Constituency-Level Namibian Index of Multiple Deprivation 2001 Report

Constituency‐level Namibian Index of Multiple Deprivation 2001

Report

June 2011

Acknowledgements This project was undertaken by Professor Michael Noble, Dr Gemma Wright, Ms Joanna Davies, Dr Helen Barnes and Dr Phakama Ntshongwana of the Centre for the Analysis of South African Social Policy at the Oxford Institute of Social Policy at the University of Oxford in the UK. The project was funded by the United Nations Development Programme Namibia. The contents of this report do not necessarily reflect the views of the UNDP. Members of the Khomas Regional Council and the Namibian Central Bureau of Statistics are thanked for their contributions during the consultation period. David Avenell is thanked for producing the maps.

Disclaimer The University of Oxford has taken reasonable care to ensure that the information in this report is correct. However, no warranty, express or implied, is given as to its accuracy and the University of Oxford does not accept any liability for error or omission. The University of Oxford is not responsible for how the information is used, how it is interpreted or what reliance is placed on it. The University of Oxford does not guarantee that the information in this report is fit for any particular purpose and does not accept responsibility for any alteration or manipulation of the report.

Contents

Executive Summary ...... 6 Section 1: Introduction ...... 7 1.1 Background ...... 7 1.2 Defining poverty and deprivation ...... 8 1.3 The concept of multiple deprivation ...... 8 1.4 Data and geography ...... 8 Section 2: Domains and Indicators ...... 10 2.1 An introduction to the domains and indicators ...... 10 Domains ...... 10 Indicators ...... 10 2.2 Material Deprivation Domain ...... 11 Purpose of the domain ...... 11 Background ...... 11 Indicators ...... 11 Combining the indicators ...... 11 2.3 Employment Deprivation Domain ...... 12 Purpose of the domain ...... 12 Background ...... 12 Indicator ...... 12 Combining the indicators ...... 12 2.4 Health Deprivation Domain ...... 13 Purpose of the domain ...... 13 Background ...... 13 Indicator ...... 13 2.5 Education Deprivation Domain ...... 13 Purpose of the domain ...... 13 Background ...... 13 Indicators ...... 14 Combining the indicators ...... 14 2.6 Services Deprivation Domain ...... 14 Purpose of the domain ...... 14 Background ...... 14

Indicators ...... 15 Combining the Indicators ...... 15 2.7 Housing Deprivation Domain ...... 15 Purpose of the domain ...... 15 Background ...... 15 Indicators ...... 15 Combining the indicators ...... 15 Section 3: Methodology ...... 16 3.1 Consultation ...... 16 3.2 Constructing the domain indices ...... 16 3.3 Standardising and transforming the domain indices ...... 16 3.4 Weights for the domain indices when combining into an overall Index of Multiple Deprivation ...... 16 Section 4: Findings – Namibia ...... 17 4.1 Multiple deprivation ...... 17 4.2 Domains of deprivation at national level ...... 21 4.3 Domains of deprivation at regional level ...... 21 4.4 Domains of deprivation at constituency level ...... 22 Section 5: Findings ‐ Khomas ...... 34 5.1 Focusing on the Khomas Region ...... 34 Section 6 Conclusion and policy implications ...... 45 Conclusion ...... 45 Policy implications ...... 45 Annex 1: Indicators included in the NIMD 2001 ...... 46 Material Deprivation Domain ...... 46 Employment Deprivation Domain ...... 46 Health Deprivation Domain ...... 47 Education Deprivation Domain ...... 47 Services Deprivation Domain ...... 47 Housing Deprivation Domain ...... 47 Annex 2: The Shrinkage Technique ...... 48 Annex 3: Exponential Transformation ...... 50 Annex 4: NIMD 2001 constituency ranks ...... 51 Annex 5: Regional boundaries ...... 54

References ...... 55

Executive Summary

This report presents a profile of multiple deprivation in Namibia. The analysis is based on data from the 2001 Census of Population and is presented at constituency level.

A Namibian Index of Multiple Deprivation (NIMD) was constructed comprising six dimensions or ‘domains’ of deprivation: Material Deprivation, Employment Deprivation, Health Deprivation, Education Deprivation, Services Deprivation, and Housing Deprivation. Each domain measures the extent of deprivation at an area level for each constituency across Namibia. The domains, and the indicators that make up each domain, are described in detail in Section 2. The domains and indicators were selected in consultation with members of Khomas Regional Council and the Namibian Central Bureau of Statistics. Although for the purposes of this exercise the indicators had to be restricted to issues that are measured in the Census, every effort was made to capture the range of deprivations experienced by people in Namibia.

The aim was to produce a profile of relative deprivation across Namibia in order for the most deprived areas to be identified. As such it is possible for policy makers to consider a particular domain of deprivation, or to refer to the overarching NIMD which is a weighted combination of all six domains. The methodological techniques that were used for the construction of the domains and the NIMD are described in Section 3.

If all the constituencies are compared alongside one another, the most deprived constituency in Namibia using the NIMD 2001 is Okaku constituency in Oshana, followed by Kongola in Caprivi, and Omuntele in Oshikoto. The least deprived constituency in Namibia using the NIMD 2001 is Windhoek East. A number of maps of Namibia are presented, showing the NIMD 2001 as well as maps of each of the six domains of deprivation.

Although Khomas contains the least deprived constituency in Namibia (Windhoek East) it nevertheless does contain areas with high levels of deprivation. Using the NIMD 2001 the most deprived constituency in Khomas is Moses Garoëb. Maps are presented for Khoma Region, again showing the NIMD 2001 and each of the six domains of dperivation.

The NIMD 2001 is a useful tool for profiling deprivation across the country, alongside other sources of information and local knowledge. Though it relates to a 2001 time point, it serves as a benchmark against which change can be measured when the 2011 Census is released. The next stage of the research programme is to produce a sub‐constituency level NIMD so that pockets of deprivation within constituencies can be identified.

Section 1: Introduction

This report presents the constituency‐level Namibian Index of Multiple Deprivation 2001 (NIMD 2001). It was produced at constituency level using data from the 2001 Census of Population. The conceptual framework which underpins the model of multiple deprivation is described in this introductory section. In Section 2 the domains and indicators that were included in the NIMD are introduced and Section 3 summarises the methodological approach that was used. Section 4 presents the national, regional and constituency‐level results across Namibia while Section 5 focuses on the results for Khomas Region.

1.1 Background The main objective was to construct a summary measure of multiple deprivation in Namibia, which would enable deprivation to be quantified and profiled at a sub‐regional level. This would enable Khomas Regional Council – and other regional councils across Namibia – to rank their constituencies in order of deprivation, and also to set them in the context of all of Namibia’s constituencies.

This project will complement research undertaken on deprivation in Nambia (e.g. Central Bureau of Statistics, 2008; UNDP Namibia, 2007) as well as country‐level analysis (e.g. UNDP, 2009) by providing a sub‐regional (i.e. constituency‐level) profile of deprivation across the country. It also builds on extensive research undertaken by members of the Centre for the Analysis of South African Social Policy (CASASP) at the University of Oxford to develop small area level indices of deprivation in South Africa and the UK. In South Africa this includes the ward‐level Provincial Indices of Multiple Deprivation developed with Statistics South Africa and the Human Sciences Research Council (Noble et al., 2006a and 2009b), the datazone level South African Index of Multiple Deprivation (SAIMD 2001) (Noble et al., 2009a), and the municipality level SAIMD 2007 (Wright and Noble, 2009). The research group have also produced a municipality level South African Index of Multiple Deprivation for Children (SAIMDC 2001) with the Human Sciences Research Council (Barnes et al., 2007 and 2009), a datazone level SAIMDC 2001 (Wright et al., 2009a), and a municipality level SAIMDC 2007 (Wright et al., 2009b) which all focus on the population aged 0‐17 inclusive. CASASP has also undertaken a scoping exercise with HSRC about the possibility of developing small area level Indices of Deprivation in Southern Africa (Noble et al., 2007). CASASP’s sister centre, the Social Disadvantage Research Centre has produced indices of deprivation in the UK for England (e.g. Noble et al., 2006b and 2008; McLennan et al., 2011) , Scotland (Noble et al., 2003), Wales (Noble et al., 2000) and Northern Ireland (Noble et al., 2001).

