SOCIAL AND ECONOMIC POLICY WORKING PAPER CONSIDERING CROSS-NATIONAL EQUITY Children in Highland Populations in South-East Asia Sólrún Engilbertsdóttir Martin C. Evans Ishani Shrestha May 2013 UNICEF POLICY AND STRATEGY Considering Cross-National Equity: Children in Highland Populations in South-East Asia © United Nations Children’s Fund (UNICEF), New York, 2013 Cross-Sectoral Policy, Division of Policy and Strategy UNICEF 3 UN Plaza, New York, NY 10017 This is a working document. It has been prepared to facilitate the exchange of knowledge and to stimulate discussion. The findings, interpretations and conclusions expressed in this paper are those of the authors and do not necessarily reflect the policies or views of UNICEF or of the United Nations. The text has not been edited to official publication standards, and UNICEF accepts no responsibility for errors. The designations in this publication do not imply an opinion on legal status of any country or territory, or of its authorities, or the delimitation of frontiers. The editor of the series is Jingqing Chai of UNICEF Policy and Strategy Division. For more information on the series, or to submit a working paper, please contact jchai@unicef.org i UNICEF SOCIAL AND ECONOMIC POLICY WORKING PAPER May 2013 CONSIDERING CROSS-NATIONAL EQUITY Children in Highland Populations in South-East Asia Sólrún Engilbertsdóttir Policy Analyst, Policy and Strategy, UNICEF Martin C. Evans Social Policy and Economic Analysis Specialist, Policy and Strategy, UNICEF Ishani Shrestha Mt. Holyoke Intern, Policy and Strategy, UNICEF Comments may be addressed by email to the authors: sengilbertsdottir@unicef.org and mcevans@unicef.org ii ACKNOWLEDGEMENTS We are grateful to the comments and suggestions made by Andrew Claypole, Beatrice Duncan, Christina Popivanova, David Anthony, Etona Ekole, Jeffrey O’Malley, Jingqing Chai, Mahesh Patel, Mizuho Okimoto-Kaewtathip, Thi Van Anh Nguyen, Lena Nguyen, Nicholas Rees, Nicola Brandt, Valentina Calderon and Yoshimi Nishino. The findings, interpretations and conclusions expressed in this paper are those of the authors and do not necessarily reflect the policies or views of UNICEF or of the United Nations. iii Table of Contents Executive Summary ................................................................................................................... vi 1. Introduction ......................................................................................................................... 1 2. Equity and Ethnicity ........................................................................................................... 2 3. Hill People of Southeast Asia ............................................................................................. 2 3.1 Viet Nam....................................................................................................................... 3 3.2 Lao PDR ....................................................................................................................... 5 3.3 Thailand ........................................................................................................................ 6 4. Hill People Populations....................................................................................................... 7 4.1 Identifying Hill People in Survey Data ........................................................................ 8 4.2 Population and Sample Sizes ........................................................................................ 9 4.3 Household Size and Composition............................................................................... 11 4.4 Locational Characteristics .......................................................................................... 12 5. Hill People and Ethnic Equity Profiling ........................................................................... 12 5.1 Asset Quintiles ............................................................................................................ 12 5.2 Deprivation, MDGs and Child Well-Being Measures................................................ 13 a. Drinking Water ........................................................................................................... 14 b. Sanitation .................................................................................................................... 15 c. Nutrition...................................................................................................................... 15 d. Health.......................................................................................................................... 16 e. Education .................................................................................................................... 16 f. Early Childhood Development ................................................................................... 17 6. Equity Monitoring and Target Setting .............................................................................. 18 6.1 Equity Target Option 1: Equality of Ethnic-Related Risk ................................................ 18 6.2 Equity Target Option 2: Achieving Fair Outcomes ............................................................ 20 a. Basing fairness on ‘average’ (mean) levels of incidence. .......................................... 20 b. Equalizing shares of deprivation to match population share. ..................................... 21 7. Concluding remarks .......................................................................................................... 22 iv Tables Table 1. Difference in Hill People household size and composition ............................................ 11 Table 2: Children in households with unimproved sanitation in Viet Nam ................................. 19 Table 3: Hill People deprivation and population shares for water and sanitation ........................ 21 Figures Figure 1: Population/children in poorest quintile % ..................................................................... 13 Figure 2: Children in households with unimproved sanitation ..................................................... 15 Figure 3: Children 0-4 yrs stunted % ............................................................................................ 15 Figure 4: Children under 5 yrs who have received no vaccination............................................... 16 Figure 5: Children aged 12 to 14 not progressed to secondary education % ................................ 17 Figure 6: Children not attending early childhood education programme % ................................. 17 v Executive Summary In line with UNICEF’s equity focus this paper discusses and demonstrates approaches to the empirical profiling of children in ethnic minorities. Using an equity lens, our analysis demonstrates how survey data can be used in child deprivation profiling to identify the ‘most disadvantaged’. Ethnic minority children, like all other children, are entitled without discrimination to all the rights enshrined in the Convention on the Rights of the Child and other relevant international human rights instruments. Investing in better information on children from minority communities means improving our understanding of ethnic classifications, alongside issues of human and economic development. Our research addresses the challenge of using Multiple Indicator Cluster Survey (MICS) data to drill down to potentially small sub-groups of the population. The approach moves ethnic analysis from one based on aggregated profiles that tend to present just two groups, the ‘ethnic majority’ and ‘ethnic minorities’, to a more nuanced approach that addresses equity issues between ‘ethnic minorities’. We focus on children from Hill People minority communities across three countries in South East Asia: Viet Nam, Laos and Thailand. By looking at a common ethnic cluster of populations, Hill People, we address how to consistently identify, measure and report equity issues for ‘indigenous peoples’ (using the UN definition that covers Hill People of SE Asia) and thus of cross-national ethnic minorities as well as national populations. The audience for this paper is primarily those who need to identify equity issues for ethnic minorities and other populations in order to accurately assess needs, differences in needs and design programme responses and determine resource allocations. In particular, the paper is a resource for country level Situation Analysis profiles conducted by UNICEF and other UN sister organisations, and similar profiling exercises undertaken by national government departments and NGOs. At its simplest we demonstrate how to use more detail in locally available data to identify the most disadvantaged children. This paper presents three country analyses alongside each other which are designed to be an exemplar that can be replicated in other countries concerned with child focused ethnic minority profiling. It provides UNICEF and other child rights partners and governments with a useful tool for equity related targeting for programming purposes. In addition, focusing on a cross-national ethnic minority group provides an opportunity for inter-regional learning and inter-country cooperation for addressing inequities. Our findings demonstrate three important areas of ethnic minority profiling: 1) The importance of rich profiling for a comprehensive understanding of ethnically based inequities, going beyond comparing the majority population against a single aggregate ethnic minority group vi 2) Robust methods and approaches to profiling small cross-national sub-groups (within the specified limitations of survey design) which can be adopted more widely in child focused ethnic profiling work 3) The potential of inter-country learning and cooperation for addressing cross-national