90523 National Statistics Bureau The World Bank Royal Government of Bhutan Bhutan Poverty Assessment 2014 Copyright © National Statistics Bureau, 2014 www.nsb.gov.bt ISBN 979-99936-28-26-2 Design by Loday Natshog Communications (tashdezyn@gmail.com) Contents Acknowledgements iv Foreword v Foreword vi Abbreviations and Acronyms vii Executive Summary viii CHAPTER 1: Introduction 01 CHAPTER 2: Evolution of Poverty, Shared Prosperity and Inequality in Bhutan 05 2.1.  Consumption Poverty, Multidimensional Poverty and Happiness 05 2.1.1.  Decline in Multidimensional poverty between 2007 and 2012 07 2.1.2.  Shared Prosperity 09 2.1.3.  Mobility in and out of Poverty between 2007 and 2012 09 2.1.4.  Growth in Bhutan has been Pro-Poor 10 2.2.  Stable Inequality 12 2.2.1.  Uneven Poverty Reduction across Dzongkhags 13 CHAPTER 3: Changing Profiles of the Poor and Bottom 40 Percent of the Population 17 3.1.  Welfare Indicators (Assets and Amenities) 17 3.2.  Health and Nutrition 18 3.3.  Gender and Poverty 22 3.3.1.  Is Poverty in Bhutan Gender-Blind? 22 3.4.  Land Ownership and Poverty 23 CHAPTER 4: Enlarging Opportunities for Children 27 4.1.  Inequality of Opportunity in Bhutan 27 4.2.  Social Outcomes for Children in Relation to Birth Circumstances 29 4.3.  Measuring Inequality of Opportunity 35 4.4.  Drivers of Change 42 CHAPTER 5: Key Drivers of Poverty Reduction in Bhutan 45 5.1.  Trading Out of Poverty 47 5.2.  Roads Out of Poverty 50 5.3.  The Hydro Effect 53 5.4.  Who were Better Able to Escape Poverty between 2007 and 2012? 53 5.5.  The Main Drivers of Poverty Reduction from People’s Perspective 54 5.6.  Better Returns on Individual’s Assets Underpin Faster Reduction in Poverty 55 5.7.  Composition versus Structure 56 CHAPTER 6: Poverty Reduction in Bhutan: Sustainability, Vulnerability and Suggested Remediation 61 Annex A: Sources of Variation in Poverty Outcomes in Bhutan 68 Annex B: Poverty Dynamics with Synthetic Panels – Framework and Results 88 Annex C: Qualitative Assessment of Poverty 101 Bhutan Poverty Assessment 2014 iii Acknowledgements This report is the first poverty assessment for Tenzin Thinley (Dzongdag), Lhawang Dorji and Bhutan prepared by the World Bank jointly with Thinley Wangchuk (Gups) in Dagana, Karma the Royal Government of Bhutan through the Drukpa (Dzongdag), Tashi Rabten (GAO), National Statistics Bureau (NSB). It builds on Rinchen Lungten (Gup) in Zhemgang, Karma Bhutan Poverty Analysis 2012, published by the Wangdi (Dzongrab), Lepo and Chedup (Gups), NSB with technical assistance from the World Dawa Zangmo (GAO) in Pema Gatshel, Sonam Bank. Wangyel (Dzongdag), Karma and Tshering The study is led by Srinivasan Thirumalai Samdrup (Gups) and Kinley Phuntsho (GAO) (Senior Economist, SASEP) with team members in Lhuentse. Focus group participants shared drawn from the World Bank, NSB, and outside candid views and debated priority issues with consultants. The Bank team comprised of Peter endearing keenness. Lanjouw and Hai-Anh Dang (DECRG), Namgyel The team benefited from the comments of Wangchuk (SASEP), Minh Cong Nguyen peer reviewers Sabina Alkire, Vikram Nehru, (ECSP3), and Smriti Seth (DECWB). From the Dean Jolliffe, and Sonam Tobgyel at concept NSB’s Socio-Economic Analysis and Research stage and also at the final stage. We are Division, Lham Dorji (Dy. Chief Research grateful for comments by David Newhouse and Officer), Dorji Lethro (Senior Statistical Genevieve Boyreau for comments on an initial Officer), Sonam Gyeltshen  (Research Officer), draft. We are thankful to the participants at and Cheku Dorji (Dy. Chief Statistical Officer, the final review meeting held in Washington Coordination and Information Division) D.C and a consultative group meeting held in worked closely with the World Bank team. Thimphu to discuss the findings of the report. Consultants who contributed to the study are Overall guidance for the study was provided Essama-Nssah, Nar Bahadur Chhetri, Krishna by Ernesto May, Director (SASEP), Kuenga Parajuli and Sanjana Dulal. Tshering, Director General (NSB), Vinaya We give special thanks to local officials and Swaroop (SASEP), Robert Saum (Country elected representatives (Gups) who facilitated Director) and Genevieve Boyreau (Resident focus group discussions and shared their Representative). views on development issues at the local level. In this regard, we wish to especially thank iv Bhutan Poverty Assessment 2014 Foreword The National Statistics Bureau is pleased to One of the factors contributing to poverty present the Bhutan Poverty Assessment 2014 reductions is due to the noble Royal Kidu Program report prepared in collaboration with the World where many landless households were able to Bank. This report is a complement to the earlier get land permanently registered in their names. Poverty Analysis Report (PAR) 2012 which The findings from the participatory assessments was prepared with the World Bank’s Technical listed small land holdings and landlessness as key Support. The PAR 2012 provides estimates constraints to achieving economies of scale in of consumption poverty, identified the trend agricultural production. and narrates the profile of the poor in terms This assessment report also shows that, of demographics and basic needs. Most of the among others, Bhutan’s poverty reduction has poverty reduction in Bhutan has occurred in the been rapid, broad-based and inclusive; in the rural areas with little change in urban poverty long-term, sustainable poverty reduction depends rates. Inequality has not changed significantly. on addressing persistent shocks, engendering Poverty reduction in dzongkhags have been found private sector led development and defining clear to be uneven. target groups for poverty reduction. The main This report identifies the key drivers of rapid drivers of prosperity in rural Bhutan appear to poverty reduction in Bhutan over the recent be increasing commercialization of agriculture, years, explaining why some dzongkhags are stuck an expanding rural road network and beneficial in poverty or reducing poverty is not significant spillovers from hydroelectric projects. while others prospered, and whether female I hope that this report becomes a headed households have a harder time reducing comprehensive source of information towards poverty. The exercise draws mainly on data from further reduction of poverty especially in sections the two rounds of Bhutan Living Standards and areas where the poverty still remains high. Survey (2007 and 2012) supplemented with focus Finally, I wish to sincerely thank the World group discussions carried out for the report in Bank for their continued support and would like select dzongkhags. to acknowledge the efforts of all officials and The report presents a more detailed analysis experts who were involved in this important of the evolution of poverty, its distributional exercise. characteristics including inequality, mobility estimates; changing profiles of the poor and bottom 40 percent of the population; issues in expanding opportunities for children. This report probes the vulnerabilities in spreading prosperity (Kuenga Tshering) in Bhutan and discusses the steps to be taken for Director General sustained poverty reduction in Bhutan. Bhutan Poverty Assessment 2014 v Foreword This report presents the first Poverty in our "Country Partnership Strategy 2014- Assessment carried out in Bhutan by the World 2019". The identified drivers of poverty reduction Bank in close collaboration with the National and recommendations have translated into : (i) Statistics Bureau, Bhutan. a focus on agriculture commercialization and Bhutan well known as a pioneer of the Gross marketization, and more broadly the sustainable National Happiness concept has a noteworthy contribution of green assets to socio-economic record in reducing poverty as well. Poverty development, (ii) supporting a social protection reduction in Bhutan, as the report finds, has been strategy, with targeted safety nets build rapid, broad-based and inclusive. Prosperity has household's resilience, (iii) a continued focus on been shared well in Bhutan with the bottom 40 the private sector development to create jobs percent enjoying faster growth than the rest. which improve living standards and are also a There are potentially useful lessons for other critical element of social cohesion, (iv) a renewed countries aspiring for poverty reduction with attention to transport and trade infrastructure, shared prosperity. recognizing its critical role in reducing poverty; Poverty reduction in Bhutan is well-founded (v) improving fiscal and spending efficiency in long-term economic development efforts of to enable the Royal Government to continue commercialization of agriculture, expanding improving the delivery of public services for the rural road networks and beneficial spillovers benefit of all. from hydroelectric projects. A good governance We - at the World Bank - are committed to infrastructure underpins the successes on support shared prosperity and the fight against poverty front. The pace of poverty reduction poverty throughout the world, and in Bhutan in appears sustainable if the emerging risks and particular, where we look forward to building on vulnerabilities are managed carefully. a strong partnership with the Royal Government Sustaining the examplary record of Bhutan's and all stakeholders. poverty reduction in the long-term would require mitigating risks from persistent shocks facing the agricultural sector, increasing reliance on private sector led development and building formal social protection for clearly identified population Genevieve Boyreau groups most vulnerable to poverty. Resident Representative The findings of the report have directly World Bank, Bhutan influenced our engagement in Bhutan as reflected vi Bhutan Poverty Assessment 2014 GOVERNMENT FISCAL YEAR July 1 – June 30 CURRENCY EQUIVALENTS Currency Unit = Bhutanese Ngultrum (Nu) US$1 = Nu 61.8 (March 25, 2014) Abbreviations and Acronyms BIMSTEC Bengal Initiative for Multisectoral MDG Millennium Development Goal Technical and Economic Cooperation MIC Middle Income Country BLSS Bhutan Living Standards Survey MoAF Ministry of Agriculture and Forest CBS Centre for Bhutan Studies MPI Multidimensional Poverty Index CPIA Country Policy and Institutional NPL National Poverty Line Assessment NSB National Statistics Bureau DDS Dietary Diversity Score OGTP One Gewog Three Products FANTA Food and Nutrition Technical PAR Poverty Assessment Report Assistance RGoB Royal Government of Bhutan FAO Food and Agricultural Organization RNR Renewable Natural Resources FGT Foster-Greer-Thorbecke SAARC South Asian Association for Regional FYP Five Year Plan Cooperation GAO Gewog Administrative Officer SAFTA South Asia Free Trade Agreement GDP Gross Domestic Product SAR South Asia Region GIC Growth Incidence Curve SL Standard of Living GNH Gross National Happiness TIP Three “I”s (incidence, intensity and HDDS Household Dietary Diversity Score inequality) of Poverty HIV Human Immunodeficiency Virus HOI Human Opportunity Index Bhutan Poverty Assessment 2014 vii Executive Summary Bhutan’s poverty reduction has been rapid, of the population enjoying faster growth than broad-based, and inclusive. Between 2007 and the rest, save for the top 10 percent (Figure 0.2). 2012, the percentage of consumption poor halved Inequality remained stable, allowing the full to 12 percent. Bhutan has nearly ended extreme effect of growth on poverty reduction. poverty1 within the living memory of a generation Yet some have stayed poor and some non- – extreme poverty touched a low of two percent poor fell into poverty. The rapid reduction in in 2012 (Figure 0.1). Broader multidimensional poverty bypassed nearly half of those found poverty indices, that include education and to be poor in 2007. Further, notwithstanding health outcomes besides standards of living, the cherished community support, families do also indicate a steep decline in the percentage of fall through cracks: for every two families that deprived population –by two-thirds, from about managed to escape poverty, one previously 25 percent to 12.7 percent. Growth in Bhutan non-poor family fell into poverty. Though Figure 0.1  Fast-Paced Poverty Reduction in Bhutan by Figure 0.2  Growth in Annual per-capita mean any Measure, 2003-2012 Consumption, by Cumulative Decile Groups 70 5.6 60 5.4 50 Annual growth in percent Percentage of Poor 5.2 40 30 5.0 20 4.8 10 4.6 0 2003 2007 2012 10 20 30 40 50 60 70 80 90 100 Per-capita consumption cumulative decile groups PPP US$2.50 National Poverty Line PPP US$1.25 Source: World Bank staff estimates based on Bhutan Living Standards Source: Poverty Analysis Reports from National Statistics Bureau, Surveys 2007 and 2012, applying nominal poverty line as deflator. Bhutan; PovcalNet: the online tool for poverty measurement developed by the Development Research Group of the World Bank: http://iresearch.worldbank.org/PovcalNet/index.htm?0 mobility of the poor in Bhutan is one of the better international examples, there is room for has been pro-poor in a substantive way –not only reducing vulnerability of the poor and near-poor. has the headcount poverty rate declined, but the The risk of falling back into poverty is greatest for poverty gap also declined across all the poverty Bhutanese in rural areas, those holding informal bands. Prosperity has been widely shared among jobs, with low education, and resident especially all income classes, with the bottom 40 percent in Pema Gatshel, Trashigang, or Dagana. Food security improved in terms of access, 1  Based on a consumption poverty line of US$1.25 per capita per but the poor still lag behind. On average, day in purchasing power parity terms. viii Bhutan Poverty Assessment 2014 Bhutanese increased dietary diversity by attendance of 84 percent. Higher completion consuming from 10 food groups (out of 12) in rates alone would help to build comparative 2012, from just seven groups in 2007. Protein advantage for Bhutanese youth in skilled labour. sources, especially, have increased to include meat, The main drivers of prosperity in rural Bhutan fish, and pulses. However, the poor lag behind appear to be increasing commercialization of in access to diversified food groups, particularly agriculture, an expanding rural road network, protein sources. In addition, inadequate food is and beneficial spillovers from hydroelectric reported by 10 percent of poor households – more projects. Helped by the recently renewed free trade than double the non-poor’s four percent. agreements with India and preferential market Female-headed households are not access to Bangladesh, Bhutanese agricultural on par with male-headed households in exports of commercial crops (notably oranges, enjoying fruits of growth. Thirty percent of cardamom, potatoes, and apples) have increased Bhutanese households are headed by females. sharply (Figure 0.3). Increasing trade has been While the poverty incidence (consumption or pro-poor. The eight-fold expansion in farm roads multidimensional poverty rate) is found to and progressive construction of highways linking be equal for both male- and female-headed with the Southern East-West highway, that runs households, some female groups (notably married along the Indian border, and new north-south and divorced) have a greater incidence of poverty links have all helped to create construction jobs than the corresponding male groups; the bottom and lowered the travel time and costs for goods 40 percent of female-headed households enjoyed and people. The four hydroelectric projects that a smaller rise in consumption compared to that of began construction in the last five years (adding their male-headed counterparts. The persistence 3 GW to the current 1.6 GW generation capacity) of the livelihood handicap for female-headed are spreading good spillovers by expansion in households, despite matrilineal inheritance and roads, jobs, and business in the project areas. a non-discriminatory labour market, suggests Individuals in lower economic deciles have reaped that disproportionate household burdens may be better rewards for their education and land. diminishing opportunities for women. Land gift under the Royal kidu program has also Opportunities for children are equalizing regardless of birth circumstances but inequities Figure 0.3  Rising Agricultural Exports from Bhutan 1,800 9.0 in completion of secondary education persist. 1,600 8.0 Bhutanese children have better and improving 1,400 7.0 opportunities in education and infrastructure 1,200 6.0 Million Nu Percent services than those of other South Asian 1,000 5.0 800 4.0 countries, and these opportunities are becoming 600 3.0 more equal across income classes. The public 400 2.0 policy of extending coverage for all and targeting 200 1.0 interventions with electricity and gas provision 0 0.0 2007 2008 2009 2010 2011 2012 have narrowed inequalities among children. Nevertheless, it is important to note that inequity Share of non-energy exports in completion of secondary education remains an Agricultural exports issue–the inequality-adjusted completion rate Source: Data from Annual Report 2012/13 of the Royal Monetary was only 32 percent in 2012, with an adjusted Authority of Bhutan Bhutan Poverty Assessment 2014 ix helped the previously landless to escape poverty. working age adults, labour-intensive horticulture Education appears to be the most important will become increasingly difficult. Contract route by far to escape poverty. farming by large-scale land owners may be a way The current pace of poverty reduction to sustain exports but benefits to poorer farmers appears sustainable in the medium term. might diminish. The current problems faced by Trade intensification with neighbors is set to farmers such as the incurable “greening disease” continue, road infrastructure is posed for more of oranges, diseases of the cardamom plants and expansion, and more hydroelectric project regular raids into farms by elephants (in low land), construction is planned to continue to 2020; the monkeys, and wild boars have persisted. The current free trade agreement with India, due for plan for introducing disease-resistant cultivars renewal in 2016, is most likely to be renewed. is not proceeding swiftly. It takes years to bring The bilateral agreement with Bangladesh, that horticultural crops to harvest and equally long has benefited Bhutan by preferential duty-free to shift to other profitable forms of production. access to 74 mostly agricultural exports, is also As a consequence of increasing commercial crop due for renewal, in 2018. In addition bilateral production, Bhutan dependence on food imports agreements with Thailand and Nepal are also on has been rising over the years, making it more the anvil. Bhutan is a net exporter of fruits and vulnerable to food price shocks. A 12 percent cardamom in the north-east region of the Indian increase in food prices – the average annual sub-continent and should be able to sustain increase in recent years - for example, can increase fruit exports to Bangladesh even with future the percentage of poor in the short-term by about preference erosion under the South Asian Free two percent points. With all petroleum products Trade Agreement (SAFTA). Growth dynamism in imported, Bhutan’s poor also face risk from fuel India and Bangladesh should be able to further price shocks. A sharp rise in the consumer prices accommodate expansion of differentiated of LPG and kerosene of the order that occurred agricultural exports from Bhutan which are well in July 2013 (quickly reversed, however) had the known for superior quality. Completion of the potential to push 0.5 percent of population into Southern East-West highway in Bhutan and poverty. Bhutan’s social protection is mainly expansion of the rural roads network would through the Royal Kidu welfare program. Risks of help to draw out the comparative advantages of downward mobility are greater than average for Bhutanese agriculture. The hydroelectric projects rural residents, male-headed households, people now under construction are expected to continue in informal jobs (the casually and self-employed), to 2016/17, and more projects are planned that and those with low education and particularly would continue to boost rural incomes indirectly. high for those living in select dzongkhags such as Despite rapid urbanization – one percent of Pema Gatshel, Dagana, Samtse, Trashigang, and rural population moving every year to urban Tsirang). areas – urban poverty has remained under two Formal social protection programs may percent, indicating that migration is not biased be necessary to help individuals cope with particularly to the poorer sections of the society. adverse economic and financial shocks. At For sustained poverty reduction, risks and present, individuals cope with shocks mostly by vulnerabilities need to be managed carefully. drawing on own savings if they are non-poor, With limited land, increasing fragmentation of or by borrowing from friends, suppliers, and land holdings, and rural-to-urban migration of money-lenders if they are poor. Because of the x Bhutan Poverty Assessment 2014 inadequacy and inelasticity of these sources for the poor and vulnerable segments of the “In my opinion over the years population, we suggest the introduction of formal the community has benefitted social protection mechanisms and possibly well- because we now have access targeted micro-credit programs. to road, electricity and mobile In the long-term, sustainable poverty services. Electricity has brought reduction depends on addressing persistent many benefits – we do not have shocks, engendering private sector led to spend time fetching firewood development and defining clear target groups for cooking, household sanitation for poverty reduction. The feasibility of crop insurance for farmers may be examined to protect has improved as we use electric the harvests from perils of diseases. Other perils, utensils to prepare meals. Mobile such as those associated with wild-life predation, connectivity has made our life easier have also persisted and evaded viable solutions. due to faster communication.” – An What poor people want to better their living FGD participant from Lhuentse standards in the long term can be summed up as dzongkhag. access to roads, electricity, public transportation, irrigation, land and higher education. Sustained poverty reduction depends on job opportunities hydropower sector by Private Public Partnerships and wage earnings of the poor. The development and subcontracting in order to create jobs. paradigm for a renewable resource rich country The Royal Government of Bhutan seems to like Bhutan would call for engendering private favor complementary use of consumption and sector led growth actively enabled by the public multidimensional poverty. But the overlap of sector. Successful agribusiness – an emerging the two approaches identifying the poor is small. sector in Bhutan - will require development of Therefore defining a clear target group for poverty value chain system (from farm to market) that reduction is important. Also, with success in will identify and remove the bottlenecks that reducing extreme consumption poverty rapidly, farmers encounter including constraints related the goal could be now shift to shared prosperity to finance and availability of crop insurance. The defined for example as the welfare of the bottom government could engender private investment in 40 percent of the population. Bhutan Poverty Assessment 2014 xi 1 Chapter Introduction Bhutan’s location provides opportunities and governance. The country transitioned from an challenges. Land-locked in the eastern Himalayas, absolute monarchy to a constitutional monarchy Bhutan is bordered by two Asian giants, China with a multi-party democracy in 2008. The second and India. Its population density, at 19 persons democratic process has evolved, with multiple per sq. km, is the lowest in South Asia. Elevation political parties participating in the July 2013 ranges from 200 meters in the southern foothills parliamentary elections for the lower house. Its to some peaks in the north that are around 7,000 Country Policy and Institutional Assessment meters above sea level. Bhutan’s picturesque (CPIA) rates fairly well and so do international topography consists of tall mountains, thick measures of governance and corruption. With forests and tumultuous rivers. In keeping with its regard to corruption perception, Bhutan ranks philosophy of sustainable development, Bhutan’s 31st among 177 countries and scores better than constitution requires that 60 percent of its land Israel, Spain, and Poland. Domestic perception area be covered by forests (around 72 percent of corruption is on the decline, as reflected in the of the land was under forest cover in 2011). increase of the Bhutan Comprehensive integrity Renewable fresh water availability, at 106,933 score in recent years. cubic meters per capita, is the third-highest in Bhutan is on its way towards Middle Income the world. Challenges include difficult terrain Country (MIC) status, has a unique poverty (tall mountains with sheer drops) that makes reduction record in international context. Its GDP connecting remote areas difficult and expensive, per capita is already US$ 2,584 in 2012 and it is and a small and dispersed population that limits poised for eight percent growth over the coming economies of scale. Moreover, Bhutan is located five years. It has done well on poverty alleviation on the Indian and Eurasian tectonic plates, and providing service to citizens. whose movements and collisions cause frequent Bhutan has made stellar progress in meeting earthquakes in the area. It also experiences other MDGs and extending gains beyond GDP growth. disasters such as landslides, forest fires, and Of the eight MDGs, seven are actionable to glacial lake outburst floods. national policies. Among these seven, Bhutan Bhutan has enjoyed continued political has already achieved or over-achieved four goals stability, strong institutions and good in halving extreme poverty, reaching gender Bhutan Poverty Assessment 2014 01 parity in education, ensuring environmental Recognizing the fact that political democracy sustainability, and reducing by three-fourths and economic empowerment do reinforce maternal mortality. Notably, in many of these, each other, the government has made poverty Bhutan’s initial conditions were worse than its reduction the central theme and main objective neighbors’ but had surpassed the neighbors’ of the 10th Five Year Plan. It intends to pursue by 2011. For instance, maternal mortality in this objective through industrial development, 1990 was at 1,000 per 100,000 – much worse national spatial planning, and integrated rural- than India’s 600, but by 2011 it had fallen urban development, a strategic expansion of to 180 while India’s was still around 200. In infrastructure, human capital development, the remaining three goals –universal primary and enhancing the enabling environment. The enrollment, halting and reversing the spread of formulation of the 10th Five Year Plan builds communicable diseases, and reducing by one- on the strong achievements of the Ninth Plan third infant and under-5 mortality – Bhutan is (2002-2007) which sought to improve the on track. Some areas still requiring attention quality of life and income, with a special focus under the MDGs are gender parity in tertiary on the poor, by promoting good governance education, detection of HIV cases, and youth and private sector-driven economic growth in unemployment. Significant progress has been addition to preserving cultural heritage and the made towards gender equality: in education, natural environment. female enrollment in primary schools stood at The purpose of this report is to provide an 88 percent in 2008 compared to one girl enrolled account of the poverty outcomes observed under for every 50 boys in 1970; maternal mortality the 10th Five Year Plan. This account is based has dropped dramatically (as noted above, to 180 on data from the 2007 and 2012 rounds of the from 1,000 in 1990), and women are almost at Bhutan Living Standards Survey (BLSS). Other parity in the labour force. quantitative data came from Labour Force Surveys The Royal Government of Bhutan (RGoB) and Renewable Natural Resource (RNR) statistics. has made the pursuit of national happiness the This is supplemented by qualitative report from a overarching goal of its development strategy. series of focus group discussions (FGD) held for In that context, it is committed to improving the study in four dzongkhags of Bhutan. Given the quality of life for the citizens through the period of this plan, the 2007 data provide a inclusive and sustainable economic growth, the valid baseline for an assessment of the poverty conservation of the natural environment, the outcomes of this plan. Similarly, the 2012 data preservation of the country’s cultural heritage are considered end-line observations reflecting and good governance. These focal areas constitute the outcome of the implementation of the 10th the four pillars spanning the concept of Gross Five Year Plan since the plan ends in 2013. National Happiness (GNH), and are being This report builds on Bhutan Poverty Analysis, implemented through a series of five year plans. 