Geographic Differentials in Multidimensional Poverty in Nepal: - - PDF document

Geographic Differentials in Multidimensional Poverty in Nepal: Rethinking

Dimensions and Method of Computation

First Author: Srinivas Goli Assistant Professor Center for the Study of Regional Development (CSRD) Jawaharlal Nehru University, New Delhi Email:sirispeaks2u@gmail.com; Mobile: +91-7042181232. Second Author: [Corresponding Author] Nagendra Kumar Maurya Assistant Professor Department of Applied Economics University of Lucknow, Lucknow, Uttar Pradesh, INDIA Email:nagendrainsearch@gmail.com

Mobile: +91-9450138773

Third Author: Prem Bhandari Assistant Research Scientist Population Studies Center, Institute for Social Research University of Michigan 426 Thompson Street, Ann Arbor, MI 4810 Phone: 734-469-6349 Email: prembh@umich.edu

Abstract: This paper examines the extent of geographic inequality in multidimensional poverty in Nepal

using the nationally representative 2011 Nepal Demographic Health Survey data. We estimate a more robust method of multidimensional poverty index (MPI), particularly in terms of indicators, their definitions and aggregation procedure than those of the previous studies. The findings suggest that despite the relatively better economic progress and a considerable reduction in education and health poverty, there is a wide inequality across the geographic regions. While, a far less has been achieved in the case of reducing the standard of living poverty i.e. wealth poverty and inequalities across the regions. Thus, the paper suggests that development policies and poverty reduction programmes in Nepal must aim to reduce multidimensional poverty, of which deprivation in education, health and basic amenities must be an integral component, along with their efforts to improve economic growth and reduce income poverty. Keywords: Multidimensional Poverty, Nepal, MPI, DHS, Geographic Differentials.

Geographic Differentials in Multidimensional Poverty in Nepal: Rethinking

Dimensions and Method of Computation

1 Introduction

During the last one decade, Nepal has gone through a major political transition. Abolition of monarchy, the establishment of a Federal Democratic Republic, and the election of Constituent Assembly in 2008 (and re-election in 2014) and adoption of the new constitution in September 2015 are the landmarks in the political history and economic planning of Nepal. The country is making every effort to move out of an extended political transition, also aiming to become a developed country in the world by 2022 (United Nations Nepal 2014). The ambitious journey of transition from a least developed country to a developing and then to a developed nation demands a concerted effort for a holistic approach to development. This is not possible without a significant reduction in the incidence of both absolute and relative multi-dimensional poverty in the country. Measurement of poverty itself is complex and highly debatable. Social scientists in different countries have adopted different dimensions to determine poverty status of the populations (see, Oxford Poverty & Human Development Initiative [OPHI] 2013; Yu 2013; Nowak and Schleicher 2014; World Bank 2014; Rangarajan and Dev 2015; Dutta 2015; Monotoya and Teixeria 2016; Reyles 2010; Rogan 2016; Wang and Wang 2016; Hanandita and Tampubolon 2016; Dhongde and Haveman 2016; Bader et

• al. 2016; Angulo et al. 2016; Guio et al. 2012; Guio et al. 2016). Nepal uses the concept of absolute

poverty and has followed its own definition: according to which a person is earning less than 1 US$ a day is termed as poor. By this definition, the latest official figures suggest that more than 35% of the total population is living under the poverty line in Nepal. However, the application of the concept of relative poverty is virtually absent in Nepal (Central Bureau of Statistics [CBS] 2013; Alkire et al. 2013; Nepal Human Development Report [NHDR] 2014; Uematsu et al. 2016; World Bank 2016). Earlier to achieve Million Development Goals (MDGs) and now to achieve Sustainable Development Goals (SDGs), Nepal has been investing in social policies including poverty reduction and active society engagement (Government of Nepal 2011; Uematsu et al. 2016). This has resulted in a significant reduction in poverty (Uematsu et al. 2016). For instance, the percentage of multidimensional poor in Nepal has dropped significantly from 64.7 percent to 44.2 percent in between 2006 and 2011 i.e. by 4.1 percentage points per year (Alkire et al. 2013). In fact, the country has been able to reduce the national poverty much faster than its neighbouring countries such as India, Pakistan and Bangladesh (Dreze and Sen 2013). In spite of considerable progress in poverty reduction in the recent years, Nepal remains one of the poorest countries in the world. With a human development index of 0.548 in 2014, Nepal is ranked 145th out of 187 countries listed in the United Nations Development Programme - 2015 (UNDP 2015). The National Living Standards Survey (NLSS) conducted in 2010-2011 reported that more than 30 percent of Nepalese live on less than US$14 per person per month using the income-based poverty estimation. However, this figure rises to 44.2 percent in the case of multidimensional poor. Further, there is a large inequality in the prevalence of poverty within the nation. While the overall poverty rate for Nepal is 30 percent, this figure rises to 45 percent in the mid-western region and 46 percent in the far-western region (NLSS 2010-11).Thus, a national level figure often obscures the within- country inequality in poverty (Uematsu et al. 2016).

2 Background and Rationale

Measuring poverty is a complicated process (Townsend 1954; 1971; 1979; Sen 1979). Early efforts of measuring poverty involved uni-dimensional indicators based on income or consumption expenditure (Bosanque 1903; Townsend 1954; 1971; 1979; Abel-Smith and Townsend 1965; Atkinson 1969; Sen 1976; 1981, 1987; 1989; Dominique 1979; Kakwani 1980; Clark et al. 1981; Atkinson 1987; Hagenaars 1987; Ravallion and Huppi 1991; Ravallion 1998). Later it was recognised that no single indicator alone could capture the multiple aspects of poverty (Townsend 1979; Foster et al. 1984). Poverty is much more than having a low income or low consumption expenditure (Townsend 1954; Sen 1970; Anand and Sen 1997; Bourguignon and Chakravarty 2003; Dreze and Deaton 2015). Realising the significance of multiple indicators, there have been some efforts to include multiple indicators in measuring poverty. The first multidimensional measure can be traced back to Townsend (1979), and the underpinnings of the Multidimensional Poverty Index (MPI) were set out by Foster et al.

(1984). The Global Multidimensional Poverty Index (MPI) was designed in 2010 by the Oxford Poverty& Human Development Initiative (OPHI) and The United Nations Development Programme using different indicators to determine poverty beyond income-based measures (OPHI, 2013). This MPI replaced the previous Human Poverty Index in subsequent Human Development Reports of world countries. Following this, numerous studies in many countries have used various procedures to estimate multidimensional poverty of individuals and households(Tsui 2002; Guio et al. 2012; Yu 2013; World Bank 2014; Nowak and Schleicher 2014; Rangarajan and Dev 2015; Dutta 2015; Monotoya and Teixeria 2016; Rogan 2016; Wang and Wang 2016; Hanandita and Tampubolon 2016; Dhongde and Haveman 2016; Bader et al. 2016; Angulo et al. 2016; Guio et al. 2016). A majority of these studies used methods designed for global MPI calculations (Alkire and Foster 2007; Alkire et al. 2011b; Alkire and Santos 2013). Advancing the existing methodology of selection of parameters in multidimensional poverty, Guio et al. (2012) and Guio et al. (2016) proposed an analytical framework for developing robust material deprivation indicators for the whole population in the context of European Union. They carried out a systematic item by item analysis at country levels to identify material deprivation items which satisfactorily meet suitability, validity, reliability and additive criteria across the European Union. There have been some efforts to include multiple indicators in measuring poverty in Nepal (CBS 2013; Alkire et

• al. 2013; NHDR 2014; Mitra 2016). However, we understand that these Nepal-specific estimates of

multidimensional poverty have methodological limitations both in terms of estimation procedures and the use of indicators and their units of measurement. This study is an attempt to refine the measure of multidimensional poverty both regarding its dimensionality and estimation procedure and fill this gap in the literature, specifically in the context of Nepal. Below we provide a theoretical background and evidence of multi-dimensional poverty measures used in previous studies. Then, we describe data, proposed method and its advantage over existing method, indicators of alternative dimensions and the procedure used to estimate the multi-dimensional poverty in this study compared to other studies in Nepal (CBS 2013; Alkire et al. 2013; NHDR 2014; Mitra 2016; Uematsu et al. 2016).