ethnic inequities In terms of the empirical results from this approach, we find for the three countries, children from Hill People communities have consistently worse outcomes (in water, sanitation, nutrition, health, education and early childhood development) than children from the ethnic majority, and also worse outcomes than children from other ethnic minorities (except in Laos, where children from other minority groups sometime have the worst outcomes). The evidence indicates that Hill People children are well placed as a group to benefit from equity re-focus based on ethnicity. The paper encourages cross-national sharing of good policy and programming practices in overcoming the challenges children of Hill People communities face. vii 1. Introduction UNICEF’s refocus on equity in recent years requires improved identification and analysis of disadvantaged groups of the population, in order to re-prioritise programming to meet the needs and address underlying inequities of these disadvantaged groups. This paper considers ethnic equity issues as they pertain to Hill People, a cross-national ethnic minority group living in the upland regions across many countries in mainland South-East Asia. We focus on a crossnational ethnic minority group for three reasons. First, although ethnic identity and inequity are commonly linked, they are not always distinguished by national borders. Other studies in Latin America have identified cross-national indigenous populations for specific profiles of poverty and deprivation (UNICEF/CEPAL, 2010)1 and in Europe the Roma community have similarly been the subject of cross-national concern (UNICEF, 2011). Second, inter-regional learning and inter-country co-operation may be needed to identify the needs and approaches of common interest in communities that are not only cross-national but also straddle national border zones. Differences in deprivation profiles for such cross-national groups may represent differences in national policy and practice. This means that the approaches that seem to work better for a certain community in one country may well hold important lessons for neighbours with similar underlying constraints of remoteness and topography. This paper does not provide such comparative cross-national policy analysis, but rather the results highlight the importance of such future analysis. Third, measuring success in reducing inequities can be improved if done crossnationally. Consistent measures of inequity and targeting approaches for reducing inequities allow a clearer understanding of common and different factors, rather than relying solely on country-level measures and approaches for very similar interventions and circumstances. UNICEF’s programming for such communities may thus benefit from consistent cross national approaches to ethnic inequity related issues. UNICEF’s equity approach leads to some of the following questions on measurement: How to consistently and accurately identify ethnic population sub-groups How to robustly identify and measure inequity in potentially small sub-groups of the population How to set targets for redressing ethnic inequity This paper focuses on these empirical questions of identification, measurement and target setting and thus can provide useful guidance to UNICEF’s Monitoring Results for Equity (MoRES)based approach, as well as other monitoring frameworks. The aim of this paper is therefore not only to provide an overview of children belonging to a particular cross-national ethnic minority, but also demonstrate methods and approaches to profiling small sub-groups of the population that can be adopted more widely. We consider the circumstances and characteristics of children in ‘Hill People’ communities in three out of the many countries in continental South-East Asia in which they live: Viet Nam, Laos and Thailand. 1 UNICEF/CEPAL (2010) Porbreza Infantial en América Latina y el Caribe, Santiago: CEPAL & UNICEF 1 2. Equity and Ethnicity When considering the drivers of national child level disparities, ethnic inequity is a prominent feature of disparity analysis. However, in many countries there are several ‘ethnic minority’ groups and using an ‘ethnic minority’ aggregate group to profile inequity can hinder an appreciation of how heterogeneous ethnic minorities often are, as well as hinder a comprehensive understanding of ethnically based inequity. Ethnic minority communities diverge from the ethnic majority population in different ways, and treating them in an aggregate can, for example, mix high income minorities with poor minorities, as well as highly deprived children with less deprived. However, once the need to separately assess ethnic minorities is acknowledged, it leads to problems of accurate identification of these various minorities. Reporting on, for example, all the 49 ethnic groups currently identified in Laos is complicated, hence some sort of aggregation is necessary. A major empirical problem is how to consistently and accurately draw ethnic identity from survey data, such as MICS, DHS or other household surveys. National statistical offices may have ethnic definitions that differ from these surveys, as well as different conventions on aggregating small groups. Some of these approaches may reflect politicised definitions of racial or ethnic difference and must be carefully considered before being used in profiling and analysis: some groups may be assimilated into the majority2 and some groups assigned to an ‘other’ category, for instance. In this paper our approach explores and exemplifies a consistent analysis that clearly defines ethnic groups, within the specified limitations of survey design. 3. Hill People of Southeast Asia Hill People refer to communities with common characteristics in South-East Asia, but go by many names and languages and cultures. Collectively they can be known as ‘Hill tribes’, ‘Montagnards’ (especially in the Central Highlands of Vietnam), ‘Southeast Asian Massif populations’ and ‘Highland ethnic minorities’. 3 The majority live above an elevation of 500 meters, across many countries and an area of approximately 2.5 million square kilometres (approximately the size of Western Europe). They are extremely diverse, geographically dispersed and politically fragmented communities with linguistic, cultural and religious differences. Their many linguistic and cultural identities include for instance the Akha, BodoFor instance in Malaysia, where the so-called ‘Bumiputera’ majority definition conflates indigenous and nonindigenous Malay ethnicities and includes those of mixed Malay ethnicity, and is distinguished from ‘Indian’ and ‘Chinese’ minorities. Recent UNDP analysis shows that indigenous groups, subsumed within the ‘Bumiputera’ definition are amongst the very poorest populations in Malaysia (United Nations Country Team, 2011 The Millennium Development Goals at 2010) 3 Other countries inhabited by Hill People populations are, India (north), China (south) and other similar groups live in Afghanistan Kyrgyzstan, Pakistan and Tajikistan. 2 2 Kachari, Lahu, Karen, Hmong Mien, Mizo and Lisu who are all Tibeto-Burman peoples who migrated over recent centuries into the areas from southern China and Tibetan regions, as well as groups who speak Malayo-Polynesian, Tai, and Mon–Khmer based languages. Hill People are recognised as being distinct from lowland living communities in several important ways. Lowland communities, while diverse, generally share a common language and culture and have developed and maintained the dominant social institutions in national political environments. Hill People do not share that heritage, and historically persistent conflicts exist between highland and lowland groups over many issues, including land ownership, language and cultural preservation, access to education and resources, as well as political representation. The commonality across Hill People is they are ethnic and cultural political minorities and live on both the topographic and social fringes of society, in elevated and rugged terrain. In these terrains they have tried to preserve their local cultures from state control and influence of lowland majorities4. They are mostly non Buddhist/Confucian in religious beliefs and traditionally animist and shamanist, but in some areas they have adapted to major religions such as forms of Christianity. Their economic systems do not rely on paddy rice or industrial production (which the majority group most commonly relies on), instead their systems traditionally ranged from hunting and gathering, forest horticulture, swidden farming and upland rice growing, and trade with lowland communities – particularly in forest and mineral products (gems and silver especially). Prior to prohibition, traditional agricultural activity also involved opium poppy production.5 Our analysis focuses on Laos, Thailand and Viet Nam as all these countries have household survey data from the Multiple Indicator Cluster Survey (MICS) from 2005/6, and all countries participated in the MICS4 round allowing for potential further analysis. We now turn to consider the position of Hill People profiling in these countries 3.1Viet Nam Most ethnic, social and economic inequity profiling to date in Viet Nam has compared all ethnic minorities against the Kinh majority (and often with the Chinese ethnic group aligned with the majority Kinh group), and not explicitly identified Hill People. Poverty profiling has demonstrated that ‘ethnic minorities’ living in the Northern and Central Highlands (and thus disproportionately Hill People) have substantially higher risk of poverty than other ethnic minorities.6 While the Vietnamese government has directed significant resources and designed specific programmes aimed at ethnic minorities, as well as officially prohibiting discrimination against ethnic minorities, there