2012–earlier collaborative work between the NSB The vision underlying this strategic framework and the World Bank. While the previous report has been enshrined in the 2008 Constitution,2 presented new estimates of consumption-based adopted at the beginning of the 10th Five Year poverty and characteristics of the poor in 2012, Plan (2008-2013). the current report offers a more detailed analysis of the evolution of poverty, its distributional 2  This Constitution marks a transition in the system of government characteristics including inequality, mobility from absolute monarchy to a parliamentary democracy. 02 Bhutan Poverty Assessment 2014 estimates (Chapter 2), changing profiles of the poor and bottom 40 percent of the population (Chapter 3), issues in expanding opportunities for children – Human Opportunity Indices (Chapter 4), identifies key drivers of poverty reduction in Bhutan (Chapter 5) and examines what are the vulnerabilities and steps to be taken for sustained poverty reduction in Bhutan (Chapter 6). Bhutan Poverty Assessment 2014 03 2 Chapter Evolution of Poverty, Shared Prosperity and Inequality in Bhutan 2.1.  Consumption Poverty, Multidimen- amounts to ending extreme poverty in 22 years, sional Poverty and Happiness or within the living memory of a generation. In just over five years, 2007-2012, poverty in Bhutan Ending extreme poverty in a generation is within was cut by half, according to the National Poverty reach for Bhutan. By any poverty measure – be it Line, a more generous measure than the US$1.25 the National Poverty Line or the PPP US$1.25 or per day line. Judging by comparable surveys and US$2.50 –Bhutan has achieved rapid reduction in methodology, the percentage of poor was cut poverty in the last decade (Figure 2.1). By measure from 23 percent in 2007 to 12 percent in 2012. If of the international norm of US$1.25 per day for more distributionally sensitive measures are used extreme poverty, Bhutan had almost eliminated (poverty gap, poverty severity), the reduction is poverty to under two percent by 2012. That even greater. Happiness is more than consumption: Figure 2.1  Fast-Paced Poverty Reduction in Bhutan by Bhutan’s unique Gross National Happiness any Measure, 2003-2012 measure. The term “gross national happiness” 70 (GNH) was first formulated in 1972 to sig- 60 nal the country’s commitment to building an 50 economy that would serve Bhutan’s unique cul- Percent of Poor ture based on Buddhist spiritual values. The 40 Centre for Bhutan Studies developed a sophis- 30 ticated survey instrument to measure GNH. 20 Four pillars support the concept: Fair socio- 10 economic development (better education and 0 health), conservation and promotion of a vibrant 2003 2007 2012 culture, environmental protection, and good gov- PPP US$2.50 National Poverty Line PPP US$1.25 ernance. These four pillars are further elaborated Source: Poverty Analysis Reports from National Statistics Bureau, in nine equally important domains: psychologi- Bhutan; PovcalNet: the online tool for poverty measurement cal well-being, living standard, health, culture, developed by the Development Research Group of the World Bank: http://iresearch.worldbank.org/PovcalNet/index.htm?0 education, community vitality, good governance, Bhutan Poverty Assessment 2014 05 balanced time use, and ecological integration. In than its neighbors, but it had surpassed those accordance with these nine domains, Bhutan has neighbors by 2011. For instance, maternal developed 33 clusters and 124 variables that are mortality in 1990 was, at 1,000 per 100,000, used to define and analyze the happiness of the much worse than India’s 600, but by 2011 it Bhutanese people. The GNH concept serves as a fallen to 180 while India’s was still around 200. unifying vision for Bhutan’s five year planning For the remaining three goals –universal primary process and all the derived planning documents enrollment, halting and reversing the spread of that guide the economic and development plans communicable diseases, and reducing by one- of the country. Proposed policies in Bhutan must third infant and under-5 mortality – Bhutan pass a GNH review based on a GNH impact as- is on track. Some areas requiring attention sessment. under the MDGs are gender parity in tertiary Bhutan has made stellar progress in meeting education, detection of HIV cases, and youth the MDGs and extending gains beyond GDP unemployment. growth. Of the eight MDGs, seven are actionable Bhutan’s poverty reduction record is to national policies. Of these seven, Bhutan unique. Using the internationally comparable has already achieved or over-achieved four US$1.25 per day poverty line, Bhutan stands goals: halving extreme poverty, reaching gender out for the pace of its poverty reduction parity in education, ensuring environmental compared to other South Asian countries and sustainability, and reducing maternal mortality the select cohort of countries with similar initial by three-fourths. Notably, in many of these poverty levels in 1990. Starting from about the factors Bhutan’s initial conditions were worse same level as that of the South Asia region in Figure 2.2  Bhutan Outpaces South Asia Region in Figure 2.3  Bhutan Poverty Reduction Leads Countries Poverty Reduction with Similar 1990 Poverty Levels 60.0 70 60 50.0 50 40.0 40 30 30.0 20 20.0 10 0 10.0 1990 1999 2010 0.0 Lesotho India* Ethiopia 1990 1999 2010 LaoPDR Ghana Sudan Indonesia* Cambodia Pakistan Developing World South Asia Bhutan China* Bhutan Source: PovcalNet: the online tool for poverty measurement developed by the Development Research Group of the World Bank: http:// iresearch.worldbank.org/PovcalNet/index.htm?0 Note: In Figure 2.3, asterisks indicate estimates aggregated from rural and urban data. 06 Bhutan Poverty Assessment 2014 1990,and with more than half of its population in poverty, Bhutan had managed to reduce the “We see almost 50 percent percentage of poor to a mere four percent by development in the recent years. 2010,while the whole of South Asia’s poverty There is change in life. Before we had level had fallen to 30 percent (Figure 2.2). to walk for 10 days with our horses Among those countries in the developing world to bring essentials, sleep in the cave, that had poverty in the 50-60 percent range in walk barefoot.” 1990, Bhutan’s poverty reduction has been the steepest (Figure 2.3). livestock ownership). None of the households 2.1.1.  Decline in Multidimensional poverty were observed to be deprived in all SL indicators, between 2007 and 2012 but nearly 70 percent were deprived in at least The 10th Five-Year Plan adopted by the one, and more than 32 percent were deprived in Bhutanese government prioritizes poverty at least half of all the indicators. Fifty percent reduction in a multidimensional way. Given the of all the consumption-poor were observed to limitations of consumption poverty measures in be deprived in at least half of the SL indicators. capturing overall deprivation, the government The highest headcount ratio (36 percent) was also estimates a more holistic measure called observed in the use of solid cooking fuel, followed the multidimensional poverty index (MPI). The by no access to improved sanitation facilities (29 MPI, which is based on the concept of capability percent). Over 10 percent of the population was deprivation, uses the Alkire Foster methodology MPI-poor and used dung, wood, or charcoal for with three equally weighted dimensions – cooking. health, education, and standard of living – each Education deprivation was the highest of which is further split into two, two, and nine in all three dimensions, with 2.5 percent of sub-indicators, respectively. A household that the population deprived in both forms of the is deprived 4/13of the weighted indicators is education indicators (schooling of household MPI-poor. members and child attendance), and 27 percent In 2012, 12.7 percent of the country’s deprived in at least one. Further, 7 percent of population was MPI-poor –not different from the consumption-poor were deprived in both the 12 percent headcount ratio for consumption education indicators, while 37 percent was poverty but only 3.2 percent of the population deprived in at least one. Among the income- was both consumption and MPI-poor at the poor households, nearly 30 percent had no same time. This huge mismatch between the adult with at least five years of education, two measures illustrates the importance of two and 15 percent had school-aged children not measures. However, there was greater overlap attending school. in the standard of living (SL) one-seventh of the By comparison, less than one percent of the weight of this dimension is assigned to each of six total population was deprived in both health indicators: electricity, sanitation, water, housing indicators (food security and child mortality) material, cooking fuel and road access, and the and 15 percent was deprived in at least one. remaining one-seventh of the weight is equally These deprivations were deepest among the distributed among assets, land ownership and income-poor, where 23 percent of the population Bhutan Poverty Assessment 2014 07 Figure 2.4  Multidimensional Poverty and Consumption Poverty Compared, 2012 Multidimensional poverty Consumption poverty 40.0 35.0 30.0 25.0 20.0 15.0 10.0 5.0 0.0 a ng el se aa a g ro pu ng a g e ng ha g e ar a ar gs as an an an ts an ts kh sh Pa gg kh nt ga H im ira ra uk ng m on G ig ag na rp th at ue od ng on Ts Th hh em Sa h Ya m Sa G Tr D Pu Lh as Ph Jo M C Bu Zh a hi Tr m up ue as Pe dr Tr gd m an Sa W Source: Bhutan Multidimensional Poverty Index 2012, NSB was deprived in at least one indicator. Further, a Figure 2.5  Trends in Multidimensional and Consumption Poverty Headcount Ratios higher incidence of child mortality was observed 35 among income-poor households (at 15% of the population) compared to food shortage (9%). 30 While none of the consumption-poor households were deprived in all health and education 25 indicators, nearly 8 percent of them (i.e., 0.9% of Percent 20 the total population) were deprived in at least one health and one education indicator. 15 The differences between MPI and 10 consumption-poverty estimates were ampli- fied at the dzongkhag level, particularly in Gasa 5 (Figure 2.4). For instance, Gasa had the highest MPI headcount ratio (37.6%) but the lowest con- 0 sumption poverty ratio (0%), while Lhuentse had 2007 2012 the highest consumption headcount ratio (32%) % MPI poor % Consumption poor with a low MPI headcount (10%). Within the SL Source: Bhutan Multidimensional Poverty Index 2012, NSB dimension, over 60 percent of Gasa’s population Note: The MPI constructed for the comparison is not the same as was deprived in at least half of the indicators, that used for the national MPI because of indicator availability in BLSS 2007. compared to Pema Gatshel, where less than 46 percent of the population was deprived in at was also poorly related to consumption-poverty. least half of the SL indicators, but it had the Dzongkhags such as Wangdue Phodrang and second-highest headcount of consumption poor, Haa, which had the highest education poverty at 27 percent. Education and health deprivation headcounts (based on deprivation in both indi- 08 Bhutan Poverty Assessment 2014 cators), were also among the ten richest in terms Figure 2.6  Growth in Annual per-capita mean Consumption, by Cumulative Decile Groups of consumption poverty. Similarly, Zhemgang and Pema Gatshel had the lowest health dep- 5.6 rivation headcounts, but featured among the highest income-poverty headcounts. Further, Annual growth in percent 5.4 in Haa and Trashi Yangtse, there were signifi- cant differences in their relative performance on 5.2 education and health deprivations, highlighting greater inconsistencies across different meas- ures of deprivation. 5.0 Time-trend mapping of multidimensional poverty conforms with the decline in 4.8 consumption poverty. Using comparable sets of indicators, the trend decline in MPI and the 4.6 headcount of MPI-poor show similar reduction 10 20 30 40 50 60 70 80 90 100 to that of the consumption poverty headcount Per-capita consumption cumulative decile groups using the same cutoff value as national MPI Source: World Bank staff estimates based on BLSS 2007 and 2012, – deprivation in one-third of the indicators, applying nominal poverty line as deflator. although the decline in the MPI-poor headcount ratio is steeper (Figure 2.5). the poor were chronically poor and how many 2.1.2.  Shared Prosperity moved in and out of poverty. Bhutan does not Bhutan’s poorer half enjoyed greater have panel data – same households surveyed over prosperity than the rest, save for the richest time. But a synthetic panel can be put together decile. In the most recent five-year period, by looking at households with age of head of 2007-12, Bhutan as a whole enjoyed an annual households restricted to between 25 and 55 per-capita consumption growth of about 5.5 for 2007 and increased by 5 years for 2012 for percent in real terms. The bottom 40 percent – the analysis of mobility of households over time3. definition used by the World Bank for measuring Using synthetic panel approach, it is found that shared prosperity – of Bhutanese raised their of the 12.4 percent poor in 2012 and 8.4 percent consumption by five percent per year. Aside from - two-thirds of all poor were poor also in 2007 the richest decile, therefore, growth has favored (Table 2.1). While 10.5 percent of the population the lower classes in Bhutan (Figure 2.6). exited poverty between the two periods, 4 percent of the population dropped into poverty from non- 2.1.3.  Mobility in and out of Poverty be- poor status. tween 2007 and 2012 Mobility in Bhutan compares well with other countries, with twice as many of the For every two families that escaped poverty, one fell into poverty. Two-thirds of the poor in 2012 were also poor in 2007.The usual poverty 3  Hai-Anh Dang and Peter Lanjouw. 2013. “Measuring Poverty Dynamics with Synthetic Panels Based on Cross-Sections”, World estimates based on cross-section data provide Bank Policy Research Paper number 6504, June 2013. Estimation a snap-shot and do not inform how many of of point estimates using synthetic panel is a new approach, earlier research by the authors used bound-estimates. Bhutan Poverty Assessment 2014 09 Table 2.1  Mobility In and Out of Poverty between 2007 distribution of per-capita expenditure lies and 2012 nowhere above the initial one. This first-order Poverty Status in 2012 stochastic dominance relation between the two (Percentage distribution of population) distributions implies that all additively separable Poverty Status Poor Non-Poor All poverty measures satisfying monotonicity4 will 2007 Poor 8.3 10.5 18.8 agree that poverty has decreased between 2007 Non-Poor 4.1 77.2 81.3 and 2012. Thus, distributional change observed All 12.4 87.7 100.0 in Bhutan between those two years is pro-poor in the sense of Ravallion and Chen (2003) and Kray Source: Staff estimates based on analysis of synthetic panel data constructed using cross-section data of Bhutan Living Standards (2006). For these authors, a distributional change Surveys 2007 and 2012. “Does a rising tide lift all boats? An investigation of the nexus between poverty reduction and poverty is pro-poor if it involves poverty reduction for mobility in Bhutan in the late 2000s”, Hai-Anh, Pete Lanjouw, and some choice of poverty index.5 T.G. Srinivasan, forthcoming. Note: The estimates here are from the synthetic panel, not reflecting Growth has been pro-poor in a substantive the entire cross-section, and therefore they differ from poverty estimates of cross-section data used elsewhere. way. The poverty implications of the above dis- tributional change are presented in Figure 2.7 population escaping poverty as entering it. that summarizes the variation in poverty out- Bhutan’s upward mobility during 2007-2012 comes in Bhutan between 2007 and 2012 on is comparable to Vietnam’s during 2004-06 the basis of TIP curves associated with poverty (Table 2.2). Mobility for the bottom 40 percent measures that bare members of the FGT (Foster- in Bhutan is diminished compared to the poor Greer-Thorbecke) family. The TIP curve6provides group. Three-fourths of the bottom 40 percent a graphical summary of incidence, intensity remained in poverty whereas almost the same and inequality dimensions of aggregate poverty proportion (10 percent) of population left the based on the distribution of poverty gaps nor- bottom 40 percent and rejoined it between 2007 malized by the poverty line7 (Jenkins and and 2012. Lambert, 1997). The curve is obtained by par- tially cumulating individual contributions to 2.1.4.  Growth in Bhutan has been Pro-Poor overall poverty from the poorest individual to the richest.8 The fact that the TIP curve for 2007 Poverty reduction in Bhutan has been lies above the 2012 curve suggests economic pro-poor. The change in the distribution of per growth in Bhutan has been pro-poor to the capita expenditure between 2007 and 2012 can also be characterized by the growth incidence 4  Monotonicity requires that, other things being equal, an increase curve (Figure 2.9). Recall that this curve shows in the living standard of any person will reduce poverty (Foster, Greer, and Thorbecke, 2010). the growth rate of an indicator of the living 5  The fact that the rate of growth at every percentile up to the 92nd standard (e.g., income or expenditure) at each is less than the average annual growth rate of per-capita expenditure quantile of the size distribution of that indicator means that economic growth in Bhutan has not been pro-poor if it is defined to exceed the average growth for the entire population. (Ravallion and Chen, 2003). The fact that the 6  TIP stands for “three ‘i’s of poverty”, that is incidence, intensity, GIC depicted in Figure 2.9 is greater than zero and inequality. for all expenditure percentiles means that the 7  The curve may also be based on absolute poverty gaps. distribution of per-capita expenditure in 2012 8  This curve is constructed in four steps: (i) rank individuals from dominates the distribution in 2007 to the first poorest to richest; (ii) compute the relative poverty gap of each order. In other words, it means the posterior individual; (iii) form the cumulative sum of the relative poverty gaps divided by population size; and (iv) plot the resulting cumulative sum of poverty gaps as a function of the cumulative population share. 10 Bhutan Poverty Assessment 2014 Table 2.2  Mobility in Select Countries   Peru (2004-05) Vietnam (2004-06) Senegal (2006-11) US (2005-07) Poor in both periods 32.7 11.0 26.3 7.2 Poor who became Non poor 9.7 7.8 21.3 3.8 Non-poor who became poor 11.2 3.9 20.8 3.1 Non-poor in both periods 46.4 77.3 31.7 85.8 All 100.0 100.0 100.0 99.9 Memo items Percentage of poor in period 2 43.9 14.9 47.1 10.3 Percentage of poor who remained 74.5 73.8 55.8 69.9 chronically poor Source: World Bank Policy Research Paper WPS 6504 (2013) and a draft paper on Senegal by the same authors. second-order. Second-order pro-poor judgments that all members of the FGT family of poverty are based on second-order stochastic dominance measures along with the Watts index agree that which is a necessary and sufficient condition for poverty in Bhutan fell significantly between additively separable poverty measures satisfying 2007 and 2012. the strong transfer axiom to agree on the pro- poorness of a distributional change (Atkinson, 1987; Ravallion, 1994). In particular, we find Figure 2.7  A Picture of Poverty in Bhutan, 2007-2012 7 2007 6 Cumulative sum of poverty gaps 5 4 3 2012 2 1 0 0 10 20 30 40 50 60 70 80 90 100 Cumulative population share Source: Author’s calculations Bhutan Poverty Assessment 2014 11 Table 2.3  Distribution of Real per capita Expenditure in Bhutan, 2007-2012 Lowest Year Mean 2nd 3rd 4th 5th 6th 7th 8th 9th 10th Decile 2007 2313.69 2.73 3.92 4.91 5.89 6.98 8.32 10.00 12.15 15.73 29.30 2012 4603.24 2.75 4.00 4.92 5.86 6.93 8.12 9.69 11.72 15.32 30.61 Source: Author’s calculations Figure 2.8  Change in Relative Inequality in Bhutan, 2007-2012 Expenditure Shares (percent) by Decile Lorenz Curves 35 100 30 80 25 Expenditure Shares 60 20 15 40 10 20 5 0 0 0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10 Expenditure Percentiles 2007 2012 2007 2012 Source: Author’s calculations 2.2.  Stable Inequality less the same over time, while that of the richest increased a little bit. These results show that the The growth process in Bhutan has been growth process in Bhutan has been distribution distribution-neutral. Table 2.3 is a summary of neutral between 2007 and 2012 (Figure 2.9 shows the distribution of per capita expenditure based national growth incidence). on the 2007 and 2012 rounds of the BLSS. The overall Gini coefficient for 2007 is The results are based on individual level estimated at 38.09 percent. In 2012 this measure data for both 2007 and 2012. The 2007 sample of relative inequality stood at 38.75 percent. includes observations on 9,798 households; the Data show that inequality between groups has 2012 dataset contains observations on 8,968 been quite stable9 (Table 2.4). This pattern of households. The summary information includes, for each round, mean per capita expenditure in 9  These results are based on a simple decomposition approach real terms and the decile distribution of that applied by Benjamin, Brandt and Giles (2005) to the case of per capita expenditure (see also Figure 2.8). inequality in rural China. The approach entails estimating a regression of the log of the welfare indicator (income or expenditure This information shows that real household per per capita) on a set of location dummies. The resulting R-squared shows the proportion of the variation of the log of the welfare capita expenditure almost doubled in the span indicator that is accounted for by the location dummies. In other words, this is the amount of variation that is “explained” by of five years. It also shows that the share of each differences in average level of living. The residual variance is linked decile below the richest has remained more or to within-location inequality. In our application for Bhutan we use Dzongkhag dummies as location variables. 12 Bhutan Poverty Assessment 2014 Figure 2.9  Growth Incidence Curve for Bhutan, 2007-2012 24 22 Annual Growth Rate (%) 20 18 16 14 12 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Source: Author’s calculations Table 2.4  Between-Group (Dzongkhag) Inequality in Bhutan set of location dummies. The resulting R-squared by Area of Residence shows the proportion of the variation of the log Year Urban Rural Bhutan of the welfare indicator that is accounted for by 2007 13.3 20.4 26.0 the location dummies. In other words, this is 2012 14.5 22.4 25.0 the amount of variation that is “explained” by Source: Author’s calculations differences in average level of living. The residual Note: Between-group inequality is measured by the proportion of the variance is linked to within-location inequality. variance of the log of per capita expenditure explained by dzongkhag of residence. This is the R2 of the regression of log per capita In our application for Bhutan we use dzongkhag expenditure on a set of dummy variables representing the dzongkhag (see Benjamin and Brandt. 2005.“The Evolution of Income Inequality dummies as location variables. in Rural China.” Economic Development and Cultural Change). 2.2.1.  Uneven Poverty Reduction across distributional change suggests that, overall, Dzongkhags the observed reduction in poverty was driven exclusively by the size effect. Incidences of unhappiness and poverty are Inequality between dzongkhags remained higher in the eastern dzongkhags. The 2010 GNH stable in Bhutan. These results are based on survey estimates that 59 percent of Bhutanese a simple decomposition approach applied by are not-yet-happy; even in the least-poor Paro Benjamin, Brandt and Giles (2005) to the case close to 47 percent are not-yet-happy. Using the of inequality in rural China. The approach entails national average for the percentage of poor in estimating a regression of the log of the welfare 2012 and not-yet-happy in 2010 as dividing lines indicator (income or expenditure per capita) on a (Figure 2.10), we note that unhappiness tends Bhutan Poverty Assessment 2014 13 Figure 2.10  Poverty and Unhappiness across Dzongkhags 35 Lhuentse 30 Zhemgang Pema Gatshel 25 Dagana Poverty Incidence 2012 (%) Samtse Samdrup Jongkhar 20 15 Tsirang Trongsa Trashi Yangtse Wangdue Phodrang Trashigang Chhukha 10 Punakha Monggar Haa 5 Sarpang Bumthang Paro Thimphu Gasa 0 40 45 50 55 60 65 70 75 80 Population who are not-yet-happy, 2010 (%) Source: Poverty Analysis Report 2012 and Gross National Happiness Report 2010 Note: Not-yet-happy are those who lack sufficiency in indicators identified in the GNH report. Figure 2.11  Uneven Poverty Reduction across Dzongkhags, 2007-2012 60 Zhemgang 50 Samtse Monggar Lhuentse Samdrup Jongkhar 40 Percent of poor in 2007 Dagana Trashigang 30 Pema Gatshel Trongsa Chhukha Sarpang 20 Punakha Wangdue Phodrang Haa Tsirang Bumthang 10 Trashi Yangtse Thimphu 0 Percent of poor in 2012 Source: Poverty Analysis Reports, 2007 and 2012 Note: Bubble size is proportional to the number of poor. For instance, Samtse is home to the largest number for a dzongkhag, 12,000, and Pema Gatshelhas 6,000. 