The South African and UK Indices are used extensively by national and local governments – and by civil society, academia and business – to identify the most deprived areas and in many instances to focus support in these areas, in order to supplement more mainstream forms of social support. It is hoped that a Namibian Index of Multiple Deprivation will similarly be a useful tool for Namibian policy makers within Regional Councils and National Government.

1.2 Defining poverty and deprivation A leading poverty researcher sets out the case for defining poverty in terms of relative deprivation as follows:

‘Individuals, families and groups can be said to be in poverty if they lack the resources to obtain the types of diet, participate in the activities and have the living conditions and amenities which are customary or at least widely encouraged or approved in the societies to which they belong’ (Townsend, 1979, p31).

Though ‘poverty’ and ‘deprivation’ have often been used interchangeably, many have argued that a clear distinction should be made between them (see for example the discussion in Nolan and Whelan, 1996). It could be argued that the condition of poverty means not having enough financial resources to meet need, whereas deprivation refers to an unmet need, which is caused by a lack of resources of all kinds, not just financial resources.

In his article ‘Deprivation’ Townsend also lays down the foundation for articulating multiple deprivation as an accumulation of several types of deprivation (Townsend, 1987). Townsend’s formulation of multiple deprivation is the starting point for the model of small area deprivation which is presented here.

1.3 The concept of multiple deprivation The starting point for the NIMD is a conceptual model of multiple deprivation. The model of multiple deprivation is underpinned by the idea of separate dimensions of deprivation which can be recognised and measured, and are experienced by individuals living in an area. Multiple deprivation is therefore conceptualised as a weighted combination of distinct dimensions or domains of deprivation. An area level score for each domain is produced and these are then combined to form an overall Index of Multiple Deprivation.

The area itself can be characterised as deprived relative to other areas, in a particular dimension of deprivation, on the basis of the proportion of people in the area experiencing the type of deprivation in question. In other words, the experiences of the people in an area give the area its deprivation characteristics. The area itself is not deprived, though the presence of a concentration of people experiencing deprivation in an area may give rise to a compounding deprivation effect, but this is still measured by reference to those individuals. Having attributed the aggregate of individual experience of deprivation to the area, it is possible to say that an area is deprived in that particular dimension. Having measured specific dimensions of deprivation, these can be understood as domains of multiple deprivation.

1.4 Data and geography The constituency‐level NIMD was produced using the 2001 Namibian Census of Population which was supplied by the Namibian Central Bureau of Statistics for the purposes of this project. Whilst the intention is always to be concept‐led rather than 'data‐driven', the project team was restricted to selecting indicators from the range of questions included within the Census.

The NIMD was produced at constituency level. There are 107 constituencies in Namibia; the digitised boundaries were supplied to the research team by the Namibian Central Bureau of Statistics. Great

8 care was taken to ensure that the three datasets (relating to individuals, households, and mortality) were based on a set of common geographical boundaries.

Ideally, the NIMD would be produced at a smaller geographical level in order to facilitate the identification of pockets of deprivation within each constituency.

Section 2: Domains and Indicators

2.1 An introduction to the domains and indicators

Domains The NIMD contains six domains of deprivation:

Material Deprivation Employment Deprivation Health Deprivation Education Deprivation Services Deprivation Housing Deprivation

Following on from the definition of multiple deprivation outlined in Section 1 each domain is presented as a separate domain index reflecting a particular aspect of deprivation. Each domain seeks to measure only one dimension of deprivation, avoiding overlaps between the domains and providing a direct measure of the deprivation in question. Individuals can experience more than one type of deprivation at any given time. Therefore it is appropriate that the same person can be captured in more than one domain. So, for example, if someone was unemployed, had no qualifications and had no access to basic material goods they would be captured in the Employment Deprivation, Education Deprivation and Material Deprivation domains.

During the consultation process a number of other domains were discussed. These included: access to recreation facilities, level of participation in community activities, crime, food security, provision of emergency services, and availability of affordable transport. Unfortunately data relating to these issues was not available within the Census; these issues could be incorporated into further iterations of the NIMD if appropriate administrative or geographical data becomes available.

There was extensive discussion within the consultation process as to whether housing and services should be combined into a single domain which focused on the 'living environment'. However, the preference was to separate housing from service provision and treat them as two distinct domains.

Indicators Each domain index contains a number of indicators, thirteen in total. Annex 1 lists the indicators for each domain.

The aim for each domain was to include a parsimonious (i.e. economical in number) collection of indicators that comprehensively captured the deprivation for each domain, but within the constraints of the data available from the Census. The indicators selected were discussed at length within the consultation group and every effort was made to ensure that they were appropriate for the Namibian context.

When identifying indicators for the domains, it was important to ensure that they are direct measures of the domain of deprivation in question and specific to that domain.

The domains need to allow different geographical areas to be distinguished from one another; therefore it would be unhelpful to identify a deprivation which is experienced by most people in most areas as this would not enable the areas to be ranked relative to each other in terms of deprivation. The model is designed to be updated in two ways: first, to allow for the re‐evaluation of the number and nature of the dimensions of deprivation; second, to allow for new and more direct measures of those dimensions to be incorporated as data becomes available.

2.2 Material Deprivation Domain

Purpose of the domain The purpose of this domain is to capture the proportion of the population experiencing material deprivation in an area by reference to the percentage of the population who are deprived of access to basic material possessions.

Background In other indices that have followed this model (e.g. UK Indices), an Income Deprivation Domain was created. However, there is an argument that such a domain is inappropriate within an Index of Multiple Deprivation, because ‐ as explained above ‐ deprivations can be regarded as the outcomes of lack of income rather than the lack of income itself. To follow Townsend, within a multiple deprivation measure only the deprivations resulting from a low income would be included so low income itself would not be a component, but lack of material possessions would be included.

In any event, the Census does not have an income question and so an income poverty indicator, if included, would need to be modelled from a different data source such as the Namibian Household Income and Expenditure Survey. Such modelling work is being undertaken separately for CBS by Lux Development and will provide a complementary small area measure of income poverty.

For these reasons, a material deprivation domain was produced. A lack of access to basic material goods can be understood as a proxy for low income. The Census includes questions about access to material goods (e.g. television, radio, newspaper, telephone and computer). Internationally these are widely used measures of variations in living standards.

For the purposes of producing this domain three of the possible five household durables were included as indicators. Access to a television/radio, and telephone/cell phone represent important modes of communication and a means of accessing information crucial to one’s life and livelihood, though the quality of the services provided cannot be taken into account.

Indicators  Number of people living in a household with no access to a television or a radio; or  Number of people living in a household with no access to a telephone/cell phone.

Combining the indicators A simple proportion of people living in households experiencing one or more of the deprivations was calculated (i.e. the number of people living in a household with no access to a television/radio; or with no access to a telephone/cell phone divided by the total population).

2.3 Employment Deprivation Domain

Purpose of the domain This domain measures employment deprivation conceptualised as involuntary exclusion of the working age population from the world of work by reference to the percentage of the working age population who are unemployed.

Background The Census records employment status in line with the International Labour Organisation (ILO) ‘labour force framework’ and the ‘priority rules’ which give precedence to employment over all other activities ‘regardless of the amount of time devoted to it, which in extreme cases may be only one hour’ (Hussmanns, 2007, p6). Therefore a person is considered to be employed if during the 7 days prior to the census night they worked for at least one hour for pay, profit or family gain.