2 Measuring Poverty: Past Efforts and Our Approach

Debates on measuring poverty were intensified in the 1970s but these discussions were mainly about measuring income poverty and defining poverty line. During this period, the identification of poor was exclusively by family-size-adjusted household income, concerning a specified income poverty line. Some contributions are worth mentioning here, for example, Townsend (1954; 1971; 1979), Sen (1970, 1972, 1973, 1992, 2000b), Bardhan (1970, 1971), Dasgupta, Sen, Starrett (1973), Gordon and Townsend (1990). Sen (1976) in his seminal article 'Poverty: An ordinal approach to measurement' has emphasised

• n the theoretical soundness of the income poverty measurement. He has suggested an ordinal approach

based on ordinal axioms for measuring poverty. However, he admitted that such approach is difficult to replicate in reality as required data may not be collected. From the mid-1970s, it was recognised that poverty is much more than just having a little income (Townsend, 2010). During this time, the ‗basic needs approach‘; social exclusion and ‗capability approaches‘ gained prominence in complementing the process of identification of poor or deprived population. Studies have shown that income does not represent the non-monetary multi-dimensional deprivations of households (such as lack of access to, such as, nutritious food, health services, quality education, portable water, livable house, sanitation facilities, electricity, basic information and more), thus fail to identify the poor correctly (Sen 1970, 1972, 1973, 1976; Townsend 1979). Consequently, researchers have introduced various non-monetary measures of deprivation, supplementing these multidimensional analyses with monetary measures to create a better

• verall picture of poverty (Townsend 1979; Foster et al. 1984). For instance, Townsend (1979)

highlighted 'relative deprivation' by which he meant absence or inadequacy of those diets, amenities, standards, services and activities which are common or customary in society. This could be understood as the initial debate on a multidimensional aspect of poverty and deprivation. His measure of deprivation included a list of sixty indicators of the standard of living. The indicators covered diet, clothing, fuel and light, house and housing amenities, etc. Sen (2000a: 18) endorsing the need to take a multidimensional approach to poverty, said, ―Human lives are battered and diminished in all kinds of different ways, and the first task is to acknowledge that deprivations of very different kinds have to be accommodated within a general overarching framework.‖ Thus, a poverty measure based on multiple indicators is more robust than the one measured based on income poverty (Gordon and Townsend 1990; Dreze and Sen 2013;

Deaton 2013; Guio et al. 2012; Guio et al. 2016). This suggests that poverty debate has moved from uni- dimensional (income) to multidimensional approach, which led to the estimation of MPI. Foster et al. (1984) proposed a systematic methodology for estimating multidimensional poverty for the first time. Their approach fueled the debate of measuring poverty and provided a framework for decomposing poverty. They showed new poverty measure that is additively decomposable with population share weights. Although their work was one of the most cited in poverty literature due to its notable methodological contribution for decomposing aggregate poverty measure, it, however, did not contribute much in measuring multidimensional poverty. Later on, Alkire and Foster (2007) made a significant methodological improvement. In 2010, Alkire and Santos developed MPI for various countries for the first time under OPHI (Alkire and Foster 2010). Multidimensional poverty comprises of a number of indicators showing the deprivations experienced by people – such as poor health, lack of education, inadequate living standard, lack of income, disempowerment, poor quality of work and threat from violence (Alkire and Foster 2011a, 2011b). These indicators may vary depending upon the context of countries, cultures or societies and their state of affairs (Alkire et al. 2011b).Thus, a key criticism of Global MPI used by UNDP is a lack

• f examination of its suitability within countries, dimensional structure reliability and validity and

measurement invariance apart from the ambiguity in their household level aggregation procedure of individual and household indicators. Some of these are discussed in detail in the method section of this study (also see Guio et al. 2012 and Guio et al. 2016). In Nepal, to the best of our knowledge, the present exercise is the third attempt to measure MPI at a disaggregated level. The first attempt was from Alkire and her team at OPHI who estimated MPI for 104 countries including Nepal (Alkire et al. 2011a). In this study, they included: years of schooling and school attendance for education dimension; child mortality and nutritional status for health dimension; and cooking fuel, sanitation, water, electricity, and floor and asset ownership for the standard of living. They estimated MPI by disaggregating individuals by place of residence and developmental region (OPHI 2013). Their findings suggested that 44.2 percent of people were multidimensional poor in 2011. This estimate was much higher than income based poverty (25.2 percent) of NLSS (2010-11). The second study is the Nepal Human Development Report, 2014 (NHDR, 2014) released by the Government of Nepal in collaboration with United Nations Development Programme (UNDP). In this report, Human Poverty Index was estimated (31.1 percent) using the percentage of people not expected to survive beyond age 40 for a long and healthy life; adult literacy rate for knowledge; and the percentage of people without access to safe water, percentage of malnourished children under age five and deprivation in economic provisioning for a decent standard of living. Human poverty index was estimated separately for a rural-urban place of residence, five development regions, three ecological zones, sub-regions and districts. In this study, as used in previous approaches, our measure of multidimensional poverty includes three basic dimensions of human development—health, education and economic status. However, our estimates of MPI differ significantly from the NHDR (2014) and Alkire et al. (2014) regarding inclusion and coverage of indicators in each dimension of MPI. First, we include net enrollment ratio instead of gross enrolment ratio under education dimension. Second, wealth index is used to measure economic status as compared to individual standard of living indicators. Third, each indicator under different dimensions has been measured at the individual level rather than at household level. We also carried out a rigorous screen of variables to identify multiple deprivation items which acceptably meet suitability, validity and reliability of the assumed latent dimensions of MPI. Below, we have explained in detail about robustness and inclusiveness of our method of estimation of MPI over NHDR (2014) and Alkire et

• al. (2011a) in the context of Nepal, which is also applicable in the Global context. This study is largely

benefited from the conceptual and methodological advancements made in recent studies (e.g. Guio et al. 2012; Guio et al. 2016; and Notten and Mendelson 2016).We also disaggregate multidimensional poverty estimates by place of residence, developmental region, ecological zone and sub-regions. 3 Data This study used data from a nationally representative Nepal Demographic Health Survey (DHS) collected in 2010-11. The sample was designed to provide estimates of most key variables for the 13 eco- development regions and three ecological regions (for further details on sampling design see Ministry of

Health and Population (MOHP) [Nepal], New ERA, and ICF International Inc. 2012). The survey covered a nationally representative sample of 10,826 households which yielded completed interviews with 12,674 women aged 15-49 years in all selected households and with 4,121 men aged 15-49 in every second

• household. Of particular interest to this study, this data provides detailed information on mortality,

nutrition, socio-demographic characteristics, access to basic amenities and household assets. 4 Measure of Multidimensional Poverty: Rethinking Dimensions and Computation We estimate MPI using information from indicators of three main dimensions – education, access to health and standard of living. However, within these dimensions, our indicators are not same as used by Alkire et al. (2011a) and NHDR (2014). The dimensions and indicators within each dimension are presented in Table 1.We describe the measurement of dimensions and indicators below. Each item with dimension was selected based on reviews of extensive literature in Nepal. For example, in the case of first two dimensions, items were selected qualitatively that represent a single latent dimension: education or health. However, the next dimension, standard of living was derived based on factor analyses because there were a large number of items that represented different kinds of material deprivations. Thus, it was important to confirm whether they are representing a single latent dimension of the standing of living or not. For each of these three aspects, item selection criteria were drawn from the literature. The items that have successfully passed principles of suitability, validity, reliability and additive were then used in aggregation to MPI. Table 1.Dimensions, Indicators and Measures of MPI and their Reliability Coefficients

Dimensions Indicators NHDR (2014) Alkire et al. (2011a) Present study – Measures Reliability Analyses†

RC

• No. of

Items AIC

Education Adult literacy rate Yes

• 0.8078

2 0.3786 Years of schooling

• Yes

Yes

• If completed less than

five years of schooling School Enrollment

• Yes

Yes

• If a child (age 6-10 years)

is not enrolled in school Health % of people not expected to survive beyond age 40 Yes

• 0.2327

3 0.0127 Under-five mortality

• Yes

Yes

• If died before the 5th

birthday Nutritional status

• Yes

Yes

• If child is underweight
• If a woman has BMI

<18.5 kg/m2 Standard of Living Percent of malnourished children under age five Yes

• Yes

*

• % without access to

safe water Yes Yes Yes #

• Sanitation
• Yes

Yes #

• Electricity
• Yes

Yes #

• Flooring
• Yes

Yes #

• Cooking Fuel
• Yes

Yes #

• Assets ownership
• Yes

Yes #

• Wealth Index
• Yes

It is a composite index of multiple household amenities and assets 0.5685 33 0.0872 Note: * This indicator in the present study was used as a part of health dimension. #These indicators are being used in the construction of wealth index, a widely known proxy measure of economic status. † Reported RC: Reliability Coefficient and AIC- Average Inter-item Covariance is the indicators used in the present study.