14 Bhutan Poverty Assessment 2014 to increase with poverty and the eastern part of the country is highly prone to unhappiness and poverty. Poverty reduction across dzongkhags has been uneven. Of the 20 dzongkhags in Bhutan, poverty reduction touched all except two, Pema Gatshel and Tsirang (Figure 2.11). The pace of poverty reduction was slower in Dagana and Lhuentse, but much faster in initially-very-poor Monggar, Samtse and Zhemgang. Even among the poor eastern dzongkhags, Zhemgang has been more successful than Lhuentse in poverty reduction. Bhutan Poverty Assessment 2014 15 3 Chapter Changing Profiles of the Poor and Bottom 40 Percent of the Population This chapter presents the change in profiles The improvements in households’ assets of the poor and bottom 40 percent of the indicators are illustrated by the following three population between 2007 and 2012. The charts, in seven key dimensions of welfare – profiles are presented in terms of asset and livestock ownership, type of dwelling wall, type of amenities, health and nutrition, gender, and dwelling roof, safe latrine access, electricity access, land ownership. This analysis complements television ownership, and access to mobile phone. discussion of the profile of the poor in Bhutan For all households (Figure 3.1), between 2007 and Poverty Analysis 2012. 2012 there were improvements in six dimensions of welfare, except for livestock ownership which 3.1.  Welfare Indicators (Assets and showed a small decline. The same pattern held Amenities) true for the poorest households and those in the bottom four deciles of the real per-capita All non-consumption indicators of welfare consumption distribution (Figures 3.2, 3.3). Both showed significant improvements between 2007 of these categories experienced relatively large and 2012, both for the general population and improvements in asset ownership in the same six the poor (Table 3.1). Basic asset and amenity indi- dimensions of welfare between 2007 and 2012, cators used in BLSS survey years 2007 and 2012 reflecting the ongoing pattern of improvement show that the biggest improvements occurred in in asset accumulation in the country at large. By mobile phone ownership, housing, and electricity far the largest improvement in both the poorest connections. In particular, there have been signi­ households and those in the bottom four deciles ficant increases in the percentages of households was in mobile phone ownership – from 11 percent with metal sheet roofs, electricity connections, and access to mobile phones. There has been a significant increase in the poorest households’ “Mobile connectivity has also access to mobile phones and electricity. Improve- benefited us in communicating faster ment in their housing conditions, specifically in during emergencies.” roof quality, has been dramatic. Bhutan Poverty Assessment 2014 17 Table 3.1  Trends in Basic Assets and Amenities, 2007-2012 All households Poor households Bottom 40 percent 2007 2012 2007 2012 2007 2012 Livestock ownership (%)1 59.6 51.8 93.0 86.9 90.3 80.7 Wall of dwelling* 22.3 26.4 3.3 6.1 5.3 12.1 Roof of dwelling** 74.0 89.3 53.4 78.8 58.7 84.4 2 48.2 62.0 21.1 32.8 23.5 41.7 Safe latrine use (%) Electricity connection (%)◊ 69.1 88.3 39.9 69.5 46.6 77.4 TV ownership (%) 37.7 58.5 6.4 21.6 8.8 33.3 Mobile Phone ownership (%) 39.3 92.8 6.6 81.7 11.2 87.8 Source: BLSS 2007 and 2012 Notes: * Percent with cement-bonded bricks/stone (external wall) ** Percent with metal sheets (roof) ◊ Electricity % “from the grid” 1 Household ownership of livestock – specifically pigs, cattle, goat, buffaloes, horses, sheep, yaks, and poultry –is included in this category. 2 Percent of flush & pit latrine (2012). The safe latrine use consists of flush to piped sewer system, flush to septic tank (without soak pit); flush to septic tank (with soak pit) and flush to pit (latrine) in 2012, and flush toilet and pit latrine with septic tank in 2007. An “improved sanitation facility” is one that hygienically separates human excreta from human contact. (WHO, 2013) To assess whether the latrine used is safe or not, the type of toilet used by the households are identified in the BLSS for years 2012 and 2007. Figure 3.1  Improvements in Households’ Assets Figure 3.2  Improvements in Households’ Assets Ownership – All Households, 2007-2012 Ownership – Poor Households, 2007-2012 Livestock Livestock ownership (%) ownership (%) 100 100 Mobile Phone 80 Wall of dwelling* 80 ownership (%) 60 Mobile Phone 60 Wall of dwelling* 40 ownership (%) 20 40 0 20 TV ownership (%) Roof of dwelling* 0 TV ownership Roof of dwelling* (%) Electricity Safe latrine use connection (%) (%) Electricity Safe latrine use connection (%) (%) 2007 2012 Source: BLSS 2007 & 2012, NSB 2007 2012 Source: BLSS 2007 & 2012, NSB in 2007 to 88 percent, in 2012, for the bottom four 3.2.  Health and Nutrition deciles. The increase in mobile phones among the Bhutan’s nutrition indicators have poorest households was similarly dramatic. Other improved in recent years and are better now significant improvements for the poor came in than in other South Asian countries, accord- electricity connections and the roofs of dwellings. ing to World Bank (2013),10 yet they remain a 10  Nutrition in Bhutan: Situational Analysis & Policy Recommendations, World Bank, 2013 18 Bhutan Poverty Assessment 2014 Figure 3.3  Improvements in Households’ Assets Ownership – Bottom 40 percent, 2007-2012 “In my opinion, community has Livestock ownership improved in terms of accessibility 100 (%) of drinking water. Now almost all Mobile 80 the households have drinking water Phone Wall of ownership 60 dwelling* in vicinity of their house and do not (%) 40 have to walk distance to fetch water. 20 The community has also a school.” 0 TV Roof of ownership dwelling* (%) under five years stunted and 15 percent under- nourished. The under nutrition problem is preva- lent in the eastern part of the country and among Electricity Safe connection latrine use children of mothers with no education. Across (%) (%) the wealth quintiles improvements in nutrition are gradual and change only from the third quin- 2007 2012 tile (Figure 3.4). Source: BLSS 2007 & 2012, NSB The BLSS 2012 did not collect anthropometric data, but using a dietary diversity score it may Figure 3.4  Percent of Wasting and Underweight Children, be possible to evaluate indirectly the direction of 2010 change and equity in outcomes. Dietary diversity 45 is a useful indicator of nutritional access by 40 itself. The nutritional diversity is measured by 35 household dietary diversity scores (HDDS).11 For 30 the purposes of our analysis, we define dietary 25 diversity as the number of different food groups 20 consumed by the household during the week prior 15 to being surveyed for the BLSS. 10 There has been an improvement in the 5 0 dietary diversity score between 2007 and Poorest S econd Third Fourth R ichest 2012. Comparison of the dietary patterns among Bhutanese households shows that nearly Stunting Underweight all households consumed food in the cereals, Source: Bhutan Multiple Indicator Survey, 2010, NSB. milk and milk products, and vegetables groups Notes: Stunting (height- for-age) less than -2SD from the reference (Figure 3.5). The main sources of energy for most population. Under nutrition (weight-for-height) less than -2SD from reference population. households are cereals, vegetables, and milk and milk products. There has been significant increase cause of concern. Bhutan Multiple Indicator Sur- in protein intake over the years. For protein, vey, 2010 showed that under nutrition is higher households tend to rely primarily on meat and fish in poorer households. The bottom 40 percent of the population had about 40 percent of children 11  For details about the construction of the HDDS, see Box 1 Bhutan Poverty Assessment 2014 19 Figure 3.5  Percent of Households Consuming a Particular Food Group in 2007 and 2012 120 2007 2012 100 80 60 40 20 0 Source: Staff estimates based on BLSS 2007 and 2012 Figure 3.6  Fractions of Households Consuming Food Figure 3.7  Fractions of Households Consuming Food Items – by Specific Groups, 2012 Items – by a Specific Group, 2007 Cereals Cereals 100 100 Sugar & Roots & Sugar & Roots & Honey 90 Tubers Honey 80 Tubers 80 60 70 40 Oils & fats Vegetables Oils & fats Vegetables 60 20 50 0 Milk & Milk Milk & Milk Fruits Fruits products products Pulses, Pulses, Meat, Poultry, Meat, Legumes Legumes Offal Poultry, Offal &Nuts &Nuts Fish Fish Non-poor Poor Subsistence poor Non-poor Poor Subsistence poor Source: Staff estimates based on BLSS 2007 and 2012 – the proportion of households consuming meat Notes: (i) Total number of observations: 9,798 for 2007 and 8,968 for 2012 increased from 46 percent to 93 percent between (ii) Households are stratified into three groups according to poverty status: “Non-poor” – households with per-capita consumption levels 2007 and 2012. Households’ consumption of above the poverty line; “Poor” – households with per-capita consump- tion levels below the poverty lines; “Subsistence poor” – households pulses, legumes, and nut products rose from 11 with per-capita consumption levels below the food poverty line. percent to 86 percent in the same period. The shifting patterns of household diets among the poor, subsistence poor, and non- 3.7). Dietary components that distinguished the poor between 2007 and 2012 are set out in diets of the poor from the non-poor were pulses, the following two diagrams (Figure 3.6 and legumes and nuts, meat products, fish, sugar 20 Bhutan Poverty Assessment 2014 Figure 3.8  Household Dietary Diversity Scores by Consumption Deciles, 2007 &2012 11.5 2007 2012 10.0 Mean Dietary Diversity Score (HDDS) 8.5 7.0 5.5 4.0 2.5 1.0 1 2 3 4 5 6 7 8 9 10 Consumption Decile Source: Staff estimates based on BLSS 2007 and 2012 Figure 3.9  Household Dietary Diversity Score by Area, products, fruits, and milk and milk products. For 2007 and 2012 three food groups – cereals, vegetables, oil and 9.6 9.5 10.0 fats – no differences existed between the poor and non-poor in 2012, while there had been some 8.5 7.6 distinction between them in 2007. 7.0 6.5 Mean HDDS There has been a significant improvement 5.5 in household dietary diversity in Bhutan. In addition, the differences in HDDS among deciles 4.0 of per-capita consumption diminish between 2.5 2007 and 2012.HDDS increased to 9.6 in 2012 from 6.6 in 2007. This compares favorably to 1.0 2012 2007 that of Nepal (8.8) and Pakistan 9.1 (Tiwari et al., 2013).12. Urban Rural The rural-urban differences between HDDS Source: Staff estimates based on BLSS 2007 and 2012 also diminished to near equivalence between 2007 and 2012, possibly because of improvements in either sickness or injury in 2007 and 2012. The market access. out-of-pocket expenditure on health at household The incidence of sickness/injury remained level increased five-fold in nominal terms between stable in Bhutan. One in six individuals reported 2007 and 2012. However, for the poor the increase was slightly smaller. Out-of-pocket expenditure 12  For comparability over time, Bhutan HDDS uses 10 food groups. as a share of consumption expenditure was five The estimates quoted for Pakistan and Nepal use 11 food groups, percent for poor households compared to 15 and therefore using same number of comparable food groups, Bhutan should be even further ahead. Bhutan Poverty Assessment 2014 21 percent for non-poor households, indicative of Figure 3.10  Trend in Poverty Incidence, by Gender of Household Head, 2007 and 2012 the equity in access to health services. 30 3.3.  Gender and Poverty 25 3.3.1.  Is Poverty in Bhutan Gender-Blind? Poverty Indicence (%) 20 The incidence of poverty in female-headed households was no different from that of male- 15 headed households in 2012. But in 2007, the female-headed households fared better; the rate 10 of decline in poverty incidence of male-headed 5 households has been faster than those of females, to bring them level (Figure 3.10). The generally 0 sanguine assessment of economic status of MHH MHH FHH FHH female-headed households may originate in the benefits ensuing from the matrilineal inheritance 2007 2012 2007 2012 of land holdings, as noted in World Bank (2013).13 Source: Staff estimates based on BLSS data Comparison by marital status of the heads of households show heightened poverty incidence for female-headed households (compared to Table 3.2  Poverty Incidence in 2012, by Marital Status similarly placed male headed households) for and Gender the never-married, married, and divorced. Only 2012 for the widowed is the poverty incidence smaller   (Percentage of poor) (Table 3.2). A disproportionate burden of family Marital Status Male Female chores, including child care by women, may Never married 3.7 10.5 restrict their choices to low-quality jobs even if Married 11.8 13.8 there is no difference in rewards for labour. The Divorced 4.4 6.2 higher poverty incidence for households headed Widower/widow 18.9 12.5 by never-married females is puzzling, however. Male-headed and female-headed house- Source: Staff estimates based on BLSS data holds have experienced a reduction in poverty between 2007 and 2012 (Figure 3.11). The growth and female-headed households have experienced incidence curves by gender of heads of households a reduction in poverty between 2007 and 2012. (left panels) and by area (urban-rural) of resi- This is the case for urban and rural households dence (right panels) show patterns of growth in as well. each sub group similar to the overall pattern. One There is a considerable heterogeneity of therefore would expect similar poverty outcomes. impact across quantiles. The information con- In particular, all additively separable poverty tained in Figure 3.11 reveals the following: The measures that satisfy both monotonicity and average annual growth rate of mean per capita the transfer axiom will agree that male-headed expenditure is virtually the same for male-headed 13  and female-headed households, but this hides Bhutan Gender Policy Note, World Bank, 2013. 22 Bhutan Poverty Assessment 2014 Figure 3.11  Economic Growth in Bhutan, by Sex and Area of Residence, 2007-2012 28 28 24 Annual Growth Rate (%) Annual Growth Rate (%) 24 20 20 16 16 Female Rural 12 Male 12 8 Urban 8 4 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Expenditure Percentiles 4 10 Female Growth Rate minus Male Growth Rate (%) Rural Growth Rate minus Urban Growth Rate (%) 2 8 0 6 -2 4 -4 2 -6 0 -8 -10 -2 10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 Source: Author’s calculations considerable heterogeneity of impact across quan- to land ownership in rural areas for households tiles; the bottom left panel of Figure 3.11 shows, with primary activity as agriculture in per-capita at each percentile, the difference between the terms, higher poverty incidence is not associated growth rate for male-headed households and that with the landless or people with smaller for female-headed households. Female-headed holdings. Even then, the poverty incidence is not households located in the lower 35 percent of the distinguishable between landless and small or distribution and above the 96th percentile expe- marginal land holders. Part of the reason could be rienced higher growth rates than male-headed that landless households engaged in agriculture households. are able to lease-in land and the share of produce to tenant is reported to have risen sharply in focus group discussions as the lessors have found 3.4.  Land Ownership and Poverty non-farm occupations or have emigrated from The relationship between land ownership rural areas. and poverty is complex in Bhutan. Poverty incidence is lower among the landless, as noted in Bhutan Poverty Analysis 2012 report. Landless households are spread across all quintiles of per- capita expenditure and unless we restrict analysis Bhutan Poverty Assessment 2014 23 Table 3.3  Incidence of Poverty, by Land Ownerships Status   2012 2007 Per-capita Dry Per-capita Dry Per-capita Total Per capita Total Per-capita Total Per capita Total   and Wet land and Wet land land holding land operated land holding land operated holding holding Landless 0.19 0.17 0.20 0.36 0.29 0.38 >0-3.0 acres 0.18 0.18 0.18 0.33 0.34 0.33 >3-5 acres 0.11 0.09 0.11 0.14 0.11 0.18 >5-10 acres 0.03 0.05 0.03 0.09 0.16 0.10 10+ acres 0.11 0.00 0.11 0.00 0.00 0.00 Total 0.18 0.18 0.18 0.33 0.33 0.33 Source: BLSS 2007 and 2012 Note: In rural Bhutan for households engaged in agriculture, in proportion Box 1: Household Dietary Diversity Score The HDDS is based on the food groups proposed by USAID’s Food and Nutrition Technical Assistance Project (FANTA). Twelve food groups are proposed for the HDDS. The potential score range is 0-12 for HDDS. The food groups used to calculate HDDS are listed in Table 3.4. Dietary diversity scores (DDS) are calculated by summing the number of food groups consumed in the household over a given reference period. For the purposes of our analysis, the food items consumed by the household in the last seven days were categorized into 12 food groups, each representing a special class of nutrients. If a household consumed an item from a particular food group, the household was assigned a value of “1” for that food group and “0” otherwise. Hence, for each household, a set of twelve parameters indicates whether or not a certain food group was consumed by the household on each day during the seven-day period. Summing of the 12 indicators for each household yields the HDDS. The higher the DDS score the better the diversity of food intake and better the quality of diets. For the analysis presented, the sixth and last food group (“Eggs & Misc.”) is dropped. There are no established cut-off points in terms of number of food groups to indicate adequate or inadequate dietary diversity for the HDDS. Because of this it is recommended to use the mean score or distribution of scores for analytical purposes and to set program targets or goals (FAO, 2011). 24 Bhutan Poverty Assessment 2014 Table 3.4  Classification of Food Groups Food Group Food group Name Examples Number corn/maize, rice, wheat, sorghum, millet or any other grains or 1 CEREALS foods made from these (e.g. bread, noodles, porridge or other grain products) white potatoes, white yam, white cassava, or other foods made from 2 WHITE ROOTS AND TUBERS roots pumpkin, carrot, squash, or sweet potato that are orange inside VITAMIN A RICH VEGETABLES AND and other locally available vitamin A rich vegetables (e.g. red sweet TUBERS pepper) dark green leafy vegetables, including wild forms and locally 3 DARK GREEN LEAFY VEGETABLES available vitamin A rich leaves such as amaranth, cassava leaves, kale, spinach other vegetables (e.g. tomato, onion, eggplant) and other locally OTHER VEGETABLES available vegetables ripe mango, cantaloupe, apricot (fresh or dried), ripe papaya, dried VITAMIN A RICH FRUITS 4 peach, other locally available vitamin A rich fruits OTHER FRUITS Other fruits, including wild fruits ORGAN MEAT liver, kidney, heart or other organ meats or blood-based foods 5 beef, pork, lamb, goat, rabbit, game, chicken, duck, other birds, FLESH MEATS insects 6 EGGS eggs from chicken, duck, guinea fowl or any other egg 7 FISH AND SEAFOOD fresh or dried fish or shellfish dried beans, dried peas, lentils, nuts, seeds or foods made from 8 LEGUMES, NUTS AND SEEDS these (e.g., hummus, peanut) 9 MILK AND MILK PRODUCTS milk, cheese, yogurt or other milk products 10 OILS AND FATS oil, fats or butter added to food or used for cooking sugar, honey, sweetened soda or sweetened juice drinks, sugary 11 SWEETS foods such as chocolates, candies, cookies and cakes spices (black pepper, salt), condiments (soy sauce, hot sauce), 12 SPICES, CONDIMENTS, BEVERAGES coffee, tea, alcoholic beverages Source: FAO, 2011 Notes: (i) The vegetable food group is a combination of vitamin A rich vegetables and tubers, dark green leafy vegetables and other vegetables (ii) The fruit group is a combination of vitamin A rich fruits and other fruits (iii) The meat group is a combination of organ meat and flesh meat Bhutan Poverty Assessment 2014 25 4 Chapter Enlarging Opportunities for Children 4.1.  Inequality of Opportunity in Bhutan life (such as education), discounted by how unequally the services are distributed between Research shows that access to a basic set the different groups of the population. The HOI of goods and services during childhood can be focuses exclusively on children, as opposed to the an important predictor of future outcomes. whole population, to endeavor to plot the course Access to quality basic services such as education, of poverty in the future. Analysis in this chapter health care, and essential infrastructure (such as electricity, improved water, and sanitation), and early childhood development provides children the opportunity to advance and reach “Most of the households have tap their potential, irrespective of their background. water for drinking. I have seen Equality of opportunity is about giving all children a lot of improvements related to the same chance to succeed in life. health: improved cleanliness and Equality of opportunity seeks to level the sanitation.” playing field so that circumstances such as sex, ethnicity, birthplace, or family background, which are beyond the control of an individual, do is restricted to educational outcomes and access not influence a person’s life chances. Success in to infrastructure. Health outcomes and early life should depend on people’s choices, effort, and childhood development are not covered in the talents, not their circumstances of birth. Reaching analysis because the BLSS data does not provide consensus on an agenda for reducing inequality of requisite data. Bhutan’s high GDP growth has opportunity is politically more viable than trying been successful in realizing improved outcomes to find agreement on redistributive policies to in education and infrastructure. Enrollments in reduce inequality of income or wealth. primary and secondary are increasing and doing The Human Opportunity Index (HOI) is so at a faster rate than other countries with the designed to gauge the progress of a country same level of GDP per capita (Figure 4.2). At the in equalizing opportunities for children in same time, access to improved water sources education. The HOI measures the availability and sanitation facilities have been increasing. of services that are necessary to progress in Figure 4.1 illustrates how a low HOI in the past Bhutan Poverty Assessment 2014 27 Figure 4.1  Human Opportunity Index (HOI) of Past and Current Poverty 35 Lhuentse Lhuentse 30 Pema Gatshel Zhemgang Pema Gatshel Dagana Dagana Samdrup Jongkhar Samdrup Jongkhar 25 Samtse Poverty rate in 2012 (%) 20 Chhukha Trongsa Chhukha Trashi Yangtse 15 Wangdue Phodrang Trashi Yangtse Monggar 10 Punakha Punakha Haa Haa 5 Thimphu Bumthang Thimphu Sarpang 0 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% HOI of Primary school completion HOI of Secondary school completion (ages 14-18 years) in 2007 (ages 20-24 years) in 2007 Figure 4.2  Time Trend Indicators Relative to per-capita GDP 100 100 Bangladesh Pakistan India Bhutan Improved water source (% of population with access) Improved sanitation facilities (% of population with access) Nepal 80 Sri Lanka 80 Sri Lanka 60 Bangladesh 60 Afghanistan Pakistan Nepal Bhutan 40 40 India 20 20 Afghanistan 0 0 1K 2K 3K 4K 5K 6K 7K 1K 2K 3K 4K 5K 6K 7K GDP per capita PPP in constant 2005 $ PPP GDP per capita PPP in constant 2005 $ PPP Nepal 100 Sri Lanka School enrollment, secondary (% gross) India School enrollment, primary (% gross) 125 Afghanistan 80 Bhutan 100 Pakistan Bhutan Afghanistan India Maldives 60 75 Sri Lanka Maldives 40 50 Bangladesh 20 Bangladesh 25 Pakistan 0 1K 2K 3K 4K 5K 6K 7K 8K 0K 1K 2K 3K 4K 5K 6K 7K 8K GDP per capita PPP in constant 2005 $ PPP GDP per capita PPP in constant 2005 $ PPP 28 Bhutan Poverty Assessment 2014 points to a higher incidence of poverty in the present. Despite Bhutan’s rapid, broad-based, “Education has made things better.” and inclusive growth, there are disadvantaged groups of children across the country who deserve and age group. At the country level, coverage a better future. for education indicators are improving, where Bhutan’s high GDP growth has been successful growth incompletion rates are higher than that in realizing improved outcomes in education of attendance rates. By a set of circumstances, and infrastructure. Enrollments in primary and the coverage rates show a remarkable disparity. secondary are increasing and doing so at a faster A child born to a family in the bottom quintile rate than other countries with the same level of of the consumption distribution has a consider- GDP per capita (Figure 4.2). At the same time, ably lower likelihood of attending and completing access to improved water sources and sanitation school than if he/she were born into a higher facilities have been increasing. quintile. Such gaps are wider in school completion than in school attendance. In 2003, for instance, 4.2.  