is a persistent lag in development and 4 Scott, J. (2009) The Art of Not Being Governed. New Haven:Yale University Press. Michaud, J. (2006). Historical Dictionary of the Peoples of the Southeast Asian Massif 6 World Bank (2009) Country Social Analysis Ethnicity and Development in Vietnam: Summary Report. Washington DC: The World Bank. 5 3 longstanding societal discrimination against Hill People. Their higher poverty risk and lower agricultural productivity is often ascribed to ‘backwardness’ by the majority. The issue of land rights and relocation has affected Hill People in Viet Nam, many of whom have lost productive gains from land for several reasons, such as programmes of ‘sedentarisation’ and the loss of customary land, as well as resettlement of Hill People villages. As said, remoteness in upland areas is one distinguishing feature of Hill People. However, Viet Nam analysis has shown that the independent roles of remoteness and topography can only partly explain differences in poverty causation and inequality, and thus factors relating to ethnicity play a large part.7 Detailed survey and assessment of ethnic differences focus on low assets (both in physical and human development terms); barriers to access for education, healthcare8, labour market and commerce; very low migration rates to urban centres; poor cultural appropriate in the design of anti-poverty and other programmes; and the lack of social and political voice.9 Recent profiles of ethnic minorities in Viet Nam suggest that other rurally located minorities such as Cham and Khmer communities have experienced some narrowing of the gap with the Kinh majority, however for highland minorities, and especially those in the Central Highlands, the gap has widened. Industrialization and foreign direct investment are mostly concentrated in the major cities where the Kinh ethnic majority mostly resides and therefore benefit the most from the associated rapid economic growth.10 Development projects in the rural and remote areas have also been largely dominated by ethnic majority based finance for migrant settlers and commercial agricultural projects. The 2008 Viet Nam Household Living Standard Survey showed only modest progress in reducing poverty levels among ethnic minority populations, which fell from 52.3 per cent in 2006 to 49.8 per cent in 2008. This is much less than the reduction seen in the poverty rate of the Kinh population (down from 10.3 to 8.5 per cent). Children belonging to ethnic minority groups are particularly at greater risk of being poor compared to children from the Kinh/Chinese ethnic majority, and this is especially the case in rural areas.11 The 2011 Viet Nam MICS demonstrates substantial ethnic disparities; ethnic minority children are three times as likely as Kinh/Hoa children to die before their first and fifth birthdays, and likewise there are large disparities between the nutritional status of ethnic minority children as compared to ethnic majority children, and one in every four Vietnamese living in ethnic minority households defecate in the open.12 7 Epprecht M., Müller D. and Minot N (2011) Annals of Regional Science Vol 46, #2, pp 349-368. Swinkels R. and C. Turk (2006) Explaining Ethnic Minority Poverty in Viet Nam: a summary of recent trends and current challenges, mimeo, Washington DC: The World Bank 9 World Bank (2009) op-cit. 10 Baulch B., Hung T. Pham and B. Reilly (2007) Ethnic minority and household welfare in Viet Nam: Empirical Evidence from 1993 to 2004, London/Brighton: DfiD & Institute of Development Studies. http://www.dfid.gov.uk/r4d/PDF/Outputs/ESRC_DFID/60423-ethnicity_household_welfare_vietnam.pdf 11 MOLISA and UNICEF (2008) Children in Viet Nam – who and where are the poor? 12 General Statistical Office (GSO), Viet Nam Multiple Indicator Cluster Survey 2011, Final Report, 2011, Ha Noi, Viet Nam 8 4 An evaluation of various government programs to support ethnic minorities suggests that they have been insufficient and often poorly targeted towards ethnic minorities, as well as poorly matched to their specific needs of language and culture, for example only one out of seven vocational training projects and two out of five housing projects focus on the specific needs of ethnic minorities13. 3.2 Lao PDR Lao PDR is among the poorest South-East Asian countries and its 5 million population is very ethnically diverse. Forty nine distinct ethnicities with around 200 ethnic subgroups are spread across four broader ethno-linguistic family groups: the majority group of Lao-Tai, who are 67% of the population, the Mon-Khmer, 21%, and Hill People from the Hmong-Lu Mien (8%), and the Chine-Tibetan (3%) linguistic groups.14 A simpler and cruder classification frequently used in the past and to this day remains in unofficial categorisation, splits the population by elevation: lowland, middle land and highland dwellers. This rough categorisation means that the Lao-Thai live in the productive paddy-rice growing and more urbanised lowlands, the Mon-Khmer in the midland rural areas, and Hill People in the highlands. These simplistic divisions reflect and reinforce widely held appreciation of ethnic and cultural difference between settled lowland paddy rice growers and hill-living dry rice growing, swidden and sometimes nomadic Hill People communities. Laos government policy on redressing ethnic inequalities since the 1980s has placed emphasis on assimilation through schemes such as resettlement of communities to be nearer government services and facilities, to stop opium production, to improve security, to promote ‘nation building’ and to reduce levels of swidden agriculture. In many ways the Lao approach to highland and Hill People can be seen as the opposite to Viet Nam’s – to move the inhabitants to lower lying areas rather than sponsoring lowland people to migrate to the highlands. Since 2003, one half of Laos districts have been ‘priorities’ for development in an attempt to narrow ethnic and regional disparities but results have been mixed, and moving highland people to lower lying and more crowded places has often increased problems of vector borne and infectious disease, reduced access to assets and increased poverty risk as well as heightening land disputes in the destination areas.15 Most recent and thorough analysis of ethnic differences in poverty have compared the Lao-Thai majority to the remaining aggregate group of ‘ethnic minorities’.16 Older analysis shows Hill People poverty headcounts in 2002/03 at 45% for Hmong-Mien and 40% for Chine-Tibet Jones R., Tran Thi Hanh, Nguyen Anh Phong and Truong Thi Thu Trang (2009) A Mapping Exercise – Poverty Reduction Programmes and Policies in Viet Nam 14 King E and van de Walle D (2012) Laos: Ethno-linguistic Diversity and Disadvantage in Hall G and Patrinos H (eds) Indigenous Peoples, Poverty and Development, Cambridge: Cambridge University Press. 15 Baird I and Shoemaker B (2007) Unsettling Experiences: Internal Resettlement and International Aid Agencies in Laos, Development and Change, 38(5) pp 865-888. 16 King and van de Walle (2012) op-cit. 13 5 language groups, much higher than the 29% for the Lao-Thai majority, but at lower levels than the Mon-Khmer group, 54%.17 Detailed decomposition of underlying differences for ethnic disparities in poverty incidence in Laos support findings in Viet Nam: that Hill People’s poverty is partly explained by lower access to productive land, larger family size and poorer infrastructure but that their efficient use of lower productive land and other scarce natural resources compared to lowland agriculturalists and this protects them from many poverty shocks.18 More generally, Laos ethnic minorities are less educated, partly a reflection of lack of access to both primary and secondary schools, and in rural areas ethnic minority children spend less time at school per day compared to their ethnic majority counterparts. Ethnic minorities also have a higher incidence of temporary health problems and are less likely than the ethnic majority to seek treatment for health ailments, reflecting both limited access and demand for health services among ethnic minorities in Lao PDR.19 20 The percentage of underweight children below the age of five is also higher among ethnic minorities, for the Hmong- Mien it is 37% and for the ethnic majority it is 34.3%.21 Ethnic minorities generally have less land the ethnic majority in Laos, and produce crops between two fifths to one quarter less per capita22. 3.3 Thailand The position of Hill People (known as ‘Hill Tribes’ or ‘Highlanders’ in Thailand) as a distinct sub-group of ethnicities minority is clearly recognized by the Thai government, and programmes have been in place since the mid-1970s to address their social and economic development. There are several distinct communities: The Karen, Hmong, Mien, Akha, Lahu, Yao, Lua H’tin, Khamu and Lisu, who together make up some 800-900,000 people. However, population counts are uncertain as registration of individuals, households and ‘villages’ for official Thai local administration is not consistent due to a range of factors.23 Many in these communities have ‘stateless’ identity as they have not registered or obtained Thai citizenship. This gives rise to distinct and separate treatment in terms of civil and political rights and increases the risk of both exclusion from social services (with identity cards needed to access health and education) and exclusion from limits placed on geographical mobility within the country. ‘Non-citizens’ are prohibited from traveling outside their home districts without prior permission from the district 17 Engvall A. (2006) Ethnic Minorities and Rural Poverty in Lao PDR (background paper for World Bank Poverty Assessment Report), Stockholm: Stockholm School of Economics (mimeo). 