4.1 Education NHDR (2014) used adult literacy rate for measuring educational deprivation. This measure has been severally criticised for being not a true indicator of measuring educational deprivation (Smith 1992; UNDP 1993). Therefore, studies are increasingly using years of schooling and school enrolment to measure educational deprivation. Alkire et al. (2011a) used both years of schooling and school enrolment as indicators of educational deprivation. While the present study also uses years of schooling and child enrolment ratio, but unlike Alkire et al. (2011a) who used gross enrolments, we used net enrolments into

• consideration. We used net enrolment to overcome the limitations of gross enrollment ratio which could

be more than 100 percent and often fails to measure the true enrollment ratio. Reliability analyses for the items included in the dimension of education suggest a high positive reliability coefficient (RC= 0.8078) and AIC (0.3786), meaning that items considered under education dimension probably measure the same underlying concept ―education‖. 4.2 Health While NDHR (2010) used percent of people not expected to survive beyond age 40 as an indicator of health status, both Alkire et al. (2011a) and we used two indicators of health: under-five mortality and nutritional status. However, we also used weight-for-age of a child as an additional indicator of nutritional status of children. Under-five mortality is measured as whether a child experienced death before his/her fifth birthday. Our other measures of health are the nutritional status of the child and the mother. The standard method

• f measuring nutritional status of a child is his/her physical growth or the weight-for-age. Weight-for-age

is a composite index of height-for-age and weight-for-height. It takes into account both acute and chronic

• malnutrition. This index provides information about growth and body composition and is measured in

terms of standard deviation units (Z-scores) from the median of the reference population. For the purpose, weight-for-age of children under five years of age is considered. Children whose standard score (Z-score)

• f weight-for-age is below minus two standard deviations (–2SD) from the median of the reference

population are considered as malnourished. For analysis, a dichotomous variable whether an under-five child‘s growth was below –2SD (coded 0) or above –2SD (coded 1) for weight-for-age was created. Mother‘s nutritional status is measured by Body Mass Index (BMI). The BMI is categorised into three components: thin (BMI less than 18.5), normal (BMI in between 18.5 to 24.9) and obese (BMI more than 25). In this study, we considered thin women with a BMI less than 18.5 as malnourished. Our indicators are same as taken by the Alkire et al. (2011a). In the case of the dimension of health, reliability analyses for the items included reveals a positive reliability coefficient and AIC, meaning that items considered under health dimension probably measure the same underlying concept ―health‖. 4.3 Wealth Index Household wealth index is used as an indicator of standard of living. Wealth index is commonly used in DHS and other country-level studies (Rutstein et al. 2000). For the purpose of this study, wealth index computed in Nepal DHS (2011) has been considered as an indicator of standard of living. The coverage of wealth index (Rutstein 1999; 2008; Ministry of Health and Population (MoHP) [Nepal] New ERA, and ICF International Inc. 2012) is much more than the six standard of living indicators (i.e. access to safe water, sanitation, electricity, flooring, cooking fuel and assets ownership) used by Alkire et al. (2011a). The wealth index in its current form, which takes a better account of urban-rural differences in the scores and indicators of wealth in the Nepal DHS (2011), is created in three steps. First, a subset of indicators common to urban and rural areas was used to create wealth scores for households in both areas (for a detailed list of indicators see Ministry of Health and Population (MOHP) [Nepal], New ERA, and ICF International Inc. 2012 & Rutstein 2008). Categorical variables were transformed into dichotomous (0-1) indicators. These indicators were then examined using Principal Component Analysis (PCA) method to produce a common factor score for each household. Second, separate factor scores were produced for households in urban and rural areas using an area-specific indicator. Third, the separate area-specific factor scores were combined to produce a nationally applicable wealth index by adjusting area-specific scores through a regression on the factor scores. This three-step procedure permits greater adaptability of the wealth index in both urban and rural areas. The resulting combined wealth index has a mean of zero and a standard deviation of one.

Once the index is computed, national-level wealth quintiles (from lowest to highest) were obtained by assigning the household score to each de jure household member, ranking each person in the population by his or her score, and then dividing the ranking into five equal categories, each comprising 20 percent of the population. In this study bottom, two quintiles poorest and poorer (40%) form the poor. This classification is widely accepted in demographic research because it has been repeatedly proven in the DHS data based analyses that these two quintiles show deprived status in several demographic indicators than other three quintiles (Rutstein 2008). The wealth index is particularly valuable in countries which lack reliable data on income and expenditures, which are the traditional indicators used to measure household economic status (Chakraborty et al. 2016). Developing countries like Nepal, with extreme geographical difficulties, high poverty incidence and not so strong statistical information system, wealth index may serve a better purpose for measuring economic status (standard of living) than direct income (Gwatkin et al. 2007; Rutstein & Johnson 2004). However, the recent studies (Guio et al. 2012; Guio et al. 2016) have suggested for testing the reliability and validity of dimensional structures and suitability of items used in the construction of material deprivation index like wealth index. Following them, we have performed the reliability analyses for the 33 items used in the wealth index construction. The results of reliability analyses suggest a high positive reliability coefficient (0.57) between the items selected, meaning they correlate to an assumed latent concept of ‗standard of living‘ in the study. 4.4 Estimation of Multidimensional Poverty Index Compared to the initial period of Townsend (1979) and Foster et al. (1984), the measure of multi- dimensional poverty has been advanced in terms of its dimensional spread and the number of items used in each dimension. In particular, the introduction of Demographic Health Survey (DHS), a homogeneous and reliable household surveys over 80 developing countries have facilitated the researchers to construct the index on unit level datasets using a wider number of indicators related to both households and

• individuals. The Global MPI becomes widely used measure since 2010 after its regular publication in

Human Development Reports of UNDP. Alkire et al. (2011a) have estimated Global MPI using three dimensions: education, health and living standard. As per this approach, a person on each indicator is identified as deprived or not deprived using information for any one household member. Then, it is aggregated across all the household members. This criterion of identifying poor and calculating MPI in their method has serious drawbacks. Researchers have often raised questions about measurability and aggregation process of indicators in the method of MPI calculation proposed by Alkire and her colleagues (Rangarajan and Dev 2015). For instance, if one person in a household is undernourished that does not mean all household members are undernourished. Similarly, if one child has not attended school for five years or more that does not mean that other children did not go to school. Moreover, if a household does not have under-five children, such households will not be included in that particular dimension of MPI. According to Nepal DHS (2011), the proportion of such households was as high as 11 percent. This is the main reason behind not estimating MPI using Alkire et al. (2011a) identification approach for neighbouring developing country ‗India‘(Rangarajan and Dev 2015). Also, Rangarajan and Dev (2015) also questioned the aggregation method of individual and household level indicators by Alkire et al. (2011a).Also, some of the recent studies by Guio et al. (2016) and Notten and Mendelson (2016) have made conceptual and methodological advancements in the construction of material deprivation index like MPI. We concur with Rangarajan and Dev (2015), Guio et al. (2016) and Notten and Mendelson (2016), therefore, made some modifications in Alkire and others' criterion of aggregation of dimension indices in

• rder to overcome the above-said limitations. In particular, we have estimated indicators at the population

level for rural-urban, ecological zones and sub-national level instead of household level in all three dimensions viz. health, education and standard of living. This method has merit over the previous method for two reasons. First, when we estimate child nutritional and child mortality indicators with the entire sample of child population as a denominator instead of households, it will overcome the limitation of excluding the households which do not have under-five children at the time of the survey. Second, this process avoids the duplication of deprivation in terms of health and education across household members. In the case of the index of standard of living, we have estimated it at the household level and generalised the household index value to all the household members living in that particular household by exporting this variable to person file.

While estimating the composite index, the UNDP Human Development Report (2010) used geometric mean (GM) to obtain Human Development Index (HDI). We believe that this method of estimating geometric means of each dimension of indices is a better procedure to derive MPI rather than the simple mean as used by Alkire and Foster (2007) and Alkire et al. (2011a). Because GM makes sure that a low achievement in one dimension is not linearly compensated by high achievement in another

• dimension. The GM reduces the level of substitutability between dimensions and at the same time ensures

that one percent decline in the index, say, life expectancy at birth has the same impact on the MPI as one percent decline in education or income index. Thus, we estimate MPI by using GM method as, MPI = (EPI*HPI*SLPI)1/3 (1) Where, EPI is education poverty index; HPI is health poverty index, and SLPI is the standard of living poverty index. For comparison, we also estimated MPI by using simple mean as: MPI = (EPI+HPI+SLPI)/3 (2) Thus, our methodology is different from Alkire et al. (2011a) and NHDR (2014) in terms of indicators used, estimation and aggregation approach. NHDR (2014) used traditional indicators to measure human poverty index which is severely criticised by the scholars now (Gordon 2000; Rutstein 2008; Dutta 2015; Dhongde and Haveman 2016; Mitra 2016; Rogan 2016; Wang and Wang 2016). For instance, NHDR (2014) used adult literacy rate for education; percentage of people not expected to survive beyond age 40 (life expectancy) for health and percentage of people not having access to safe drinking water and child undernourishment for the standard of living. While adult literacy rate is replaced by the mean years of schooling at UNDP human development index because it does not depict the current scenario in a country

• r state. Therefore, UNDP includes school enrollments which represent current scenario. We considered

both mean years of schooling and net school enrollment ratio. Among health indicators, NHDR (2014) fails to take more sophisticated and sensitive measures like child mortality and nutrition. Standard of living poverty is measured by deprivation in access to drinking water, which is acceptable; but the inclusion of child undernourishment cannot be justified to be included in the standard of living poverty unless they had tested for its suitability and reliability in this dimension. It also fails to incorporate several