Social Outcomes for Children in the primary school attendance rate of the poor- Relation to Birth Circumstances est quintile was 50 percent, but for the richest it Coverage of school attendance and com- was 93 percent. However such gaps had narrowed pletion are improving, but continue to vary in 2007 and 2012. For instance, in the bottom according to factors such as income quintiles, quintile 20 percent of 15-19 year-olds completed residences, sex, and characteristics of house- primary school while 75 percent of those in rich- hold head such as education, employment, sex est quintile did so. Similarly large variances exist Figure 4.3  Education Coverage, by Wealth Quintiles Primary school attendance Primary school completion Secondary school attendance Secondary school completion Year By (6-12 years) (15-19 years) (13-16 years) (20-24 years) 2003 Q1 50.5% 19.6% 35.1% 3.1% Q2 65.1% 31.4% 55.8% 5.3% Q3 76.7% 44.5% 73.3% 9.7% Q4 86.5% 63.4% 80.4% 28.5% Q5 92.6% 75.0% 85.3% 40.4% 2007 Q1 71.3% 25.2% 55.4% 7.4% Q2 82.8% 45.4% 71.8% 17.5% Q3 89.8% 64.7% 82.1% 27.7% Q4 94.8% 75.2% 89.4% 44.4% Q5 97.4% 90.2% 92.8% 68.0% 2012 Q1 90.8% 55.5% 82.5% 19.2% Q2 93.8% 64.5% 83.9% 30.6% Q3 96.4% 77.5% 89.6% 42.7% Q4 95.5% 83.2% 90.0% 55.2% Q5 97.9% 87.9% 93.6% 69.3% 0% 50% 100%0% 50% 100% 0% 50% 100%0% 50% 100% Coverage Coverage Coverage Coverage Q1 Q2 Q3 Q4 Q5 Bhutan Poverty Assessment 2014 29 Figure 4.4  Multidimensional Inequalities of Education, 2003-2012 4.0 2003 Legend 2003 13 Quintile - Q5/Q1 2007 Location - Urban/Rural 12 Employment - Yes/No 3.5 Secondary/No Education 11 Secondary school completion (20-24 years) Primary school completion (14-18 years) 10 2007 3.0 9 8 2.5 2003 7 2003 6 2003 5 2003 2.0 4 2007 3 2012 2012 2007 1.5 2012 2 2007 2012 2012 2012 2003 2003 1 2007 2012 2012 1.0 2007 0 Primary school attendance (6-12 years) Secondary school attendance (13-16 years) in urban and rural education, household heads’ areas; and children in households where the levels of education, and employment. head is employed vs. those in households where While there are still notable gaps between the head is unemployed. Second, relationships school attendance and school completion between those indicators haveadjusted positively. among the various categories surveyed, those For primary education, inequalities of attendance gaps have narrowed rapidly in the last decade vs.completion have gonefrom about 1.8 and (Figure 4.4).The illustration shows inequalities 3.7 respectively to roughly equal, at1.1 and 1.6, measured in terms ofratios betweentwo suggesting that there was an improvement in extreme categories in the same groups in the the transition from school attendance to school survey period. Each point represents the level completion. The situation is also improving, but of inequality by a specificcircumstance; for quitedifferently,insecondary education, where instance, the ratio of primary school attendance inequalities in school attendance are improving for children who are in the top quintile to that faster than those of completion. This suggests of the bottom quintile. First, it shows clearly that the completion rates aremuch lower in that the levels of inequality have been reduced secondary school than in primary school. ineach of the four inequality ratios: children in In addition, children typically are subject to the richest quintile vs.children in the poorest multiple deprivations or disadvantages. Those quintile; children in households with uneducated multiple deprivations futher reduce opportunities heads vs. children in households with secondary in the long run. As shown in many instances, education; children in citiesvs.children in rural a child does not belong to a single group with 30 Bhutan Poverty Assessment 2014 Figure 4.5  Multiple Deprivations – Proportional Distribution of Children 6-12 years Multiple deprivations - Primary School Attendence (6-12 Years) 2003 2007 2012 Gender Female 50.94 49.79 49.33 Male 49.06 50.21 50.67 Location Rural 82.70 77.53 72.40 Urban 17.30 22.47 27.60 Wealth Q1 28.82 36.48 27.62 Quintiles Q2 24.39 22.94 23.94 Q3 21.21 17.98 20.42 Q4 15.72 15.00 16.91 Q5 9.85 7.59 11.10 Grou p Rural Q1 28.63 35.76 26.36 (Urban Rural Q2 22.66 20.48 20.26 Quintile) Urban Q1 0.19 0.72 1.26 Urban Q2 1.73 2.46 3.68 Group Rural Q1 Female 15.21 17.94 13.41 (Urban Rural Q1 Male 13.42 17.82 12.95 Quintile Rural Q2 Female 11.49 10.34 9.73 Sex) Rural Q2 Male 11.17 10.14 10.53 Urban Q1 Female 0.12 0.31 0.68 Urban Q1 Male 0.07 0.41 0.59 Urban Q2 Female 0.79 1.29 1.80 Urban Q2 Male 0.95 1.17 1.89 Distribution (%) Distribution (%) Distribution (%) Figure 4.6  Multiple Deprivations – Proportional Distribution of Children 13-16 years Multiple deprivations - Secondary School Attendence (13-16 Years) 2003 2007 2012 Gend er Fema le 49.69 51.57 50 .26 Ma le 50.31 48.43 49 .74 Locati on Rural 83.45 77.13 74 .00 Ur ba n 16.55 22.87 26 .00 Wealth Q1 28.02 31.10 27 .98 Quin ti les Q2 25.80 24.04 22.93 Q3 19.41 19.25 20.21 Q4 16.22 15 .46 16.90 Q5 10.55 10.16 11.98 Grou p Rural Q1 27.81 30.51 26 .85 (Urb an Rural Q2 24.52 21.76 20.03 Quin ti le) Ur ba n Q1 0.21 0.59 1.13 Ur ba n Q2 1.28 2.28 2.90 Grou p Rural Q1 Fem ale 13.86 15.11 13.47 (Urb an Rural Q1 M ale 13.95 15 .40 13.38 Quin ti le Rural Q2 Fem ale 11.97 10 .99 9.70 Sex) Rural Q2 M ale 12.54 10 .77 10 .33 Ur ba n Q1 Fem ale 0.09 0.42 0.58 Ur ba n Q1 M ale 0.12 0.17 0.55 Ur ba n Q2 Fem ale 0.67 1.24 1.47 Ur ba n Q2 M ale 0.61 1.04 1.44 Distribution (%) Distribution (%) Distribution (%) Bhutan Poverty Assessment 2014 31 Figure 4.7  Coverage of Basic Infrastructure, by Education of Household Head Using gas and Access to improved Access to improved Access to improved Access to piped Year By Access to electricity electricity for private sanitation sanitation water water cooking 2003 No Education 33.2% 78.8% 90.7% 80.8% 78.6% 18.8% Primary 57.3% 64.6% 95.7% 88.6% 86.2% 49.0% Secondary 89.3% 71.5% 98.5% 97.8% 97.8% 84.4% Post-Secondary 91.5% 88.3% 100.0% 99.0% 99.0% 88.5% 2007 No Education 63.8% 80.9% 94.9% 88.4% 87.3% 45.1% Primary 79.4% 71.7% 95.9% 94.2% 93.5% 65.3% Secondary 94.7% 73.3% 99.3% 98.2% 98.1% 91.5% Post-Secondary 96.0% 88.1% 99.6% 99.8% 99.7% 95.2% 2012 No Education 87.6% 67.4% 73.7% 97.6% 96.3% 78.6% Primary 95.3% 72.0% 84.4% 98.2% 96.5% 90.4% Secondary 98.3% 78.2% 94.7% 99.0% 98.3% 97.3% Post-Secondary 98.5% 85.3% 96.6% 99.6% 99.4% 99.0% 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% 0% 50% 100% Coverage Coverage Coverage Coverage Coverage Coverage No Education Primary Secondary Post-Secondary disadvantaged circumstances, but to a variety provision, location of residence, education level ofsuch groups. For example, Figure 4.5 shows of the household head, and wealth quintle index that among children aged 6-12 years in primary are important factors. Coverage of electricity,gas school in 2003, about 83 percent lived in rural usage, and electricity for cooking acrossthe areas, 29 percent were in the poorest quintile, country depends on the location of the household and 15 percent were female. Such children, (urban vs rural) as well as the education of the therefore,are have multiple disadvantages, and household head. Figure 4.7 shows the coverage the numbers haverisen in the last decade. Thus, in of infrastructure indicators by education level of cases of multiple deprivation the child’s access to household head. It shows clearly that household opportunities such as education is signfinicantly heads with no education or primary or lesser worse than his or her peers who are subject to just education have a lower coverage rate of access a single deprivation or disadvantage. Whilethere to basic infrastructure than do households are various differentiated contributions to the with heads with secondary or post-secondary inequalities of circumstance, some factors might education. be more important than others. Coverage of infrastructure varies significantly There have been improvements in in degree ofimprovement acrossthe country, opportunities for access to infrastructure, and betweendifferent categories of circumstance specifically electricity, gas, and electricity for during the survey period. In addition, overall cooking. While the gender of the household inequalityiesas well as those of different head does not matter as much for infrastructure categories of circumstances have been reducing in 32 Bhutan Poverty Assessment 2014 Figure 4.8  Multidimensional Inequalities of Infrastructure, 2003-2012 2003 1.4 1.4 1.3 1.3 2003 2012 2003 Access to improved private sanitation 1.2 1.2 2007 Access to piped water 2012 2003 2012 2007 2012 2007 1.1 1.1 2012 2007 2012 2007 2007 2003 2012 1.0 2012 1.0 2003 2003 2007 By 26-44 Years/62 Years and over 2003 Male/Female Q5/Q1 0.9 0.9 2007 2007 Secondary/No Education Urban/Rural 2007 Yes/No 0.8 2003 0.8 Access to improved sanitation Access to improved water 2003 30 25 Using gas and electricity for cooking 20 15 10 By 2003 26-44 Years/62 Years and over 2007 Male/Female 5 Q5/Q1 2003 Secondary/No Education 2007 Urban/Rural 2007 2012 Yes/No 0 2012 Access to electricity Bhutan Poverty Assessment 2014 33 Figure 4.9  Primary School and Gas/Electricity Coverage at District Level Primary school completion (14-18 years) - 2012 Legend 50-60 60-70 70-80 80-90 Gasa 90-100 No data Lhuentse Bumthang 64% Punakha Trashi Yangtse 93% Thimphu 61% 76% 89% Paro 87% Wangdue Phodrang Trongsa 53% 58% Haa Monggar Trashigang 77% 67% 71% Zhemgang Tsirang Samtse Chhukha Dagana 74% 61% Sarpang Pema Gatshel Samdrup Jongkhar 63% 61% 66% 69% 81% 72% Coverage - Using gas and electricity for cooking - 2012 Legend 30-40 50-60 Gasa 60-70 34% 70-80 80-90 90-100 Lhuentse Bumthang 86% Punakha Trashi Yangtse 97% Thimphu 98% 90% 99% Wangdue Phodrang Paro 92% 99% Trongsa 76% Haa Monggar Trashigang 93% 74% 89% Zhemgang Samtse Tsirang Chhukha Dagana 56% 69% 71% Sarpang Pema Gatshel Samdrup Jongkhar 86% 69% 91% 84% 72% 34 Bhutan Poverty Assessment 2014 this time. Figure 4.8 shows the multidimensional equitably the available service is distributed inequalities measured by the ratio of extremes in among groups, differentiated by circumstances. circumstances. There is a clear trend of significant This approach is more robust than traditional decreasing inequalities between 2003 and 2012. ones –looking at only one dimension, the strength For instance, the ratio of urban to rural for access of the approach lies in providing a single measure to improved water has been reduced from around encapsulating both coverage and distribution of 1.2 to nearly 1.0, suggesting that households that coverage (Box 2). in urban and rural areas have nearly the same The distribution of opportunity for children access to water. Inequalities in access towater to access to education is improving in Bhutan. and sanitation showsomewhat similar reduction As indicated in Figure 4.10, the HOI for primary rates. However, the indicators for electricity, school attendance has increased significantly, gas and electricity for cooking shows clearly that reaching universal primary attendance where not all households have the same, or any, access; secondary school attendance is lower even though inequalites are fourtimes higher in the same its coverage increased from 60 percent in 2003 to dimensions between access to electricity and 87 percent in 2012. These results can be expected using gas and electricity for cooking. because the opportunity costs of sending children Across the country, the coverage of education to school at these ages are much higher at the and infrastructure varies significantly, with some secondary than at the primary level. This also district units performing relatively better than implies that some financial incentives, such as average at times. As Figure 4.9 shows, coverage conditional cash transfer programs could be more for primary school completion rates were above effective in targeting older children to stay in 50 percent for all districts in 2012, but thatthe school for learning longer. situationwas far worse for secondary schools at Despite the high coverage for attendance, the the district level. It is clear that in many districts story is much different for school completion: the Bhutan has made significant progress in meeting HOI and coverage for school completion is far some MDGs foreducation, such as universal from universal. Besides the opportunity costs, primary enrollment. In addition, reductions in there are other important and unobserved factors poverty have also contributed to the progress contributing to completing school, such as ability of improvededucation indicators. However, and determination.14 In addition, opportunities those improvements are unequal atsub-national of children to access and complete secondary levels. Similarly, in the use of gas and electricty school or higher are much more limited, since the for cooking, coverage is highest in the middle of opportunities for them to move between grades the country (more than 90 percent), a somewhat or levels are low, especially for different groups similar and the pattern to that of primary school across circumstances. completion. From the perspective of infrastructure, basic infrastructure services make significant 4.3.  Measuring Inequality of Opportunity contributions to wellbeing. Necessary services In order to analyze the variations in access such as access to improved water and sanitation across multiple circumstances, this paper makes use of the Human Opportunity Index (HOI). The 14  It is important to note that the HOI provides a lower bound HOI measures in a single indicator the coverage on the inequality prevalent in a given place, calculated by rate of a particular service, adjusted by how circumstances that are measurable and for which the data is available. Bhutan Poverty Assessment 2014 35 Box 2: The Human Opportunity Index The Human Opportunity Index (HOI) measures the availability of services that are necessary to progress in life (such as nutrition), discounted by how unequally the services are distributed among different groups in the population. Two countries that have identical coverage of nutrition services for infants, for instance, may have a different HOI if the infants that lack this service systematically share a personal circumstance beyond their control, such as gender, caste, parental income, or place of birth. Put simply, the HOI is coverage corrected for equity. In theory, it can be changed by providing more services to all (“scale effect”), or distributing services more fairly (“equalization effect”). Calculating the HOI The calculation of the HOI focuses on the dissimilarity index/inequality index, originally a demographic measure of evenness widely used in the analysis of social mobility, sociology in general and typically applied to dichotomous outcomes. Paes de Barro et al. (2009) define the dissimilarity index (D-index) as the weighted average of absolute differences of group-specific access rates (pi) from the overall average access rate (̅), or: (1) The D-index takes a value between 0 and 1. A value of zero indicates that access rates for all groups considered are the same, while positive values indicate that certain groups of individuals have a lower probability of access to the infrastructure service considered. In practical terms, the dissimilarity index reflects the percentage of the coverage rate of a particular opportunity that has to be discounted in order to obtain the HOI, i.e., (2) As equation (2) shows, the HOI can be improved either by an increase in coverage (which is still bounded at 100 percent, universal access, and so the more people have access, the less likely that a particular segment of population is being left behind), or by a closer-to-zero dissimilarity index. At higher levels of coverage for a service, there is less room for unequal distribution of access across groups. However, the dissimilarity index, as we shall see, varies across opportunities and units of analysis, even at similar levels of coverage. Caveats It is important to note that the HOI provides a lower bound on the inequality prevalent in a given place. Any calculation of the HOI can only include those circumstances that are measurable and for which data exist. Having a lower bound measure complicates comparison between countries or geographical regions, especially if the purpose is to investigate which country or region is more inequitable overall, and not which one has a lower minimum level of inequality. Illustrating the HOI by Example Consider two countries, A and B, and consider a basic opportunity such as access to primary education. Suppose that in both countries, 50 percent of all children go to school. From the perspective of overall coverage, both countries look alike. Now suppose that in country A, no girl attends school, but in country B, 50 percent of both girls and boys attend school. The HOI discounts the coverage rate of 50 percent in country A through D since access is more unequal, based on the circumstance, gender. For country B, since there is no inequality based on gender, there is no discounting, making the HOI 50 percent, or equal to the coverage. Since country B has a higher HOI, it is more equal than country A, even though the average enrollment rate is the same in both countries. 36 Bhutan Poverty Assessment 2014 Figure 4.10  Human Opportunity Index Bhutan Poverty Assessment 2014 37 “We only have a primary school that was established some 30 years ago. Upgrading of the school will have benefits such as we can sell some of our local produce to the school and we do not have to send our children to far away school. It incurs huge additional cost on transportation, living arrangements, frequent buying of school uniforms and shoes so frequently school uniforms and shoes last for many years while children can stay with us and attend classes” – A male FGD participant, Phangkhar community, Zhemgang. “There is no up gradation of the school in the Gewog due to which we have to send our children to other far off school. We face financial problems. When health facility like BHU has male staff we women face problem in discussing our health issues” – A female FGD participant, Phangkhar community, Zhemgang. have a direct impact on health status and wellbeing level, and shows significant improvement in the of all members in the household. Having access to last half-decade, with some districts performing electricity and piped water also help households considerably better than others. For example, increase their productivity, and let their children Dagana went from 12 percent in 2007 to 58 have more time for education since they don’t percent in 2012, and Trashigang from 39 percent have to collect wood and water. to 65 percent in the same period. Compared to basic education indicators, Besides the appealing feature of HOI being able infrastructure indicators show that Bhutan to summarize inequality of opportunity without has improved in providing basic infrastructure tracking different circumstances one by one, the services over the last decade. Having access HOI framework also allows the decomposition to electricity and using gas and electricity for of the contribution of particular circumstances, cooking were a great challenge in 2003, however such as education of household heads and wealth the coverage and HOI index had significantly quintile index overtime. This is very useful improved by 2012, with the HOI moving from 87 since individuals and households are typically percent to 77 percent. Having access to improved characterized by a variety of circumstances, and water and sanitation is not much of a problem it is important to know the relative importance for Bhutan, with HOIs high on average and still of the various factors in explaining inequality of increasing, and the coverage much higher than opportunity. that of other countries in the region. When looking at opportunities in education Across the country there is wide variation across the entire country, family wealth and in the performance of the various districts. For location (urban versus rural) explain a large example, the HOI for primary school completion fraction of the observed inequality. In all in 2012 in Bumthang, Thimphu, and Paro years, except 2012, the two factors combined districts were highest, at more than 80 percent, contributed at least 60 percent to the inequality, and much higher than the country as a whole, at even though their relative importance has been 65 percent. Figure 4.11 maps the HOI at district declining over the last decade. In 2012, the 38 Bhutan Poverty Assessment 2014 Figure 4.11  HOI at District Level HOI - Primary school completion (14-18 years) - 2007 Legend 30 or less 30-40 40-50 50-60 Gasa 60-70 20% Lhuentse Bumthang 33% Punakha Trashi Yangtse 49% Thimphu 41% 47% 67% Paro 65% Wangdue Phodrang Trongsa 40% 44% Haa Monggar Trashigang 56% 31% 39% Zhemgang Tsirang Samtse Chhukha Dagana 40% 31% Sarpang Pema Gatshel Samdrup Jongkhar 22% 41% 12% 35% 38% 49% HOI - Primary school completion (14-18 years) - 2012 Legend 40-50 50-60 60-70 70-80 Gasa 80-90 No data Lhuentse Bumthang 52% Punakha Trashi Yangtse 89% Thimphu 49% 68% 86% Paro 83% Wangdue Phodrang Trongsa 45% 45% Haa Monggar Trashigang 65% 59% 65% Zhemgang Tsirang Samtse Chhukha Dagana 67% 51% Sarpang Pema Gatshel Samdrup Jongkhar 56% 51% 58% 59% 75% 67% Bhutan Poverty Assessment 2014 39 Figure 4.12  Inequality Index and Contributions by Circumstances Inequalit y Index Primary school attendence (ages Primary school completion (ages Secondary school attendence (ages Secondary school completion (ages Year 6-12 years) 14-18 years 13-16 years) 20-24 years) 2003 10.1% 22.9% 15.5% 47.5% 2007 5.5% 19.9% 8.8% 34.4% 2012 1.4% 8.9% 3.2% 23.5% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% Inequality Index Inequality Index Inequality Index Inequality Index Relat ive Cont ribut ion Primary school attendence (ages Primary school completion (ages Secondary school attendence (ages Secondary school completion (ages Year 6-12 years) 14-18 years 13-16 years) 20-24 years) 2003 53.83 46.26 51.36 40.38 2007 49.18 54.74 50.68 43.81 2012 31.25 24.16 15.04 40.27 15.38 18.66 20.49 34.91 16.13 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 Contribution Contribution Contribution Contribution Circumst ance Wealth Urban Number of children Head's gender Head's employment Head's education Head's age Child gender Inequalit y Index Using gas and Access to improved Access to improved Access to improved Year Access to electricity Access to piped water electricity for cooking sanitation private sanitation water 2003 36.5% 51.3% 3.1% 4.8% 6.5% 7.4% 2007 16.4% 28.5% 1.5% 4.9% 3.8% 4.1% 2012 5.1% 8.8% 8.3% 7.3% 0.7% 1.1% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% 0% 20% 40% 60% Inequality Index Inequality Index Inequality Index Inequality Index Inequality Index Inequality Index Relat ive Cont ribut ion Using gas and Access to improved Access to improved Access to improved Year Access to electricity Access to piped water electricity for cooking sanitation private sanitation water 2003 27.9 27.3 23.8 24.2 28.8 19.7 14.5 42.7 25.2 34.6 24.3 19.8 38.9 20.2 17.8 39.7 21.1 17.2 2007 35.6 24.7 20.6 28.0 25.1 24.5 34.5 18.7 19.9 45.3 13.09.6 29.8 25.9 21.2 15.0 29.7 26.1 20.9 15.2 2012 37.2 22.3 20.9 14.2 29.9 22.7 23.9 14.7 26.2 24.9 16.0 19.5 40.7 18.2 20.1 37.4 15.7 20.0 34.0 19.3 19.9 0 50 100 0 50 100 0 50 100 0 50 100 0 50 100 0 50 100 Contribution Contribution Contribution Contribution Contribution Contribution Circumst ance District Urban Wealth Head's gender Head's employment Head's education Head's age 40 Bhutan Poverty Assessment 2014 Figure 4.13  Drivers of Change Annualized Growth Rates in HOI (Bars show Growth Rate of HOI, Line Divides Growth into that Due to Growth in Coverage and Growth in Equality) Indicator Period Primary school attendence (6-12 years) 2003-2012 4.4% Primary school completion (14-18 years) 2003-2012 8.2% Secondary school attendence (13-16 years) 2003-2012 5.5% Secondary school completion (20-24 years) 2003-2012 17 .7% 0% 5% 10% 15% 20% Annualized growth rate (percent) Annualized Growth Rates in HOI (Bars show Growth Rate of HOI, Line Divides Growth into that Due to Growth in Coverage and Growth in Equality) Indicator Period HH Access to electricity 2003-2012 13 .0% Using gas and electricity for 2003-2012 18 .5% cooking Access to improved 2003-2012 -2 .1% sanitation Access to improved 2003-2012 -1 .0% private sanitation Access to improved 2003-2012 2.4% water Access to piped 2003-2012 2.6% water Annualized growth rate (percent) Bhutan Poverty Assessment 2014 41 public or private investment in electricity and “Before, we did not have electricity. water services and projects –albeit that such Now after we received electricity environment exists mostly in urban areas. wellbeing has improved. Road has also been connected. Before, we used 4.4.  Drivers of Change horses to carry our loads. Previously For most opportunities in education and we had to travel long way even to get infrastructure the main driver of change has been kerosene for lighting purposes.” increased coverage, but not much in the way of equity-enhancing programs. Figure 4.12 shows the annualized growth rates in HOI over the contribution of a household head’s education is last decade; growth decomposable into growth becoming an important factor, being now the in coverage and growth in equality. Notably, third-largest contributor to unequal opportunities in education and infrastructure most of the in education. Other factors that seem to growth in these indicators has come from “scale matter include the number of children in the effect” improvements; in other words increased household, and children’s gender for secondary coverage, as opposed to the targeting of specific school attendance and completion. The fact that groups or circumstances. On the other hand, for circumstances matter less and less for school access to electricity and use of gas and electricity attendance and completion across the country is for cooking indicators a significant proportion good news. It speaks to the perspective of school of the improvements in HOI has come from universal policies over the last decade. Those policies targeted at underserved groups rather policies are expanding and improving inputs, than increased overall coverage, which indicates such as hiring quality teachers, and building more a more fair distribution of services (the so-called and better school infrastructure that ultimately “equalization effect”). close the gap of unreached population. The driver of change is much the same at the From the perspective of infrastructure, it district level, where most of the HOI growth is the location (district and urban/rural) and has come from growth in coverage. Figure 4.14 family wealth that explains over 70 percent of the shows the HOI growth maps for primary school inequality of opportunity across infrastructure completion and access to electricity in 2007 indicators. This result is consistent with findings and 2012. Numbers at the districts first denote in other countries in the region, where factors of growth of HOI, and secondly growth of coverage. location and family wealth (measured by quintiles It shows that in most of the districts there has of per capita household expenditure) are the most been improvement in HOI, although there are important circumstances. For most indicators, the still large variations in the magnitude of these inequality index has declined significantly over improvements across districts. The figure also the last decade, by four-to-six-fold, suggesting shows a large proportion of HOI growth has been that access to basic infrastructure indicators is due to coverage growth in spite of its unequal becoming more equal by different circumstances distribution. (from 36% to 5% for access to electricity and from 7% to 1% for access to piped water). This is due to an enabling environment that encourages 42 Bhutan Poverty Assessment 2014 Figure 4.14  Coverage Growth has been the Main Driver of HOI Growth Annualized growth in HOI (%) - Primary school completion (14-18 years) - (2007-2012) Legend 5 or less 5-10 10-15 15-20 Gasa 30-35 Null Lhuentse Bumthang 9.0%/ Punakha Trashi Yangtse 12.0%/ 7.7% Thimphu 3.3%/ 7.2%/ 8.9% 5.0%/ 2.5% 6.3% Paro 3.3% 4.9%/ Wangdue Phodrang Trongsa 2.9% 2.2%/ 0.6%/ Haa 1.5% 0.5% Monggar Trashigang 2.8%/ 3.1% 13.1%/ 10.4%/ 10.2% 7.4% Zhemgang Tsirang Samtse Chhukha Dagana 10.1%/ 10.1%/ Sarpang Pema Gatshel Samdrup Jongkhar 18.7%/ 4.6%/ 32.0%/ 8.1% 7.5% 10.5%/ 13.4%/ 6.2%/ 12.4% 2.2% 21.9% 7.6% 10.4% 4.1% Annualized growth in HOI (%) - Access to electricity - (2007-2012) Legend 5 or less 5-10 10-15 15-20 Gasa 20-25 17.1%/ 40-50 11.5% Null Lhuentse Bumthang 15.6%/ Punakha Trashi Yangtse 8.5%/ 13.0% Thimphu 6.0%/ 6.0% 0.3%/ 5.5% Paro 0.2% Wangdue Phodrang Trongsa 6.8%/ 17.8%/ Haa 5.4% 13.2% Monggar Trashigang 6.8%/ 5.8%/ 5.5% 4.7% Zhemgang Tsirang Samtse Chhukha Dagana 12.9%/ 42.1%/ Sarpang Pema Gatshel Samdrup Jongkhar 11.5%/ 2.7%/ 20.2%/ 9.5% 31.8% 8.5%/ 14.5%/ 11.5%/ 8.2% 1.7% 16.6% 6.0% 11.9% 7.8% Bhutan Poverty Assessment 2014 43 5 Chapter Key Drivers of Poverty Reduction in Bhutan This chapter uses four alternative approaches to understanding the drivers of Bhutan’s rapid “We have access to services like RNR. poverty reduction, since there is no direct, Through a group formation such as comparable data available on the sources of vegetable group has really helped household income for two points in time. The us. Provision of services is mainly first, economy-wide approach is to analyze the through technical assistance like key factors that have driven growth and their marketing, harvesting, provision of potential pro-poor bias. The second approach seeds, fertilizers, etc.” is to construct a synthetic panel using cross- section data in the BLSS 2007 and 2012 to determine the profiles of people who escaped poverty. The third approach is to identify key “At the household level over the drivers of change from an in-depth review of last few years definitely there is focus group discussions. Finally, an econometric improvement in income. Many estimation of welfare change at the percentile households now sell dairy products in level is analyzed in order to shed light on the the local market although it may be contribution from the accumulation of assets in small quantity.” and returns on assets. During 2007-2012, rapid poverty reduction has been driven mostly by growth in per-capita Table 5.1  Shapley Decomposition of Change in Poverty in consumption expenditure than redistribution. Bhutan, 2007-2012 For the whole of Bhutan, by every poverty Overall Growth Redistribution measure, growth in per-capita consumption Headcount -11.16 -10.66 -0.05 has been the dominant factor behind poverty Poverty Gap -3.45 -3.28 -0.06 reduction (Table 5.1). One notable exception is Squared Poverty Gap -1.38 -1.35 -0.05 Pema Gatshel where growth was negative, but the Watts Index -4.50 -4.31 -0.01 reduction in inequality there largely compensated for this adverse effect. In Lhuentse and Trashi Source: World Bank staff estimates Bhutan Poverty Assessment 2014 45 Figure 5.1  Decomposition of Change in Poverty, by dzongkhag Monggar Zhemgang Samtse Trashigang Samdrup Jongkhar Sarpang Lhuentse Chhukha Bumthang Trongsa Haa Dagana Punakha Wangdue Phodrang Gasa Paro Thimpu Trashi Yangtse Pema Gatshel Tsirang Growth Redistribution Source: World Bank staff estimates Yangtse, meanwhile, the redistribution effect Table 5.2  Poverty Outcomes in Bhutan, by Area of Residence, 2007 on poverty was adverse enough to undercut the positive effects of growth (Figure 5.1). Squared Poverty Population Distribution Much of the poverty reduction in Bhutan Headcount Poverty Gap Share of the Poor Gap has taken place in the rural areas. Rural Bhutan Urban 1.4 0.3 0.1 26.0 1.6 cut poverty by more than half between 2007 and Rural 31.1 8.1 3.0 74.0 98.4 2012; the more distributionally sensitive the Bhutan 23.3 6.1 2.3 100.0 100.0 measure, the greater was the reduction in poverty. Contrastingly, poverty increased slightly, by all Source: World Bank staff estimates measures in urban Bhutan (Tables 5.2 and 5.3). Rural-urban migration has played a part in the slight urban poverty increase; a migration rate of working life starting at the bottom of economic about 1.2 percent a year out of rural Bhutan has class in urban areas. The number of poor increased swelled the urban population. Most migrants are in urban Bhutan by 800 persons while in the rural likely to be young and in the early part of their areas it dropped by 77,000. 