance, the female literacy and child survival maps (Maps 3 and 4, respectively). Considering the well-known in-fluence of women’s education on child health, one might ex-pect the two maps to look much the same. But in fact, they are quite different. For instance, districts with high child mortality rates are concentrated in and around Madhya Pradesh, but the low-literacy belt (for the age group of 10-14 years) actually goes around Madhya Pradesh. The state combines high child mor-tality rates with relatively high literacy rates in the younger age groups, at least by north Indian standards. It is, thus, important to remain alive to the details of various indicators, and to ac-knowledge that “problem areas” are indicator-specific (beyond some broad patterns, such as Kerala doing well all-round).14

The child nutrition map (Map 6) is also intriguing. In parti-cular, this map does not display the “lumpy” configuration of the other maps, where contiguous districts tend to have similar levels of child development. In Maps 3 to 5 (and earlier maps), there are relatively few “discrete jumps” across district boundaries: contiguous districts tend to be in identical or adjacent quintiles of the relevant scale, so that the shades change in a smooth manner across the map. As noted earlier, there are occasional jumps across state boundaries, possibly reflecting differences in state policies. But generally, neighbouring districts tend to have relatively similar human or child development indicators.

The child nutrition map, however, is much less lumpy, and further, it shows many discrete jumps across district bounda-ries. This includes “long jumps”, from (say) top to bottom quin-tile, as one crosses some district borders, including borders between districts that have relatively similar social indicators otherwise, such as Sidhi (north-eastern Madhya Pradesh) and Koriya (north-western Chhattisgarh). There are other surpris-ing features in Map 6, such as high levels of undernutrition in Uttarakhand, which is doing quite well on other indicators, and low levels of undernutrition in a string of districts of eastern Madhya Pradesh and western Chhattisgarh (Rewa, Satna, Shahdol, Koriya, Bilaspur) that are doing very poorly otherwise,

SPECIAL ARTICLE


with big inter-district jumps around that area (e g, between Dindori and Shahdol).

It is not clear whether these odd features of Map 6 are “real”, or whether they reflect high margins of error in the district-level estimates of child nutrition from DLHS-2 data. To probe this fur-ther, we aggregated the DLHS-2 nutrition indicators to the level of “regions”, and compared them with the corresponding esti-mates from the second National Family Health Survey (NFHS-2,

conducted in 1998-99). The comparison is presented in Figure 1 (p 43) in the form of a scatter diagram. Clearly, there are major discrepancies between the two surveys as far as child nutrition indicators are concerned, even after allowing for the difference in reference years. Given that NFHS-2 data are considered rea-sonably credible in this respect (the NFHS-based regional pat-terns are fairly consistent with independent evidence), this casts some doubt on the accuracy of DLHS-2 estimates. The absence of reliable child nutrition data at the district level is an important gap in India’s statistical system, all the more serious as child undernutrition rates in India are among the highest in the world.

Map 3: Female Literacy (Age 10-14, 2001)

Proportion (%) literate girls, age 10-14 years

Top quintile (>.92%)

Second quintile (.86-.92%)

Third quintile (.77-.85%)

Fourth quintile (.64-.76%)

Bottom quintile (<.64%)

Missing values

Source: Census of India 2001.

Map 4: Child Mortality Rate (2001)

Source: Population Foundation of India (2008).

Probability (%) of dying before age five


Second quintile (.75-94%)



Bottom quintile (<50%)

Missing values

Percentage of children (aged 12-23 months)who have received full immunisation


Second quintile (.59-.76%)



Bottom quintile (<.25%)

Missing values

Source: Second District Level Household Survey, 2002-04.

Map 5: Child Immunisation (2002-04)

Map 6: Child Nutrition (2002-04)

Proportion (%) of underweight children, 12-23 months


Second quintile (.52-58%)



Bottom quintile (<37%)

Missing values

Source: Second District Level Household Survey, 2002-04.

SPECIAL ARTICLE


5 UpdateBy way of a quick update, Tables 2 and 3 present a simple HDI for Indian states, using more recent data. The component indi-cators are given in Table 2, with 2005 as the reference year (this is the latest year for which comprehensive data are avail-able). Since the results are sensitive to the choice of indicators used to capture education, health and economic poverty (the three standard components of HDI-type indexes), we use three indicators for each aspect of human development, and give them equal weight. The normalisation and aggregation formu-las are the same as before, with nine indicators instead of three. In Table 3, the HDI is calculated in two ways: (1) with 2005 as the reference year, using the data given in Table 2; (2) using the latest available data for each indicator. Both ways, the results are much the same. We tried many variants of this HDI, but the basic patterns are fairly robust, as long as education, health and standard of living get roughly equal weights.