It follows that unemployment is defined as a situation of a total lack of work. The definition of unemployment adopted by the 13th International Conference of Labour Statistics (ICLS) stipulates three criteria which must be simultaneously met for a person to be considered unemployed. According to this official definition, the unemployed are those persons within the economically active population (aged 15‐65 inclusive) who during the reference period (for the 2001 Census this is the 7 days prior to census night) were:

1. Without work, i.e. in a situation of total lack of work; and 2. Currently available for work, i.e. not a student or homemaker or otherwise unavailable for work; and 3. Seeking work, i.e. taking steps to seek employment or self‐employment.

Using the Census it was not possible to measure whether unemployed people were available for work and seeking work.

Though other indices have also included people of working age who cannot work because of illness or disability, as they are involuntarily excluded from the world of work and internationally are regarded as the 'hidden unemployed' (Beatty et al., 2000), the consultation group wanted to limit this domain to the economically active population and therefore disabled or long‐term sick people were not included. The age band was modified to 15‐59 inclusive to reflect a concept of working age relevant to Namibia.

Indicator  Number of people aged 15‐59 who are unemployed.

Combining the indicators The domain was calculated as those identified as unemployed and aged 15 to 59 inclusive divided by the number of people who are economically active in that age group

2.4 Health Deprivation Domain

Purpose of the domain This domain identifies areas with relatively high rates of people who die prematurely. This domain measures premature mortality but not aspects of behaviour or environment that may be predictive of forthcoming health deprivation.

Background Though the consultation process raised the importance of measuring people's health status, and access to health facilities and healthcare, these issues could not be measured using the Census data. It was therefore not possible to include any measures of morbidity or access to health services.

Instead we use a form of standardised mortality ratio known as Years of Potential Life Lost (YPLL) which is an internationally recognised measure of poor health. This is the level of unexpected mortality weighted by the age of the individual who has died (For details about how this indicator was constructed see Blane and Drever, 1998). An area with a relatively high death rate in a young age group (including areas with high levels of infant mortality) will therefore have a higher overall YPLL score than an area with a similarly relatively high death rate for an older age group, all else being equal.

The YPLL indicator is a directly age and gender standardised measure of premature death (i.e. death under the age of 75)1. The YPLL measure is related to life expectancy in an area. Areas with low life expectancy will have high YPLL scores.

There are a number of factors which will contribute to low life expectancy/high YPLL scores in an area. High levels of infant mortality, high levels of perinatal mortality as well as high levels of serious illness such as HIV/AIDS and tuberculosis will all contribute to reduced life expectancy in an area and therefore high YPLL scores. Thus, although the YPLL is a mortality measure it does reflect the extent of serious ill‐health in an area. Although it would have been possible to have used infant mortality, under‐five mortality, and life expectancy as indicators, YPLL in effect combines all these issues into a single indicator and is therefore a broader and more useful overview of health deprivation in an area.

Indicator  Years of potential life lost

2.5 Education Deprivation Domain

Purpose of the domain This domain measures deprivation in educational attainment for people aged 15 to 59 inclusive.

Background Elsewhere in the SADC Region it has been shown that the level of educational attainment in the working age adult population is closely linked to an individual’s employment status and future opportunities for those individuals and their dependents (Bhorat et al., 2004).

1 Because the direct method of standardisation makes use of individual age/gender death rates it is often associated with small numbers. An empirical Bayes or ‘shrinkage’ technique is therefore used to smooth the individual age/gender death rates in order to reduce the impact of small number problems on the YPLL.

The 2001 Census includes a record of the level of education completed and a record of illiteracy. These two questions provide the best available measures of educational attainment and make up the indicators for this domain.

The consultation process additionally raised the importance of affordable education and availability of tertiary education opportunities, but again this could not be adequately captured using the Census.

Indicators  Number of 15‐59 year olds (inclusive) with no schooling completed at secondary level or above; or  Number of 15‐59 year olds (inclusive) who are illiterate.

Combining the indicators A simple proportion of the working age population (aged 15 to 59 years old inclusive) who had not completed schooling at secondary level or who are illiterate was calculated (i.e. the number of people with no schooling completed at secondary level or above, or who are illiterate divided by the population, all aged 15 to 59 inclusive).

2.6 Services Deprivation Domain

Purpose of the domain The purpose of this domain is to measure deprivation in basic services to the home.

Background The Census questionnaire provides indicators on household access to basic amenities. These aspects of the immediate environment in which people live impact on the quality of their life and provide good measures of deprivation in terms of access to services.

Measuring access to electricity as a basic amenity is a useful indicator of living environment deprivation. To measure access to electricity three Census indicators were considered; main source of energy for: cooking, lighting and heating. Although cost, availability and effectiveness are factors in the consumption of all energy supplies, it has been argued that the choice of fuel for cooking may be influenced by cultural preference rather than availability alone, whereas the use of electricity for lighting would generally be the preferred choice if available and therefore provides a more valid measure of deprivation in terms of access to energy (Bhorat et al., 2004).

Access to clean drinking water and sanitation facilities is essential for the good health of the population and thus an important indicator to include in this domain. Having no access to piped water within the home or within 400 metres of the home is considered a deprivation, to be in line with the MDG. However, the consultation group argued that the distance should be set lower at 200 metres. It was also decided that toilet facilities other than flush toilets or VIP should be considered inadequate.

A further indicator about refuse collection services was included, in order to measure those without access to regular refuse collection services.

Indicators  Number of people living in a household without the use of electricity or solar power for lighting; or  Number of people living in a household without access to a flush toilet or VIP; or  Number of people living in a household without piped water/borehole/ borehole with covered tank (but not open tank)/protected well inside their dwelling or yard or within 200 metres; or  Number of people living in a household without a regular refuse collection service

Combining the Indicators A simple proportion of people living in households experiencing one or more of the deprivations was calculated (i.e. the number of people living in a household without electricity or solar power for lighting/or without adequate toilet/or without adequate water provision/or without regular refuse collection divided by the total population).

2.7 Housing Deprivation Domain

Purpose of the domain The purpose of this domain is to measure those whose housing is inadequate in terms of construction and overcrowding.

Background The quality of housing construction provides an important indicator for the quality of day‐to‐day life and vulnerability to shocks such as adverse weather conditions (Bhorat et al 2004; Programme of Action Chapter 2 Word Summit for Social Development Copenhagen 1995; SADC 1998).

There was much discussion in the consultation process about traditional dwellings and their adequacy. Though the census contains fairly precise information about materials used in the construction process there is no way of identifying whether the resultant buildings were of a high quality or not; it was therefore agreed that only shacks could be reliably identified as inadequate.

The crowding indicator is calculated by dividing the number of people in the household by the number of rooms excluding bathrooms, toilets, kitchens, stoops and verandas. Different versions of the crowding indicator were considered. It was felt that the most appropriate measure of crowding was to classify three or more people per room as a deprivation. Setting the capacity cut‐off at two or more people per room was considered. However, it was felt that this lower capacity would capture too many non‐deprived people, for example relatively well‐off couples sharing a one room urban apartment.

Indicators  Number of people living in a household that is a shack; or  Number of people living in a household with three or more people per room.

Combining the indicators A simple proportion of people living in households experiencing either one of these deprivations was calculated (i.e. the number of people living in a shack/or in overcrowded conditions divided by the total population).

Section 3: Methodology

3.1 Consultation The domains and indicators contained in the NIMD were discussed with members of the Central Bureau of Statistics and civil servants from Khomas Regional Council in Windhoek in January 2011. This two‐day consultative process enabled the Census to be scrutinised in detail by the group in order to ensure that the most appropriate domains and indicators for the Namibian context were identified and incorporated into the NIMD.