46 Bhutan Poverty Assessment 2014 Table 5.3  Poverty Outcomes in Bhutan, by Area of as the main source of income in 2007, by 2012 Residence, 2012 it had shifted to wages and salaries, although we Squared cannot distinguish if this came from agriculture Poverty Population Distribution Headcount Poverty or non-agriculture sectors. Gap Share of the Poor Gap Urban 1.8 0.3 0.1 34.0 3.1 There are three key drivers behind Rural 16.7 3.6 1.2 66.0 96.9 the dynamism of rural Bhutan: increasing Bhutan 12.0 2.6 0.9 100.0 100.0 agricultural trade, expanding road networks, and spreading spillovers from hydroelectric projects. Source: World Bank staff estimates 5.1.  Trading Out of Poverty Table 5.4  Bottom Quintile’s First-Ranked Income Source (Percentage of Rural Households) Despite the limited land available for 2007 2012 cultivation, land productivity in value terms Wages and salaries 18.5 41.5 has been rising in Bhutan. Bhutan has less Own business 4.8 8.1 than 3 percent of land area under cultivation, Own farm 63.5 34.3 and this may have been decreasing.16However, Remittance 2.1 6.8 crop production per land area has been rising Pension 0.0 0.3 in constant prices. The FAO (2014) estimates 8 Real estate / rent 2.1 0.4 percent annual growth in crop production per all others 9.0 8.5 hectare over 2006 to 2011, on top of a 7 percent 100.0 100.0 annual increase over the preceding five-year period.17Moreover, real GDP from agriculture Source: World Bank staff estimates shows lackluster growth over the same period. With production shifting fast to high-value crops, the GDP for agriculture which is based on 2000 Since most poverty reduction has been driven figures, is likely to underestimate the real GDP by growth and took place in the rural areas (93 contribution from agriculture. percent, with the population shift to lower- A shift to high-value commercial crops poverty urban areas contributing 13 percent15), has been an important reason for the rise in we examine, in the rest of this chapter, the production value per acre. Spurred by trade factors that could have been responsible for the agreements with India and Bangladesh, Bhutan transformation of the rural sector. has been shifting to crops more suited to its Most rural households in the bottom comparative advantage. The area under cereals quintile now rank wages and salaries as their has been on the decline with substitution by fruit main source of income. It appears that most of and vegetable crops. Further, the commercial the employment expansion in the rural sector has been pro-poor. In the bottom quintile, there 16  has been a dramatic shift in the main income A time-trend for cultivated areas is hard to assess because assessment methodology has improved. Factors such as fluctuation source between 2007 and 2012. While own farm in snow cover, urbanization, and fallowing of tseri land have also contributed to the altering of land-cover figures, according to the enterprise was identified by the bottom quintile Land Cover Mapping Project, Ministry of Agriculture and Forests, 2010. 17  FAOSTAT country profile for Bhutan was accessed on February 7, 15  The interaction term increased poverty by 6 percent. http://faostat.fao.org/site/666/default.aspx Bhutan Poverty Assessment 2014 47 Table 5.5  Share in Value of Total Production   2008 2009 2010 2011 2012 Cereals 28.7 24.8 19.8 17.5 15.0 Vegetables 15.1 17.1 10.8 13.7 12.1 Fruits & nuts 21.4 30.4 26.9 30.7 31.9 Meats 2.2 1.6 1.7 12.5 11.4 Eggs 0.6 0.9 18.0 5.7 11.8 Others 32.0 25.2 22.8 19.9 17.8 Value of Prod’n. (Nu m) 9,232 9,875 11,708 14,162 15,450 Source: World Bank staff estimates based on Renewable Natural Resource Statistics, 2012 Box 3: The Mountain Hazelnut Project Mountain Hazelnut Project is the first 100 percent FDI in Bhutan. In terms of the structure, they currently have a Holding Company in the British Virgin Islands, with a sub-entity in Hongkong and the operating entity in Bhutan. The company now has 25,000 acres planted small holder hazelnuts in 5 Dzongkhags in Bhutan and expanding to Punakha and Wangdue Phodrang in the west and Zhemgang in the center. For each of the last few years they have been planting over a million trees (plan was to have 10 million trees producing 40,000 tonnes of hazelnut). The first batch of trees has started to produce and they seem to have a very stable management team. The supply chain of seedlings as well as the extension / tree distribution system / monitoring seems to be well developed. The company provides 500 jobs (directly and to significant number of women) and impacts welfare in areas with about 15percent of Bhutan’s population. crops fetch higher value per acre. The proportion 16 land, air, and sea routes. In 2008, Bangladesh of cereals has nearly halved in value and that of signed a 10-year bilateral trade agreement with fruits and vegetables is now triple the proportion Bhutan, opening new routes through eastern of cereals (Table 5.5). Bhutan and expanding preferences to Bhutan Agricultural exports are rising fast, for duty-free exports for 74 items, of which facilitated by trade agreements. Agricultural 49 are agriculture-based. It is estimated that exports have grown at 22 percent annually for key exports from Bhutan (oranges, apples, while non-agricultural exports have stagnated. and cardamom) the preference margin is large, Electricity exports have also been flat with no exceeding 50 percent in 2012/13,18 although this new hydroelectric plant coming on stream during could be eroded when SAARC tariffs fall under the period. Bhutan renewed its 10-year free trade agreement with India, the largest trading 18  Based on Bangladesh Tariff schedule 2012/13 including partner, in 2006, ensuring tariff-free trade across customs duty, supplementary duty, regulatory duty, advances income tax, value added tax, and advanced trade VAT. Custom duty alone is 25 percent. 48 Bhutan Poverty Assessment 2014 Figure 5.2  Rising Agricultural Exports from Bhutan Box 4:  The “One Gewog Three 1,800 9.0 Products” Initiative 1,600 8.0 1,400 7.0 “One Gewog Three Products” (OGTP) 1,200 6.0 is an initiative to achieve food self- Million Nu Percent 1,000 5.0 800 4.0 sufficiency and poverty alleviation 600 3.0 through large-scale production of 400 2.0 200 1.0 at least three different renewable 0 0.0 natural resource products. As a part 2007 2008 2009 2010 2011 2012 of the initiative, every gewog identifies Share of non-energy exports between one and three commodities Agricultural exports for production and marketing, based on market availability and potential Source: Data from Annual Report 2012/13 of the Royal Monetary Authority of Bhutan production. The basic idea is to capitalize on the Table 5.6  Export Crops are Pro-Poor potential of each region in order to Decile Proportion of Farmers Sales as percent of Cash enhance food self-sufficiency and Income rural livelihoods. Indicative evidence   Oranges Apples Potatoes Oranges Apples Potatoes suggests that the OGTP initiative has 1 47.1 2.6 45.5 57.1 21.6 16.8 played an important role in boosting 2 46.5 3.6 47.8 59.7 21.9 19.3 the product diversification of the 3 46.6 4.0 48.6 62.8 26.3 19.4 country. It has provided farmers 4 46.0 4.3 45.9 61.9 17.6 19.5 5 44.1 5.1 45.5 62.9 9.5 20.7 with more opportunities to improve 6 42.5 6.8 45.6 61.4 23.5 25.9 their living standards. For example, 7 41.1 7.6 46.6 62.2 30.5 24.2 farmers in the Dunglagang gewog of the 8 37.9 11.2 45.2 65.4 27.9 28.9 Tsirang dzongkhag decided on broiler 9 30.1 18.1 47.0 62.2 33.5 32.0 production as a part of their OGTP 10 17.5 34.1 46.7 63.0 43.0 29.9 initiative, and targeted the production of 8 metric tons of chicken per year, Source: Based on Renewable Natural Resource Census, 2008 Note: Deciles are based on household-purchased food expenditure. which they achieved easily during the 10th FYP; in FY 2009/10 they sold the South Asian Free Trade Agreement. But 9.97t of chicken. Their total profit Bhutan has the time to further consolidate its earned from chickens by June 2010 was exports to Bangladesh. Nu 319,800. The promotion of poultry Principal export crops are pro-poor. The farming with this initiative has brought principal export crops of oranges, potatoes an increase in farming households’ and apples account for nearly two-thirds of all income and a simultaneous reduction in agricultural exports from Bhutan. Oranges are poultry imports to the country. exported mostly to Bangladesh and potatoes Bhutan Poverty Assessment 2014 49 to India. Among these crops, half or more of carefully. High quality rural roads reduce poverty, the farmers in poorer deciles grow oranges and raise consumption, provide access to off-farm jobs, potatoes, with apples a distant third. A significant and increase school enrolment and completion share of the cash income of the farmers in poorer rates (Dercon et al., 2009; Yamauchi et al., 2009). deciles is sourced from these export crops Khandker et al. (2009) investigated the impacts (Table 5.6), with orange farmers in poorer deciles of rural road projects using household-level panel reaping close to 60 percent of their cash income data from Bangladesh. Findings suggest that rural from the crop. The poverty incidence has fallen road investments reduce poverty significantly sharply, with close to 50 percent of households through higher agricultural production, higher engaged in agriculture, second only to those wages, lower input and transportation costs, and engaged in construction. higher output prices. Rural poverty incidence The 10th Five Year Plan has been an important in Laos declined by 9.5 percent of the rural aid to the agricultural sector. The strategies population between 1997/98 and 2002/03, and of the plan give impetus to the commercial approximately 13 percent of this decline can orientation of agriculture by aiming to: be attributed to improved road access (Warr, 2010). The road investment gains were reported a. Enhance sustainable rural livelihoods proportionately higher for the poor than for the through improved agricultural and live- non-poor, in other words road investments are stock productivity and expansion of pro-poor. Evidence from Sri Lanka, Indonesia, commercial prospects of agriculture and and the Philippines confirm that the poor and other natural resource endowments; very poor benefit, substantially so, from rural a. Conserve and promote sustainable com- roads (Hettige, 2006). In India, the expenditure mercial utilization of forest and water on roads was found to have the largest impact resources; on rural poverty compared to other types of b. Promote sustainable utilization of arable public expenditure. For every 1 million rupees agriculture and pasture land resources; (US$22,000) invested in rural roads, 163 people c. Enhance food security through sustain- were lifted out of poverty. In Vietnam, for every able and enhanced food production and dong invested in roads, the value of agricultural availability, improved access to food and production would increase by three dongs (World enabling effective distribution, market- Bank, 2010). ing and import of food; and Bhutan embarked on big programs of road d. Transform subsistence agriculture to infrastructure building in the 10th Five Year small-scale commercial agriculture with- Plan (2008-2013). There were two initiatives in out compromising food security. “One the road sector one from the Ministry of Works Gewog Three Products” (OGTP) is a and Human Settlement and the other on rural declared policy to diversify production access under the Ministry of Agriculture and activities at the gewog level (Box 4). Forests: a. Ensure that 85 percent of the rural 5.2.  Roads Out of Poverty population lives within a half-day’s walk International evidence on the impact of from the nearest road; rural roads on poverty is positive, although b. Connect Phuentsholing to Samtse thro­ attribution problems need to be handled ugh the Southern East-West Highway; 50 Bhutan Poverty Assessment 2014 “Our vegetables are not competitive against the one imported from border town of India because it is said that our cabbages contains lots of water inside, the cabbages are not green, the leaves are yellowish in colour. Owing to poor quality of road, vegetables get damaged while transporting them to longer distance” – A male FGD participant, Drujeygang gewog, Dagana. Figure 5.3  Highways Added, 2007-2012 Dzongkhag Boundry Road Network Gasa Lhuentse Thimphu Punakha Bumthang Trashi Yangtse Paro Trongsa Wangdue Phodrang Haa Trashigang Monggar Dagana Tsirang Sarpang Samtse Chhukha Zhemgang Samdrup Jongkhar Pema Gatshel Source: NSB c. Construct the Lhamoizinkha-Dagana roads to facilitate access to hydropower highway (75 km); projects. d. Complete construction of two north- The new highways and farm roads benefit south highways, Gyelpoizhing-Nganglam the poor. The highways built in the past five (64.3km) and Gomphu-Panbang (56 km); years connect some of the lagging and poorer and dzongkhags to the Southern East-West highway. e. Construct (28 km) and upgrade (329 km) This road links and helps by cutting travel times Bhutan Poverty Assessment 2014 51 Figure 5.4  Increase in Farm Road Density, 2007-2012 and Initial Population Density of Poor, 2007 25 Samtse 20 Density of poor, number per area (Sq km) 15 Chhukha Monggar 10 Samdrup Jongkhar Trashigang Pema Gatshel Sarpang 5 Zhemgang Thimphu Tsirang Dagana Punakha Lhuentse Trongsa Trashi Yangtse Haa Wangdue Phodrang Paro 0 Gasa Bumthang 0 5 10 15 20 25 30 35 40 45 50 Increase in farm road density 2007-2012, kms per area Source: World Bank staff estimates to cross-border trade with India and Bangladesh The road expansion targets have been and to the central and western parts of Bhutan. over-achieved in terms of improving access to In Zhemgang dzongkhag, the poorest of all the poor of rural communities. One-quarter dzongkhags in 2007, at 58 percent, poverty fell of the 10th Five Year Plan’s capital expenditures by more than half. Focus group discussants have been on the roads sector, and amount to a highlighted particularly the roads opened cumulative 18 percent of GDP. Nearly 4,000 kms from Gomphu to Panbang as improving their of farm roads have been built (at a norm rate of market access to Gelephu, the trading post with Nu 3 million per km) – an eight-fold increase from India. The Samtse-Phuentsholing connection the baseline of 2007. The farm road density has eliminates the erstwhile 2.5 hour detour through increased dramatically across all dzongkhags and India, now reachable in one hour for the poorer a pro-poor bias is noticeable (Figure 5.4). Focus residents of Samtse, which has twice the national group discussants valued road construction as the average incidence of poverty. The Dagana- most beneficial in creating jobs in construction, Lhamoizinkha connection would benefit the poor and improving access to markets, schools, and in Dagana, a laggard in poverty reduction, with health centers. It appears that per km expenditure prosperous Chhukha, and give it easier access to on roads may have been less at Nu 1.5 per km thus cross-border trade. far. With a job-generation norm of about one job 52 Bhutan Poverty Assessment 2014 Table 5.7  Hydroelectric Projects under Construction, 2007-2012 Investment Project Start End Location Nu billion Punatsangchhu-I 35.15 2006 2016 Wangdue Punatsangchhu-II 37.78 2010 2017 Wangdue Mangdechhu 33.82 2010 2017 Trongsa Dagachhu 12.0 2009 2013 Dagana Source: Ministry of Economic Affairs (the investment cost is as per the Detailed Project Report and does not account for the cost-of-completion of projects) for every US$1500 US$19 spent, the cumulative Some positive benefit to all the local suppliers, short-term jobs created could be around 65,000 even at the lower end, can be expected – though between 2007 and 2012. It is likely that much this is hard to quantify. of the job generation could have benefited to Bhutanese labour local to the communities. It is 5.4.  Who were Better Able to Escape also likely that working age poor could have been Poverty between 2007 and 2012? taken on the manual labour of road construction. Using a synthetic panel to look at the profile of those who showed greater upward mobility 5.3.  The Hydro Effect than the average for the country as a whole, we Spillover effects from hydroelectric projects find that residence, gender, and education matter serve also to boost local economies in the most. Urban residents, residents in Thimphu, project site dzongkhags. During 2007-2012, Paro (capital region), Punakha, Wangdue, and work on four hydroelectric power projects was Trongsa (proximity to hydroelectric power plant ongoing (Table 5.7). location) and those employed in service sector The expenditure phasing of these project had greater chances of escaping poverty. Female- imply about Nu 13 billion spent in the year 2012. headed households had a better chance of upward Though most of the machinery and equipment mobility. Education was the by far the best vehicle are imported, construction work is expected to to exit poverty of all the characteristics. have significant local expenditures for transport services and for the foreign workers who live in Bhutan. As of 2012, there was one foreign worker “Some households also are into non- for every eight Bhutanese counterparts. Hosting farm activities by being contractors the living expenses of about 50,000 foreign and carpenters . Some household workers (7% of population), valued at the poverty members are employed nearby and line, amounts to Nu one billion per year. This is some even are working in PHPA a significant boost to the local economies of the hydropower project area and they dzongkhags. The estimated total consumption of do send money to the households (as the three dzongkhags was about Nu three billion remittances)” in 2012, and the presence of foreign workers is estimated to augment spending by one-third. 19  Citation awaiting permission. Bhutan Poverty Assessment 2014 53 Figure 5.5  Upward Mobility Based on Synthetic Panels, 2007-2012 90 80 Percentage (%) 70 60 50 Rural Urban Female Male 8 Yrs or less Regular paid Casual paid Unpaid family worker Self-employed No education 9-12 Yrs >12 Yrs No remitance Received remitance Landless 2 Acres or less 2-5 Acres 5 Acres or more Agriculture Manufacturing Services Bumthang Chhukha Dagana Gasa Haa Lhuentse Monggar Paro Pema Gatshel Punakha Samdrup Jongkhar Samtse Sarpang Thimphu Trashigang Trashi Yangtse Trongsa Tsirang Wangdue Phodrang Zhemgang Population Groups Less than average More than average Note: dashed line is national average Source: Author’s calculations 5.5.  The Main Drivers of Poverty investment in building infrastructure (farm Reduction from People’s Perspective road, electricity, mobile connectivity) in the 10th FYP enhanced connectivity, and accessibility. It In all of the four Dzongkhags visited, focus also helped to diversify activity into non-farm group discussants spoke of positive changes sectors as people could earn income by working to their livelihood in the past five years. Public as daily wage workers, small contractors, and as shopkeepers. Households could also earn cash by selling whatever they could produce, such “Road and electricity are the two as vegetables. The Agriculture Ministry’s RNR main things we received. We extension services, provision of better seeds, perceive development because we and technical support in marketing of vegetables have drinking water and produce helped farmers in raising their earnings. vegetables. Electricity has also many Dzongkhags that experienced rapid poverty benefits like weaving can be done reduction (Zhemgang and Lhuentse): during night. We don’t have to fell Zhemgang • Cash income as daily wage workers down trees and we save time for on the roadside. Local people both other chores” skilled and unskilled employed by small contactors engaged in construc- 54 Bhutan Poverty Assessment 2014 tion of houses, building of roads, etc. transformation that underlies development • Households earned cash income by and ultimately drives distributional change. In selling vegetables in the community, particular, Bourguignon, Ferreira and Lustig such as to the staff in the gewog and (2005) identify three key forces that determine agriculture department RNR centers, observed changes in the distribution of living and by supplying to boarding schools, standards: (i) endowment effects or population teachers, and hospital staff. effects due to changes in socio-demographic • In lower Zhemgang (lower Kheng), characteristics of the population (e.g., area of oranges are the main cash crop. It was residence, age, education, and ownership of not affected by pests or diseases as physical and financial assets, (ii) price effects due badly as in Dagana and Pema Gatshel. changes in returns on factors of production, and The construction of the national (iii) occupational effects due to changes in the highway brought opportunities occupational structure of the population. which greatly increased the market Given data and other constraints, the focus accessibility for orange as well is on two types of effects: the endowment effect employment opportunities as daily and the effect associated with changes in returns wage workers. on those endowments. The latter effect is also Lhuentse known as the structural effect. In this reduced- • Accessibility has improved due to form framework, it is likely that the behavioral construction of farm roads and other effect is mixed up with the price effect. Household development activities. per capita expenditure is our living standard Dzongkhags that experienced slow poverty indicator, and is a function of both observable reduction (Dagana and Pema Gatshel): and non-observable individual or household • Overall improvement in income in the characteristics. By applying the Oaxaca-Blinder last five years due to developments in method20 to decompose the growth incidence infrastructure, farm roads, accessibility curve (GIC) into a component associated with to health, education, and RNR extension the endowment effect and another related to services. the structural effect. This decomposition entails • Increasing opportunities in the non- running unconditional quintile regressions (Firpo farm sectors as wage workers, small et al., 2009) for the first 99 percentiles of the contractors, etc. distribution of log per capita expenditure. The broad categories of characteristics 5.6.  Better Returns on Individual’s considered includes: (i) Demographics (e.g., age, Assets Underpin Faster Reduction in marital status, female-headed household, and Poverty household size); (ii) Household and community The living standard of an individual is a pay-off 20  from participating in the life of society. This pay- The standard Oaxaca-Blinder decomposition seeks to decompose changes in the unconditional mean outcome into off is a function of endowments, behavior and the the composition or endowment effect, and the price or structural effects. The method relies on the conditional expectation function circumstances that determine the returns on these (CEF) to summarize the relationship between individual outcomes and individual characteristics, and then on the law of iterated endowments from any social interaction. Changes expectations to link the unconditional mean to characteristics. in these elements are the result of the structural Fortin et al. (2011) have extended this logic to the decomposition of changes in other distributional statistics beyond the unconditional mean, such as quantiles. Bhutan Poverty Assessment 2014 55 assets (e.g., years of education, durable goods, Figure 5.6  A Decomposition of Growth Incidence in Bhutan, 2007-2012 such as fridge, electric iron, TV, etc., land ownership, ownership of livestock, distance to 40 nearest agricultural extension service center, 30 Annual Growth Rate (%) Endowment Effect distance to nearest hospital, distance to nearest 20 tarred road, distance to nearest feeder road, Overall Incidence distance to dzongkhag headquarters, and distance 10 to nearest bank); (iii) Sector of employment (e.g., 0 Structural Effect primary, secondary, non-public services, public -10 sector, non-paid labour); and (iv) Area/dzongkhag -20 of residence.21 Durables are included among the 0 10 20 30 40 50 60 70 80 90 100 characteristics because they are excluded from Expenditure Percentiles consumption expenditure (see RGoB, 2013). Source: Author’s calculations We therefore use the analogy between growth accounting and the counterfactual decomposition of the distribution. The endowment effect has of the GIC considered to link the endowment roughly an inverted U-shape. It is upward sloping effect to notion of accumulation (of factors of up to the 77th percentile and therefore increases production) and we take the structural effect to inequality over much of the distribution. The be an indicator of productivity in socioeconomic structural effect dominates the endowment effect interaction. Accumulation and productivity are at the low end of the distribution up to the 28th indeed the two basic ideas that structure the percentile. It turns negative between the 60th and study of economic growth. 95th percentile. The endowment effect is mostly positive and overwhelms the structural effect 5.7.  