18 Engvall (2006) op-cit and van de Walle D. and Gunerwardena D (2001) Sourcs of Ethnic Inequality in Viet Nam, Journal of Development Economics, Vol 65 pp 177-207. 19 World Bank (2006) Lao PDR Poverty Assessment Report From Valleys to Hilltops — 15 Years of Poverty Reduction, Washington DC: The World Bank http://unpan1.un.org/intradoc/groups/public/documents/apcity/unpan035358.pdf 20 King and van de Walle (2012) op-cit 21 World Bank (2006) op-cit 22 ibid 23 Aguettant J (1996) Impact of Population Registration on Hilltribe Development in Thailand, Asia-Pacific Population Journal, Vol 11 #4 pp47-72. 6 office or outside their home provinces without permission from the provincial governor. Those without cards may not travel at all and have difficulty accessing any services, owning land, and accessing bank credit. There are few recent sources of published data on Hill People’s poverty or other deprivations in Thailand. The 2001 World Bank Poverty Assessment Report gives no ethnic breakdowns of poverty incidence, but confirmed that factors associated with Hill People were linked to higher risk: being in North East and North regions, rural, of low education status and with small landholdings.24 The most recent Human Development Reports for 2003, 2007 and 2009 all discuss various issues concerning Hill Peoples’ rights and development but provide no ethnic breakdowns of poverty or Human Development Indicators, and their discussion of ethnicity focuses on issues of migrants and refugee populations that include Hill People in part. A specialized survey in 2006 by UNESCO and the Bureau of Social Development found more than 30% of Hill People did not have citizenship and that lack of knowledge of rights or language problems explained 50% of the low birth registration. Those children without official citizenship were 73% less likely to enter primary school compared to their citizen peers and 94% less likely to enter secondary school. Overall, when compared to ethnic majority Thais living in the same areas, Hill People were approximately 95% less likely to enter primary school. Underlying economic discrimination resulted in Hill Peoples being between 16% (Lahu) and 93% (Mien) less likely than the ethnic Thai majority to have land ownership rights. 4. Hill People Populations While Hill People communities are ‘minorities’ when considered in national populations they actually represent significant cross-national populations in Southeast Asia. For example, the Hmong community total around 5 million in Asia, almost equal to the population of Lao PDR.25 Overall, Hill People in the South-Eastern portion of the Asian landmass total to around 80 million.26 Hence, some care is needed in considering how national ‘ethnic minorities’ link to considerable cross-national populations of similar cultures and ethnicities. Of the three countries we focus on in this report, the largest, in terms of population, is Viet Nam, where the 2009 Population and Housing Census shows a total population of 85.8 million, of which the Kinh majority was 86%. Fifty three ethnic groups make up the remaining 14%, of which the larger ethnic groups are Thai: 1.6 million (1.8%); Muong: 1.3 million (1.5%); and Mong: 1.1 million (1.2%).27 The 2005 Lao PDR census reports a total population of 5.6 million and identified 49 different ethnic groups. The majority, 55%, were Lao-Thai and Hill People such as the Hmong, comprise 24 World Bank (2001) Thailand Social Monitor: Poverty and Public Policy, Washington DC: The World Bank. Michaud, J. (2006). op-cit 26 ibid 27 http://www.gso.gov.vn/default_en.aspx?tabid=515&idmid=5&ItemID=10799 25 7 8% of the population and the Khamu, who are an ethnic minority group that mainly resides in middle altitudes in the north, comprise 11% of the population.28 The 2010 Thailand census reported a total population of 65.5 million29 but there is no information on ethnic minorities in the English census report30. Estimates of Hill People population numbers are uncertain as previously discussed, but other sources such as the FAO, report 3,527 Hill People villages in 20 provinces of Thailand comprising more than 750 thousand people. This suggests Hill People are around 1% of the whole population. The largest Hill People group in Thailand is Karen (46 percent), followed by Hmong (16 percent) and Lahu (11 percent).31 From these census numbers we estimate that 19% of Laos’, 11% of Viet Nam’s and 1% of Thailand’s populations are of Hill People ethnicity. 4.1 Identifying Hill People in Survey Data We now turn to address the first of our empirical questions outlined earlier, “How to consistently and accurately identify ethnicity-related population sub-groups” when profiling child level disparities for UNICEF equity profiling. Our previous overview of Hill People populations in the three countries gives us good grounds to base our equity profiling to identify children in Hill People populations as potentially being among the ‘most disadvantaged’. But how far are we able to take forward such equity concerns about children in Hill People communities using national MICS survey datasets? MICS datasets all have ethnic identifiers (based on languages spoken in the household), but each national dataset has differing interpretations of groupings of ethnic and linguistic identities. The three country level MICS datasets produced different problems and solutions to identifying Hill People: In the Thai dataset, a category for ‘Hill tribes’ was already defined as a specific identity (alongside Cambodian (Khmer), Lao, Myanmar, Chinese and Thai). In Lao PDR data, the ethnicity flags identified just four groups: ‘Mong’, ‘Khamu’, ‘Lao’ and ‘other’. Tabulations showed distributions by region that put the majority of ‘Mong’ in the North along with the Khamu group, but when the definitions were matched to more detailed breakdowns of ethnic and linguistic groups, there were two main problems in interpreting ethnic difference. Firstly, the ‘Khamu’ definition could well refer to the Lao practice of distinguishing between ‘midlanders’ (or Lao Theung) and ‘highlanders’ (Lao Sung) and thus including Khamu within a ‘Hill Tribe’ grouping could confuse a local readership. We also knew from the earlier literature review that poverty rates were higher for the Khamu than Hmong groups and thus conflating them into a single group may obscure clear analysis. 28 http://www.nsc.gov.la/Products/Populationcensus2005/PopulationCensus2005_chapter1.htm http://popcensus.nso.go.th/en/ 30 http://popcensus.nso.go.th/upload/popcensus-08-08-55-E.pdf 31 http://www.fao.org/docrep/004/ak216e/ak216e04.htm 29 8 Secondly, there were other smaller communities of Hill People such as the Akha, Mien and Lahu that were not separately identified but subsumed within the ‘other ethnicity’ group. In the absence of clear ethnic identifiers, we chose to use the best proxies available in the data and drew Hmong (Mong) communities and all the ‘other ethnicity’ communities that lived in the North region into a combined group of ‘Hill People’, and separately identified the Khamu community. Our results should be interpreted bearing in mind the uncertainties that spring from this optimal but unsatisfactory classification. A comparison of these groups to the national level documentation from the Statistical Office and other sources seemed to suggest that large errors would be minimal. The Vietnam MICS had less aggregated and more accurate identification of ethnic communities. We used the specified ethnicity flags to compare with ethnographic and other profiling materials such as location flags such as region and district to attribute ‘Thai’, ‘Kho Mu’, ‘H’Mong’, ‘Dao’, ‘Gia-Rai’, ‘Bana’, ‘Xo Dang’, ‘Mnong’, ‘Bru’, and ‘Khang’ groups to Hill People status and left the remaining ethnic communities in a residual ‘other minority ethnic groups’ category once Kinh and Chinese communities had been aggregated into the ‘majority ethnic group’. These different types and levels of data difficulties reinforce the recommendations made recently by the World Bank Country Social Analysis in Viet Nam which apply to other countries in South-East Asia, “More and better information is needed on minorities, from how they are classified by ethnic group to where they are located in the country to their levels of economic development. …….. Poor classification and overly general data can lead to inaccurate targeting of resources, while more detailed local data can help identify the most vulnerable.”32 4.2 Population and Sample Sizes Consideration of ethnic equity issues relies not only on the sub-group characteristics and correct identification but also on the underlying population sizes and how these are captured in survey data. Hill People are small minorities within in each country and thus lead to potential ‘small numbers’ problems in survey data. This is particularly problematic when considering indicators of child deprivation and wellbeing where the specific age-groups for measuring, for example, immunization coverage, enrollment in programmes and education may result in small samples compared to ‘all children’ for such groups. This leads to potential problems in obtaining consistent measurement (for instance for MDGs) for small population sub-groups The problem across the three countries considered in this paper is worst for Thailand, where MICS3 survey data indicates that 0.6% of the individual sample is from Hill People communities, less than would be expected to match Census profiles. Even the very large overall sample size for MICS3 in Thailand of over 137,000 individuals means that only around 750 are of Hill People ethnicity, and, once we start looking at children by age-groups to capture health and education and other 32 World Bank (2008) op-cit p. 5 9 deprivations, then sub-samples become very small indeed. This is less of a problem for Laos and Viet Nam where MICS3 samples of Hill People are 4,444 (13.3% of survey sample) and 4,594 (12.6% of survey sample) respectively. This brings us to the second of our major empirical questions outlined earlier: “How to robustly identify and measure inequity in potentially small sub-groups of the population?” We employ three approaches to ensure that our profiling recognizes the problems of small sub-samples Using population weights for all profiling but being careful to acknowledge a) small underlying numbers and b) that population weights are not designed to make ethnic subgroup’s representative, but are usually based on sampling frames and strata based on region, district or urban/rural populations. For instance, using population weights in Thailand reduces the proportion of Hill People from 0.6% in un-weighted to 0.3% in weighted proportions of the total population. For all of the analysis presented in this paper we use population weights. Showing statistical significance of small sub-samples is crucial when reporting simple profiling results – for instance, it is not robust to report the proportion of Hill People children aged 12-14 who are married in Thailand because the underlying sample for this age group is below 10 observations. We use statistical tests, t-tests, to ensure that small sample sizes are large enough to report significant profiles (that they are not showing results that could have occurred randomly). Comparing sub-groups to show ‘significant difference’ is key to equity profiling: to demonstrate that children in Hill People communities have, for example, different rates of immunization, school progression and malnutrition when compared to the ethnic majority, and, perhaps, when compared to other minorities. However, comparing subgroup populations of different sample sizes means that each estimate has different level of significance. We therefore show the 95% confidence intervals that are either side of the point estimate when comparing Hill People to other population sub-groups. This allows a distinction between cases where the 95% confidence intervals overlap (thus frustrating any finding of significant differences) and those cases where differences are robust and reliable. Even when we adopt such careful profiling practices there remains the problem of how to interpret differences that are found between Hill People minorities and other populations. How far do differences reflect different characteristics rather than issues of human rights, such as different access to or treatment by public services is always wise to bear in mind. For example, are Hill People communities different in the way they form households and families, as well as their particular location and environmental situations? 10 4.3 Household Size and Composition Hill People live in relatively larger households when compared to ethnic majority households, as shown in table 1: the mean household size of Hill People in Viet Nam is 6.1, compared to 4.9 for the Kinh majority. In Lao PDR, mean household size for Hill People is 7.7 compared to 6.1 for Lao majority but overlapping confidence intervals suggest that differences with other minorities are not significant. In Thailand, mean household size for Hill People is 4.5 compared to 4.3 for the ethnic majority but the wider confidence intervals for the small sub-group mean that it could be between 4.3 and 4.6 and thus not statistically different from the majority. Considering children only, Hill People households, on average, have larger numbers of children than the ethnic majority. Hill People households in Viet Nam have 2.8 children on average as opposed to 1.7 for the Kinh majority; in Lao PDR, 4.4, compared to 2.7 for the Lao-Thai majority and in Thailand 1.6 compared to 1.4 for the Thai majority. Larger numbers of children between Hill People and other ethnic minorities are significant in Viet Nam and in Laos but only when comparing with Khamu minority; otherwise overlapping confidence intervals mean that clear differences between Hill People and’ other minorities’ in Thailand and Laos are not discernible. Larger numbers of children on average is accompanied by having higher incidence of household with children in their communities. 94% of Hill People households in Viet Nam have children compared to 83% of the Kinh majority; in Lao PDR, 98% do so compared to 94% of the Lao majority. In Thailand, no difference is discernible; 72% of both Hill People and Thai majority households have children. Children in Hill People communities are younger on average than ethnic majority children, children’s mean age in Viet Nam is 8.0 years, compared to 9.0 years in the Kinh ethnic majority households; in Laos 7.8 compared to 9.0 for Lao and in Thailand, 8.0 compared to 8.8. Table 1: Difference in Hill People household size and composition Mean number of household members Viet Nam 95% CI Laos 95% CI Thailand 95% CI Hill People 6.1 6.0 – 6.2 7.7 7.6 – 7.8 4.5 4.3 – 4.6 Other minorities 5.1 5.1 – 5.2 7.7 7.5 – 7.8 4.2 4.1 – 4.4 Khamu minority -- -- 6.8 6.7 – 6.9 -- -- Majority 4.9 4.9 – 4.9 6.1 6.0 – 6.1 4.3 4.3 – 4.3 Mean number of children per household Hill People 2.8 2.8 - 2.9 4.4 4.4 - 4.5 1.6 1.5 – 1.8 Other minorities 1.9 1.9 -2.0 4.3 4.2 – 4.4 1.4 1.3 – 1.6 Khamu minority -- -- 3.5 3.5 – 3.6 -- -- Majority 1.7 1.7 -1.7 2.7 2.7 – 2.7 1.4 Average child age by ethnic group 11 1.4 -1.4 Hill People 8.0 7.8 – 8.3 7.8 7.7 – 8.1 8.0 7.3 – 8.8 Other minorities 9.1 8.8 – 9.5 7.8 7.6 – 8.0 7.3 6.5 – 8.2 Khamu minority -- -- 8.2 8.1 – 8.5 -- -- Majority 9.0 8.9 – 9.1 9.0 8.9 – 9.2 8.8 8.7 – 8.9 In short, Hill People communities have larger households, with more and younger children, than the ethnic majorities. 4.4 Locational Characteristics We do not have the ability from MICS data to empirically capture good measures of ‘remoteness’ for Hill People, for instance the distance of communities from paved and unpaved roads, from health facilities, from schools of different types and levels. We have already seen from the early summary discussion that there are restrictions on mobility and migration for Hill People in all three countries combined with policies of relocation and ‘sedenterisation’. Location is a key issue to understanding underlying ethnic differences not just because of underlying geographical characteristics of remoteness and topography, but also due to policy reasons related to service provision in such locations and to relocation of populations in and out of them. In Viet Nam, only 0.7% of Hill People live in ‘urban’ designated areas, compared with 23% of the majority ethnic population. In Lao PDR the same comparison is 1.3% compared to 22% and in Thailand, it is 0.03% compared to 30%. Some care is needed in interpreting the actual urbanity of Hill People because they may travel to trade in market towns, that may or may not be designated as ‘urban’ areas, and there may be ‘illegal’ settlement in provincial towns (or cities such as Chiang Mai in Thailand) that is not captured in survey sampling. But the overall position is quite clear. Not only are Hill People predominantly rural, they are also in based in hilly terrain and at distance from formal employment opportunities, access to financial services, trade and from social and education service hubs. 5. Hill People and Ethnic Equity Profiling Are Hill People the ‘most disadvantaged’ in terms of ethnic equity in these countries? To answer this we compare the incidence of several UNICEF programme-related deprivations across ethnic communities, with Hill People distinguished from ‘other minorities’ and the ethnic majority group in each country. For all the following analysis we use data from Multiple Indicator Cluster Surveys (MICS) for the three countries from the year 2006. 5.1 Asset Quintiles 12 Figure 1 shows the proportion of all ethnic communities that are in the poorest quintile of asset holdings (at the individual level) for all three countries. There is a clear ethnic propoor gradient across all three countries with Hill People concentrated in the poorest quintile.33 The starkest concentration is in Viet Nam, where 88% of Hill People are in the poorest quintile, as opposed to 12% of those belonging to the majority ethnic group and 62% for other ethnic minorities. Similarly, in Thailand, 69% of Hill People are in the poorest asset quintile, far higher than other ethnic minority groups, 28%, and the majority Thai ethnic group, 19%. In Lao PDR, Hill People are highly over-represented in the poorest quintile, 38%, but not more so than other ethnic minorities, also 38%, and less so than the Khamu minority, 49%. However, all Lao minorities are overrepresented compared to the Lao majority ethnic community of whom only 5% of are in the poorest quintile. Figure 1: Population/children in poorest quintile % Ethnic inequity is more pronounced when children (aged 0-17 inclusive) are considered; we see that the proportion of children from Hill People populations in the poorest asset quintile rises to very high levels. In Viet Nam, 90% are in the poorest asset quintile; in Thailand, 75% (with 95% confience intervals of 68% and 82%, so we can not definitely say that there is a higher proportion than adults), and in Lao PDR 39%, while 51% of the Khamu minority children are in the poorest quintile. Part of the explanation for these higher proportions is the previously explained demographic profile of Hill People: they have more children in larger households as well as having low levels of assets. 