• ther important variables under the dimension of the standard of living: sanitation, electricity, housing,

and many others. Therefore, our indicator, especially in the case of the standard of living poverty (measured as wealth index) which incorporates multiple household assets, is a better indicator. From methodology (dimensional structure and aggregation process) point of view, we differ with Alkire et al. (2011a) on two points: number and type of indicators and estimation process used in the computation of MPI. In terms of indicators, we differ only at two places. First, we have used net enrollment ratio rather than gross enrollment ratio in the dimension of education, and also we have wealth index based on 33 assets instead of Alkire and colleagues‘ standard of living index based on six

• indicators. Second, we also differ with Alkire and colleagues in terms of MPI estimation procedure. They

have estimated deprivation basically at a household level even for indicators which conventionally supposed to estimate at the individual level. As said previously, they have applied deprivation to the whole household even if only one member is deprived. For instance, if one child is malnourished then all children are considered as malnourished. However, we estimated each indicator based on the nature of indicator and considered the unit of analysis accordingly, i.e. if it is an individual kind of variable then measured at the individual level or if it is household variable then measured at the household level. 5 Results 5.1 Descriptive Statistics Table 2 presents descriptive statistics of indicators covered under each dimension of MPI by place of residence, ecological regions, developmental regions and eco-development sub-regions. The descriptive statistics show a large difference in each of the specific indicators of all three dimensions of MPI viz. education, health and standard of living.

5.1.1 Distribution of Indicators by Geographic Regions: Education By place of residence, about 66.2 percent people do not complete five years of schooling in rural areas as compared to 45.7 percent in urban areas. On the other hand, children between 6-10 years of age not enrolled are 11 percent in rural areas as compared to 6.2 percent in urban areas. The result shows a huge gap in rural-urban educational poverty. Differences were also observed across ecological zones (viz. Mountain, Hill and Terai). A large number of people across the ecological regions had less than five years of schooling. The figure varies from 60.5 percent in hills to 70.5 percent in the mountain areas. Surprisingly, children not enrolled (Age 6-10 years) reflects the reverse situation. The highest number of children not enrolled (Age 6-10 years) found to be highest in Terai region (12.30%) and lowest in Mountain region (6.80%). By development regions, the highest proportion of individuals with less than 5 years of education were reported in the mid-western development region (67.5%) followed by far western region (67.4). The lowest proportion was found in the eastern development region (60.7%). Children not enrolled among 6- 10 year old were the lowest in figures. Except for central region (15.90%), in other regions children not enrolled (Age 6-10 years) were below 10 percent. The far-western region recorded the lowest (6.30%) number of children not enrolled age 6-10 years. Further disaggregation based on eco-developmental region suggests that over 70% individuals had less than 5 years of education in the Central Mountain (71.6%), Western Mountain (73.4%), far-western hills (70%) and central Terai (71.7%). On the other hand, the proportion of children not enrolled in school was among the highest in Central Terai (20.5%) followed by a western hill (9.9%) and mid-western hill (9.8%). These results show that there is variation in education poverty by rural-urban residence, development regions, and eco-developmental sub-regions. 5.1.2 Distribution of Indicators by Geographic Regions: Health By place of residence, among health poverty indicators, as far as under-five mortality is concerned, rural areas still face a dearth of health facilities as reflected in high under-five mortality (6%) as compared to considerably moderate figure (4.40%) in urban areas. Gaps are noticeable in the case of nutritional indicators also. More than one-third of total children in rural areas are poor regarding nutritional status whereas in urban areas less than one-fourth of total children are poor. However, as compared to children, nutritional status of adult females is slightly better. Female with BMI<18.5 has been considered as poor in terms of nutritional status. As per our estimates, 18.8 percent females in rural areas and 14.1 percent females in urban areas have poor nutritional status. By ecological regions, we found maximum adult females with BMI<18.5 in the Terai region (22.7%) but children with poor nutritional status were maximum in the mountain region (46.3%). Under- five mortality was the highest in the mountain region (8.10%). As against of this, hills were found to be the least poor in terms of select health indicators. Among developmental regions, broadly mid- and far-western region show the greatest extent of deprivations on an average in all health indicators. In the case of under-five mortality, children underweight, and adult females with BMI<18.5 found to be highest in the mid-western (Mortality<5 - 6.50%; Children Underweight – 43.0%; & BMI<18.5 – 19.3%) followed by far-western (Mortality<5 - 7.50%; Children Underweight – 40.4%; & BMI<18.5 – 23.9%) regions. In the case of eco-developmental sub-region, Western Mountain, along with the highest number of people who attained less than five years of schooling, is also the region which has experienced the highest number of under-five deaths (10%) and underweight children (53%). However, central Terai region had the highest number of adult females with BMI<18.5 (26%). Under-five mortality was the least in the central hill (3.70%) whereas underweight children and adult females with BMI<18.5 were lowest in eastern Terai (31%) and Western Hill (8.3%). Thus, there is no single sub-region which has performed better in all health indicators. 5.1.3 Distribution of Indicators by Geographic Regions: Standard of Living By place of residence, huge poverty differences are evident in rural-urban areas in the case of wealth status. The share of poorer and poorest in terms of wealth status is more than seven times in rural areas than urban areas. About 6.3 percent people are found to be in poorer and poorest wealth quintile in urban areas as against of 44.6 percent in rural areas.

By ecological regions, high variation was noticed in the case of wealth status. Around 75% people are poor in mountain region followed by hills (52.5%) and Terai region (25.5%). Wealth poverty is three times higher in mountain region than Terai. In the case of developmental regions, the incidence of wealth poverty was just double in the mid- and far-western regions as compared to other developmental regions. About 60 percent people were poor in terms of wealth in mid- and far-western regions as compared to around 35 percent in other three

• regions. This indicates the unequal distribution of wealth across developmental regions.

Things are not different in eco-development sub-regions. The degree of variation in wealth poverty can be imagined from the number of people belonging to poor and poorest wealth quintile ranges from 16.1 percent (eastern Terai) to 84.4 percent (Western Mountain). Far-western hills (79.4%) and mid- western hills (72.9%) were other two sub-regions with huge wealth poverty. All Terai related sub-regions exhibit low wealth poverty. 5.2 Multidimensional Poverty Index Table 3 provides estimates of MPI by place of residence, ecological regions, development regions and sub-regions. MPI by Rural-urban Place of Residence: Overall MPI estimate for rural areas (0.25) is about 2.5 times greater than in urban areas (0.10). While 25 percent people in the rural areas are experiencing multidimensional poverty in rural areas, this proportion is only 10 percent in the urban areas. Specifically, the results show that rural areas are at a disadvantageous position in comparison with urban areas in all three dimensions viz. education, health and wealth status. The highest difference in rural-urban poverty was found in wealth status. Poverty in terms of wealth status in a rural area (0.45) was nearly seven times higher than their urban counterparts (0.06). MPI across Ecological Regions: Estimates by ecological region reveal that MPI is the highest in the mountain region followed by hills and Terai regions. About one-third, people are experiencing multidimensional poverty in the mountain region (0.29) as compared to one-fifth in the Terai region (0.21). Surprisingly, educational poverty is the highest in the Terai region (0.28). Again, stark differences can be noticed in the case of wealth poverty. About three-fourth people were poor in mountain region as compared to only one-half and one-fourth in Hill and Terai regions respectively. MPI across Development Regions: The results of MPI by development region reveal that mid- western (0.27) and far-western regions have the highest incidence of poverty, whereas, eastern region (0.20) the least. Educational poverty is the highest in the central region (0.32) followed by the western region (0.23). Here again, the educational poverty is found to be lowest (0.21) in the far-western region, but in contrast to it, this region has the highest health poverty (0.15) and second-highest standard of living poverty (0.58). Similar trends for enrolment were also observed under different survey reports (Asia- Pacific Cultural Center for UNESCO [ACCU] 2001; DHS 2011; NLSS 2011). However, differences in health poverty among other regions than far-western are not very significant. But pronounced divergence can be noticed again in the case of poverty in terms of standard of living. Poverty in terms of standard of living is nearly double in mid-western (0.61) and far-western (0.57) regions as compared to other three

• regions. Furthermore, differences can also be seen in the overall MPI score by development region.

MPI across Eco-Development Sub-regions: Table 3 presents the MPI and relative rank of sub-

• regions. Western mountain region is the poorest (0.33) among all followed by a far-western hill (0.30)

and mid-western hill (0.29). Trends in the individual dimensions suggest greater variations in the case of educational and standard of living poverty. Poverty differences between least and most educationally poor are 0.20, while in the standard of living poverty are 0.18. However, differences are much less in the case

• f health poverty which is only 0.06. Standing of sub-regions in terms of individual dimensions also

differs significantly. For instance, central Terai (0.38) and far-western Terai (0.18) in education, far- western hill (0.15) and Western Hill (0.84) in health and Western Mountain (0.33) and eastern Terai (0.15) in the standard of living poverty are the most and least poor regions respectively.