Composition versus Structure past the 28th percentile. The configuration of Figure 5.6 shows a decomposition of the the three curves presented in Figure 5.6 implies total variation in the distribution of log per that the level of the GIC is determined mainly capita expenditure (essentially the GIC) into by the composition or endowment effect. In two components. The first component is due to particular, the gains achieved by people located changes in the distribution of characteristics at the bottom of the distribution up to the 28th while the second represents the contribution of percentile are due to the structural effect while changes in the distribution of returns to those the gains beyond that point are due mainly to characteristics. The structural effect is roughly U-shaped. The fact that it is downward sloping up to the “The opening of school has overall 77th percentile means that the structural effect benefitted the community. reduces inequality in that part of the distribution and tends to increase it in the upper segment Previously we have to send our children to school which were far off. Sending children far off places adds 21  Our choice of dummy variables implies that the reference case (conditional on characteristics represented by continuous variables) to additional expenditure on travel, is landless, does not own any of the durables listed in the equation, living arrangement cost” resides in Thimphu in a male-headed household, and the sector of employment is listed as other. 56 Bhutan Poverty Assessment 2014 Figure 5.7  Accounting for the Endowment and Structural Effect 30 80 Overall Endowment Effect 25 60 Annual Growth Rate (%) Annual Growth Rate (%) 20 40 15 Reference Group 10 20 5 Durable Goods Overall Structural Effect 0 0 -20 -5 Demographics -10 -40 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Source: Author’s calculations Figure 5.8  The Effects of Land Ownership Endowment Effect Structural Effect 1.6 8 1.2 Ownership of more than 5 acres of land 7 Annual Growth Rate (%) Annual Growth Rate (%) 0.8 6 0.4 5 0.0 4 Five acres or less -0.4 -0.8 Ownership of 5 acres of land or less 3 -1.2 2 -1.6 1 -2.0 More than five acres 0 -2.4 0 10 20 30 40 50 60 70 80 90 100 -1 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Expenditure Percentiles Source: Author’s calculations the composition effect. The pro-poorness of the basis of sets of covariates. Figure 5.7 shows the distribution change that occurred in 2007-2012 key covariates that shape both the endowment is due mainly to the structural effect while the effect and the structural effect. The left panel of increase in inequality observed over the same Figure 5.7 compares the full composition effect to period is driven by the endowment effect. Since the contribution of ownership of durable goods. the structural effect represents the change reward It is evident that these characteristics are the for participation in socioeconomic arrangements, main drivers of the composition effect. The right these results suggest that socioeconomic panel compares the overall structural effect and arrangements in Bhutan may have gradually the contributions of household demographics become more progressive. and the coefficient of the reference group. To determine the factors driving both the These results show that both the level and the composition and structural effects we further dispersion of the full structural effect are closely disaggregate these two components on the tracked by household demographics. A further Bhutan Poverty Assessment 2014 57 Figure 5.9  The Effects of Years of Schooling “The gewog as a community has 1.4 improved in terms of accessibility in 1.2 Annual Growth Rate (%) the last five years. It has access to 1.0 Structural Effect 0.8 motorable road and all chiwogs are 0.6 connected by farm road. However, 0.4 the farm road is not accessible 0.2 throughout the year especially 0.0 Endowment Effect during monsoon season because of -0.2 0 10 20 30 40 50 60 70 80 90 100 landslides.” Expenditure Percentiles Source: Author’s calculations decomposition, not shown here, revealed that the ownership is concerned, the configuration of key driver is the household size. the endowment effects for small and large land While ownership of durable goods and the holdings shown on the left panel is due to two household demographics can certainly serve as facts: (1) the returns to land were negative in 2007; targeting variables in the formulation of policy (2) small land holdings increased between 2007 interventions, it is useful to consider the effects of and 2012 (most likely due to land redistribution) some other covariates that are directly subject to while large land holdings decreased. That is why intervention. Focusing on a couple of productive the composition effect of small land holdings assets: land ownership and years of education, the is negative while that of large land holdings is findings from a recent participatory assessment positive. The structural effect for both types list small land holding as a key constraint to of landholdings is shown on the right panel of achieving economies of scale in agricultural Figure 5.8. The returns on both types of holdings production. The Royal Government has also increased over time. While the overall structural granted land to about 61,339 beneficiaries (Over effect tends to dampen inequality, the structural 1 acre per head) under the Kidu program for effect of land ownership increases inequality. socio-economically disadvantaged groups during Figure 5.9 shows that the structural effect of 2009-2013 (National Land Commission). Finally, years of education dominates the endowment while acknowledging important achievement in effect across all quintiles. Both effects are more the domain of education, the 10th Five Year Plan significant beyond the median. This demonstrates deplores the fact that low adult literacy constrains clearly that schooling is a contributing factor to improvement in the HDI. inequality. While returns on years of education The endowment and structural effects have increased over time, note that in the lower of land ownership and of years of schooling are half of the distribution it is lower. presented in Figures 5.8 and 5.9. As far as land 58 Bhutan Poverty Assessment 2014 Bhutan Poverty Assessment 2014 59 6 Chapter Poverty Reduction in Bhutan: Sustainability, Vulnerability and Suggested Remediation The current pace of poverty reduction these fruits in the north-east region of the Indian appears sustainable in the medium term. sub-continent and the trade logistical gains made The policy foundations of the current poverty by then should be able to sustain fruit exports reduction achievement lie in trade intensification to Bangladesh. In addition, tariff reduction with neighbors, rural infrastructure expansion, under SAFTA is scheduled on a graduated annual and development impetus from hydroelectric basis over eight years. India’s north-east region power. At the household level, the emphasis on (Arunachal Pradesh, Assam, Manipur, Meghalaya, education of children seen across all economic Mizoram, Nagaland, and Tripura) with an agro- classes is a welcome sign that families will be climate similar to Bhutan’s is likely to become a able to seek out these opportunities to escape potential competitor in horticultural products. poverty. But growth dynamism in India and Bangladesh Trade intensification with neighbors is set should be able to accommodate expansion of to continue. The existing, 10-year free trade differentiated products from Bhutan. agreement with India is due for renewal in 2016, The infrastructure within and beyond and this is likely to be concluded. The bilateral Bhutan is set for significant improvement. agreement with Bangladesh, which has benefited The 20-year road sector master plan for Bhutan, Bhutan by preferential duty free access to 74 2007-2027, has a prioritized plan for expansion mostly agricultural exports, is due for renewal of improving inter-Dzongkhag connectivity in 2018. In addition, bilateral agreements (feeder roads, highways) and the completion of with Thailand and Nepal are also on the anvil. the Southern East-West Highway. An additional Under the South Asian Free Trade Agreement, 2,654 km of feeder roads (connecting gewogs Bangladesh is grouped with Bhutan as “least to dzongkhag headquarters), 537 km of inter- developed members” with slower pace of tariff dzongkhag highways, including tunnels, and reduction to 0-5 percent by 2016, and this could 794 km of Southern East-West highway are erode the 15 percent preference margin currently planned to be completed. These developments in enjoyed by Bhutan in orange and apple exports to conjunction with the 4,000 km expansion, “from Bangladesh. However, Bhutan is a net exporter of the bottom up”– farm roads first –during the 10th Bhutan Poverty Assessment 2014 61 “The income from the oranges has gone down from average Nu 50,000 to Nu 20,000 so we do not know what to do next. May be we should plant mountain hazelnut as an alternative. Heard that a weather condition of our area is similar to that of Lingmithang in Monggar and it might work here” – A male FGD participant, Shumar, Pema Gatshel “Many orange orchards are damaged by disease these days. Now oranges are not even available to eat. First it affected the trees in Denchi village and shifted upwards. Many people of our locality believe that dust from Gypsum powder factory leads to dying of the crops as well as orange” – A female FGD participant, Shumar, Pema Gatshel “I didn’t see the insect but the root of the orange tree has been damaged”– A female FGD participant, Shumar, Pema Gatshel. “According to the agriculture sector the solution to the disease is after many rounds of discussion we have been advised to cut down all the orange trees even if all the trees in the orchard are not affected. If even one tree is affected rest of the trees also have to be cut down and burnt. The government is providing free orange saplings. Now farmers are apprehensive to the advice because the question is how they would manage without income until the new trees start bearing fruits. It takes at least five years to start bearing fruits” – A key informant, Shumar, Pema Gatshel “Because of water scarcity we depend on monsoon rain to transplant paddy. When there is monsoon everyone in the village start transplanting paddy and therefore we cannot exchange labour and sometimes we have to keep our land fallow” – A male participant, Nangkhor community, Zhemgang FYP, if built according to plan, have the potential Initiative for Multi-Sectoral Technical and to greatly improve long-term welfare of especially Economic Cooperation (BIMSTEC). There are rural households. The international competitive as many initiatives to develop infrastructure as advantage of Bhutan will be enhanced by the there are regional cooperation agreements. completion of the Southern East-West Highway Expansion of hydroelectric power in Bhutan that runs parallel to the Indian border. Bhutan through 2020 will aid poverty reduction both could benefit from the development of a road directly and indirectly. Bhutan’s plans to corridor connecting South Asia with South-East increase six-fold its current capacity for electric Asia through Myanmar under the Bay of Bengal power to 10 GW by 2020, with India as the main 62 Bhutan Poverty Assessment 2014 “One of the main problems is that we are not able to protect our crop from wild animals such as wild boars and monkeys. On top of that important cash crops like oranges and cardamom have been affected by diseases and their yield have declined over the years. These problems have also resulted some of the households to migrate to urban areas. When one household migrates other households also tend to migrate thus turning farms into thick jungle making it easier for wild animals to attack crops in the locality” – A male FGD participant, Kana Community, Dagana export market. Big projects under construction Besides risks at societal level, some (Punatsangchhu I and II, Mangdechhu, and population groups face a variety of challenges. Dagachhu) will add 3 GW of capacity to the Bhutanese can take justifiable pride in community current capacity of 1.6 GW, reaching half-way support for individual families. BLSS 2012 social to the goal of 10 GW by 2017. If more new capital module analysis showed that 70 percent plant agreements are signed and implemented of rural households and 60 percent of urban from now until 2020, they will continue to give households had five or more individuals to turn development impetus during the construction to for support in case of emergencies. Despite this and the generation phases. mutual support, for every 10 families that moved The current trend toward commercial crop out of poverty two were falling into poverty. production carries common risks. With limited While the churning around the poverty line is land, increasing fragmentation of land holdings less pervasive in Bhutan than in many other (“as children get married they demand their own countries, reliance on community support is not land”), and rural to urban migration of working proving adequate for some. Bhutan has no formal age adults, labour-intensive horticulture will social protection mechanisms to help individuals become difficult. Contract farming by large-scale ride out of hard times. Risk of downward mobility land owners may be a way to sustain exports but is greater than average (Figure 6.4) for rural benefits to poorer farmers might diminish. The residents, male-headed households, people in current problems faced by farmers, such as the informal jobs (casual and self-employed), those incurable “greening disease” of oranges, diseases with low education levels, and particularly high for of the cardamom plants, and regular raids into those living in poorer dzongkhags (Pema Gatshel, farms by elephants (in low land), monkeys, Dagana, Samtse, Trashigang, and Tsirang). and wild boars have been difficult to solve. The Female-headed households have pointed out alternative of planting disease-resistant plants is a in focus group discussions that no one was willing running battle. The option of genetically modified to participate with them in labour exchange plants is also against the organic farming trend. It arrangements at times of peak farm labour needs. takes years to bring horticultural crops to harvest The suggested remedial measures for adverse and equally long to shift to other profitable shocks are formal social protection programs production. Some farmers (mainly the rich) have for individuals. At present, individuals cope with shifted to walnuts, but these, too, take many shocks mostly by drawing on own savings if they years to harvest. are non-poor, and by borrowing from friends, Bhutan Poverty Assessment 2014 63 Figure 6.1  Figure 46: Downward Mobility, by Population Groups, 2007-2012 8 6 Percentage (%) 4 2 0 Rural Urban Female Male Regular paid Casual paid Unpaid family worker Self-employed No education 8 Yrs or less 9-12 Yrs >12 Yrs No remitance Received remitance Landless 2 Acres or less 2-5 Acres 5 Acres or more Samtse Agriculture Manufacturing Services Bumthang Chhukha Dagana Gasa Haa Lhuentse Monggar Paro Pema Gatshel Punakha Samdrup Jongkhar Sarpang Thimphu Trashigang Trashi Yangtse Trongsa Tsirang Wangdue Phodrang Zhemgang Population Groups Less than average More than average Source: Author’s calculations suppliers, and money-lenders if they are poor. Because of the inadequacy and inelasticity of these “Before wildlife used to attack crops sources for the poor and vulnerable segments of in the dark, but I believe that now the population, we suggest the introduction of with electricity there is gradual formal social protection mechanisms and possibly reduction in damage to crops by wild micro-credit programs that are well targeted. animals” For sustained poverty reduction, risks and vulnerabilities need to be managed carefully. With limited land, increasing fragmentation of of oranges, diseases of the cardamom plants and land holdings, and rural-to-urban migration of regular raids into farms by elephants (in low land), working age adults, labour-intensive horticulture monkeys, and wild boars have persisted. The will become increasingly difficult. Contract plan for introducing disease-resistant cultivars farming by large-scale land owners may be a way is not proceeding swiftly. It takes years to bring to sustain exports but benefits to poorer farmers horticultural crops to harvest and equally long might diminish. The current problems faced by to shift to other profitable forms of production. farmers such as the incurable “greening disease” As a consequence of increasing commercial crop 64 Bhutan Poverty Assessment 2014 “For this community maize has been principal food from time immemorial. People use to grow maize and eat as a special diet to work in the fields but due to drought we could not harvest like before which has affected our food security” – A male FGD participant, Drujeygang, Dagana “We have lot of wet land for paddy cultivation but now the water sources have started drying up and there is limited volume of water left for sharing among households. Lack of irrigation channel is a problem on top of that due to which wet land remains fallow” – A female FGD participant, Drujeygang, Dagana “Nowadays we have been experiencing hot weather with rise in temperature and may be this is because of lot of constructions works going on and building of factories elsewhere which causes pollution. Water sources have been drying up because may be we are using excessive wood for construction of houses and blasting. Even the taste of oranges is not that sweet like before may be because of the heat” – A female FGD participant, Drujeygang, Dagana “The dust from the mining might have affected the oranges. Drinking water was really not a problem but may be because of the dust and the blastings the source of drinking water is shifting downwards. Dust has affected the trees, animal fodder and because of no rain dust does not settle” – A male FGD participant, Shumar community, Pema Gatshel production, Bhutan dependence on food imports in informal jobs (the casually and self-employed), has been rising over the years, making it more and those with low education and particularly vulnerable to food price shocks. A 12 percent high for those living in select dzongkhags such as increase in food prices – the average annual Pema Gatshel, Dagana, Samtse, Trashigang, and increase in recent years - for example, can increase Tsirang). the percentage of poor in the short-term by about In the long-term, sustainable poverty two percent points. With all petroleum products reduction depends on addressing persistent imported, Bhutan’s poor also face risk from fuel shocks, engendering private sector led price shocks. A sharp rise in the consumer prices development and defining clear target groups of LPG and kerosene of the order that occurred for poverty reduction. The feasibility of crop in July 2013 (quickly reversed, however) had the insurance for farmers may be examined to protect potential to push 0.5 percent of population into the harvests from perils of diseases. Other perils, poverty. Bhutan’s social protection is mainly such as those associated with wild-life predation, through the Royal Kidu welfare program. Risks of have also persisted and evaded viable solutions. downward mobility are greater than average for What poor people want to better their living rural residents, male-headed households, people standards in the long term can be summed up as Bhutan Poverty Assessment 2014 65 access to roads, electricity, public transportation, government could engender private investment in irrigation, land and higher education. Sustained hydropower sector by Private Public Partnerships poverty reduction depends on job opportunities and subcontracting in order to create jobs. and wage earnings of the poor. The development The Royal Government of Bhutan seems to paradigm for a renewable resource rich country favor complementary use of consumption and like Bhutan would call for engendering private multidimensional poverty. But the overlap of sector led growth actively enabled by the public the two approaches identifying the poor is small. sector. Successful agribusiness – an emerging Therefore defining a clear target group for poverty sector in Bhutan - will require development of reduction is important. Also, with success in value chain system (from farm to market) that reducing extreme consumption poverty rapidly, will identify and remove the bottlenecks that the goal could be now shift to shared prosperity farmers encounter including constraints related defined for example as he welfare of the bottom to finance and availability of crop insurance. The 40 percent of the population. 66 Bhutan Poverty Assessment 2014 Bhutan Poverty Assessment 2014 67 Annex A: Sources of Variation in Poverty Outcomes in Bhutan Abstract This annex uses data from the 2007 and 2012 growth incidence curve (GIC) reveals that the rounds of the BLSS to assess the poverty outcomes growth process was clearly pro-poor in the sense associated with the implementation of the 10th that socioeconomic institutions rewarded the Five Year Plan (2008-2013). During the period poor more than the non-poor. Furthermore, a under consideration, poverty incidence in Bhutan decomposition of the urban-rural differential fell significantly, from about 23 percent in 2007 also shows that the reduction in urban bias is due to about 12 percent in 2012. Inequality remained mainly to the fact that the growth process favors almost constant with the Gini coefficient hovering the rural sector relative to the urban. Overall, around 38 percent. It is clear that this impressive these findings suggest that socioeconomic reduction in poverty is due mainly to growth in arrangements in Bhutan have become more per-capita expenditure. A decomposition of the progressive over time. Introduction the least, at an estimated average of 1.3 percent, about one-half of the target rate. Trade and other The RGoB has made the pursuit of services performed better, achieving an average national happiness the overarching goal of its growth rate of 13 percent, higher than the development strategy. In that context, it is target rate. committed to improving the quality of life for The process of economic growth led to the citizens through inclusive and sustainable significant changes in the composition of GDP. economic growth, the conservation of the natural During the Ninth Plan, the share of the primary environment, the preservation of the country’s sector fell from 29 percent to 20.3 percent. By cultural heritage and good governance. These the end of that Plan, the secondary and tertiary focal areas constitute the four pillars spanning sectors accounted for 43 percent and 36 percent the concept of GNH, and are being implemented of GDP respectively. The importance of the through a series of five year plans. The vision secondary sector is linked mainly to growth in underlying this strategic framework has been electricity and construction; it does not reflect enshrined in the 2008 Constitution adopted at any significant developments in manufacturing the beginning of the 10th Five Year Plan. (IMF, 2010a). The RGoB has made poverty reduction the A Joint World Bank-IMF staff advisory note central theme and main objective of the 10th Plan. assessing the 2009 Poverty Reduction Strategy It is pursuing this objective through industrial (IMF, 2010b) notes that the strong performance development, national spatial planning, of the economy under the Ninth Plan, along integrated rural-urban development, strategic expansion of infrastructure, human capital with improvements in governance, have put development, and enhancement of the enabling Bhutan firmly on track to achieve most of environment. The formulation of the 10th Plan the Millennium Development Goals (MDGs). builds on the strong achievements of the Ninth Poverty incidence fell from 36 percent in 2000 Plan (2002-1007) which sought to improve the to about 23 percent in 2007. The Gross Primary quality of life and income, with a special focus Enrolment Rate (GPER) in schools increased on the poor, by promoting good governance from 81 percent in 2002 to 109 percent in 2007. and private sector-driven economic growth in The country has achieved gender parity in both addition to preserving cultural heritage and the primary and basic education due to the fact natural environment. that the enrolment rate was growing faster for During the Ninth Plan, real GDP grew on girls than for boys at the primary and secondary average by 9.6 percent between 2003 and 2007, levels of education. During the same plan period, increasing real GDP from Nu 23.5 billion in sustained investments in both human resources 2002 to Nu 37.5 billion in 2007. This impressive and physical capital in the health sector led to growth, driven mainly by a continuous and significant improvement in the health status of sustained expansion of the electricity sector, the population. Under-five mortality dropped caused the GDP per capita to reach its highest significantly from 84 per thousand to 60 level ever recorded, estimated at US$1,414 in per thousand live births. Maternal mortality 2006 compared to US$835 in 2002 (IMF, 2010a). decreased from 255 to about 215 per hundred In terms of sectoral growth performance, thousand live births. Access to safe drinking available information suggests that the primary water and sanitation expanded considerably sector (agriculture, livestock, and forestry) grew under the Ninth Plan. Bhutan Poverty Assessment 2014 69 The socioeconomic improvements brought considered also indicate rapid poverty reduction. about by the plan implementation are reflected The key question addressed in this paper therefore also in changes in the Human Development is: What drives this impressive poverty reduction? Index (HDI), which combines three indicators A reliable answer to this question might shed of aggregate living standard: (i) life expectancy, some light on why some areas or socioeconomic (ii) educational achievement, and (iii) GDP per groups lag behind despite this strong aggregate capita. The HDI has almost doubled in value performance. over the past 20 years or so. By 2006, Bhutan The importance of the key question above ranked 131st among all countries surveyed. In stems from the fact that an aggregate judgment the 1980s, improvements in the HDI were due about changes in poverty outcomes may hide mostly to increases in life expectancy and real more than it reveals about the heterogeneity of per capita GDP. During the Ninth Plan, life impacts underpinning the aggregate outcome. expectancy stagnated at around 66 years so that Yet a deeper understanding of this diversity of improvements in the HDI observed over that impact is required if one is to better calibrate period were due mostly to increases in enrolment interventions for poverty reduction. Resource- rates in primary and secondary education, and in allocation mechanisms adopted in the 10th Plan GPD per capita (IMF, 2010a). thus strive to account for the poverty status The purpose of this annex is to provide an of the potential beneficiaries, among other account of the poverty outcomes observed under considerations. the 10th Plan, which serves also as the RGoB’s The credibility and hence the usefulness of Poverty Reduction Strategy Paper (PRSP). This an answer to a policy question hinges critically account is based on data from the 2007 and on the quality of its informational basis, consist- 2012 rounds of the BLSS. Given the period of ing of available facts (data) and the logic used to this plan, the 2007 data provide a valid baseline analyze and interpret those facts. The approach for an assessment of the poverty outcomes of followed in this paper is motivated by the fol- this plan. Similarly the 2012 data are considered lowing considerations. Poverty measures and endline observations, reflecting the outcome of all other distributional statistics are computed the