5.2 Deprivation, MDGs and Child Well-Being Measures Keeping our focus on children, the following analysis of some of the main pillars of child wellbeing (that are directly linked to UNICEF programme activity) illustrate the ethnic equity 33 The Wealth Index is a composite measure of cumulative living standard of a household, calculated using data on household’s ownership of selected assets, such as televisions and bicycles, materials used for housing construction, and types of water access and sanitation facilities. Generated with a statistical procedure known as principal components analysis, the Wealth Index places individual households on a continuous scale of relative wealth 13 position for Hill People, namely and in order; water, sanitation, nutrition, health, early childhood development and education. Unfortunately the sample size was too small to look at child protection indicators, such as child marriage and birth registration. Where possible, we use indicators that reflect international standards for dimensions for child well-being, such as the Convention on the Rights of the Child, the World Summit for Children, and the Millennium Development Goals. To be consistent and help interpretation we profile using ‘deprivations’, i.e. the absence or lower quality of a desired state of well-being or the direct incidence of ‘negative’ indicators of child well-being. This means that ethnic inequity is captured in ‘higher’ scores for minority ethnic communities such as Hill People when compared to other ethnicities. a. Drinking Water MDG 7 on environmental sustainability includes commitments to improve water and sanitation facilities. The relationship between water and child health and survival is well established, where unsafe drinking water severely impacts children’s health. The relationship between poverty and poor access to improved34/safe source of drinking water is also well established. In Viet Nam, 30% of Hill People children compared to just 9% of Kinh/Hoa majority 80% 64% 59 children use unimproved drinking water 51 60% 51** sources, but other ethnic minorities have 36 40% 30** 27 broadly similar disparity (27% to 9% (the 14 20% majority)). In Thailand, 68% of Hill 9 5 People children use unimproved water 0% sources compared to 5% of children from the majority ethnic group. In Lao the picture is almost reversed, where 59% of children from other ethnic minorities use VIET NAM LAOS THAILAND (ave 12%) (ave 50%) (ave 6%) unimproved drinking water source, and ** t-test significant at 0.05 or higher 51% of both Hill People and the ethnic Figure 2: Children in households with unimproved drinking water majority children use unimproved drinking source water source. A large percentage of both the ethnic majority and Hill People children get their drinking water from unprotected wells, which is considered an unimproved source of drinking water, while a large percentage of the Khamu minority get their water from public stands which are considered an improved source of drinking water, explaining why the Khamu have the best access to improved drinking water source of all the ethnic groups. 34 Improved water source: Piped water into dwelling, plot or yard; public tap/standpipe; tubewell/borehole; protected dug well; protected spring; rainwater collection. Unimproved water source: Unprotected dug well; unprotected spring; cart with small tank/drum; bottled water; tanker-truck; surface water (river, dam, lake, pond, stream, canal, irrigation channels). Bottled water is considered improved only when the household uses water from an improved source for cooking and personal hygiene 14 b. Sanitation 100% 87** 72** 80% 84 67 53 60% 49 33 40% 13** 8 20% 0% VIET NAM LAOS (ave 39%) (ave 61%) 1 THAILAND (ave 1%) Poor access to improved sanitation35 is linked to various child related illnesses, most notably diarrhea which is one of the main killers of children under five years. Rural populations, such as Hill People, generally have poorer access to improved sanitation than urban populations, and this is substantiated in these countries. In Viet Nam 87% of Hill People children have unimproved sanitation compared to 33% for Kinh majority children. In Lao PDR ** t-test significant at 0.05 or higher 72% of Hill People children have unimproved sanitation, while 49% of the Lao majority have unimproved sanitation. The sanitation situation is worst for other ethnic minority groups, where 84% of children have unimproved sanitation. The difference is not as stark in Thailand, where 13% of Hill People children have un-improved sanitation, as opposed to just 1% of the Thai majority. Figure 3: Children in households with unimproved sanitation A large percentage of Hill People children have no access to any kind of sanitation facility in comparison to children from the ethnic majority across all three countries. In Viet Nam, 69% of Hill People children have no access to any sanitation facility as opposed to 9% of Kinh majority children. The disparities are less evident in Lao PDR but the percentage of Hill People children without access to any sanitation facilities is still high, or 68%, and this is also high, or 43% for children belonging to the majority ethnic group. More disturbingly, 84% of children from other minority groups have no access to any kind of sanitation facility. In Thailand 10% (95% confidence intervals between 4%, 16%) of Hill People children have no access to any kind of sanitation facility as opposed to 1% of the ethnic majority children. c. Nutrition 60% 49** 51 40% 49 31 18** 9 20% 12 0% Insufficient nutrition in childhood has been linked to stunted physical growth, poorer learning outcomes, poorer health and so forth. Malnutrition is clearly linked to poverty. Viet Nam has a separate General Nutrition Survey and results by ethnicity 35 Improved sanitation: Flush toilet/latrine; VIP latrine; pit latrine with slab; composting toilet. Unimproved: flush to elsewhere; pitLAOS latrine without slab/open pit;THAILAND bucket; hanging toilet/hanging latrine; no facilities, bush or field (ave 41%) (ave 12%) ** t-test significant at 0.05 or higher Figure 4: Children 0-4 years stunted % 15 are not available36, however regional disparities that can be linked to Hill People populations suggest higher rates of stunting compared to lowland majority ethnic areas. In Lao and Thailand the MICS includes various anthropometric information, among them height for age (stunting) which is indicative of long-term insufficient nutrient intake and frequent infections. In Lao stunting rates of 49%, for Hill People children (aged 0-4), are higher than that of the Lao majority (31%), but rates are even higher for other ethnic minorities (51%). In Thailand, stunting rates appear to be highest for Hill People children at 18% but 95% confidence intervals at 8% to 28% show that these cannot be interpreted as significantly different from children from the ethnic majority group (12%). d. Health Immunisation rates are an important measure of child well-being as over 60% of child deaths in low income countries are attributable to infectious diseases, and the most effective prevention is full immunization for all major diseases. The best immunisation indicator is based on a denominator of children aged up to 23 months. But small sample sizes make this denominator 100% problematic when considering small 73** 80% ethnic sub-groups such as Hill People 55 58 51 60% children. We therefore choose a larger 40% age-cohort of those aged under 5 years to 14 12** 6 20% 8 assess those who have never received 7 2 0% any vaccination. Again, overall rates of -20% immunisation vary, with higher coverage in Viet Nam and Thailand compared to Lao PDR. However, evidence of ethnic VIET NAM LAOS THAILAND disparity is clear. In Viet Nam, 12% of (ave 4%) (ave 58%) (ave 8%) **t-test significan at 0.05 or higher Hill People children aged 0-4 have never Figure 5: Children under 5 years who have received no vaccination received any vaccination, compared to 7% for other ethnic minorities and 2% for the ethnic majority. In Lao PDR, much lower overall coverage is even lower for Hill People children, where 73% of children have never received any vaccination, worse than 55% for other ethnic minorities and worse than 51% for the ethnic majority. For Thailand, again, the problem of small numbers confounds an equity analysis as the numbers of those never vaccinated are small and thus comparing rates for the different ethnic groupings cannot be done robustly. e. Education 36 http://www.unicef.org/Viet Nam/resources_18459.html 16 Enrolment in compulsory primary education from the age of six37 is high across all groups in all three countries, with little obvious 92 66** 87 difference between ethnic groups in 85** 100% 80% 40 enrolment and reported attendance. Of 57 60% course, quality of schooling and 40% 25** 10 6 4 20% subsequent drop out may differ 0% ethnically in unobserved ways. However, when we consider an older group of children, those aged between the ages of 12 to 14 years who should VIET NAM LAOS THAILAND (ave have completed primary and be (ave 68%) 10%) (ave 5%) ** t-test significant at 0.05 or higher attending lower secondary education, Figure 6: Children aged 12 to 14 not progressed to secondary we see significant ethnic disparities. In education % Viet Nam 25% of Hill People children of this age have Primary education or lower as their highest level of education, compared to 4% of the majority, suggesting huge differential fall out from schooling. In Lao, overall progression to lower secondary schooling is less common for all ethnic groups, but even so, Hill People children are more likely to have primary education or lower as their highest level of education, 85%, compared to the ethnic majority children, 57%; but other ethnic minorities seem to have the greatest disadvantage at 92%. Results for Thailand are not statistically robust for the very small Hill People samples of this age group. f. Early Childhood Development Early childhood is considered the most critical period for children’s current 100% 86 66 62 and future well-being, and early 80% 69** 60% childhood educational programmes can 39 35 38 40% improve children’s cognitive and 20% physical development. However, the 0% implementation of such programmes across the three countries differs very greatly, with higher coverage in Thailand and Vietnam across all THAILAND VIET NAM LAOS (ave 39%) children than in Lao PDR. In terms of (ave 39%) (ave 92%) ** t-test significant at 0.05 or higher ethnic disparity in provision, in Viet Nam 69% of Hill People children do Figure 7: Children not attending early childhood education programme % not attend early childhood education (ECD) programmes as opposed to 35% of the Kinh majority children. In Lao PDR 98% of Hill People children do not attend ECD programmes as compared to 86% of children of Lao majority 98** 99 97 37 http://stats.uis.unesco.org/unesco/TableViewer/tableView.aspx?ReportId=163 17 group. However, results for Thailand once again are not robust due to small sample sizes except for the majority population where just 38% of Thai majority children do not attend ECD, as opposed to rates to over 60% for ethnic minorities in general. 