5.3 Income-based Poverty vs. Multidimensional Poverty In this section, we compare the income-based poverty measure with MPI. While poverty is the lack of resources over time and MPI is a consequence of deprivation in various resources. There is a complementary and dynamic relationship between these two indicators. In developed countries context, there is considerable variation were observed between income-based poverty estimates and MPI. But existing literature in developing countries like Nepal hardly gives any evidence on the nature of the relationship between income poverty and multi-dimensional poverty. This study fills this decisive gap. Table 4 compares estimates of multidimensional poverty by our study (by the arithmetic mean method and geometric mean method) with income-based poverty estimates by Central Bureau of Statistics (CBS) 2010-11, Nepal and human poverty estimates by NHDR (2014). Our estimates of overall MPI (Geometric Mean) are similar to income-based poverty. For instance, 23.2 percent people are poor as per our estimates, and the corresponding figure by income based measure is 25.2 percent. Meanwhile, MPI estimates by the arithmetic mean are similar to NHDR (2014) estimates. As per NHDR (2014), poverty is 31.1 percent whereas as per our estimates it is 31percent. This gives us confidence in our result. Rural-urban poverty differences by MPI (GM) are about 15 percent which is around 12 percent in the case of income based poverty by NHDR. Our poverty estimates for the urban area (9.8%) show lesser poverty as compared to CBS 2010-11 (15.5%) and NHDR 2014 (18.5%) but in the case of rural poverty, the differences are not very significant between our estimates (24.8%) and CBS 2010-11 (27.4%). However, rural poverty shown by NHDR 2014 is about 10 percent higher than our MPI (GM) estimates. In the case of classification based on ecological regions mountain region showed the highest poverty by all measures but high variability in estimates across the methods is visible. Only 28.7 percent people are experiencing multidimensional poverty as per MPI (GM) whereas the corresponding figure by MPI (Arithmetic Mean) is 43.4 percent, by CBS 2010-11 is 42.3 percent and 38.5 percent by NHDR (2014). The same is the case with Hill and Terai regions. A similar contrast in terms of variability of estimates by different methods is also visible in development regions. In the case of sub-regions, there is a change in relative position of sub-regions in terms of least and most multidimensional poor. As per MPI (Arithmetic Mean& Geometric Mean), Western Mountain sub- region (49.5% & 32.7%) is the region with the highest multidimensional poor and Eastern Terai (21.4% & 15.2%) is the least multidimensional poor sub-region. On the other hand, Far-Western Hill (42.1%) and Central Hill sub-regions (24.7%) were the most and least multidimensional poor respectively as per NHDR 2014. Overall, compared to developed countries, in Nepal, we found less variation in the levels of income-based poverty and multidimensional poverty. This may be because with an increase in absolute income levels with economic growth, the relative disparities in material access increases. With the introduction of every new technology and resources, richer tend to access faster than their counterpart, thus leads to rising in disparities which also give rise to increase in relative deprivation or poverty (Deaton, 2013). However, in low-income countries still, the overall material access relative to developed countries is less. Therefore, there is less gap between incomes based poverty and multidimensional deprivation. 6 Discussion and Conclusion Unlike several recent studies on multidimensional poverty which used Global MPI procedure of estimation, we have proposed an alternative approach for measuring MPI. In this approach, poverty in terms of each indicator has been measured at the individual level rather than at household level. Moreover, the aggregation procedure used in this study is similar to the procedure used in human development index calculation. As pointed out in the methods section, this procedure overcomes the limitations of Global MPI used by UNDP HDR reports. Our results are methodologically robust in terms

• f both numbers of indicators taken for each dimension and method of aggregation. Overall, in the case of

Nepal, our estimate MPI (0.232) is higher than OPHI (2013) estimates (0.217). The estimates at sub-national level suggest that geographical location still works as a major determining factor in poverty as mountain region has the highest multidimensional poverty. The disparity between rural and urban poverty is significant. Despite substantial progress in reduction of income poverty in recent years, multidimensional poverty in rural areas of Nepal remains very high. It may have serious implications because more than 80% of the population in Nepal still lives in rural areas. However, a careful observation of different dimensions of multi-dimensional poverty shows that it is the rural

deprivation in the standard of living which is a major contributor to rural-urban differences in the total

• MPI. These results are in tune with some of the previous studies (Jerve 2001; Wagle 2007; NHDR 2009;

Bhurtel 2013), which also noted the huge income poverty in rural areas which contributes majorly to multi-dimensional rural poverty in the country. However, the education status seems to be not related to the geographical location as our study indicates the highest net enrollment in the mountain region followed by hills and Terai regions. While the NHDR (2014) shows that hills have lowest educational poverty. The educational poverty seems to be indicator sensitive where NHDR (2014) poverty index takes illiteracy rate into consideration to calculate it, while this study had taken years of schooling and net enrollment ratio which are better indicators than adult literacy and gross enrollment ratio. Similar contrasting results in educational poverty are also

• bserved by development region and sub-development region. This result assumes importance in the

context where Nepal government fixed their targets of educational attainment under development plans in terms of enrollment ratios, not literacy; our results provide the more robust basis for effective policy planning and implementation. The results are also robust because regions with a high number of children not enrolled (6-10 years age) coincide with the high number people attending less than five years of schooling. It is also noticeable that educationally poor regions (Terai) are the regions which border with the most backwards states of the neighbouring country ‗India‘ (Uttar Pradesh and Bihar) regarding education and economic progress. There may be some common explanations and linkages behind the low development process in the regions which require deeper empirical investigation (Samuels et al. 2011; Srinivasan 2012). One possible explanation is that migration is very high in the Terai region (49% highest among all regions), especially, the male migration (49.8%) which is also the case with neighbouring Indian states (Nepal DHS 2011). Nepalese leave their homes at an early age in expectation of earning a livelihood and, thus, do not give due attention to educational attainment (Samuels et al. 2011). Though inter-region differences can be noticed in health poverty, it is comparatively lesser than educational and standard of living poverty. The heavy presence of international agencies i.e. International Centre for Integrated Mountain Development (ICMOD), World Health Organization (WHO) and United Nations Children‘s Fund, etc. and the introduction of many health related progressive schemes by the Nepal government in the last two decades has brought the much-needed improvement in health status across all the regions, especially the disadvantageous region like mountain (Bentley 1995; Engel et al. 2013). Engel et al. (2013) noted: "A consistent policy focus and sustained financial commitment by the government and donors throughout the past two decades, including substantial increases in funding for maternal health since the early 1990s, has allowed for widespread improvements in access to medical services, particularly in remote areas". Thus, Nepal‘s experience in the health sector can provide important lessons for other developing countries, especially, the South Asian Association of Regional Cooperation (SAARC) nations struggling with high levels of maternal mortality and poor health facilities, mainly within a circumstance of difficult terrain and high-income poverty. Regarding wealth status, poverty levels are higher across all the regions. The high poverty in terms

• f wealth status indicates poor conditions of housing, sanitation, electricity and drinking water and other

basic amenities. Wealth poverty contributes to significant differences in the overall multidimensional poverty of different place of residence and regions. Wealth poverty in rural areas is seven times higher than urban areas; while it is three times higher in mountain than Terai region. Similarly, it is almost double in the far-western and mid-western region as compared to other developmental regions. Stark differences were also observed across the sub-regional classification. These findings from the study are corroborated by other studies which also noted the significant rise in income and wealth inequalities during last three decades across the rural-urban and regions (World Institute for Development Economics Research [WIDER] 2005; World Bank 2006; Wagle 2007; NHDR 2014). Bhurtel (2013) argued that it implies two things "First, the labour share of the national income has declined over time while the share

• f capital has rapidly increased. The dull growth of agriculture and stagnant manufacturing has mainly

contributed to the growing economic inequalities. Secondly, the government (Nepal) has failed to take fiscal measures to reduce income inequality. Monetary measures such as providing cheap and easy credit to the poor have been largely ineffective". High incidence of the standard of living poverty in the mountain areas is also a reflection of poor provision of basic needs of life as well as a confirmation of the fact that natural features still is an important determinant of multidimensional poverty. This shows that the government policies in Nepal mainly focus on the economic growth and employment generation and

highly ignore to bridge the gap between rich and poor. At the same time, the government also failed to remove income inequalities as stated in Tenth Plan document (Wagle 2007; NHDR 2014). In terms of comparability across the methods, it is important to note that CBS 2010-11 estimates of poverty are essentially income based poverty which is comparable to our standard of living poverty estimates rather than overall MPI estimates. Income is the source as well as the outcome of the wealth of a household. Past wealth helps in income generation and thus generated income helps in wealth