implementation of the 10th Plan although the on the basis of a distribution of living standards plan formally ended in 2013. across individuals or households. This observa- Policy analysis can be considered a process tion implies that changes in poverty outcomes designed to produce evidence to answer reflect variations in the underlying distribution important questions that policymakers and of individual outcomes. Thus, a distributional other key stakeholders might have about design, change is pro-poor if it involves poverty reduc- performance, or results. In the context of tion for some choice of poverty index. Given evidence-based decision-making, policymakers that a distribution is fully characterized by are interested in what works and why. There its mean and the degree of inequality, several is evidence that poverty reduction under the authors (e.g., Datt and Ravallion, 1992) have 10th Plan has been as impressive as under the proposed counterfactual decomposition meth- preceding plan. In particular, poverty incidence ods to identify the contribution of changes (based on household per capita expenditure) fell in the mean and in inequality to variations in again significantly from about 23 percent in 2007 overall poverty. Within this framework, the to 12 percent in 2012. All other poverty measures contribution of a change in the mean is known 70 Bhutan Poverty Assessment 2014 as the size effect while the effect of a change in endowment and structural effects. However, we relative inequality is the redistribution effect find it more informative to conduct the analysis in (Essama-Nssah, 2012). terms distributional change driving the variation The usefulness of the size and redistribution in poverty outcomes. This approach is supported effects in policymaking is severely limited by by the fact that, for the class of additively separable the fact that these effects account for changes poverty measure, which includes FGT measures in poverty on the basis of variation in summary and the Watts index, a change in poverty over statistics that are hard to target with policy time can be written as a weighted sum of points instruments. Therefore, there is a need to link along the growth incidence curve (GIC) up to the observed poverty outcomes to factors associated poverty line (Essama-Nssah and Lambert, 2009). with deep structural elements that drive individual Basically, we apply the Oaxaca-Blinder method1 to behavior and social interaction. The living standard decompose the GIC into a component associated of an individual is a pay-off from her participation with the endowment effect and another related to in the life of society. This pay-off is a function of the structural effect. This decomposition entails endowments, behavior, and the circumstances that running unconditional quantile regressions determine the returns on these endowments from (Firpo et al., 2009) for the first 99 percentiles of any social interaction. Changes in these elements the distribution of log per capita expenditure. are the result of the structural transformation The rest of this annex is organized as follows: that underlies development and ultimately drives Section 1 describes the observed outcomes in distributional change. In particular, Bourguignon, poverty, growth, and inequality, and confirms Ferreira, and Lustig (2005) identify three key that the implementation of the 10th Plan led to an forces that determine observed changes in the impressive poverty reduction; Section 2 focuses on distribution of living standards: (i) endowment the endowment and structural effects associated effects or population effects due to changes with distributional changes over time and across in socio-demographic characteristics of the areas of residence, particularly the urban-rural population (e.g., area of residence, age, education, differential (the available data could not support and ownership of physical and financial assets, (ii) the same type of analysis at dzongkhag level); and price effects (also known as structural effects) due Section 3 carries concluding remarks. The annex to changes in returns on factors of production, concludes with a set of Data Tables. and (iii) occupational effects due to changes in the occupational structure of the population. 1.  Underlying Distributional Change Given data and other constraints, this paper and Associated Poverty Outcomes focuses only on two types of effects: (i) the A fundamental step in answering the key endowment effect and (ii) the structural effect question raised in the introduction to this paper associated with changes in returns on those entails describing what happened to consumption endowments. In this reduced-form framework, it is likely that the behavioral effect is mixed up with 1  The standard Oaxaca-Blinder decomposition seeks to decompose the price effect. Household per capita expenditure changes in the unconditional mean outcome the composition or endowment effect and the price or structural effect. The method is our living standard indicator, and is a function relies on the conditional expectation function (CEF) to summarize the relationship between individual outcomes and individual of both observable and non-observable individual characteristics, and then on the law of iterated expectations to link the unconditional mean to characteristics. Fortin et al. (2011) or household characteristics. It is possible to have extended this logic to the decomposition of changes in other decompose directly changes in poverty into the distributional statistics beyond the unconditional mean, such as quantiles. Bhutan Poverty Assessment 2014 71 Table A–1  Distribution of Real per-capita Expenditure in Bhutan, 2007-2012 Year Mean Lowest Decile 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 2007  2313.69 2.73 3.92 4.91 5.89 6.98 8.32 10.00 12.15 15.73 29.30 2012  4603.24 2.75 4.00 4.92 5.86 6.93 8.12 9.69 11.72 15.32 30.61 Source: Authors’ calculations based on BLSS 2007 and 2012 poverty in Bhutan between 2007 and 2012. This that, overall, the observed reduction in poverty section describes variation in poverty outcomes was driven exclusively by the size effect. not only over time but across areas of residence. The change in the distribution of per-capita It also provides a characterization of the pattern expenditure between 2007 and 2012 can also of growth as reflected by the GIC. be characterized by the growth incidence curve Table A–1 presents a summary of the presented in Figure A–2. Recall that this curve distribution of per-capita expenditure based shows the growth rate of an indicator of the on individual-level data in the 2007 and 2012 living standard (e.g., income or expenditure) rounds of the BLSS conducted by the NSB. The at each quantile of the size distribution of that 2007 sample includes observations on 9,798 indicator (Ravallion and Chen, 2003). The fact households and the 2012 dataset for 8,968 that the GIC depicted in Figure 1.2 is greater households. The summary information includes, than zero for all expenditure percentiles means for each round, mean per-capita expenditure in that the distribution of per-capita expenditure real terms and the decile distribution of that per- in 2012 dominates the distribution in 2007 to capita expenditure. It shows that real household the first order. In other words, it means the per-capita expenditure almost doubled in the span posterior distribution of per-capita expenditure of five years. It also shows that the share of each lies nowhere above the initial one. This first-order decile below the richest has remained more or stochastic dominance relation between the two less the same over time, while that of the richest distributions implies that all additively separable increased a little. These results show that the poverty measures satisfying monotonicity3 will growth process in Bhutan has been distribution- agree that poverty has decreased between 2007 neutral between 2007 and 2012. The overall Gini and 2012. Thus, distributional change observed coefficient for 2007 is estimated at 38.09 percent. in Bhutan between those two years is pro-poor in In 2012 this measure of relative inequality stood the sense of Ravallion and Chen (2003) and Kray at 38.75 percent. Table A–2 shows also that (2006). For these authors, a distributional change between-group inequality has been quite stable.2 is pro-poor if it involves poverty reduction for This pattern of distributional change suggests some choice of poverty index. How pro-poor is the observed distributional change in Bhutan over the period under 2  These results are based on a simple decomposition approach applied by Benjamin, Brandt, and Giles (2005) to the case of consideration? Osmani (2005) argues that any inequality in rural China. This entails estimating a regression of poverty-reducing change should not be considered the log of the welfare indicator (income or expenditure per capita) on a set of location dummies. The resulting R-squared shows the automatically pro-poor. He recommends that a proportion of the variation of the log of the welfare indicator that is accounted for by the location dummies. In other words, this is the amount of variation that is “explained” by differences in average 3  level of living. The residual variance is linked to within-location Monotonicity requires that, other things being equal, an increase inequality. In our application for Bhutan we use dzongkhag dummies in the living standard of any person will reduce poverty (Foster, as location variables. Greer, and Thorbecke, 2010). 72 Bhutan Poverty Assessment 2014 Figure A–1  Change in Relative Inequality in Bhutan, 2007 and 2012 Expenditure Shares (percentage) by Decile Lorenz Curves 35 100 30 80 25 Expenditure Shares 60 20 15 40 10 20 5 0 0 0 10 20 30 40 50 60 70 80 90 100 1 2 3 4 5 6 7 8 9 10 Expenditure Percentiles 2007 2012 2007 2012 Source: Author’s calculations distributional change be considered pro-poor if it Table A–2  Between-Group (Dzongkhag) Inequality, by Area of Residence achieves an absolute reduction in poverty greater than would occur in a benchmark case. Such a Year Urban Rural Bhutan benchmark could be a counterfactual or some 2007 13.3 20.4 26.0 socially desirable outcome. In other words, the 2012 14.5 22.4 25.0 pro-poorness of a distributional change depends Source: Author’s calculations on the chosen standard of comparison. One may chose a relative standard of comparison defined by a factor (1+ρ) which indicates the minimum summarizing the variation in poverty outcomes change in living standard that society would in Bhutan between 2007 and 2012 on the basis like the poor to experience given the change in of TIP curves associated with poverty measures the overall distribution (Duclos, 2009). Thus an that bare members of the FGT family. The TIP overall distributional change would be considered curve4 provides a graphical summary of incidence, pro-poor if the outcomes of the poor change by intensity, and inequality dimensions of aggregate a factor of at least (1+ρ). First-order pro-poor poverty based on the distribution of poverty judgments imply that this condition is satisfied gaps normalized by the poverty line5 (Jenkins for all acceptable poverty lines. Let ρ equal the and Lambert, 1997). The curve is obtained by annual growth rate of the average per-capita partially cumulating individual contributions to expenditure. The fact that the rate of growth overall poverty from the poorest individual to the for most percentile up to the 35th is greater than the average annual growth rate of per capita expenditure means that economic growth in Bhutan has been pro-poor to the first-order. 4  TIP stands for “three ‘i’s of poverty”, that is incidence, intensity, The poverty implications of the above and inequality. distributional change are presented in Figure A–3 5  The curve may also be based on absolute poverty gaps. Bhutan Poverty Assessment 2014 73 Figure A–2  Growth Incidence Curve for Bhutan, 2007-2012 24 22 Annual Growth Rate (%) 20 18 16 14 12 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Source: Author’s calculations richest.6 The fact that the TIP curve for 2007 lies subgroups deviates from the overall pattern. above the 2012 curve suggests economic growth Figure A–4 shows growth incidence curves by sex in Bhutan has been pro-poor to the second order. of head of household (left panels) and by area Second-order pro-poor judgments are based on (urban-rural) of residence (right panels). The second-order stochastic dominance which is a pattern of growth in each sub-group is similar to necessary and sufficient condition for additively the overall pattern. We therefore expect similar separable poverty measures satisfying the strong poverty outcomes. In particular, all additively transfer axiom to agree on the pro-poorness of a separable poverty measures that satisfy both distributional change (Atkinson, 1987; Ravallion, monotonicity and the transfer axiom will agree 1994). In particular, we find that all members of that male-headed and female-headed households the FGT family of poverty measures, as well as experienced reduction in poverty between 2007 the Watts index agree that poverty in Bhutan fell and 2012. This is also the case for urban and rural significantly between 2007 and 2012. households. Our discussion so far has focused on aggregate The information contained in Figure poverty and distributional outcomes. We now A–4 reveals the following facts. While the consider disaggregated results to see the extent average annual growth rate of mean per capita to which the experience of some population expenditure is virtually the same for male- headed and female-headed households, this hides 6  This curve is constructed in four steps: (i) rank individuals from considerable heterogeneity of impact across poorest to richest; (ii) compute the relative poverty gap of each individual; (iii) form the cumulative sum of the relative poverty gaps quantiles. The bottom left panel of Figure A–4 divided by population size; and (iv) plot the resulting cumulative sum shows, at each percentile, the difference between of poverty gaps as a function of the cumulative population share. 74 Bhutan Poverty Assessment 2014 Figure A–3  A Picture of Poverty in Bhutan, 2007-2012 7 2007 6 Cumulative sum of poverty gaps 5 4 3 2012 2 1 0 0 10 20 30 40 50 60 70 80 90 100 Cumulative population share Source: Author’s calculations the growth rate for male-headed households of additively separable poverty measures. An and that for female-headed households; female- estimation of the size and redistribution effects headed households located in the lower 35 associated with the growth process shows that percent of the distribution and above the 96th the two effects have opposite signs. The size percentile experienced higher growth rates than effect is negative and leads to poverty reduction. male-headed households. A similar comparison of The redistribution effect is positive and tends to the GIC ordinates for the rural and urban areas counter the size effect. The observed increase in indicates that for all percentiles up to the 98th, inequality is the reason why a relative standard the growth rate of per-capita expenditure was based on average growth rate would not declare higher in the rural areas than in the urban areas. the observed distributional change pro-poor. The This suggests that, even though poverty remains latter is the main determinant of the observed essentially a rural phenomenon, the 10th Plan’s pro-poorness of growth Bhutan because it strategy of channeling investments to rural areas dominates the redistribution effect in absolute may have worked to an extent. terms. For the design and implementation of targeted interventions that might enhance 2.  The Endowment and Structural Effects the effectiveness of poverty reduction policies, By definition, the variation in poverty it is crucial to have a clear understanding the outcomes reflects the underlying distributional heterogeneity that might underlie this aggregate change as depicted by the GIC. The description outcome. This section attempts to identify some of the poverty implications of the process of of the factors that shape the pattern of growth economic growth in Bhutan in 2007-2012 depicted in Figure A–4. This identification relies clearly demonstrates that economic growth led on the logic of the Oaxaca-Blinder decomposition to poverty reduction as indicated by a wide class to split the GIC into two components related to Bhutan Poverty Assessment 2014 75 Figure A–4  Incidence of Growth by Gender and Area of Residence, 2007-2012 28 28 24 Annual Growth Rate (%) Annual Growth Rate (%) 24 20 20 16 16 Female Rural 12 Male 12 8 Urban 8 4 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Expenditure Percentiles 4 10 Female Growth Rate minus Male Growth Rate (%) Rural Growth Rate minus Urban Growth Rate (%) 2 8 0 6 -2 4 -4 2 -6 0 -8 -10 -2 10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100 Source: Author’s calculations distance to nearest hospital, distance to nearest the endowment and structural effects. The use of tarred road, distance to nearest feeder road, linear regression offers an opportunity to consider distance to dzongkhag headquarters, and distance the contribution of specific covariates to these to nearest bank); (iii) Sector of employment effects. The analysis focuses on the overall growth (primary, secondary, non-public services, public incidence and on the urban-rural differential. sector, and non-paid labour); and (iv) Area/ dzongkhag of residence.7 We include durables 2.1.  Overall Growth Incidence among the characteristics because they are Returns on Household Characteristics: As excluded from consumption expenditure (RGoB, explained earlier, we use regression analysis to 2013). link log per-capita expenditure to individual or Table A–7 (see Data Tables at the end of household characteristics. The broad categories this annex) presents sample regression results of characteristics considered includes: (i) for both 2007 and 2012. The table shows the Demographics (age, marital status, female-headed coefficients and the associated standard errors household, and household size); (ii) Household for OLS and selected unconditional quantile (RIF) and community assets (years of education, durable goods such as fridge, electric iron, TV, etc., land 7  Our choice of dummy variables implies that the reference case (conditional on characteristics represented by continuous variables) ownership, ownership livestock, distance to is landless, does not own any of the durables listed in the equation, nearest agricultural extension service center, resides in Thimphu in a male-headed household, and that the sector of employment is listed as other. 76 Bhutan Poverty Assessment 2014 regressions.8 Focusing first on OLS results, we find dummies are negative and statistically significant that among the demographic characteristics age coefficients. These returns are relative to the and household size are the only covariates that omitted dzongkhag (i.e., Thimphu). We note are statistically significant. However, the effect of that the coefficients for Gasa, Paro, and Trashi- age is very small. As expected, household size is Yangtse are positive but not significant in 2007 negatively correlated with per capita expenditure. while that of Tsirang is positive and significant. Similarly, returns on education are positive and Composition versus Structure: The OLS statistically significant. Most durable goods have estimates give only average impacts for the a positive and significant effect on per-capita characteristics under consideration. We will expenditure. But, ownership of a heater water therefore base the decomposition of the GIC boiler, rice cooker, and TV does not have the on the results from RIF regressions in order to same sign across both datasets (negative in 2007 appreciate the extent of heterogeneity in these and positive in 2012). Similarly, land ownership impacts across quantiles. This decomposition switches from negative in 2007 to positive in is analogous to growth accounting, which is an 2012. We interpret this as an improvement in exercise designed to identify the key drivers of productivity, possibly linked to availability of economic growth by decomposing growth in complementary inputs and the development of output into two components: one attributable road infrastructure. to changes in factors of production such as On average, the returns on employment in physical and human capital, and a residual not the primary sector were significant and negative related to changes in output levels. This residual in 2007, but insignificant in 2012. The returns is commonly taken to stand for change in total on employment in the secondary sector are factor productivity (TFP). We consider the living statistically significant in both years, but positive standard of an individual or a household as an only in 2012. Employment in the public sector outcome of participation in the life of society. This does not seem to pay. The associated regression outcome is a function of individual characteristics coefficient is negative and significant only in and returns on those characteristics. We therefore 2007. Employment in the non-public service use the analogy between growth accounting sector is associated with positive and significant and the counterfactual decomposition of the coefficient only in 2012. Residence in the rural GIC considered to link the endowment effect to area is negatively correlated with the welfare notion of accumulation (of factors of production) indicator. This negative correlation is statistically and we take the structural effect to be an indicator significant in 2007. Similarly, most dzongkhag of productivity in socioeconomic interaction. Accumulation and productivity are indeed the two 8  basic ideas that structure the study of economic RIF stands for Recentered Influence Function. The influence function of a distributional statistic such as the mean or a quantile growth. measures the impact on the statistic of a small change in the underlying distribution. The RIF is equal to the statistic in question Figure A–5 shows a decomposition of the plus its influence function, if it exists (Firpo et al., 2009). Because the expected value of the influence function is equal to zero, total variation in the distribution of log per the RIF offers a simple way of linking a distributional statistic capita expenditure (essentially the GIC) into to individual or household characteristics, using the conditional expectation function (CEF). Oaxaca-Blinder decomposition requires two components. The first component is due unconditional expectation of the statistic of interest. This can be obtained by applying the law of iterated expectations to the RIF to changes in the distribution of characteristics regression. As it turns out, the OLS implements the RIF regression for the unconditional mean of an outcome variable. See Essama- while the second represents the contribution of Nssah and Lambert (2013) for the derivation of RIFs for a variety of changes in the distribution of returns on those distributional statistics used in policy impact evaluation. Bhutan Poverty Assessment 2014 77 Figure A–5  A Decomposition of Growth Incidence in endowment effect. Since the structural effect Bhutan, 2007-2012 represents the change reward for participation 40 in socioeconomic arrangements, these results 30 suggest that socioeconomic arrangements in Annual Growth Rate (%) Endowment Effect Bhutan may have become more progressive 20 Overall Incidence over time. 10 What are the factors driving both the 0 composition and the structural effects? We further Structural Effect disaggregate these two components on the basis -10 of sets of covariates. Figure A–6 shows the key -20 0 10 20 30 40 50 60 70 80 90 100 covariates that shape both the endowment effect Expenditure Percentiles and the structural effect. The left panel compares Source: Author’s calculations the full composition effect to the contribution of ownership of durable goods. It is evident that these characteristics are the main drivers of the characteristics. The structural effect has a more composition effect. The right panel compares the or less a U-shape. The fact that it is downward overall structural effect and the contributions of sloping up to the 77th percentile means that the household demographics and the coefficient of the structural effect reduces inequality in that part reference group. These results show that both the of the distribution and it tends to increase it level and the dispersion of the full structural effect in the upper segment of the distribution. The are closely tracked by household demographics. A endowment effect has roughly an inverted further decomposition, not shown here, revealed U-shape. It is upward sloping up to the 77th that the key driver is household size. percentile and therefore increases inequality over While ownership of durable goods and the much of the distribution. The structural effect household demographics can certainly serve as dominates the endowment effect at the low end of targeting variables in the formulation of policy the distribution up to the 28th percentile. It turns interventions, it is useful to consider the effects negative between the 60th and 95th percentile. of other covariates that are directly subject to The endowment effect is mostly positive and intervention – for instance, in land ownership overwhelms the structural effect past the 28th and years of education. A recent participatory percentile. The configuration of the three curves assessment found that small land holdings are presented in Figure A–5 implies that the level of an important constraint on achieving economies the GIC is determined mainly by the composition of scale in agricultural production. The Royal or endowment effect. In particular, the gains Government has also granted land to about achieved by people located at the bottom of the 61,339 beneficiaries (Over one acre per head) distribution up to the 28th percentile are due to under the Kidu program for socio-economically the structural effect while the gains beyond that disadvantaged groups during 2009-2013 point are mainly due to the composition effect. (National Land Commission). As far as education The pro-poorness of the distribution change is concerned, the 10th Plan, while acknowledging that occurred in 2007-2012 due mainly to the important achievements in education, deplores structural effect while the increase in inequality low adult literacy as a constraint on improvement observed over the same period is driven by the in the HDI. 78 Bhutan Poverty Assessment 2014 Figure A–6  Accounting for the Endowment and Structural Effects 30 80 Overall Endowment Effect 25 60 Annual Growth Rate (%) Annual Growth Rate (%) 20 40 15 Reference Group 10 20 5 Durable Goods Overall Structural Effect 0 0 -20 -5 Demographics -10 -40 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Source: Author’s calculations The endowment and structural effects of land 2.2.  