To interpret the results, it is useful to remember that the HDI is essentially an average (over different indicators) of the state’s position between the top and bottom states on a linear scale (see Section 1). It would take value 1 for a state that does best in terms of every indicator, and 0 for a state that does worst in terms of every indicator. The impressive value of 0.970 for Kerala reflects the fact that Kerala does best among all major states in terms of six of the nine indicators, and is very close to best for the remaining three. Dire warnings about Kerala’s approach to development being “unsustainable”, heard at regular intervals during the last 20 years, are yet to come true.

The highest-HDI states are Kerala, Himachal Pradesh and Tamil Nadu (in that order). An earlier ranking of states based Table 2: Selected Indicators for Major States (2005) Education-RelatedIndicators Health-RelatedIndicators Poverty-RelatedIndicators FemaleLiteracy, Proportionof Proportionof UnderFive Proportionof Proportionof Proportionof Proportionof MedianPer Age15-49,2005-06 ChildrenAged6-14 ChildrenAged MortalityRate, YoungChildren AdultWomen Populationbelow Populationin CapitaIncome, (%) AttendingSchool, 8-11WhoPassa 2005-06 FullyImmunised, withLowBMI, thePovertyLine, India’sLowest 2004-05 2005-06(%) SimpleReading (PerThousand) 2005-06(%) 2005-06(%) 2004-05(%) WealthQuintile, (Rs/Year) Test,2004-05(%) 2005-06(%)

Andhra Pradesh 49.6 81.4 50 63.2 46.0 33.5 29.6 10.8 6,241

Assam 63.0 84.4 72 85.0 31.4 36.5 34.4 19.8 6,000

Bihar 37.0 62.2 44 84.8 32.8 45.1 54.5 28.2 3,530

Chhattisgarh 44.9 81.1 61 90.3 48.7 43.4 49.4 39.6 5,306

Gujarat 63.8 83.0 64 60.9 45.2 36.3 31.6 7.2 6,300

Haryana 60.4 84.1 65 52.3 65.3 31.3 24.1 4.1 9,443

Himachal Pradesh 79.5 96.2 83 41.5 74.2 29.9 22.9 1.2 9,942

Jammu & Kashmir 53.9 87.8 40 51.2 66.7 24.6 13.1 2.8 8,699

Jharkhand 37.1 71.7 59 93.0 34.2 43.0 45.3 49.6 4,833

Karnataka 59.7 84.0 53 54.7 55.0 35.5 33.3 10.8 5,964

Kerala 93.0 97.7 82 16.3 75.3 18.0 19.6 1.0 9,987

Madhya Pradesh 44.4 89.1 46 94.2 40.3 41.7 48.6 36.9 4,125

Maharashtra 70.3 87.2 66 46.7 58.8 36.2 38.2 10.9 7,975a

Orissa 52.2 77.5 58 90.6 51.8 41.4 57.2 39.5 3,450

Punjab 68.7 85.3 66 52.0 60.1 18.9 20.9 1.4 9,125

Rajasthan 36.2 75.4 55 85.4 26.5 36.7 34.4 24.2 6,260

Tamil Nadu 69.4 93.9 79 35.5 80.9 28.4 29.4 10.6 7,000

Uttar Pradesh 44.8 77.2 39 96.4 23.0 36.0 40.9 25.3 4,300

Uttarakhand 64.6 90.4 63 56.8 60.0 30.0 32.7 6 6,857

West Bengal 58.8 79.7 51 59.6 64.3 39.1 34.2 25.2 6,250

India 55.1 79.6 54 74.3 43.5 35.6 37.2 20 5,999a Including Goa.Sources: Figures with 2005-06 as reference year are from the third National Family Health Survey (International Institute for Population Sciences 2007, and state reports for school attendance); poverty estimates 2004-05 from the Tendulkar Committee Report (as reprinted in Planning Commission 2012); reading proficiency and median per capita income from the India Human Development Survey (Desai et al 2010).