3.2 Constructing the domain indices In all domains apart from the Health Domain, the overall score is a simple proportion of the relevant population, and so can be easily interpreted.

As Censuses can be regarded as a sample from a super‐population, it is important to consider and deal with large standard errors. A technique that takes standard errors into account but still enables one to then combine the domains into an overall index of multiple deprivation is called Bayesian shrinkage estimation. Specifically, the scores for constituencies can be unreliable when the deprived population is small and so the shrinkage technique was applied to each of the domains.

The ‘shrunk’ estimate is the weighted average of the original constituency level estimate and an appropriate larger spatial unit. The weight is based on the standard error of the original constituency estimate and the amount of variation within the Region. For further details about this technique please see Annex 2 and also Noble et al., 2006b.

3.3 Standardising and transforming the domain indices Having obtained a set of Domain Indices these needed to be combined into an overall Index of Multiple Deprivation. In order to combine Domain Indices which are each based on different metrics there needed to be some way to standardise the scores before any combination could take place. A form of standardisation and transformation is required that meets the following criteria. First it must ensure that each Domain has a common distribution; second, it must not be scale dependent (i.e. conflate size with level of deprivation); third, it must have an appropriate degree of cancellation built into it; and fourth, it must facilitate the identification of the most deprived Constituencies. The exponential transformation of the ranks best meets these criteria and was applied in the NIMD 2001. For further details see Annex 3 and Noble et al., 2006b.

3.4 Weights for the domain indices when combining into an overall Index of Multiple Deprivation Domains are conceived as independent dimensions of multiple deprivation, each with their own additive impact on multiple deprivation. The strength of this impact, though, may vary between domains depending on their relative importance. As a starting point we recommended equal weights for the domains and this was supported by the consultation group. Each domain was therefore assigned a weight of 1. The NIMD was therefore constructed by adding the standardised and transformed domain indices with equal weights.

Section 4: Findings – Namibia

4.1 Multiple deprivation This subsection considers multiple deprivation at constituency level. Due to the fact the NIMD was produced at constituency level it is not possible to provide the NIMD at higher levels of spatial aggregation (i.e. regional and national levels). The rest of section 4.1 therefore presents the overall NIMD at constituency level (see Table 1 and Map 1). However, the constituency level NIMD is summarised at regional level in Chart 1 below.

The most deprived constituency in Namibia is Okaku in Oshana Region, followed by Kongola in Caprivi and Omuntele in Oshikoto . The least deprived constituency is Windhoek East in Khomas Region. Table 1 shows the five most deprived and five least deprived constituencies in Namibia showing their score on the NIMD and rank (where 1=most deprived and 107=least deprived). Annex 4 lists this information for all of the constituencies in Namibia.

Table 1 The five most deprived and least deprived constituencies on the NIMD 2001

Region Constituency NIMD Score NIMD Rank Oshana Okaku 391.4 1 Caprivi Kongola 377.1 2 Oshikoto Omuntele 371.7 3 Oshana Uukwiyu 366.2 4 Omusati Otamanzi 350.5 5

Hardap Rehoboth Urban West 56.1 103 Khomas Khomasdal North 46.8 104 Karas Oranjemund 29.1 105 Khomas Windhoek West 19.6 106 Khomas Windhoek East 7.4 107

Chart 1 shows the minimum, maximum and median rank of constituencies in each region, and the interquartile range for the overall NIMD. Before describing the results, the following page provides further details about how to interpret the chart.

How to interpret the boxplot charts: The vertical green line for each region shows the range of the ranks of the constituencies in a region (for the example chart shown on this page, results are just presented for Khomas Region). The most deprived constituency in Namibia is ranked 1, and the least deprived constituency is ranked 107. The ‘T’ at the top of the green line shows the rank of the least deprived constituency in the region. The ‘Upside‐down T’ at the bottom of the green line shows the rank of the most deprived constituency in the region. The green box for each region shows the range of the NIMD ranks of the middle 50% of constituencies in the region (the interquartile range). The horizontal line within the box for each region represents the rank of the median constituency within that region.

If the box is relatively short this indicates that constituencies are ranked in a narrow range, with similar NIMD ranks (and therefore similar levels of multiple deprivation). If this box sits towards the bottom of the chart it tells us that constituencies in the region are concentrated in the most deprived part of the national distribution of the NIMD. If the box sits towards the top of the chart it tells us that the ranks of the constituencies in the region are concentrated in the least deprived part of the national distribution.

The charts therefore provide an information rich summary – at regional level – of the constituencies in each region.

Chart 1: Namibian Index of Multiple Deprivation 2001 Constituency Level by Region: Interquartile Range 100 80 60 40 20 Rank of Constituency [where 1 = most deprived] 0

vi s o ri a na ene ena a ikot upa ap Karas h C Erongo un s Hardap Khom K Omusati Osh Kavango Omaheke O hangw tjozondj O O

Chart 1 shows that Caprivi, Kavango and Ohangwena regions have their constituencies concentrated at the most deprived end of the spectrum. In contrast, Erongo, Hardap, Karas and Khomas have their constituencies concentrated in the least deprived part of the national distribution.

Map 1 shows the NIMD for the whole of Namibia.

How to interpret the maps. The maps in this section of the report relate to the whole of Namibia and are presented at constituency level. The constituencies are divided into five equal groups – quintiles – which enables one to identify the 20% most deprived constituencies in the country (shaded deep blue) through to the 20% least deprived constituencies in the country (shaded yellow). A map of the 13 Regions in Namibia, showing their names and boundaries, can be found in Annex 5 for ease of reference.

In Map 1, it is clear that the most deprived constituencies in Namibia are located in the north of the country, corresponding with the data shown in Table 1. The most deprived constituencies are located within Omusati, Ohangwena, Oshikoto, Oshana, Kavango, Otjozondjupa and Caprivi.

4.2 Domains of deprivation at national level Although we cannot calculate multiple deprivation rates as such we can present each of the individual domains of deprivation at both national and regional levels, and for all domains except the health domain we can compare the domain scores.

At a national level, and based on the definitions of deprivation used in the NIMD:

 64.6% of the total population experience material deprivation;  31.4% of the economically active population experience employment deprivation;  76.2% of working age people (15‐59 inclusive) are educationally deprived;  81.7% of the total population experience services deprivation;  35.2% of the total population experience housing deprivation.

4.3 Domains of deprivation at regional level Table 2 below shows the rates of the individual domains of deprivation for each of the regions in Namibia.

Table 2: Regional level domain scores for five domains in the NIMD 2001

Material Employment Education Services Housing Region Deprivation Deprivation Deprivation Deprivation Deprivation Caprivi 73.0% 17.5% 76.7% 93.4% 64.2% Erongo 27.4% 34.2% 70.3% 35.3% 47.5% Hardap 42.9% 33.9% 78.4% 78.4% 48.1% Karas 37.2% 28.6% 74.7% 62.6% 48.6% Kavango 75.1% 20.4% 86.8% 99.1% 41.7% Khomas 38.0% 29.4% 58.8% 39.5% 46.6% Kunene 78.3% 23.5% 82.9% 92.8% 53.7% Ohangwena 86.0% 37.3% 84.3% 99.6% 14.4% Omaheke 62.8% 24.0% 82.2% 89.0% 66.1% Omusati 83.2% 36.5% 82.5% 99.1% 12.4% Oshana 57.0% 40.9% 74.6% 91.6% 20.6% Oshikoto 83.4% 45.2% 81.4% 95.0% 17.3% Otjozondjupa 57.0% 31.7% 77.3% 75.0% 55.2% Namibia 64.6% 31.4% 76.2% 81.7% 35.2%

In terms of material deprivation, the most deprived region is Ohangwena where 86% of the population are materially deprived. The next most deprived regions on this domain are Oshikoto (83.4%) and Omusati (83.2%). The least materially deprived region is Erongo (27.4%) followed by Karas (37.2%) and Khomas (38%).