6. Equity Monitoring and Target Setting The evidence presented in this paper indicates that Hill People children are well placed as a group to benefit from equity re-focus based on ethnicity across these three countries, and potentially in other countries of South-East Asia. By separate and distinctly profiling Hill People children and comparing them to majority ethnic populations we clearly demonstrate aspects of inequity that are obscured when ‘all ethnic minorities’ are compared to the ethnic majority. This is especially true for Viet Nam, where grouping different ethnicities into a Hill People group shows that they are ‘most disadvantaged’ when compared to other ethnic minorities as well as when compared to the ethnic ‘Kinh’ majority. In Lao and Thailand, where ‘Hill People’ ethnicity is recognized as more distinct than in Viet Nam, there is clear evidence of ethnic inequity. But data problems partly frustrate clear profiling as less detailed ethnic groupings in the Lao MICS data make it more difficult to profile ‘Hill People’ distinctly, and for Thailand small sample sizes made it more difficult to profile children from Hill People communities in terms of statistical robustness. Having demonstrated a case for considering equity refocusing towards children in Hill People communities, we now turn to look at the final question for this paper, outlined earlier: “How to set targets for redressing inequity?” While the first step in any equity re-focus is to accurately identify where inequity exists, there also needs to be a clear aim for refocused programming for an equitable outcome. These may be medium term or long term targets, depending on how large the inequity gap is to be narrowed and the volume and capacity of government programme outputs. An ‘end-point’ defined in equitable terms is a helpful policy tool even if it takes years or decades to reach it. However, defining an end-point equity outcome is not straightforward and there are several potential approaches to setting equity targets. The different approaches listed below are designed to be an exemplar that can replicated, by UNICEF and other partners, in other countries concerned with child focused equity targeting based on ethnicity. 6.1 Equity Target Option 1: Equality of Ethnic-Related Risk It is possible to think of an equity position for Hill People children based on ‘equality of ethnicrelated risk’. This approach attempts to reduce the incidence of deprivations and rights violations that directly stem from factors that relate to ethnic identity rather than other factors. This approach seeks to redress particular and systematic factors that explain ethnic differences, rather than commonly occurring factors which may be due to, for example, low income or poor education. These factors can affect all ethnic groups but are worse for the target ethnic group. 18 However, while this seems a sensible approach, we almost certainly lack a full picture on what is driving unequal incidence across ethnic communities. For instance, how far are the current child level inequities the result of discrimination and how far do they simply represent differences in characteristics that are not necessarily linked to ethnic discrimination: differences in household size and composition, fertility, location, and in characteristics such as adult literacy? One way of thinking about the particular ethnic disparities in well-being that we have observed in the earlier analysis is that these reflect ‘gross differences’ between Hill People and other populations. These differences then result from a wider range of drivers that include a sub-set of factors that are a net ‘ethnic penalty’ and that cannot be explained by other differences in characteristics with the majority population. Such an approach would seek to ensure that Hill People children’s enrolment in secondary education, risk of stunting and access to improved sanitation and safe water etc. were equal to other groups when underlying characteristics were the same (such as location, education of head of household etc.). This approach is illustrated by looking at simple consistent comparisons that take into account two major factors that can account for significant levels of disparities, namely education levels and location. Table 2 repeats the analysis previously shown in Figure 2 but this time shows the rates for all children who have no access to improved sanitation in Table 2: Children in households with unimproved sanitation in Viet Nam Viet Nam, alongside the rates for children who both live in rural areas and have mothers with no educational qualification. The table shows that the gross differences between Hill People children and the ethnic majority are based on higher overall values (an overall more deprived population) but the net differences between Hill People and the majority narrow considerably, from a 54 percentage point difference (87% compared to 33%) to a 47 percentage point difference (91% compared to 44%). At the same time this example diminishes the perceived unfairness of targeting assistance on particular ethnically identified groups that can be felt by similarly deprived people who are not in the ethnic priority group, the Hill People group are still considerably worse off even when we account for location and education of mother. However, the advantages of such target setting for accurate parameters of disparities are offset by data complexity and uncertainty to properly control for all the ‘non-ethnicity-related’ factors of potential importance. Formulating ethnic based equity targets potentially relies on complex assessments of household characteristics which may require collecting more information from potential programme recipients. =Such an approach enables a clearer understanding of how equity issues are particular to ethnic minority groups or are shared across other ethnicities – including deprived children of the majority ethnicity. It more clearly demonstrates the trade-offs 19 between identifying inequity in purely ethnic versus other associated terms that are predictive of low income or poverty. This could be crucially important when, UNICEF and other child rights partners, are deciding how to re-orientate programming towards ‘recognizable’ sub-categories based on distinct group characteristics rather than on underlying factors such as illiteracy, low income or geographic location. 6.2 Equity Target Option 2: Achieving Fair Outcomes A second and alternative approach to promoting equity, based on ethnicity, is to design interventions aimed at achieving ‘fair’ outcomes. These outcomes would be expressed in simple aggregate ‘gross’ terms in which an equitable end-point would be where deprivations and violations were equally spread across the population and there are several alternatives in this regard. a. Basing fairness on ‘average’ (mean) levels of incidence. This approach would set targets that identify the difference between Hill People and the average position. This is the approach taken, for example, by the ‘inequality of opportunity’ indices that have been developed in Latin America and elsewhere to assess how far children have differential access to services and social and economic rights38. For such an approach the target is simple: to close the difference between the current inequitable level of provision and the average. For example, for water and sanitation provision (previously shown in Figures 1 and 2 above), it would imply that the provision of improved sanitation for Hill People would have to grow by 9% in Thailand, 83% in Viet Nam and 18% in Lao PDR to bring coverage up to average levels, and the provision of improved water by 50% in Thailand and by 21% in Viet Nam, but with no remedial growth needed in Lao PDR as Hill People are at or over the average position39. However, outcomes based on averages do need to be based on a clear understanding of the overall population. For example, would sub-national differences have to be brought to a national average as part of an overall approach to reducing equity to the average position, or could ethnic disparities be reduced to reflect sub-national averages? In the case of Hill People there is the cross-national issue in addition. Considering equity issues across countries for a cross-national community would have to similarly consider the role of overall population composition and sizes of ethnic communities that partly determine national averages. Our analysis clearly showed the differences between a large sub-group and a small sub-group and their respective national averages (such as when comparing Hill People disparities in Thailand and Lao PDR respectively). Given that Hill People are a cross-national ethnic minority faced with similar 38 Barros R., Ferreira F., Vega J and Chanduvi J (2009) Measuring Inequality of Opportunities in Latin America and the Caribbean, Washington DC: The World Bank. 39 