Table 3: Human Development Index for Indian States Using2005asReferenceYeara UsingLatestAvailableDatab

Kerala 0.970 0.980

Himachal Pradesh 0.846 0.837

Tamil Nadu 0.749 0.761

Punjab 0.742 0.734

Haryana 0.670 0.640

Jammu & Kashmir 0.655 0.626

Uttarakhand 0.612 0.640c

Maharashtra 0.601 0.627

Gujarat 0.520 0.526

Karnataka 0.500 0.507

Andhra Pradesh 0.458 0.466

West Bengal 0.446 0.483

Assam 0.441 0.402

India 0.400 0.410

Rajasthan 0.301 0.318

Chhattisgarh 0.271 0.290

Madhya Pradesh 0.230 0.252

Orissa 0.229 0.268

Uttar Pradesh 0.212 0.214

Jharkhand 0.170 0.218

Bihar 0.106 0.108a Based on National Family Health Survey (NFHS) data for 2005-06, National Sample Survey (NSS) data for 2004-05, and India Human Development Survey (IHDS) data for 2004-05. See Table 2.b Reference year: 2011 for female literacy (age 7+, from Census of India 2011); 2008 for child mortality (from Government of India, 2008); 2009-10 for official poverty estimates (from Planning Commission 2012); 2005 for other indicators – as in the first column.c In the absence of a child mortality rate estimate for Uttarakhand in 2008, the child mortality rate of Himachal Pradesh has been used (both states have similar infant mortality rates).Sources: See Table 2. The index is an unweighted average of normalised values for each of the 9 indicators presented in Table 2, using the normalisation rule mentioned in the text. The states are ranked in decreasing order of HDI for 2005.

SPECIAL ARTICLE


on the ABC index, using NFHS data for 1998-99, had already placed these three states at the top (Drèze, Khera and Naray-anan 2007). During the last 10 years, all of them have made further rapid progress in terms of poverty reduction and social development (Chakraborty 2011). Each of these three states has pursued active social policies based on universalistic prin-ciples, at an early stage of development, when they were still quite poor. These policies were often criticised, at that time, for being profligate, unaffordable or “populist”. But there is no evidence that they have interfered with these states’ economic success. On the contrary, all three are now quite well-off in economic terms, aside from having much better social indica-tors than most other states.

The low-HDI states are the usual suspects: the large north Indian states that used to be known by the unflattering acronym of BIMARU (or rather BIMAROU, with Orissa added). In fact, there is a striking gap between these states and the rest: the next state – from the bottom – in this ranking is Assam, with a much higher HDI.

Bihar is the state with lowest HDI, by a long margin. This is a familiar pattern, but it is sometimes forgotten in the prevailing mood of enthusiasm for Bihar’s recent progress (or alleged progress). It is quite possible that the state’s changing political priorities, with greater emphasis on development, will show results in due course. But as things stand, there is little evi-dence of a human development breakthrough in Bihar, even in the most recent data. For instance, Bihar is the only state where poverty (especially rural poverty) did not decline to any significant extent between 2004-05 and 2009-10, according to official estimates (Planning Commission 2012).15 The accumu-lated damage of decades of indifference towards people’s basic needs is enormous, and will take a long time to repair.

Uttar Pradesh is another human development disaster, with, for instance, only 39% of children aged 8-11 being able to pass a simple reading test (going beyond liberal definitions of “literacy”), and only 23% of young children being fully immu-nised. Like Bihar, Uttar Pradesh is held up by a heavy burden of historical inequalities (of class, caste and gender), and also by decades of state apathy towards common people (Drèze and Gazdar 1996). The dismal state of human development in Uttar Pradesh is not a trivial matter, considering that Uttar Pradesh has a larger population (about 200 million) than any country in the world except China, India, Indonesia and the United States. Uttar Pradesh’s population is expected to grow to 300 million or so by 2026, and more than 400 million by the end of the century, an interesting prospect.16

As Table 3 illustrates, there are tentative indications of Chhattisgarh and Rajasthan detaching themselves a little from the rest of the northern heartland, leaving the JUMBO states (Jharkhand, Uttar Pradesh, Madhya Pradesh, Bihar and Orissa) behind. However, too much should not be made of this convenient acronym. For one thing, this pattern is somewhat sensitive to the choice of indicators. For another, Chhattisgarh’s recent achievements (e g, in terms of child health or food security) should be seen in the light of its atrocious record of state repression, human rights violations and forced displacement of tribal communities, particularly in the former Bastar region.