• accumulation. Nonetheless, this feeding loop is based on the condition that income level must be above

their consumption expenditure level. The income level of two households or individuals may be same, but their wealth level may differ significantly. Income level often does count for generational accumulation and transfer of wealth. Therefore, income may show the lesser incidence of poverty than wealth. This is exactly the case with Nepal. Poverty regarding wealth status showed the relatively higher incidence of poverty amongst different geographical areas. Even, at country level poverty in terms of wealth status differs significantly than income poverty. Our measure of standard of living poverty is comprehensive as it is based on wealth index which has been prepared by including multiple household assets. The estimates from this study show the greater extent of wealth poverty compared to OPHI (2013). This study had examined the extent of multidimensional poverty in Nepal disaggregated by geographic regions. It adopted the more robust method of MPI compared to Global MPI of UNDP, particularly regarding indicators, their definitions, dimensional structure and aggregation procedure than that of the previous studies. It also took into consideration the latest methodological improvements in calculating deprivation index measures by Guio et al. (2012), Guio et al. (2016) and Notten and Mendelson (2016). To conclude, the findings of the study suggest that although Nepal has experienced a decent economic progress and a considerable reduction in education and health poverty with a considerable increase in wealth inequalities across the regions, overall MPI remains high. Thus, a far less has been achieved in the case of reducing the standard of living poverty i.e. wealth poverty and inequalities across the regions. Thus, the paper suggests that development policies and poverty reduction programmes in Nepal must aim to reduce multi-dimensional poverty by geographic regions, of which deprivation in education, health and basic amenities must be an integral component, along with their efforts to improve economic growth and reduce income poverty. References Abel-Smith, B., & Townsend, P. (1965). The Poor and the Poorest, London, G. Bell & Sons. ACCU (2001). Regional Gender and Ethnic Disparity in Education. Asia-Pacific Cultural Center for UNESCO, Japan. http://www.accu.or.jp/litdbase/literacy/nrc_nfe/eng_bul/BUL18.pdf. Accessed

• n December 16 2014.

Alkire S., Roche J. M., Santos M. E. & Seth S. (2011a). Multidimensional Poverty Index 2011: Brief Methodological Note. Oxford Poverty & Human Development Initiative (OPHI), Oxford Department of International Development, Queen Elizabeth House (QEH), University of Oxford, November, 2011. Alkire, S. & Foster, J. (2007). Counting and multidimensional poverty measures (No. 7). OPHI Working Paper Series. Alkire, S. and Santos, M.E. (2010). Acute Multidimensional Poverty: A New Index for Developing

• Countries. OPHI Working Papers 38, University of Oxford.

Alkire, S., & Foster, J. (2011a). Counting and multidimensional poverty measurement. Journal of public economics, 95(7), 476-487. Alkire, S., & Foster, J. (2011b). Understandings and misunderstandings of multidimensional poverty

• measurement. Journal of Economic Inequality, 9(2), 289-314.

Alkire, S., & Santos, M. E. (2013). A multidimensional approach: poverty measurement & beyond. Social indicators research, 112(2), 239-257. Alkire, S., Adriana, C. & Roche, J. M. (2013). Multidimensional Poverty Index 2013: Brief Methodological Note and Results. Oxford Poverty & Human Development Initiative (OPHI), Oxford Department of International Development, Queen Elizabeth House (QEH), University of Oxford, February, 2013. Alkire, S., Foster, J., & Santos, M. E. (2011b). Where did identification go? Journal of Economic Inequality, 9(3), 501-505.

Alkire, S., Roche, J. M., and Vaz, A. (2014). Multidimensional Poverty Dynamics: Methodology and Results for 34 Countries. OPHI Research in Progress. Anand, S., & Sen, A. (1997). Concepts or Human Development and Poverty! A Multidimensional

• Perspective. United Nations Development Programme, Poverty and human development: Human

development papers, 1-20. Angulo, R., Díaz, Y., & Pardo, R. (2016). The Colombian multidimensional poverty index: Measuring poverty in a public policy context. Social Indicators Research, 127(1), 1-38. Atkinson, A. B. (1987). On the measurement of poverty. Econometrica: Journal of the Econometric Society, 749-764. Bader, C., Bieri, S., Wiesmann, U., & Heinimann, A. (2016). A different perspective on poverty in Lao PDR: Multidimensional poverty in Lao PDR for the years 2002/2003 and 2007/2008. Social Indicators Research, 126(2), 483-502. Bardhan, P. K. (1970). On the minimum level of living and the rural poor. Indian Economic Review, 5(1), 129-136. Bentley, H. (1995). The organisation of health care in Nepal. International Journal of Nursing Studies, Volume 32, Issue 3, 260 – 270. Bhurtel, B. P. (2013). Rich man's world as the wealth gap grows in Nepal. The Nation, The Kathmandu Post, Asia News Network, October 17, 2013. http://nationmultimedia.com/opinion/RICH-mans- world-as-the-wealth-gap-grows-in-Nepal-30217256.html. Accessed on December 16 2014. Bourguignon, F., & Chakravarty, S. R. (2003). The measurement of multidimensional poverty. Journal of Economic inequality, 1(1), 25-49. Central Bureau of Statistics (CBS) (2011). Poverty in Nepal. National Planning Commission Secretariat, Government of Nepal, Nepal. www.cbs.gov.np. Accessed on December 16 2014. Chakraborty, N. M., Fry, K., Behl, R., & Longfield, K. (2016). Simplified asset indices to measure wealth and equity in health programs: a reliability and validity analysis using survey data from 16

• countries. Global Health: Science and Practice, 4(1), 141-154.

Clark, S., Hemming, R., & Ulph, D. (1981). On indices for the measurement of poverty. The Economic Journal, 91(362), 515-526. Dasgupta, P., Sen, A., & Starrett, D. (1973). Notes on the measurement of inequality. Journal of economic theory, 6(2), 180-187. Deaton A. (2013). The Great Escape: Health, Wealth, and the origins of inequality, Princeton University Press, Princeton. Dhongde, S., & Haveman, R. (2016). Multi-dimensional deprivation in the US. Social Indicators Research, 1-24. Dutta, S. (2015). Identifying Single or Multiple Poverty Trap: An Application to Indian Household Panel

• Data. Social Indicators Research, 120(1), 157-179.

Engel, J., Glennie, J., Adhikari, S. R., Bhattarai, S. W., Prasai, D. P. and Samuels, F. (2013). NEPAL‘S STORY Understanding improvements in maternal health, Overseas Development Institute, UK. Available at: https://www.odi.org/sites/odi.org.uk/files/odi-assets/publications-opinion- files/8624.pdf. Foster, J., Greer, J., & Thorbecke, E. (1984). A class of decomposable poverty measures. Econometrica: Journal of the Econometric Society, 761-766. Gordon, D. & Townsend, P. (1990). Measuring the poverty line. Radical Statistics, 47, 5–12. Gordon, D. (2000). The scientific measurement of poverty: recent theoretical advances. Researching Poverty, 37-58. Aldershot: Ashgate. Government of Nepal (2011). Nepal Living Standards Survey 2010/2011. Statistical Report Volume I, Central Bureau of Statistics, National Planning Commission Secretariat, Government of Nepal, November 2011. Guio A.-C., Gordon D., Marlier E. (2012). Measuring Material Deprivation in the EU: Indicators for the Whole Population and Child-Specific Indicators. Eurostat Methodologies and Working

• Papers. Luxembourg: Office for Official Publications of the European Communities (OPOCE).

Guio, A. C., Marlier, E., Gordon, D., Fahmy, E., Nandy, S., & Pomati, M. (2016). Improving the measurement of material deprivation at the European Union level. Journal of European Social Policy, 26(3), 219-333.

Gwatkin, D.R., Rutstein, S., Johnson, K., Suliman, E., Wagstaff, A., & Amouzou, A. (2007). Socio- economic differences in health, nutrition, and population. Washington (DC): World Bank. Hagenaars, A. (1987). A class of poverty indices. International Economic Review, 583-607. Hanandita, W., & Tampubolon, G. (2016). Multidimensional poverty in Indonesia: trend over the last decade (2003–2013). Social Indicators Research, 128(2), 559-587. Jerve, A. M. (2001). Rural-urban linkages and poverty analysis. in Alejandro Grinspun et al (eds.): Choices for the poor: Lessons from national poverty strategies. Bergen/New York: CMI/UNDP

• pp. 89-120.

Kakwani, N. (1980). On a class of poverty measures. Econometrica: Journal of the Econometric Society, 437-446. Mehta, A.C. (2003). Can There Be Alternative Indicators of Enrolment: A Critical Review of Frequently Used Indicators. Presented at Regional Workshop on Education Statistics 2003, National Institute

• f Educational Planning and Administration, India.

Ministry of Health and Population (MOHP) [Nepal], New ERA, and ICF International Inc. (2012). Nepal Demographic and Health Survey 2011. Kathmandu, Nepal: Ministry of Health and Population, New ERA, and ICF International, Calverton, Maryland. Mitra, S. (2016). Synergies among monetary, multidimensional and subjective poverty: Evidence from

• Nepal. Social Indicators Research, 125(1), 103-125.