The most employment deprived regions are Oshikoto (45.2% and Oshana (40.9%), with the least employment deprived region being Caprivi at 17.5%.

In terms of education deprivation this is highest in Kavango (86.8%) and Ohangwena (84.3%). Khomas (58.8%) is the least deprived on this domain.

Ohangwena (99.6%), Omusati (99.1%) and Kavango (99.1%) are by far the most deprived regions in terms of services whereas Erongo and Khomas (35.3% and 39.5% respectively) are the least services deprived.

Omaheke (66.1%) and Caprivi (64.2%) are the regions with the highest levels of housing deprivation whereas Omusati (12.4%) and Ohangwena (14.4%) are the least housing deprived.

4.4 Domains of deprivation at constituency level The rest of this section considers deprivation at constituency level across the whole of Namibia.

The following box plot charts look at the interquartile range of the constituency ranks of deprivation for each region for each of the domains. Chart 2 shows the interquartile range for material deprivation. It is clear that the constituencies in Caprivi, Kavango, Kunene, Ohangwena, Omusati, and Oshikoto are concentrated towards the deprived end of the spectrum; this is in contrast to Erongo, Hardap, Karas and Khomas where the constituencies are concentrated towards the less deprived end of the distribution.

Chart 2: Namibian Index of Multiple Deprivation 2001 Material Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

vi s o ri a na ene ena a ikot upa ap Karas h C Erongo un s Hardap Khom K Omusati Osh Kavango Omaheke O hangw tjozondj O O

Map 2 presents the material deprivation domain at constituency level for the whole of Namibia. The most deprived constituencies (shaded deep blue) can be found in Kunene, Omusati, Ohangwena, Kavango, Oshikoto, Oshana, Caprivi and Otjozondjupa.

Chart 3: Namibian Index of Multiple Deprivation 2001 Employment Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

vi s o ri a na ene ena a ikot upa ap Karas h C Erongo un s Hardap Khom K Omusati Osh Kavango Omaheke O hangw tjozondj O O

In terms of employment deprivation, the picture is rather different. The only region to show concentrated employment deprivation with a very low median rank for the constituencies is Oshikoto (see chart 3). In many other regions including Oshana, Omusati and Ohangwena, though there is employment deprivation the range is much greater with some constituencies relatively less deprived. Only Caprivi and Kunene show relatively low levels of employment deprivation. For most other regions the range occupies the middle ground. This is reflected in Map 3 below.

Chart 4 relates to health deprivation. In almost all regions there is quite a concentration of health deprivation. Caprivi and Ohangwena show health deprivation concentrated at the more deprived end of the spectrum whereas Omaheke, Otjozondjupa and to a certain extent Khomas and Erongo have their constituencies concentrated at the least deprived end of the spectrum.

Chart 4: Namibian Index of Multiple Deprivation 2001 Health Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

p o s i a a na na oto d ma e eke k up ar Karas usat ha Caprivi Erongo unene ah shi H Kho K Om Os Kavang Om O hangw tjozondj O O

Map 4 presents the health deprivation domain at constituency level. The most deprived constituencies on this domain are located within Omusati, Oshana, Ohangwena, Oshikoto, Kavango and Caprivi.

With regard to education deprivation, in most regions the constituencies show a wide range of ranks from very deprived to less deprived (see chart 5). The best example of this is Oshana. For some regions the constituencies are more concentrated; Khomas’ constituencies are at the least deprived end of the distribution whereas Kavango’s constituencies are concentrated at the most deprived part of the distribution.

Chart 5: Namibian Index of Multiple Deprivation 2001 Education Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

vi s o ri a na ene ena a ikot upa ap Karas h C Erongo un s Hardap Khom K Omusati Osh Kavango Omaheke O hangw tjozondj O O

Map 5 presents the education deprivation domain at constituency level for the whole of Namibia. The most deprived constituencies are located within the north and north‐eastern parts of the country, particularly within Kunene, Omusati, Oshana, Ohangwena, Oshikoto, Kavango, Caprivi, Otjozondjupa and Omaheke.

In terms of services deprivation (Chart six), constituencies within Caprivi, Ohangwena, Omusati, Oshana and Oshikoto are grouped towards the most deprived end of the spectrum. On the other hand, constituencies in Erongo, Hardap, Karas and Khomas are concentrated towards the least deprived end of the spectrum.

Chart 6: Namibian Index of Multiple Deprivation 2001 Services Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

p o s i a a na na oto d ma e eke k up ar Karas usat ha Caprivi Erongo unene ah shi H Kho K Om Os Kavang Om O hangw tjozondj O O

Map 6 presents the services deprivation domain at constituency level for the whole of Namibia. The most deprived constituencies (shaded deep blue) are located in Omusati, Oshana, Ohangwena, Oshikoto, Kavango and Caprivi.

Housing deprivation is widespread across regions (see chart 7), with the exception of Ohangwena Omusati, Oshana and Oshikoto which have relatively less housing deprivation, based on the definition of housing deprivation used in the NIMD. This is not to say that these areas have no housing deprivation; rather that they have relatively low levels of deprivation in terms of overcrowding and people living in shacks based on the 2001 Census data.

Chart 7: Namibian Index of Multiple Deprivation 2001 Housing Deprivation Domain Constituency Level by Region: Interquartile Range 100 80 60 40 20 0 Rank of Constituency [where 1 = most deprived]

vi s o ri a na ene ena a ikot upa ap Karas h C Erongo un s Hardap Khom K Omusati Osh Kavango Omaheke O hangw tjozondj O O

Map 6 presents the housing deprivation domain at constituency level for the whole of Namibia. The most deprived constituencies are located in Kunene, Erongo, Oshikoto, Caprivi, Otjozondjupa, Omaheke, Hardap and to the north of Windhoek in Khomas.

Section 5: Findings - Khomas

5.1 Focusing on the Khomas Region Section 4 presented the results of the 2001 Namibian Index of Multiple Deprivation for the whole of Namibia, at national, regional and constituency levels. In this section we just consider Khomas region. From the regional profiles in Section 4, it is clear that Khomas is not one of the most deprived regions in the country. Indeed, Khomas is the least deprived region in Namibia on the education deprivation domain, the second least deprived region on the services deprivation domain, and the third least deprived region on the material deprivation domain. However, as was stressed at the beginning of Section 4, this does not mean that there is no deprivation within Khomas, rather that there are relatively lower levels of deprivation ‐ as measured in the NIMD ‐ than in many other areas of Namibia.

For policy makers in Khomas it is useful not only to contextualise the region compared to other regions and the country as a whole, as has been undertaken in Section 4, but also to explore which areas within Khomas are the most deprived. Whilst some forms of provision will be universal across Namibia or within Khomas, other interventions may require a prioritisation between areas. This report enables deprivation to be considered down to constituency level, and therefore for the ten constituencies within Khomas to be compared alongside each other.

Table 3 shows the NIMD 2011 score and rank for each of the constituencies in Khomas. The most deprived constituency in Khomas is Moses Garoëb, which is therefore given a rank of 1 among the constituencies in Khomas. If ranked alongside all constituencies in Namibia (the final column in Table 3) we see that it ranks as the 46th most deprived constituency in Namibia. The least deprived constituency in Khomas – and in Namibia as a whole – is Windhoek East.