A similar remark applies to the north-eastern states as well as to Jammu and Kashmir. For most of the indicators in Table 2, the average for the north-eastern region (approximated as a population-weighted average of state figures) is higher than the all-India average. Jammu and Kashmir, for its part, comes sixth among 20 major states in the HDI ranking – see Table 3. However, there is a dreadful trail of human rights violations and suppression of civil liberties in the northern-eastern region as well as in Jammu and Kashmir. This is a useful reminder of the limitations of “social indicators” as a barometer of development.

6 Concluding Remarks

The geography of deprivation in India varies for different as-pects of social development. Different regions emerge as the “problem areas” depending on whether one is focusing on, say, rural poverty, or child mortality, or female literacy. For in-stance, Madhya Pradesh has high mortality rates, but is now doing relatively well (at least in comparison with other north-ern states) in terms of literacy in the younger age groups.

Interestingly, however, two different summary indexes of deprivation point to very similar general regional patterns. Among other robust patterns is one of widespread deprivation across most of the northern heartland. On each side of this region, there are contiguous areas with relatively high levels of human development, in the north and the south. But there is also a smaller problem region in the south – a patch of contigu-ous districts around northern Karnataka. Within the northern heartland, areas that seem to call for special attention include southern Orissa, southern Chhattisgarh, southern Rajasthan, northern Madhya Pradesh, and large parts of Bihar, Jharkhand and Uttar Pradesh.

While these patterns are not surprising, they are perhaps worth noting at a time when the geography of deprivation in India is often confused by an exaggerated focus on indicators that are actually hard to interpret. For instance, there are frequent attempts to “rubbish” Kerala’s achievements based on indicators such as suicide rates or alcoholism. Suicide rates have also put states such as Andhra Pradesh and Karnataka in a grim light.17 Without denying that suicides and alcoholism are major social concerns, it is important not to miss the wood for the trees.

There is also a questionable narrative of rapid “catch-up” of the northern states with the rest of the country in the last 10 years or so (Institute of Applied Manpower Research 2011). The findings presented in this paper are not incompatible with the possibility of some “convergence” over time, across states, in levels of hu-man development.18 Indeed, convergence need not affect rank-ings of states or districts – the main focus of this paper. However, at least some of the alleged evidence of recent convergence is based on misleading measures of progress.19 And if there is some convergence, there is also an issue of speed, considering the enormous gap that separates the northern heartland from most other states. The updated state HDIs presented earlier suggest that whatever convergence may have taken place has not made much of a dent in this considerable gap. Hopefully, the release of detailed census data for 2011 will soon make it possible to re-examine the convergence story.

SPECIAL ARTICLE


Notes

1 Some of the limitations of the HDI were explicitly noted in the first Human Development Report (United Nations Development Programme 1990). For other useful comments and critiques, see Chakraborty (2002), Raworth and Stewart (2003), Kovacevic (2010), among others.

2 See, e g, Kumar (1991, 1996), Ghosh (2006), and Institute of Applied Manpower Research (2011); on multidimensional poverty indexes (MPIs), see Jayaraj and Subramanian (2010), Alkire and Foster (2011), Alkire and Seth (2012), among others. While MPIs are more in-formative, in some ways, than HDIs, they also have some drawbacks, e g, ignoring depriva-tions among those who are not counted as “poor”. This is one reason for the continued rel-evance of HDI-type indexes.

3 For earlier work on similar lines, see Kumar (1991, 1996), Dutta et al (1997), Debroy and Bhandari (2003), Borooah and Dubey (2007), Chaudhuri and Gupta (2009), among others.

4 In DLHS-2, each household has a standard of living (SLI) “score” on a scale of 0 to 40, based on the following criteria: source of drinking water (0 to 3 points); type of house (0 to 4 points); source of lightning (0 to 2 points); cooking fuel (0 to 2 points); toilet facility (0 to 4 points); ownership of durables (0 to 25 points). A household is deemed to have a low standard of living if its SLI score is less than 10. At the all-India level, 42% of households had a low stand-ard of living in 2002-04, based on this criterion. See International Institute for Population Sciences (2006a), pp 28-29.