Montoya, Á. J. A., & Teixeira, K. M. D. (2016). Multidimensional Poverty in Nicaragua: Are Female- Headed Households Better Off? Social Indicators Research, 1-27. Nepal Living Standard Survey [NLSS] (2011). Nepal - Living Standards Survey 2010-2011, Third

• Round. Central Bureau of Statistics - National Planning Commission Secretariat, Government of

Nepal. NHDR (2009). Nepal Human Development Report 2009: State Transformation and Human Development. UN Development Programme (UNDP), available at: http://www.refworld.org/docid/4a8a72542.html [accessed 29 May 2017]. NHDR (2014). National Human Development Report: Beyond Geography Unlocking Human Potential. National Planning Commission and United Nations Development Programme, Nepal. Notten, G. and Mendelson, M. (2016). Using low income and material deprivation to monitor poverty

• reduction. Caledon Institute of Social Policy, CANDA.

Nowak, D., & Scheicher, C. (2014).Considering the extremely poor: Multidimensional poverty measurement for Germany. Social Indicators Research, 1-24. OPHI (2013). Nepal Country Briefing - Multidimensional Poverty Index Data Bank. Oxford Poverty Human Development Initiative, University of Oxford.www.ophi.org.uk/multidimensional- poverty-index/mpi-country-briefings/. Rangarajan, C. & Dev, M. S. (2015). Counting the Poor: Measurement and Other Issues. Economic and Political Weekly, L(2), 70-74. Reyles, D.Z. (2010). The Human Development Index (HDI). Oxford Poverty and Human Development Initiative (OPHI). http://www.ophi.org.uk/wpcontent/- uploads/HDI-measure-interpretation-key- messages.pdf. Accessed on May 13 2014. Rogan, M. (2016). Gender and multidimensional poverty in South Africa: Applying the global multidimensional poverty index (MPI). Social Indicators Research, 126(3), 987-1006. Rutstein, S, & Johnson, K. (2004). The DHS wealth index. DHS Comparative Reports No. 6. Calverton (MD): ORC Macro; 2004. http://www.dhsprogram.com/publications/publicationcr6-comparative- reports.cfm. Rutstein, S. 1999. Wealth versus expenditure: Comparison between the DHS wealth index and household expenditures in four departments of Guatemala. Calverton, Maryland, USA: ORC Macro (unpublished). Rutstein, S., Jhonson, K. & Gwatkin, D. (2000). Poverty, health inequality, and its health and demographic effect. Presented at the annual meeting of the Population Association of America, Los Angeles, California. Rutstein, S.O. (2008). The DHS Wealth Index: Approaches for Rural and Urban Areas. The DHS Working Papers, No.60, Demographic and Health Research, USAID, October, 2008.

Samuels, F., Nino-Zarazua, M., Wagle, S., Sultana, T. & Sultana, M. M. (2011).Vulnerabilities of movement: cross-border mobility between India, Nepal and Bangladesh. Overseas Development Institute (ODI), Background Note, November, 2011. Sen, A. (1970). Interpersonal aggregation and partial comparability. Econometrica: Journal of the Econometric Society, 393-409. Sen, A. (1972). Interpersonal comparison and partial comparability: A correction. Econometrica (pre- 1986), 40(5), 959. Sen, A. (1973). Poverty, inequality and unemployment: some conceptual issues in

• measurement. Economic and Political Weekly, 1457-1464.

Sen, A. (1979). Issues in the Measurement of Poverty. The Scandinavian Journal of Economics, 285-307. Sen, A. (1992). Poverty Reexamined. Cambridge: Harvard University Press. Sen, A. (2000a). A Decade of Human Development. Journal of Human Development, 1(1), 18. Sen, A. (2000b). Social Exclusion: Concept, Application, and Scrutiny. Social Development Papers, 1.Office of Environment and Social Development, Asian Development Bank. Accessed from: http://hdl.handle.net/11540/2339. License: CC BY 3.0 IGO. Smith, P. (1992). Measuring Human Development. University of Southampton, Mimeo. Srinivasan, P.V. (2012). Regional Cooperation and Integration through Cross-Border Infrastructure Development in South Asia: Impact on Poverty. Asian Development Bank South Asia Working Paper Series, 14, November 2012. The Hindu (2013). Oxford study finds Nepal reducing poverty faster than India. March 8, 2013. http://www.thehindu.com/news/international/south-asia/oxfordstudy-finds-nepal-reducing- poverty-faster-than-india/article4522065.ece.Accessed on June 28 2014. Townsend, P. (1954). Measuring poverty. The British Journal of Sociology, 5(2), 130-137. Townsend, P. (1979). Poverty in the United Kingdom. Harmondsworth: Penguin Townsend, P. (ed.) 1971). The Concept of Poverty. London: Heinemann. Trani, J. F., Biggeri, M., & Mauro, V. (2013). The multidimensionality of child poverty: Evidence from

• Afghanistan. Social Indicators Research, 112(2), 391-416.

Tsui, K.Y. (2002). Multidimensional Poverty Indices. Social Choice and Welfare, 19(1), 69-93. Uematsu, H., Shidiq, A. R., & Tiwari, S. (2016). Trends and drivers of poverty reduction in Nepal: a historical perspective. Policy Research Working Paper; No. 7830.World Bank, Washington, DC. World Bank. https://openknowledge.worldbank.org/handle/10986/25146 License: CC BY 3.0 IGO.‖ UN Nepal (2014). 12-point understanding reached between the Seven Political Parties and Nepal Communist Party (Maoists). http://www.un.org.np/. Accessed on July 3 2014. UNDP (1993). Human Development Report 1993. United Nations Development Programme (UNDP), Oxford University Press, New York, 105-06. Wagle, U. R. (2007). Economic Inequality in the ‗Democratic‘ Nepal: Dimensions and Implications. Presented at the Himalayan Policy Research Conference, Madison, WI, October 11, 2007. https://ejournals.unm.edu/index.php/nsc/article/view/642/739. Accessed on December 16 2014. Wang, Y., & Wang, B. (2016). Multidimensional poverty measure and analysis: a case study from Hechi City, China. Springer Plus, 5(1), 642. Wider (2005). World Income Inequality Database (V 2.0a). World Institute for Development Economics Research, United Nations University. World Bank (2006).World Development Indicators 2006. Washington, DC: The World Bank. World Bank (2016). Moving up the ladder: poverty reduction and social mobility in Nepal. Available at: http://documents.worldbank.org/curate d/en/171641467117954924/Moving-up-the-ladder- poverty-reduction-and-social-mobility-in-Nepal. World Bank (2014). Poverty and Inequality Measures in Practice: A Basic Reference Guide with Stata

• Examples. Washington, D.C.

Yu, J. (2013). Multidimensional poverty in China: Findings based on the CHNS. Social Indicators Research, 112(2), 315-336.