Table 3: NIMD Score and Rank for constituencies in Khomas

NIMD rank (within NIMD rank Khomas) (within Namibia) NIMD Where 1=most Where 1=most Constituency score deprived deprived Moses Garoëb 215.3 1 46 Tobias Hainyeko 204.2 2 50 Samora Machel 139.9 3 78 Katutura Central 127.0 4 85 Katutura East 101.1 5 94 Soweto 86.3 6 97 Windhoek Rural 84.5 7 98 Khomasdal North 46.8 8 104 Windhoek West 19.6 9 106 Windhoek East 7.4 10 107

This information is also depicted in Map 8 below. Unlike the maps in Section 4, this map (and maps 9‐14 later in this section) just show the constituencies in Khomas and are shaded in the same colour

34 with the lightest shading relating to the least deprived constituency. The constituencies are individually coloured and the colours do not therefore relate to the quintiles in Section 4. The zoom‐ out box shows all of the constituencies in Khomas, whereas the main map zooms in so that the small (in physical size) constituencies within Khomas are visible.

Table 4 provides the domain scores for each constituency in Khomas, excluding health as the health score is not calculated as a rate. The other five domains are in the form of simple deprivation rates. So for example 63.5% of the population in Tobias Hainyeko constituency experienced material deprivation in 2001. The within‐Khomas ranks are shown as well as the domain scores, for each constituency in Khomas (where 1=most deprived). For the health domain, the rank is shown in the final column.

In terms of material deprivation, the most deprived constituencies in Khomas are Tobias Hainyeko (with 64% of the population experiencing material deprivation) followed by Moses Garoëb and Katutura Central.

In relation to employment deprivation, the most deprived constituencies in Khomas are Katutura Central (with 39% of the relevant population being employment deprived) followed by Moses Garoëb and Tobias Hainyeko.

These same three constituencies are also the three most deprived constituencies in Khomas in terms of education deprivation: Moses Garoëb is the most deprived (with 73% of the relevant population being education deprived) followed by Tobias Hainyeko and Katutura Central.

In relation to services deprivation, the three most deprived constituencies in Khomas are Tobias Hainyeko (with a very high 83% of the total population being services deprived) followed by Windhoek Rural and Moses Garoëb.

The three most deprived constituencies in Khomas in terms of housing deprivation are Tobias Hainyeko (with a very high 81% of the total population being housing deprived) followed by Moses Garoëb and Samora Machel.

The three most deprived constituencies in Khomas in terms of health deprivation are Moses Garoëb, Soweto and Tobias Hainyeko.

It should be noted that Windhoek East and Windhoek West are the two least deprived constituencies on all six of the domains.

Maps 9‐14 present each of the six domains at constituency level for Khomas. As with Map 8, the constituencies in Khomas are shaded in the same colour with the lightest shading relating to the least deprived constituency. It is intended that these maps should provide accessible profiles of the domains of deprivation in the Khomas Region, and they can be considered in conjunction with Table 4.

Table 4: Domain Ranks for each constituency in Khomas (plus domain scores in the form of rates of deprivation for all domains except health)

Material Employment Education Services Housing Deprivation Deprivation Deprivation Deprivation Deprivation Health rank rank rank rank rank Deprivation (within (within (within (within (within rank (within Khomas) Khomas) Khomas) Khomas) Khomas) Khomas) Material Where Employment Where Education Where Where Where Where Deprivation 1=most Deprivation 1=most Deprivation 1=most Services 1=most Housing 1=most 1=most Constituency Rate deprived Rate deprived Rate deprived Deprivation deprived Deprivation deprived deprived Tobias Hainyeko 63.5% 1 38.0% 3 71.3% 2 83.4% 1 81.3% 1 3 Katutura Central 49.6% 3 38.5% 1 70.2% 3 26.2% 6 46.8% 4 4 Katutura East 35.4% 6 37.1% 4 67.5% 5 34.4% 5 42.1% 6 5 Khomasdal North 26.5% 8 24.0% 7 54.1% 8 10.0% 8 32.4% 8 8 Soweto 31.5% 7 32.7% 6 60.0% 7 10.7% 7 33.9% 7 2 Samora Machel 46.4% 5 36.6% 5 68.2% 4 46.3% 4 64.9% 3 6 Windhoek East 6.4% 10 8.6% 10 23.2% 10 7.3% 10 9.5% 10 10 Windhoek Rural 48.3% 4 23.8% 8 67.0% 6 79.5% 2 46.6% 5 7 Windhoek West 11.0% 9 14.1% 9 36.2% 9 7.8% 9 13.7% 9 9 Moses Garoëb 56.4% 2 38.1% 2 72.6% 1 73.7% 3 78.8% 2 1

Section 6 Conclusion and policy implications

Conclusion A profile of multiple deprivation has been produced for all constituencies in Namibia, with additional analysis undertaken for Khomas Region.

Before considering the policy implications it should be stressed that the NIMD provides a profile of relative deprivation across Namibia. Even the least deprived areas contain deprivation; they are simply less deprived than other areas with higher levels of deprivation.

The NIMD is based on data relating to 2001 and change may have taken place since then. Forthcoming analysis using the 2011 Census will shed light on the extent to which change has occurred. It is also important to note that constituencies vary considerably by population size. Indices elsewhere internationally tend to be produced at a smaller area level (in terms of population size), using area units of roughly equal population size in order to compare the level of deprivation in areas as fairly as possible. It is hoped that this will become possible in Namibia too, thereby enabling pockets of deprivation in otherwise less deprived areas to be revealed.

Policy implications There are many ways in which a small area index of multiple deprivation can support pro‐poor policy initiatives. By providing reliable information on both the distribution of multiple deprivation and the distribution of the individual domains of deprivation, small area indices of deprivation can provide the evidence base for directing extra resources for special policies (known as area‐based policy interventions) at the most deprived areas. Area‐based policy interventions should not be seen as supplanting or replacing mainstream interventions to tackle poverty and deprivation. Rather, they should be seen as supplemental policy interventions to tackle those aspects of deprivation which are concentrated spatially.

By describing the geographical distribution and extent of individual dimensions of deprivation and of overall multiple deprivation, small area indices of deprivation can also provide a baseline map of deprivation against which progress in poverty reduction in these areas can be measured.

Thus, the constituency level NIMD provides a profile of the most deprived areas in Namibia, and within Khomas in particular. It provides policy makers with an evidence‐base on which to make decisions regarding resource allocation, the location of special programmes, and where to prioritise government support and service delivery relating to the various domains of deprivation. Whilst some policies will be rolled‐out universally across Namibia, other policies may require knowledge about the conditions in particular areas. The following examples illustrate how this could work. However, it must be stressed that these are simply illustrative and precise policy prescriptions will depend on local knowledge and policy priorities.

Taking the index on a domain by domain basis, we can first consider material deprivation. Material deprivation exists when people have insufficient resources to purchase or obtain important material possessions such as having access to a phone or to information using TV or radio. This may indicate very little employment in an area or very poorly paid employment in an area. This might suggest the

45 deployment of special job creation initiatives (see below). It might also mean that Social Security provision is not reaching those to whom it is targeted. If this is the case special take‐up measures may be needed. The South African Index of Multiple Deprivation was used in this way to promote the take‐up of the Child Support Grant.

Considering the employment domain, areas exhibiting high levels of employment deprivation might be prioritised for job creation projects or help directed to promote agricultural development.

Similarly, areas with high levels of health deprivation might be prioritised for further investigation as to what are the main drivers of high mortality in the area in order to improve health and healthcare services in the area.

If an area shows extensive education deprivation this may mean that either there are high levels of illiteracy in the area or that there is a large proportion of the population who did not attend secondary school. In these areas special adult education or literacy projects might be established.

Where there are high levels of services deprivation then this suggests the need to prioritise the rollout of services.

Similarly if there are high levels of housing deprivation this suggests either overcrowding or large numbers of shacks, or both, suggesting the need to prioritise the building of decent houses in the area.