5 See Drèze and Sen (2002), and earlier work cited there.

6 When variants of this HDI are examined, using alternative component indicators (such as age-specific literacy rates instead of adult literacy, or alternative indicators of child mortality), or different aggregation or normalisation formu-las, the precise number and list of low-HDI dis-tricts in northern Karnataka tends to change. What is not in doubt is that there is a problem of low human development in that area, even though the exact contours of this low-HDI area depend on the specification of the component indicators and summary index.

7 The last pattern has already been noted earlier, in the context of literacy rates (PROBE Team, 1999: 10).

8 Citizen’s Initiative for the Rights of Children Under Six (2006); see also Drèze, Khera and Narayanan (2007).

9 There is little gender bias in the other indicators, with the partial exception of child survival.

10 See, e g, FOCUS Report (Citizens’ Initiative for the Rights of Children Under Six 2006), Chap-ter 5; also Drèze, Khera and Narayanan (2008).

11 See Srinivasan (2010), Sinha (forthcoming), and the literature cited there.

12 Discussions of this contrast go back at least to B R Ambedkar (1955), who had a rather sweep-ing view of this matter: “There is a vast differ-ence between the North and the South. The North is conservative. The South is progressive. The North is superstitious, the South is rational. The South is educationally forward, the North is educationally backward. The culture of the South is modern. The culture of the North is ancient.” The context of this statement was a concern that the organisation of states on lin-guistic lines would lead to “consolidation of the North and balkanisation of the South”, and from there, to a revolt of the South against the dominance of the North.

13 See, e g, Karve (1965), Sopher (1980), Dyson and Moore (1983), Sinha (forthcoming); also Drèze and Sen (2002) and the literature cited there.

14 On this point, see also Borooah and Dubey (2007).

15 Also, in Table 3, if the HDI is recalculated after discarding 2005 data and using only those indicators for which more recent data are available (see note “b” in Table 3), Bihar looks even worse, with a human development index of 0.073.

16 See Dyson and Hanchate (2000) and Visaria and Visaria (2003).

17 According to the latest National Crime Records Bureau data, suicide rates in all the southern states are much higher than the national aver-age; among the major states, Bihar has the lowest suicide rate, followed by Uttar Pradesh (National Crime Records Bureau 2011: xviii).

18 For evidence of convergence in the 1980s and 1990s, see Dholakia (2003) and Ghosh (2006, 2011).

19 See Chakraborty’s (2011) critique of this aspect of the recent India Human Development Report (Institute of Applied Manpower Research 2011). Similar criticisms apply to some of the earlier research on convergence in the 1980s and 1990s.

References

Alkire, S and J Foster (2011): “Counting and Multi-dimensional Poverty Measurement”, Journal of Public Economics, 95.

Alkire, S and S Seth (2012): “Multidimensional Poverty Index (MPI) Rates in Rural and Urban Indian States”, mimeo, Oxford Poverty and Human Development Initiative, University of Oxford.

Ambedkar, B R (1955): Thoughts on Linguistic States; reprinted in Government of Maharashtra (1979), Dr Babasaheb Ambedkar: Writings and Speeches, Volume 1, Department of Education, Government of Maharashtra, Bombay.

Borooah, V K and A Dubey (2007): “Measuring Re-gional Backwardness”, Journal of Applied Eco-nomic Research, 1.

Chakraborty, Achin (2002): “Issues in Social Indi-cators, Composite Indices, and Inequality”, Economic & Political Weekly, 30 March.

– (2011): “Human Development: How Not to In-terpret Change”, Economic & Political Weekly, 17 December.

Chaudhuri, S and N Gupta (2009): “Levels of Living and Poverty Patterns: A District-Wise Analy-sis”, Economic & Political Weekly, 28 February.

Citizens’ Initiative for the Rights of Children Under Six (2006): Focus On Children Under Six (New Delhi); available at www.righttofoodindia.org.

Debroy, B and L Bhandari, ed. (2003): District Level Deprivation in the New Millennium (New Delhi: Konark).

Desai, S B, A Dubey, B Joshi, M Sen, A Shariff and R Vanneman (2010): Human Development in India: Challenges for a Society in Transition (New Delhi: Oxford University Press).

Dholakia, Ravindra H (2003): “Regional Disparity in Economic and Human Development in India”, Economic & Political Weekly, 27 September.