These are ‘growth’ targets, with the difference expressed as a percentage of the average position; absolute numbers of sanitary installations etc would be the underlying numerical difference between hill people households and the average at the household level 20 inequities, setting targets that ensure comparison across differently apportioned populations and framing targets that take into account differing population sizes may be preferable. This leads us to suggest a second potential approach to ‘fairness’. b. Equalizing shares of deprivation to match population share. Table 3 shows the comparisons of overall population share (the % of the child population that are in Hill People communities) compared to the deprivation share (the % of the Hill People children that are deprived) in both water and sanitation provision. This exemplifies the ‘population share’ approach to equity target setting40. In an equitable distribution of deprivation, any group would experience a share of overall deprivation that was equal to its population share. However, table 3 illustrates inequity in the over-representation of Hill People children in water and sanitation deprivation. Equity targets using this approach would seek to reduce any over-representation, and thus reduce the ratio score to 1 or less. This means, for example, that Hill People children in Thailand would have to have their share of sanitation deprivation lowered by a factor of 11.8, in Viet Nam by 2.2 and in Lao PDR less appears to be needed to equal population shares. Table 3: Hill People deprivation and population shares for water and sanitation These population share based figures show the relative differences, between Thailand with very high overall rates of improved water and sanitation in which Hill People have a very large relative gap, and say Lao PDR that has overall lower provision of improved water and sanitation but where Hill People are at a smaller relative disadvantage. We specifically recommend that UNICEF and other partners consider using a population share approach when considering equity refocusing, and suggest that this approach is particularly suitable for cross-national profiles of equity. Calculating population shares is relatively simple and can be captured regularly in rounds of MICS, DHS or other household surveys to establish progress. Of course there are other approaches to targeting and considering equitable outcomes. Some approaches aim for complete equality and narrowing the gap completely between the best and worst off in some elements of deprivation, especially those that have fundamental and long 40 By solely considering water and sanitation provision as an exemplar, allows us to use the whole population of children aged 0-17 rather than smaller age-groups (thus we avoid the problem of small and missing numbers for some of the other deprivation counts that we saw in our earlier analysis). 21 lasting threats to child welfare and development, such as nutrition. If so, then using the average as an end-point is inappropriate, and perhaps an intermediary target based on the average is an entirely valid point on the way to full equality. If zero incidence (or close to it) of child deprivations is also a long-term aim, then equalizing shares is also a valid short to medium term objective. Such short to medium term objectives and targets can be well suited to government expenditure cycles. 7. Concluding remarks At the outset of this paper we put forward three empirical questions on how to capture ethnic group characteristics, how to profile disparities and inequity and how to set targets for ethnic equity, using Hill People communities as an exemplar ethnic group across three Southeast Asian countries. Our profiling and analysis showed how difficult it is to accurately and consistently identify ethnic groups, particularly those who live across national boundaries where the need for consistent definitions is perhaps greatest. Nonetheless, our analysis also showed that accurate ethnic profiling is doable when detailed ethnic markers are included in household surveys and sample sizes large enough to draw significant conclusions, such as in the Viet Nam MICS. Having identified Hill People as best we could under the constraints of dataset definitions, it was then possible to consider how such a ‘finer grained’ analysis of ethnic equity, compared to the practice of comparing ‘ethnic minorities’ as an aggregate to the ethnic majority population. We found that children in Hill People communities were always seriously disadvantaged when compared to the ethnic majority across Viet Nam, Lao PDR and Thailand. We also found that they were the ‘the most disadvantaged’ in comparison with all other ethnic communities in Viet Nam and Thailand. The comparison in Lao PDR was not as clear due in part to the uncertainty in ethnic identifiers. Recommendations As stated earlier, the audience for this paper is primarily those who need to identify equity issues for ethnic minority/indigenous children in order assess priority areas for programme design and resource allocation. We have demonstrated how to use more detail in locally available data to identify the most disadvantaged. The paper can be considered a resource for country level Situation Analysis of Women and Children, commonly undertaken by UNICEF, as well as other UN sister organizations and national governments. Some of the specific recommendations emerging from this paper are for these partners to promote: Use of disaggregated data: Data collection disaggregated by ethnicity remains an issue. However, we have shown that MICS has real potential in providing extensive profiling not only of ethnic minorities in general but specific groups of ethnic minorities. Data disaggregated by ethnicity is imperative, without good data it is impossible for countries to know if the specific 22 measures and policies aimed at certain ethnic minorities are actually improving their situation. In countries such as Thailand where the MICS data does not allow to profile robustly certain ethnic minorities, the emphasis should be on how to gather data on these minority populations who historically have been among the most disadvantaged. UNICEF Thailand country office is already considering this, with the implementation of an ‘Equity MICS’ that will emphasize sampling frames and questionnaire instruments specially designed for in-depth analysis of minority populations. Clear transparent approaches to data analysis: Our analysis faced the problem, often found when considering small sub-groups of the population, of small numbers. We were able to plan for challenges regarding small sub-groups and meet them with a clear methodological approach that invested in transparency. Given that empirical verification of equity issues for sub-groups is a core issue for any equity refocus, we adopted clear statistical approaches to ensuring that out analysis was robust – by reporting ‘t-tests’ and 95 percent confidence intervals to show where statistical significance was assured and that differences between ethnic groups was ‘statistically different’. This is good practice and should be adopted in all such future work. Interpretation of results always requires care. There may be, for instance, group differences based on religion or other reason that are not captured by ‘ethnic’ definitions. Analysis of such other potential drivers of difference should be undertaken before any final determination of the profile of disadvantage is made. Consider cross-national equity issues: Our approach can be adopted in any national dataset to profile ethnic disparities. However, additionally, the cross-country ethnic minority analysis has the added potential to be adopted in other countries/regions to shed light on children living marginalized ethnic minorities, and especially indigenous communities, that reside across several countries, such as Roma populations in Europe, the Maasai in Kenya and Tanzania, the Quechua in South America, and so forth. Such analysis is of relevance to UNICEF’s equity focus, which emphasizes reviewing the equity focus of any child situation analysis. However, it is worth considering whether cross-national equity be a concern at regional rather than country level? The approach of this working paper was to analyse child well-being of certain groups across national boundaries, bringing into question the effectiveness and relevance of UNICEF’s current approach to profiling children within national boundaries. Do we miss out on understanding equity to the fullest when we limit our analysis within national boundaries? How do we ensure better targeted UNICEF programmes, which address inequities by serving these disadvantaged populations across national boundaries? The Roma analogy may be useful in that regard, as the Roma community have similarly been the subject of cross-national concern in Europe. In December 2011 under the new European Framework for National Roma Integration Strategies, the 27 member states of the EU were required to submit to the European Commission, National Action Plans (NAPs) on Roma inclusion. A similar action plan addressing the disadvantage Hill People children face in South Asia could be considered as a way forward. 23 In Laos, Thailand and Viet Nam there are already in place, to a varying degree, social policies and programmes that are meant to level the playing field, so that circumstances such as which ethnic group you are born into do not influence a child´s life chances. However, without additional policy action, Hill People children will have little chance to even out the inequality lived by their parents. As long as Hill People ethnic minority children do not have access to, or do not utilize, basic services that are considered critical for future advancement in life, inequality of opportunity will persist. 24