However, it must be emphasised that, given the very high levels of poverty and deprivation across Namibia, even the least deprived areas will contain large numbers of poor people. Put another way, just because an area does not show as very deprived on any particular domain deprivation it does not mean that there are no deprived people living in the area. It is for this reason that spatially directed policy initiatives are not the entire solution to poverty and deprivation. They should be seen as a complement to not a substitution for mainstream anti‐poverty strategies.

Annex 1: Indicators included in the NIMD 2001

Material Deprivation Domain Numerator  Number of people living in a household with no access to a television or a radio; or  Number of people living in a household with no access to a telephone/cell phone. Denominator This domain used the total population as a denominator

Employment Deprivation Domain Numerator  Number of people aged 15‐59 who are unemployed.

Denominator The denominator for this domain was the total economically active population aged 15‐59 inclusive.

Health Deprivation Domain Numerator  Years of potential life lost

Education Deprivation Domain Numerator  Number of 15‐59 year olds (inclusive) with no schooling completed at secondary level or above; or  Number of 15‐59 year olds (inclusive) who are illiterate. Denominator This domain used the population of 15‐59 year olds (inclusive) for the denominator.

Services Deprivation Domain Numerator  Number of people living in a household without the use of electricity or solar power for lighting; or  Number of people living in a household without access to a flush toilet or VIP; or  Number of people living in a household without piped water/borehole/ borehole with covered tank (but not open tank)/protected well inside their dwelling or yard or within 200 metres; or  Number of people living in a household without a regular refuse collection service Denominator The total population was used as the denominator for this domain.

Housing Deprivation Domain Numerator  Number of people living in a household that is a shack; or  Number of people living in a household with three or more people per room. Denominator The total population was used as the denominator for this domain.

Annex 2: The Shrinkage Technique2

In some areas, particularly where populations at risk are small, data may be ‘unreliable’, that is more likely to be affected by measurement error or sampling error, with particular constituencies getting unrepresentatively low or high scores on certain indicators. The extent of a score’s ‘unreliability’ can be measured by calculating its standard error. A technique known as ‘shrinkage estimation’ (i.e. empirical Bayesian estimation) has been used to deal with this problem.3

Shrinkage involves moving unreliable constituency scores (i.e. those with a high standard error) towards another more robust score. This may be towards more deprivation or less deprivation. There are many possible candidates for the more robust score to which an unreliable score could move. The region mean has been selected for this purpose but others could, in theory, include the Namibian mean or the means of areas with similar characteristics.

Arguably, the movement of unreliable scores towards the mean score for Namibia would be inappropriate because of the large variation across the country and because it would be preferable to take into account local circumstances. It was therefore concluded that shrinkage to the region mean was the best and most reliable procedure. This is in essence the same as shrinking to the population weighted constituency mean for a region.

The actual mechanism of the procedure is to estimate deprivation in a particular constituency using a weighted combination of (a) data from that constituency and (b) data from the region. The weight attempts to increase the efficiency of the estimation, while not increasing its bias. If the constituency has a high standard error and a region appears to be an unbiased estimation of the constituency score then the constituency score moves towards the region score.

Although most scores move a small amount, only unreliable scores, that is those with a large standard error, move significantly. The amount of movement depends on both the size of the standard error and the amount of heterogeneity amongst the constituencies in a region.

The ‘shrunken’ estimate of a constituency‐level proportion (or ratio) is a weighted average of the two ‘raw’ proportions for the constituency and for the corresponding region.4

* pj )p

2 This section is a modified version of that contained in the technical annex to Noble et al., 2006a. 3 For England see Noble, Smith et al, 2000a p16; Noble, Wright et al, 2004 p16; for Wales see Noble, Smith, Wright et al, 2000 p8; for Northern Ireland see Noble, Smith, Wright et al, 2001 p11; Noble, Barnes et al, 2005a p7; and for Scotland see Noble, Wright et al, 2003b p15. 4 Where appropriate the weighted average is calculated on the logit scale, for technical reasons, principally because the logit of a proportion is more nearly normally distributed than the proportion itself.

The ‘shrunken’ constituency‐level estimate is the weighted average

* z j  wj z j  (1 wj )z [1]

where zj is the constituency level proportion, z is the region level proportion, wj is the weight given to the ‘raw’ constituency ‐j data and (1wj) the weight given to the overall proportion for the region.

The formula used to determine wj is

2 1 s j w j  2 2 [2] 1 s j 1 t

2 where sj is the standard error of the constituency level proportion, and t is the inter‐constituency variance for the k constituencies in the region, calculated as

1 k 2  2 t =  ( z j z) [3] k 1 j1

Annex 3: Exponential Transformation

In order to combine the domains into an overall NIMD, the domain scores are first standardised by ranking and then the ranks are transformed to an exponential distribution. The exponential distribution has a number of properties, most importantly that it enables control over cancellation and it helps identify the most deprived constituencies.

The purpose of exponential transformation is to adjust the range of each transformed domain to make them comparable. This is important when the domains are combined because simply averaging the ranks could result in high deprivation in one domain being cancelled out by low deprivation on a different domain.

Applying the exponential transformation converts each domain to a distribution with a range of 0 to 100, and a score of 100 for the most deprived small area. Twenty five per cent of small areas have a score higher than 50. The skewness in the transformed distribution reduces the extent to which deprivation on one domain can be cancelled by lack of deprivation on another. The exponential transformation formula selected enables the most deprived constituencies to be identified. The formula distributes the scores to stretch out the 25% highest scoring (most deprived) constituencies and compress the less deprived end of the distribution.

The transformation used is as follows. For any constituency, denote its rank on the Domain, scaled to the range [0,1], by R (with R=1/N for the least deprived, and R=N/N, i.e. R=1, for the most deprived, where N=the number of small areas in Namibia).

The transformed Domain, X say, is X = ‐45.5*ln{1 ‐ R*[1 ‐ exp(‐100/45.5)]} where ln denotes natural logarithm and exp the exponential or antilog transformation.

Annex 4: NIMD 2001 constituency ranks

Region Constituency NIMD Score NIMD Rank Oshana Okaku (most deprived) 391.4 1 Caprivi Kongola 377.1 2 Oshikoto Omuntele 371.7 3 Oshana Uukwiyu 366.2 4 Omusati Otamanzi 350.5 5 Ohangwena Omulonga 349.9 6 Caprivi Sibinda 349.8 7 Kavango Ndyona 345.9 8 Ohangwena Ongenga 345.9 9 Ohangwena Endola 339.3 10 Ohangwena Ondobe 328.3 11 Oshikoto Onyaanya 326.0 12 Omusati Okahao 307.0 13 Omusati Tsandi 302.2 14 Ohangwena Omundaungilo 301.7 15 Caprivi Kabe 300.7 16 Kavango Kapako 299.6 17 Oshikoto Eengondi 297.4 18 Otjozondjupa Tsumkwe 295.9 19 Kavango Mukwe 295.3 20 Kavango Rundu Rural East 295.1 21 Caprivi Katima Mulilo Rural 294.7 22 Oshikoto Onayena 293.9 23 Oshikoto Okankolo 292.6 24 Ohangwena Epembe 276.0 25 Caprivi Linyanti 273.7 26 Kunene Epupa 271.8 27 Oshana Okatyali 269.5 28 Kunene Sesfontein 269.4 29 Ohangwena Ohangwena 267.1 30 Ohangwena Oshikango 264.7 31 Kavango Kahenge 251.5 32 Ohangwena Engela 249.5 33 Oshana Ompundja 246.7 34 Omusati Elim 246.2 35 Omaheke Otjombinde 246.0 36 Kavango Mashare 240.1 37 Omusati Etayi 234.8 38 Kavango Mpungu 232.2 39