Drèze, J P and H Gazdar (1996): “Uttar Pradesh: The Burden of Inertia” in J P Drèze and A K Sen (ed.) (1996), Indian Development: Selected Re-gional Perspectives (New Delhi: Oxford Univer-sity Press).

Drèze, J P, R Khera and S Narayanan (2007): “Early Childhood in India: Facing the Facts”, Indian Journal of Human Development, 1.

Drèze, J P and A K Sen (2002): India: Development and Participation (Oxford: Oxford University Press).

Dutta, B, M Panda and W Wadhwa (1997): “Human Development in India” in S Subramanian (ed.) (1997), Measurement of Inequality and Poverty (New Delhi: Oxford University Press).

Dyson, T and A Hanchate (2000): “India’s Demo-graphic and Food Prospects”, Economic & Political Weekly, 11 November.

Dyson, T and M Moore (1983): “On Kinship Struc-ture, Female Autonomy, and Demographic Be-havior in India”, Population and Development Review, 9.

Ghosh, Madhusudan (2006): “Economic Growth and Human Development in Indian States”, Economic & Political Weekly, 29 July.

– (2011): “Regional Disparities in Education, Health and Human Development in India”, Indian Journal of Human Development, 5.

Government of India (2008): Sample Registration System Statistical Report 2008 (New Delhi: Of-fice of the Registrar General).

Institute of Applied Manpower Research (2011): In-dia Human Development Report 2011, IAMR and Planning Commission, New Delhi.

IIPS (2000): National Family Health Survey 1998-99 (NFHS-2): India, International Institute for Population Sciences, Mumbai.

– (2006a): Reproductive and Child Health: Dis-trict Level Household Survey 2002-04, Inter-national Institute for Population Sciences, Mumbai.

– (2006b): Nutritional Status of Children and Prevalence of Anaemia among Children, Adoles-cent Girls and Pregnant Women: District Level Household Survey on Reproductive and Child Health, India 2002-2004, International Insti-tute for Population Sciences, Mumbai.

– (2007): National Family Health Survey (NFHS-3), 2005-06: India, International Institute for Pop-ulation Sciences, Mumbai.

Jayaraj and Subramanian (2010): “A Chakravarty-D’Ambrosio View of Multidimensional Depri-vation: Some Estimates for India”, EPW, 6 February.

Karve, Irawati (1965): Kinship Organisation in India (Mumbai: Asia Publishing House).

Kovacevic, Milorad (2010): “Review of HDI Critiques and Potential Improvements”, Human Deve-lopment Research Paper 2010/33, UNDP, New York.

Kumar, A K Shiva (1991): “UNDP’s Human Develop-ment Index: A Computation for Indian States”, Economic & Political Weekly, 22 October.

– (1996), “UNDP’s Gender-related Development Index: A Computation for Indian States”, Eco-nomic & Political Weekly, 6 April.

NCRB (2011): Accidental Deaths and Suicides in India 2010, National Crime Records Bureau, New Delhi.

Planning Commission (2012): “Press Note on Pov-erty Estimates, 2009-10”, Planning Commis-sion, New Delhi.

PROBE Team (1999): Public Report On Basic Educa-tion (New Delhi: Oxford University Press).

Rajan, S I, P M Nair, K L Sheela, L Jagatdeb and N R Mishra (2008): Infant and Child Mortality in India: District Level Estimates (New Delhi: Population Foundation of India).

Raworth, K and D Stewart (2003): “Critiques of the Human Development Index: A Review” in S Fukuda-Parr and A K Shiva Kumar (ed.) (2003), Readings in Human Development (New Delhi: Oxford University Press).

Sinha, Dipa (forthcoming): “Health and Human Development: Comparative Experiences of Tamil Nadu and Uttar Pradesh”, PhD thesis, Ja-waharlal Nehru University, New Delhi.

Sopher, D, ed. (1980): An Exploration of India: Geo-graphical Perspectives on Society and Culture (Ithaca, NY: Cornell University Press).

Srinivasan, Vivek (2010): “Understanding Public Services in Tamil Nadu: An Institutional Perspec-tive”, PhD dissertation, University of Syracuse.

UNDP (1990): Human Development Report 1990, United Nations Development Programme, New York.

Visaria, P and L Visaria (2003): “Long-Term Popu-lation Projections for Major States, 1991-2101”, Economic & Political Weekly, 8 November.

regional patterns of human and child deprivation in … and child...regional patterns of human and...

Documents