Table 2. Descriptive Statistics of Indicators, Nepal, 2011

Measure Education Health Standard of living Less than 5 years of schooling (n = 41120) Children not enrolled (Age 6-10 Years) (n = 4706) Under 5 mortality (n = 11192) Nutritional status Wealth status - poor and poorest wealth quintile (n = 48123) Children (WAG -2SD) (n = 2475) Adult female (BMI <18.5) (n =5800) % CI (95%) % CI (95%) % CI (95%) % CI (95%) % CI (95%) % CI (95%) LL UL LL UL LL UL LL UL LL UL LL UL Place of residence Urban 45.7 44.5 46.9 6.2 5.4 7.0 4.4 3.8 5.0 23.8 22.6 25.0 14.1 13.2 15.0 6.3 5.1 7.5 Rural 66.2 65.7 66.7 11.3 11.0 11.6 6.0 5.8 6.2 36.6 36.1 37.1 18.8 18.4 19.2 44.6 44.1 45.1 Ecological zone Mountain 70.5 68.9 72.1 6.8 5.8 7.8 8.1 7.3 8.9 46.3 44.7 47.9 16.5 15.2 17.8 71.8 70.1 73.5 Hills 60.5 59.8 61.2 9.3 8.9 9.7 5.4 5.1 5.7 34.3 33.6 35.0 12.4 11.9 12.9 52.5 51.8 53.2 Terai 64.8 64.2 65.4 12.3 11.9 12.7 5.8 5.5 6.1 34.6 34.0 35.2 22.7 22.2 23.2 25.5 24.9 26.1 Development Region Eastern 60.7 59.8 61.6 8.4 7.8 9.0 5.0 4.6 5.4 32.0 31.1 32.9 16.2 15.5 16.9 34.6 33.7 35.5 Central 65.0 62.6 67.4 15.9 14.4 17.4 5.8 4.7 6.9 34.5 32.2 36.8 20.2 18.3 22.1 32.2 29.8 34.6 Western 59.3 58.4 60.2 9.1 8.5 9.7 5.4 4.9 5.9 32.1 31.2 33.0 14 13.2 14.8 35.7 34.7 36.7 Mid-western 68.5 67.3 69.7 7.7 6.9 8.5 6.5 5.9 7.1 43.0 41.8 44.2 19.3 18.3 20.3 61.2 60.0 62.4 Far-western 67.4 66.0 68.8 6.3 5.4 7.2 7.5 6.8 8.2 40.4 39.1 41.7 23.9 22.8 25.0 57.9 56.5 59.3 Eco-Development Sub-region Eastern Mountain 64.8 61.7 67.9 6.0 4.0 8.0 5.5 4.0 7.0 32.6 29.5 35.7 10 7.5 12.5 65.4 62.2 68.6 Central Mountain 71.6 63.0 80.2 8.0 2.5 13.5 6.0 1.8 10.2 43.9 35.4 52.4 14.9 8.0 21.8 59.9 51.2 68.6 Western Mountain 73.4 70.9 75.9 6.5 4.9 8.1 10.4 9.2 11.6 53.5 51.0 56.0 22.2 20.2 24.2 84.4 81.9 86.9 Eastern Hill 63.2 61.6 64.8 9.5 8.5 10.5 6.3 5.5 7.1 34.5 33.0 36.0 11.8 10.6 13.0 61.8 60.2 63.4 Central Hill 52.7 51.4 54.0 8.8 8.0 9.6 3.7 3.1 4.3 32.2 31.0 33.4 11.5 10.5 12.5 32.8 31.5 34.1 Western Hill 60.1 58.9 61.3 9.9 9.1 10.7 5.4 4.8 6.0 26.6 25.4 27.8 8.3 7.3 9.3 48.8 47.5 50.1 Mid-Western Hill 68.3 66.5 70.1 9.8 8.6 11.0 6.2 5.3 7.1 40.2 38.4 42.0 18.6 17.1 20.1 72.9 71.0 74.8 Far-Western Hill 70.0 67.7 72.3 8.0 6.5 9.5 6.2 5.1 7.3 48.7 46.4 51.0 23.4 21.6 25.2 79.4 77.1 81.7 Eastern Terai 58.9 57.8 60.0 8.2 7.5 8.9 4.1 3.5 4.7 30.7 29.6 31.8 19.3 18.4 20.2 16.1 14.9 17.3 Central Terai 71.7 70.7 72.7 20.5 19.9 21.1 6.7 6.2 7.2 34.7 33.7 35.7 26.4 25.6 27.2 28.9 27.9 29.9 Western Terai 58.3 56.8 59.8 8.2 7.3 9.1 5.4 4.7 6.1 41.2 39.7 42.7 21.3 20.1 22.5 17.5 16.0 19.0 Mid-Western Terai 67.3 65.4 69.2 5.7 4.5 6.9 5.7 4.8 6.6 38.8 36.9 40.7 20.2 18.7 21.7 41.5 39.6 43.4 Far-Western Terai 63.9 62.0 65.8 5.1 3.9 6.3 6.9 6.0 7.8 31.0 29.1 32.9 23.7 22.2 25.2 36.2 34.3 38.1 Nepal 63.5 63.1 63.9 10.7 10.4 11.0 5.8 5.6 6.0 35.4 35.0 35.8 18.2 17.9 18.5 39.6 39.2 40.0

Source: Author‘s calculations based on Nepal DHS 2011 data. Note: LL-Lower Limit, UL-Upper Limit

Table 3. Multidimensional Poverty by place of Residence, Nepal, 2011

Note: LL-Lower Limit, UL-Upper Limit Source: Author‘s calculations based on Nepal DHS 2011 data. Background characteristics Region Education Poverty Index (EPI) Health Poverty Index (HPI) Standard of Living Poverty Index (SLPI) Multidimensional Poverty Index (MPI)

Rank Score 95% CI Score 95% CI Score 95% CI Score 95% CI LL UL LL UL LL UL LL UL

Place of Residence Urban 0.17 0.16 0.18 0.09 0.09 0.10 0.06 0.05 0.08 0.10 0.09 0.11 2 Rural 0.27 0.27 0.28 0.13 0.13 0.13 0.45 0.44 0.45 0.25 0.24 0.25 1 Ecological Region Mountain 0.22 0.20 0.23 0.15 0.15 0.16 0.72 0.70 0.73 0.29 0.27 0.3 1 Hill 0.24 0.23 0.24 0.11 0.11 0.11 0.53 0.52 0.53 0.24 0.23 0.24 2 Terai 0.28 0.28 0.29 0.13 0.13 0.13 0.26 0.25 0.26 0.21 0.2 0.21 3 Development Region Eastern 0.23 0.22 0.23 0.11 0.11 0.11 0.35 0.34 0.35 0.20 0.2 0.21 5 Central 0.32 0.30 0.34 0.12 0.12 0.14 0.32 0.30 0.35 0.23 0.21 0.25 3 Western 0.23 0.22 0.24 0.11 0.11 0.11 0.36 0.35 0.37 0.21 0.2 0.22 4 Mid-western 0.23 0.22 0.24 0.14 0.14 0.15 0.61 0.60 0.62 0.27 0.26 0.28 1 Far-western 0.21 0.19 0.22 0.15 0.15 0.16 0.58 0.57 0.59 0.26 0.25 0.28 2 Eco-Development Sub-region

Eastern Mountain

0.20 0.17 0.23 0.10 0.10 0.12 0.65 0.62 0.69 0.23 0.21 0.26 7

Central Mountain

0.24 0.16 0.32 0.12 0.12 0.18 0.60 0.51 0.69 0.26 0.19 0.34 4

Western Mountain

0.22 0.20 0.24 0.19 0.19 0.21 0.84 0.82 0.87 0.33 0.3 0.35 1

Eastern Hill

0.25 0.23 0.26 0.11 0.11 0.12 0.62 0.60 0.63 0.26 0.24 0.27 5

Central Hill

0.22 0.20 0.23 0.08 0.08 0.09 0.33 0.32 0.34 0.18 0.17 0.19 11

Western Hill

0.24 0.23 0.26 0.09 0.09 0.10 0.49 0.48 0.5 0.22 0.21 0.23 8

Mid-Western Hill

0.26 0.24 0.28 0.13 0.13 0.14 0.73 0.71 0.75 0.29 0.27 0.31 3

Far-Western Hill

0.24 0.22 0.26 0.14 0.14 0.16 0.79 0.77 0.82 0.30 0.28 0.32 2

Eastern Terai

0.22 0.21 0.23 0.10 0.10 0.11 0.16 0.15 0.17 0.15 0.14 0.16 13

Central Terai

0.38 0.37 0.39 0.14 0.14 0.15 0.29 0.28 0.30 0.25 0.24 0.26 6

Western Terai

0.22 0.21 0.23 0.13 0.13 0.14 0.18 0.16 0.19 0.17 0.16 0.18 12

Mid-Western Terai

0.20 0.18 0.21 0.13 0.13 0.14 0.42 0.40 0.43 0.22 0.2 0.23 9

Far-Western Terai

0.18 0.16 0.20 0.14 0.14 0.15 0.36 0.34 0.38 0.21 0.19 0.22 10 Total 0.26 0.26 0.26 0.12 0.12 0.12 0.40 0.39 0.40 0.23 0.23 0.24

19 Table 4. Income based Poverty vs Multidimensional Poverty in Nepal, 2011

Background Characteristics Region Poverty (%) Multidimensional Poverty (AM) 2011♣ Multidimensional Poverty (GM) 2011♣ Income based Poverty (2010- 11)# Human Poverty - 2010-11 by NHDR (2014)† Place of Residence Urban 14.6 9.8 15.5 18.5 Rural 33.4 24.8 27.4 34.0 Ecological Region Mountain 43.4 28.7 42.3 38.5 Hill 33.9 23.6 24.3 29.2 Terai 27.0 20.9 23.4 33.0 Development Region Eastern 27.9 20.3 22.3 29.2 Central 29.7 23.4 21.7 31.5 Western 28.0 20.7 21.4 27.2 Mid-western 39.4 26.8 31.7 36.6 Far-western 38.2 26.3 45.7 34.8 Sub-region Eastern Mountain 38.1 23.4

• 30.7

Central Mountain 39.1 26.1

• 37.5

Western Mountain 49.5 32.7

• 29.3

Eastern Hill 37.6 25.8

• 30.2

Central Hill 25.4 18.1

• 24.7

Western Hill 31.7 22.0

• 25.6

Mid-Western Hill 43.3 29.1

• 38.2

Far-Western Hill 46.5 30.1

• 42.1

Eastern Terai 21.4 15.2

• 29.5

Central Terai 31.2 25.1

• 39.4

Western Terai 23.0 16.9

• 29.7

Mid-Western Terai 31.9 21.7

• 32.5

Far-Western Terai 29.3 20.8

• 28.4

Total 31.0 23.2 25.2 31.1 Source: ♣ Author‘s estimates based on Nepal DHS 2011 data; #CBS (2010-11), Nepal. †Nepal Human Development Report (2014).