Region Constituency NIMD Score NIMD Rank Omusati Onesi 227.7 40 Oshikoto Oniipa 224.0 41 Oshikoto Omuthiyagwiipundi 222.9 42 Ohangwena Okongo 222.3 43 Otjozondjupa Okakarara 220.2 44 Omusati Okalongo 217.0 45 Khomas Moses ||Garoëb 215.3 46 Erongo Daures 210.1 47 Hardap Gibeon 208.0 48 Karas Berseba 205.0 49 Khomas Tobias Hainyeko 204.2 50 Oshikoto Olukonda 197.0 51 Kavango Rundu Rural West 195.9 52 Omaheke Aminuis 193.0 53 Kunene Opuwo 189.5 54 Oshikoto Guinas 188.0 55 Omusati Outapi 185.1 56 Omaheke Kalahari 183.7 57 Hardap Rehoboth Rural 181.1 58 Oshana Okatana 180.9 59 Omaheke Steinhausen 175.6 60 Omusati Ruacana 175.4 61 Kunene Khorixas 175.0 62 Ohangwena Eenhana 173.5 63 Omaheke Epukiro 173.0 64 Oshana Ondangwa 168.1 65 Hardap Mariental Rural 167.5 66 Omusati Anamulenge 164.2 67 Omusati Ogongo 159.4 68 Caprivi Katima Mulilo Urban 158.9 69 Erongo Walvis Bay Rural 154.0 70 Omaheke Otjinene 150.7 71 Kavango Rundu Urban 149.2 72 Kunene Kamanjab 149.0 73 Oshana Oshakati West 147.9 74 Oshana Uuvudhiya 145.6 75 Omaheke Gobabis 144.5 76 Karas Karasburg 142.1 77 Khomas Samora Machel 139.9 78 Hardap Rehoboth Urban East 139.6 79 Omusati Oshikuku 137.3 80 Hardap Mariental Urban 133.7 81 Otjozondjupa Okahandja 129.9 82

Region Constituency NIMD Score NIMD Rank Kunene Outjo 129.8 83 Karas Keetmanshoop Rural 129.1 84 Khomas Katutura Central 127.0 85 Otjozondjupa Omatako 121.9 86 Otjozondjupa Otavi 117.6 87 Oshana Ongwediva 111.5 88 Karas Luderitz 108.9 89 Oshikoto Tsumeb 107.9 90 Erongo Karibib 107.3 91 Oshana Oshakati East 104.8 92 Otjozondjupa Otjiwarongo 102.5 93 Khomas Katutura East 101.1 94 Otjozondjupa Grootfontein 96.7 95 Erongo Omaruru 94.9 96 Khomas Soweto 86.3 97 Khomas Windhoek Rural 84.5 98 Karas Keetmanshoop Urban 84.2 99 Erongo Swakopmund 79.9 100 Erongo Arandis 78.4 101 Erongo Walvis Bay Urban 77.4 102 Hardap Rehoboth Urban West 56.1 103 Khomas Khomasdal North 46.8 104 Karas Oranjemund 29.1 105 Khomas Windhoek West 19.6 106 Windhoek East Khomas (least deprived) 7.4 107

Annex 5: Regional boundaries

References

Barnes, H., Gutierrez‐Romero, R. and Noble, M. (2006) Multiple Imputation of Missing Data in the 2001 South African Census, CASASP Working Paper No. 4, Oxford: Centre for the Analysis of South African Social Policy, University of Oxford.

Barnes, H., Noble, M., Wright, G. and Dawes, A. (2009) ‘A geographical profile of child deprivation in South Africa’, Child Indicators Research 2 (2): 181‐199.

Barnes, H., Wright, G., Noble, M. and Dawes, A. (2007) The South African Index of Multiple Deprivation for Children 2001, Cape Town: Human Sciences Research Council Press.

BEATTY C., FOTHERGILL S. and MACMILLAN R. (2000) A theory of employment, unemployment and sickness, Reg. Studies 34, 617‐630.

Bhorat, H., Poswell, L. And Naidoo, P. (2004) Dimensions of Poverty in Post‐Apartheid South Africa, Cape Town: DPRU, University of Cape Town.

Blane, D. and Drever, F. (1998) ‘Inequality among men in standardised years of potential life lost, 1970‐93’, British Medical Journal, 317: 255‐256

Central Bureau of Statistics (2008) A Review of Poverty and Inequality in Namibia, Windhoek: Central Bureau of Statistics, National Planning Commission.

Hussmanns, R (2007) Measurement of employment, unemployment and underemployment – Current international standards and issues in their application: International Labour Organisation, United Nations.

McLennan, D., Barnes, H., Noble, M., Davies, J., Garratt, E. and Dibben, C. (2011) The English Indices of Deprivation 2010, London: Department for Communities and Local Government.

Noble, M., Babita, M., Barnes, H., Dibben, C., Magasela, W., Noble, S., Ntshongwana, P., Phillips, H., Rama, S., Roberts, B., Wright, G. and Zungu, S. (2006a) The Provincial Indices of Multiple Deprivation for South Africa 2001, Oxford: University of Oxford, UK.

Noble, M., Barnes, H., Wright, G., McLennan, D., Avenell, D., Whitworth, A. and Roberts, B. (2009a) The South African Index of Multiple Deprivation 2001 at Datazone Level, Pretoria: Department of Social Development.

Noble, M., Barnes, H., Wright, G. and Roberts, B. (2009b) ‘Small area indices of multiple deprivation in South Africa’ Social Indicators Research 95(2): 281‐97.

Noble, M., McLennan, D., Wilkinson, K., Whitworth, A. , Barnes, H. and Dibben, C. (2008) The English Indices of Deprivation 2007, London: Communities and Local Government.

Noble, M., Smith, G.A.N., Wright, G., Dibben, C. and Lloyd, M. (2001) The Northern Ireland multiple deprivation measure 2001, Occasional Paper No 18. Northern Ireland Statistics and Research Agency, Belfast.

Noble, M., Smith, G.A.N., Wright, G., Dibben, C., Lloyd, M. and Penhale, B. (2000) Welsh Index of Multiple Deprivation, London: National Statistics.

Noble, M., Whitworth, A., Allen, J., Wright, G. and Roberts, B. (2007) Developing small area indices of multiple deprivation in Southern Africa: A scoping study. Report for the Southern Africa Trust.

Noble, M., Wright, G., Lloyd, M., Dibben, C., Smith, G.A.N. and Ratcliffe, A. (2003) Scottish indices of deprivation, Edinburgh: Scottish Executive.

Noble, M., Wright, G., Smith, G.A.N. and Dibben, C. (2006b) ‘Measuring multiple deprivation at the small‐area level’, Environment and Planning A 38(1): 169‐185.

Nolan, B. & Whelan, C.T., 1996. Resources, deprivation, and poverty, Oxford University Press.

Townsend, P., 1987. Deprivation. Journal of Social Policy, 16, pp.125‐145.

Townsend, P. (1979) Poverty in the United Kingdom: A survey of household resources and standards of living, Harmondsworth, Middlesex: Allen Lane and Penguin Books.

United Nations Development Programme (2009) Human Development Report 2009 ‐ Overcoming barriers: human mobility and development. New York: Palgrave Macmillan.

United Nations Development Programme Namibia (2007) UNDP Namibia Economic Review 2007, Windhoek: UNDP Namibia.

Wright, G., Barnes, H., Noble, M. and Dawes, A. (2009a) The South African Index of Multiple Deprivation for Children 2001 at Datazone Level, Pretoria: Department of Social Development.

Wright, G. and Noble, M. (2009) The South African Index of Multiple Deprivation 2007 at Municipality Level, Pretoria: Department of Social Development.

Wright, G., Noble, M., Barnes, H. and Noble, S. (2009b) The South African Index of Multiple Deprivation for Children 2007 at Municipality Level, Pretoria: Department of Social Development.