The Urban-Rural Differential ownership and of years of schooling are presented Recall that integrated rural-urban in Figures A–7 and A–8. With regard to land development is an important pillar of the poverty ownership, the configuration of the endowment reduction strategy underpinning the 10th Plan. effects for small and large land holdings shown An earlier comparison of the GIC of the rural on the left panel reflects two facts: (i) the returns and urban sectors showed that for all percentiles on land were negative in 2007, and (ii) small land up to the 90th, the growth rate of per-capita holdings increased between 2007 and 2012 (most expenditure was higher in the rural areas than in likely due to land redistribution) while large land the urban areas (Figure A–8). We further analyze holdings decreased. That is why the composition the urban-rural differential to try to uncover effect of small land holdings is negative while that what drives this observation. Figure A–9 shows of large land holdings is positive. The structural the unconditional quantile regression coefficients effect for both types of land holdings is shown of the dummy variable indicating rural residence on the right panel of Figure A–7. The returns on in the regression of log per-capita expenditure on both types of holdings increased over time. While individual and household characteristics. The fact the overall structural effect tends to dampen that the plot of (unconditional) quantile process inequality, the structural effect of land ownership coefficients for 2012 lies significantly above that increases inequality. for 2007 reveals that, other things being equal, Figure A–8 shows that the structural effect the returns on rural residence have improved over of years of education dominates the endowment the period under consideration. This represents effect across all quantiles. Both effects are a significant reduction in the urban bias that more significant beyond the median; clearly existed in 2007 as shown by the steep decline in demonstrating that schooling is a contributing the quantile process in 2007. However, the 2012 factor to inequality. While returns on years of plot still lies mostly below zero, indicating room education have increased over time, we note for improvement. that they are much lower in the lower half of the Figure A–10 shows a comparison of the urban- distribution. rural differential in living standards in Bhutan for Bhutan Poverty Assessment 2014 79 Figure A–7  The Effects of Land Ownership Endowment Effect Structural Effect 1.6 8 1.2 Ownership of more than 5 acres of land 7 Annual Growth Rate (%) Annual Growth Rate (%) 0.8 6 0.4 5 0.0 4 Five acres or less -0.4 -0.8 Ownership of 5 acres of land or less 3 -1.2 2 -1.6 1 -2.0 More than five acres 0 -2.4 0 10 20 30 40 50 60 70 80 90 100 -1 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Expenditure Percentiles Source: Author’s calculations Figure A–8  The Effects of Years of Schooling composition of the overall distributional change 1.4 over time. The structural effect is declining across 1.2 quantiles while the composition effect is increas- Annual Growth Rate (%) 1.0 Structural Effect ing. The two effects are countering each other. The 0.8 structural effect tends to reduce the rural-urban 0.6 gap while the endowment effect tends to increase 0.4 it. In both years, there is no significant difference 0.2 between the endowment and structural effects in 0.0 Endowment Effect the lower part of the distribution. In 2007, the -0.2 0 10 20 30 40 50 60 70 80 90 100 structural effect dominates slightly the composi- Expenditure Percentiles tion effect between the 11th and 36th percentiles. Source: Author’s calculations In 2012, this dominance relation holds from the 1st up to the 26th percentile. All of these considera- tions point to the conclusion that any remaining 2007 and 2012. The fact that the curve for 2007 urban bias in the distribution of living standards dominates that for 2012 confirms the finding is mostly due to the composition effect. that the gap between the rural and urban sector is declining. 3.  Concluding Remarks To further understand the factors that may be contributing to the structure and evolution In the context of its overall development of the urban bias in Bhutan, we decompose the strategy designed to promote GNH for the people total urban-rural differential following the same of Bhutan, the RGoB has made poverty reduction Oaxaca-Blinder approach that we used earlier to the focal objective of the 10th Five Year Plan and decompose growth incidence curves into endow- hence a metric for evaluating socioeconomic ment and structural effects. The results of this performance under that plan. The beginning of this decomposition are presented in Figure A–11. For plan period coincided with the adoption of a new both 2007 and 2012, the configuration of the Constitution marking a transition from absolute results is similar to what we obtained in the de- monarchy to a parliamentary democracy. In a 80 Bhutan Poverty Assessment 2014 Figure A–9  Returns on Rural Residence Figure A–10  Evolution of the Urban-Rural Differential .1 1.0 Unconditional Quantile Regression Coefficients 2007 Difference in Log Expenditure 0.9 .0 0.8 2007 2012 -.1 0.7 0.6 2012 -.2 0.5 0.4 -.3 0.3 -.4 0.2 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Expenditure Percentiles Expenditure Percentiles Source: Author’s calculations Source: Authors’ calculations Figure A–11  A Decomposition of the Urban-Rural Differential in Bhutan 2007 2012 3 2.5 Difference in Log Expenditure Difference in Log Expenditure Endowment Effect 2.0 2 Endowment Effect 1.5 1 1.0 Total Differential Total Differential 0.5 0 0.0 -0.5 -1 Structural Effect -1.0 Structural Effect -2 -1.5 01 02 03 04 05 06 07 08 09 0 100 01 02 03 04 05 06 07 08 09 0 100 Source: Author’s calculations democratic system of government, policymaking (iii) what explains the observed outcomes. In requires transparency and accountability not this paper, we focus on the last question as we only for policy choices, but for results as well. This seek to describe the poverty outcomes associated requirement has made evidence-based decision- with the implementation of the 10th Plan and to making the bedrock of the policy cycle. identify some key factors that determine those This paper provides an assessment of the outcomes. poverty outcomes observed over the 10th Plan A key step in accounting for variation in period based on data from the 2007 and 2012 an outcome requires a plausible association rounds of the BLSS. Policy evaluation is meant to between that outcome and possible explanatory produce evidence to answer important questions factors. Almost by definition, variation in poverty that policymakers and other key stakeholders outcomes reflects the underlying distributional care about. In general, decision-makers are change depicted by the relevant GIC. interested in knowing: (i) whether they are doing Furthermore, a distribution of living standards the right things the right way, (ii) whether what is fully determined by its mean and the degree they are doing is working and worth the cost, and of inequality. Variation in poverty outcomes can Bhutan Poverty Assessment 2014 81 therefore be seen as driven by these factors as A counterfactual decomposition of the overall well. However, it is hard to target distributional GIC into the endowment and structural effects statistics such as the mean or a measure of shows that the level of the GIC is determined inequality with policy instruments. This creates mainly by the endowment effect. Furthermore, the need for deeper analysis linking distributional the pro-poorness of the distribution change change to individual or household characteristics. that occurred in 2007-2012 is due mainly to the The living standard of an individual (or structural effect while the increase in inequality household) is a pay-off from participation in the observed over the same period is driven by the life of society subject to individual endowments endowment effect. Since the structural effect and the circumstances that determine the returns represents the change reward for participation on those endowments from social interaction. in socioeconomic arrangements, these results This consideration provides a theoretical basis suggest that socioeconomic arrangements in for the link between distributional statistics Bhutan may have become more progressive and individual characteristics. We rely on the over time. A closer look at some particular Conditional Expectation Function to implement covariates revealed that: (i) the endowment this link empirically. The identification of sources effect is accounted for mostly by the ownership of variation in poverty outcomes is based on of durable goods; (ii) the structural effect is counterfactual decomposition. driven by demographic factors; (iii) the returns Analysis of the 2007 and 2012 rounds of on land improved between 2007 and 2012; (iv) the BLSS shows that real household per-capita the structural effect of land ownership tends expenditure more than doubled in the span of to increase inequality; and (v) schooling is a five years. This doubling of per-capita expenditure contributing factor to inequality through both its was also accompanied by an increase in relative endowment and structural effects. inequality; the overall Gini coefficient increased A similar decomposition analysis of the urban- from 33.37 percent in 2007 to 38.81 percent rural differential confirms that the gap between in 2012. We find that a wide class of poverty the rural and urban sectors has been shrinking measures (e.g., FGT and Watts) would agree that over time. This reduction is driven by the poverty in Bhutan declined between 2007 and structural effect. Much of the remaining urban 2012 – the poverty incidence fell from 23 percent bias is accounted for by the composition effect. to 12 percent. These findings suggest that the In any case, these findings suggest that the policy implementation of the 10th Plan has been pro- of integrated rural urban development may be poor to a certain extent. The exact extent depends working. on the chosen standard of evaluation. If one The overall conclusion emerging from this adopts a relative standard based on the annual analysis is that the implementation of the 10th growth rate of the average per-capita expenditure, Plan has been pro-poor. This result is most likely then the conclusion would be that the observed due to that fact that socioeconomic arrangements distributional change was not pro-poor enough. in Bhutan have become more progressive. 82 Bhutan Poverty Assessment 2014 Table A–3  Poverty Outcomes in Bhutan, by Dzongkhag, 2007 Squared Poverty Headcount Poverty Gap Watts Index Population Share Number of Poor Gap Bumthang 10.93 1.93 0.54 2.28 2.55 1,753 Chhukha 20.27 4.86 1.70 6.08 10.74 13,704 Dagana 31.10 8.82 3.62 11.69 3.00 5,867 Gasa 4.15 0.67 0.20 0.81 0.60 156 Haa 13.19 3.49 1.60 4.82 1.99 1,650 Lhuentse 42.97 11.92 4.58 15.32 2.49 6,749 Monggar 44.41 11.76 4.08 14.62 6.06 16,959 Paro 3.95 0.70 0.21 0.85 5.63 1,401 Pema Gatshel 26.21 5.82 1.79 7.06 3.76 6,197 Punakha 15.65 3.21 0.98 3.90 4.03 3,966 Samdrup Jongkhar 37.98 10.97 4.57 14.66 5.55 13,270 Samtse 46.76 14.68 6.17 19.55 8.85 26,056 Sarpang 19.43 4.78 1.54 5.82 6.38 7,809 Thimphu 2.39 0.46 0.11 0.53 13.77 2,073 Trashigang 29.28 7.12 2.63 9.08 7.58 13,966 Trashi Yangtse 14.33 2.22 0.54 2.56 2.89 2,610 Trongsa 22.15 6.17 2.27 7.82 2.32 3,231 Tsirang 13.89 2.84 0.91 3.49 3.01 2,635 Wangdue Phodrang 15.84 3.01 0.92 3.65 5.70 5,685 Zhemgang 52.86 15.18 5.71 19.39 3.11 10,364 Bhutan 23.20 6.06 2.26 7.75 100.00 146,101 Source: Author’s calculations Table A–4  Poverty Outcomes in Bhutan, by Dzongkhag, 2012 Squared Poverty Headcount Poverty Gap Watts Index Population Share Number of Poor Gap Bumthang 3.44 0.26 0.02 0.27 2.19 437 Chhukha 11.25 2.28 0.75 2.80 9.44 6,169 Dagana 25.10 5.84 1.98 7.25 3.33 4,857 Gasa 0.00 0.00 0.00 0.00 0.52 0 Haa 6.39 1.40 0.48 1.73 1.50 555 Lhuentse 31.89 8.43 3.18 10.78 2.45 4,545 Monggar 10.54 1.75 0.56 2.20 6.59 4,036 Paro 0.00 0.00 0.00 0.00 5.42 0 Pema Gatshel 26.88 5.52 1.68 6.70 3.84 6,004 Punakha 9.99 2.52 1.00 3.28 3.77 2,191 Samdrup Jongkhar 21.01 4.57 1.47 5.61 5.24 6,393 Samtse 22.16 4.68 1.44 5.69 9.46 12,192 Sarpang 4.17 0.68 0.19 0.80 5.92 1,436 Thimphu 0.52 0.04 0.01 0.04 15.38 464 Trashigang 11.52 2.74 0.93 3.39 7.52 5,034 Trashi Yangtse 13.48 2.82 0.99 3.52 2.76 2,165 Trongsa 14.93 3.50 1.14 4.29 2.30 1,995 Tsirang 14.83 2.53 0.72 3.01 3.26 2,809 Wangdue Phodrang 10.94 2.34 0.79 2.96 5.84 3,716 Zhemgang 26.27 7.18 2.88 9.41 3.28 5,006 Bhutan 12.04 2.61 0.87 3.24 100.00 70,005 Source: Author’s calculations Bhutan Poverty Assessment 2014 83 Table A–5  Inequality in Distribution of per-capita Expenditure, by Dzongkhag, 2007 Gini Atkinson (1) Atkinson (2) Mean Log Deviation Theil Variance Log per capita Expenditure Bumthang 30.41 14.42 24.22 15.57 18.90 26.53 Chhukha 37.74 21.08 36.01 23.68 24.99 44.67 Dagana 30.12 14.14 26.66 15.24 15.16 30.89 Gasa 25.80 10.22 18.57 10.78 11.46 20.34 Haa 28.82 13.08 25.06 14.02 13.99 28.42 Lhuentse 29.86 13.60 23.71 14.62 16.49 26.54 Monggar 32.83 15.79 26.35 17.18 19.33 30.25 Paro 31.03 14.29 25.07 15.41 16.42 28.76 Pema Gatshel 27.89 11.74 21.31 12.48 12.98 23.98 Punakha 36.06 18.98 30.94 21.04 24.14 36.29 Samdrup Jongkhar 39.06 22.22 37.11 25.13 27.42 45.98 Samtse 37.31 20.29 33.95 22.67 24.85 41.18 Sarpang 31.10 14.51 25.85 15.67 16.36 29.95 Thimphu 31.51 15.23 27.39 16.52 17.69 31.45 Trashigang 31.85 15.51 26.47 16.85 19.36 29.83 Trashi Yangtse 28.08 12.00 21.05 12.78 14.34 23.26 Trongsa 34.85 18.17 31.78 20.05 21.05 38.25 Tsirang 32.97 16.50 27.09 18.04 21.77 30.27 Wangdue Phodrang 32.29 15.59 26.62 16.94 18.92 30.46 Zhemgang 36.58 19.51 30.70 21.71 26.60 35.25 Bhutan 38.09 21.31 36.08 23.96 25.76 44.54 Source: Author’s calculations Table A–6  Inequality in Distribution of per-capita Expenditure, by Dzongkhag, 2012 Gini Atkinson (1) Atkinson (2) Mean Log Deviation Theil Variance Log per capita Expenditure Bumthang 29.67 13.44 23.40 14.43 15.92 26.31 Chhukha 35.90 19.05 32.24 21.14 23.37 38.40 Dagana 28.84 12.81 23.23 13.70 14.42 26.25 Gasa 38.93 22.31 33.72 25.25 31.76 39.07 Haa 33.58 16.93 29.59 18.55 19.69 34.87 Lhuentse 37.75 20.61 34.38 23.08 24.53 42.40 Monggar 33.48 16.97 28.80 18.60 21.71 32.59 Paro 33.74 17.32 27.99 19.01 23.89 31.11 Pema Gatshel 24.78 9.44 17.45 9.92 10.34 19.07 Punakha 34.63 18.29 32.43 20.20 21.54 38.54 Samdrup Jongkhar 41.87 25.17 39.68 28.99 34.95 49.01 Samtse 35.70 19.05 30.28 21.14 26.48 34.20 Sarpang 25.58 10.02 18.46 10.56 10.99 20.34 Thimphu 34.84 17.90 29.62 19.72 22.74 34.32 Trashigang 31.75 15.21 27.08 16.50 17.42 31.38 Trashi Yangtse 34.79 17.93 29.50 19.76 23.20 33.94 Trongsa 37.18 20.33 34.51 22.73 24.27 42.30 Tsirang 34.80 17.85 28.75 19.66 23.57 32.80 Wangdue Phodrang 31.28 15.41 27.64 16.74 19.25 30.80 Zhemgang 30.88 15.10 27.15 16.36 17.87 30.94 Bhutan 38.75 22.02 36.32 24.87 28.44 44.09 Source: Author’s calculations 84 Bhutan Poverty Assessment 2014 Table A–7  OLS and RIF Regression Coefficients on Log Expenditure, 2007-2012 OLS OLS RIF_10 RIF_10 RIF_30 RIFT_30 RIF_90 RIF_90 Eq Name: 2007 2012 2007 2012 2007 2012 2007 2012 Dep. Var: LRPCEXP LRPCEXP RIFQT_10 RIFQT_10 RIFQT_30 RIFQT_30 RIFQT_90 RIFQT_90 C  8.250  8.919  6.997  7.547  7.941  7.904  9.493  10.812 (0.0896)** (0.0552)** (0.0600)** (0.0568)** (0.0681)** (0.0600)** (0.3455)** (0.2048)** FMHEADED -0.012  0.009 -0.012 -0.002 -0.022  0.014 -0.000 -0.057 (0.0084) (0.0074) (0.0057)* (0.0076) (0.0064)** (0.0081) (0.0325) (0.0276)* AGE  0.003 -0.002  0.002 -0.002  0.001 -0.000  0.005 -0.004 (0.0008)** (0.0007)** (0.0005)** (0.0008)* (0.0006) (0.0008) (0.0030) (0.0028) AGE_SQ -0.003  0.003 -0.002  0.002 -0.001  0.001 -0.003  0.006 (0.0011)** (0.0009)** (0.0007)** (0.0009) (0.0008) (0.0010) (0.0041) (0.0034) MARRIED -0.012 -0.009 -0.034  0.009  0.038  0.016 -0.145 -0.106 (0.0294) (0.0095) (0.0197) (0.0098) (0.0223) (0.0104) (0.1133) (0.0354)** EDUYEARS  0.005  0.013  0.000  0.002  0.001  0.005  0.023  0.034 (0.0008)** (0.0008)** (0.0006) (0.0008)** (0.0006)* (0.0008)** (0.0032)** (0.0028)** HSIZE -0.055 -0.237  0.048  0.081  0.035 -0.028 -0.388 -0.710 (0.0059)** (0.0055)** (0.0040)** (0.0057)** (0.0045)** (0.0060)** (0.0228)** (0.0205)** HSIZE_SQ -0.001  0.010 -0.005 -0.010 -0.006 -0.004  0.017  0.039 (0.0004)** (0.0004)** (0.0003)** (0.0004)** (0.0003)** (0.0005)** (0.0016)** (0.0016)** HEATER -0.052  0.148  0.011  0.025 -0.034  0.093 -0.151  0.339 (0.0068)** (0.0078)** (0.0046)* (0.0080)** (0.0052)** (0.0084)** (0.0263)** (0.0289)** BUKHARI -0.038  0.079 -0.009  0.075 -0.060  0.074 -0.039  0.066 (0.0081)** (0.0084)** (0.0054) (0.0086)** (0.0062)** (0.0091)** (0.0312) (0.0310)* CHOESHAM -0.049  0.098  0.007  0.044 -0.033  0.070 -0.125  0.254 (0.0075)** (0.0078)** (0.0050) (0.0081)** (0.0057)** (0.0085)** (0.0288)** (0.0291)** FRIDGE  0.114  0.159 -0.019  0.021  0.024  0.124  0.428  0.230 (0.0111)** (0.0131)** (0.0074)** (0.0134) (0.0084)** (0.0142)** (0.0426)** (0.0485)** ELIRON  0.122  0.117  0.010  0.031  0.047  0.023  0.335  0.333 (0.0083)** (0.0075)** (0.0056) (0.0078)** (0.0063)** (0.0082)** (0.0322)** (0.0280)** W_BOILER -0.007  0.115 -0.014  0.180 -0.080  0.071  0.179  0.123 (0.0113) (0.0135)** (0.0076) (0.0139)** (0.0086)** (0.0146)** (0.0434)** (0.0500)* STOVE  0.064  0.082  0.031  0.041  0.013  0.108  0.181  0.138 (0.0073)** (0.0068)** (0.0049)** (0.0070)** (0.0055)* (0.0074)** (0.0280)** (0.0253)** CURRYCKR  0.070  0.136 -0.000 -0.018  0.070  0.072  0.245  0.373 (0.0139)** (0.0180)** (0.0093) (0.0186) (0.0106)** (0.0196)** (0.0536)** (0.0669)** RICECKR  0.348 -0.021 -0.032 -0.152 -0.063  0.118  1.081  0.110 (0.0605)** (0.0459) (0.0405) (0.0473)** (0.0460) (0.0499)* (0.2332)** (0.1705) G_MACHINE  0.005  0.094  0.034  0.032  0.051  0.042 -0.104  0.145 (0.0113) (0.0078)** (0.0076)** (0.0081)** (0.0086)** (0.0085)** (0.0434)* (0.0291)** RADIO -0.037 -0.030 -0.010 -0.028 -0.033  0.008 -0.049 -0.145 (0.0075)** (0.0066)** (0.0050) (0.0068)** (0.0057)** (0.0072) (0.0288) (0.0245)** TV -0.042  0.050 -0.062 -0.051  0.026  0.038 -0.072 -0.018 (0.0095)** (0.0104)** (0.0063)** (0.0107)** (0.0072)** (0.0113)** (0.0365)* (0.0387) PRIMARY -0.065  0.020 -0.018  0.051 -0.013  0.009 -0.178  0.042 (0.0150)** (0.0115) (0.0100) (0.0119)** (0.0114) (0.0125) (0.0578)** (0.0427) SECONDARY -0.043  0.086 -0.002  0.017  0.003 -0.026 -0.165  0.418 (0.0187)* (0.0181)** (0.0125) (0.0187) (0.0142) (0.0197) (0.0719)* (0.0673)** NONPUBL -0.013  0.105 -0.001  0.026  0.006  0.015 -0.037  0.296 (0.0114) (0.0112)** (0.0077) (0.0116)* (0.0087) (0.0122) (0.0440) (0.0418)** PUBLIC -0.054 -0.004 -0.007 -0.009 -0.013 -0.022 -0.173 -0.006 Bhutan Poverty Assessment 2014 85 OLS OLS RIF_10 RIF_10 RIF_30 RIFT_30 RIF_90 RIF_90 Eq Name: 2007 2012 2007 2012 2007 2012 2007 2012 Dep. Var: LRPCEXP LRPCEXP RIFQT_10 RIFQT_10 RIFQT_30 RIFQT_30 RIFQT_90 RIFQT_90 (0.0167)** (0.0151) (0.0112) (0.0156) (0.0127) (0.0164) (0.0643)** (0.0562) NOPAY -0.020 -0.013 -0.016 -0.000 -0.022 -0.036  0.017 -0.050 (0.0173) (0.0126) (0.0116) (0.0129) (0.0132) (0.0137)** (0.0667) (0.0467) LD5ACRES -0.102  0.111 -0.050  0.019 -0.045  0.044 -0.263  0.210 (0.0137)** (0.0079)** (0.0092)** (0.0081)* (0.0104)** (0.0086)** (0.0527)** (0.0292)** LDMORE5ACRES -0.075  0.167 -0.057 -0.041 -0.070 -0.018 -0.087  0.412 (0.0128)** (0.0125)** (0.0086)** (0.0128)** (0.0097)** (0.0136) (0.0493) (0.0463)** CATTLE -0.010 -0.030 -0.003 -0.008 -0.009 -0.050 -0.017 -0.066 (0.0011)** (0.0040)** (0.0007)** (0.0041) (0.0008)** (0.0043)** (0.0042)** (0.0147)** HORSES -0.003 -0.046  0.022  0.003  0.015  0.054 -0.065 -0.057 (0.0054) (0.0104)** (0.0036)** (0.0107) (0.0041)** (0.0113)** (0.0206)** (0.0384) YAK -0.000  0.095 -0.002  0.036 -0.001  0.067 -0.001  0.138 (0.0012) (0.0162)** (0.0008)* (0.0167)* (0.0009) (0.0176)** (0.0045) (0.0602)* T_AGRI  0.001  0.013  0.005  0.012 -0.008  0.009  0.035  0.036 (0.0052) (0.0031)** (0.0035) (0.0032)** (0.0039)* (0.0034)** (0.0199) (0.0115)** T_BANK -0.001 -0.004 -0.021  0.000 -0.001 -0.012  0.017  0.003 (0.0035) (0.0011)** (0.0023)** (0.0011) (0.0026) (0.0012)** (0.0134) (0.0041) T_TARREDROAD -0.028  0.008 -0.055  0.013 -0.046 -0.011  0.000  0.060 (0.0045)** (0.0034)* (0.0030)** (0.0035)** (0.0034)** (0.0036)** (0.0175) (0.0125)** T_FDROAD  0.004 -0.058  0.003 -0.048  0.010 -0.041 -0.011 -0.138 (0.0031) (0.0080)** (0.0020) (0.0082)** (0.0023)** (0.0087)** (0.0118) (0.0297)** T_FIREW -0.005  0.000 -0.001  0.007 -0.009 -0.008 -0.041  0.023 (0.0024)* (0.0028) (0.0016) (0.0028)* (0.0018)** (0.0030)** (0.0094)** (0.0103)* T_DZHQ -0.025 -0.002 -0.022  0.003 -0.010  0.003 -0.039  0.003 (0.0026)** (0.0015) (0.0018)** (0.0015)* (0.0020)** (0.0016) (0.0101)** (0.0054) T_HOSPITAL -0.003 -0.021  0.000 -0.031 -0.004 -0.026  0.010 -0.037 (0.0023) (0.0034)** (0.0016) (0.0035)** (0.0018)* (0.0037)** (0.0090) (0.0125)** RURAL -0.137 -0.011  0.002 -0.007 -0.052  0.024 -0.243  0.035 (0.0098)** (0.0087) (0.0066) (0.0089) (0.0075)** (0.0094)** (0.0379)** (0.0323) Bumthang -0.047 -0.305  0.021 -0.058  0.032 -0.036 -0.472 -0.929 (0.0267) (0.0208)** (0.0179) (0.0214)** (0.0203) (0.0226) (0.1028)** (0.0771)** CHHUKHA -0.038 -0.361  0.025 -0.075  0.037 -0.172 -0.359 -0.739 (0.0130)** (0.0128)** (0.0087)** (0.0132)** (0.0099)** (0.0139)** (0.0501)** (0.0475)** DAGANA -0.211 -0.491  0.051 -0.357 -0.071 -0.291 -0.573 -0.957 (0.0386)** (0.0277)** (0.0258)* (0.0285)** (0.0293)* (0.0301)** (0.1487)** (0.1028)** GASA  0.005 -0.075  0.309 -0.096  0.106 -0.065  0.291 -0.988 (0.1400) (0.0701) (0.0938)** (0.0722) (0.1065) (0.0762) (0.5399) (0.2603)** HAA -0.157 -0.458 -0.002 -0.211 -0.057 -0.135 -0.387 -0.748 (0.0276)** (0.0251)** (0.0185) (0.0259)** (0.0210)** (0.0273)** (0.1063)** (0.0932)** LHUENTSE -0.239 -0.489  0.160 -0.566 -0.138 -0.353 -0.625 -0.432 (0.0406)** (0.0280)** (0.0272)** (0.0288)** (0.0309)** (0.0304)** (0.1565)** (0.1040)** MONGGAR -0.147 -0.175 -0.007  0.052  0.037  0.021 -0.614 -0.515 (0.0236)** (0.0172)** (0.0158) (0.0177)** (0.0179)* (0.0187) (0.0908)** (0.0638)** PARO  0.002 -0.011  0.007  0.005  0.094  0.024 -0.111 -0.033 (0.0151) (0.0128) (0.0101) (0.0132) (0.0115)** (0.0139) (0.0583) (0.0476) PEMAGATSHEL -0.144 -0.658  0.159 -0.319  0.034 -0.635 -0.857 -0.941 (0.0325)** (0.0213)** (0.0217)** (0.0219)** (0.0247) (0.0231)** (0.1251)** (0.0790)** PUNAKHA -0.051 -0.251  0.052 -0.067 -0.034 -0.141 -0.201 -0.542 (0.0190)** (0.0152)** (0.0127)** (0.0157)** (0.0144)* (0.0166)** (0.0731)** (0.0566)** 86 Bhutan Poverty Assessment 2014 OLS OLS RIF_10 RIF_10 RIF_30 RIFT_30 RIF_90 RIF_90 Eq Name: 2007 2012 2007 2012 2007 2012 2007 2012 Dep. Var: LRPCEXP LRPCEXP RIFQT_10 RIFQT_10 RIFQT_30 RIFQT_30 RIFQT_90 RIFQT_90 SAMDRUPJ -0.072 -0.088 -0.044 -0.025 -0.042  0.003 -0.104 -0.151 (0.0199)** (0.0180)** (0.0134)** (0.0185) (0.0152)** (0.0195) (0.0768) (0.0668)* SAMTSE -0.223 -0.354 -0.052 -0.021  0.008 -0.099 -0.875 -0.722 (0.0172)** (0.0167)** (0.0115)** (0.0172) (0.0131) (0.0182)** (0.0662)** (0.0622)** SARPANG -0.121 -0.322  0.058  0.073  0.023 -0.048 -0.531 -0.917 (0.0164)** (0.0149)** (0.0110)** (0.0154)** (0.0125) (0.0162)** (0.0633)** (0.0554)** TRASHIGANG -0.185 -0.414 -0.039 -0.018  0.016 -0.215 -0.582 -1.092 (0.0191)** (0.0153)** (0.0128)** (0.0158) (0.0146) (0.0167)** (0.0738)** (0.0569)** TRASHIY  0.023 -0.270  0.015 -0.192  0.046 -0.099 -0.259 -0.819 (0.0361) (0.0278)** (0.0242) (0.0286)** (0.0274) (0.0302)** (0.1390) (0.1030)** TRONGSA -0.059 -0.244  0.010 -0.162 -0.045 -0.209 -0.618 -0.357 (0.0338) (0.0260)** (0.0227) (0.0268)** (0.0257) (0.0283)** (0.1304)** (0.0966)** TSIRANG  0.106 -0.291 -0.005 -0.112  0.049 -0.122  0.367 -0.802 (0.0381)** (0.0249)** (0.0255) (0.0257)** (0.0290) (0.0271)** (0.1470)* (0.0926)** WANGDUEP -0.043 -0.264 -0.006 -0.061  0.036 -0.009 -0.305 -0.655 (0.0185)* (0.0156)** (0.0124) (0.0161)** (0.0141)* (0.0170) (0.0714)** (0.0581)** ZHEMGANG -0.087 -0.359  0.037  0.039  0.041  0.100 -0.182 -0.773 (0.0322)** (0.0233)** (0.0216) (0.0239) (0.0245) (0.0253)** (0.1243) (0.0863)** Observations: 13155 21190 13155 21190 13155 21190 13155 21190 R-squared: 0.3115 0.4461 0.1799 0.1401 0.2419 0.2404 0.1525 0.2123 F-statistic: 105.8402 303.9084 51.3155 61.5096 74.6442 119.4643 42.0930 101.7009 Source: Author’s calculations Bhutan Poverty Assessment 2014 87 Annex B: Poverty Dynamics with Synthetic Panels – Framework and Results Overview of Synthetic Panel Method9 (2) Let xij be a vector of household characteristics Let zj be the poverty line in period j, j= 1 or 2. observed in survey round j (j= 1 or 2) that are also We are interested in knowing such quantities as observed in the other survey round for household i, I= 1,… N. These household characteristics can (3a) include such time-invariant variables as ethnicity, religion, language, place of birth, parental which represents the percentage of households education, and others available in the survey. The that are poor in the first period but non poor in vector xij can also include deterministic variables the second period (considered together for two such as age (which given the value in one periods), or survey round can then be determined given the time interval between the two survey rounds), (3b) or time-varying household characteristics if retrospective questions about the round-1 values which represents the percentage of poor of such characteristics are asked in the second households in the first period that escape poverty round survey. To reduce spurious changes due in the second period. Put differently, for the to changes in household composition over time, average household, quantity (3a) provides the we usually restrict the estimation samples to joint probabilities of household poverty status in household heads age, say 25 to 55 in the first cross both periods, and quantity (3b) the conditional section and adjust this age range accordingly in probabilities of household poverty status in the the second cross section.10 second period given their poverty status in the Then let yij represent household consumption first period. or income in survey round j, j= 1 or 2. The linear If true panel data are available, we can easily projection of household consumption (or income) calculate the quantities in (3a) and (3b); otherwise, on household characteristics for each survey in the absence of such data, we have to rely on round is given by synthetic panels to study mobility. Two standard assumptions are then made to operationalize the (1) framework to construct synthetic panel data. First, the underlying population being sampled in 9  We provide an overview of the synthetic panel method developed survey rounds 1 and 2 are assumed to be the same by Dang, Lanjouw, Luoto, and McKenzie (2014) and Dang and in terms of the household characteristics xij; more Lanjouw (2013) in this section. For more details, readers are encouraged to read the original papers. specifically, it is assumed that xi1= wxi2, and yi1|xi1 10  This age range is usually used in traditional pseudo-panel and yi1|xi2 have identical distributions. Second, �i1 analysis but can vary depending on the cultural and economic and �i2 are assumed to follow a bivariate normal factors in each specific setting. 88 Bhutan Poverty Assessment 2014 distribution with correlation coefficient � and Since � is usually unknown in most contexts, standard deviations and respectively.11 If we can first approximate the simple correlation � is known, quantity (3a) can be estimated by coefficient between birth cohort-aggregated household consumption between the two surveys, then estimate � using the following formula (4) where �2�.� stands for the bivariate normal (5) cumulative distribution function (cdf) ) (and 2�.� stands for the bivariate normal probability It is then straightforward to estimate quantity density function (pdf)). Note that in equality (4), (3b) by dividing quantity (4) by , the estimated parameters obtained from data in both survey rounds are applied to data from where ��.� stands for the univariate normal the second survey round (x2) (or the base year) cumulative distribution function (cdf). for prediction. 11  In other words, this assumption implies that households in period 2 that have similar characteristics to those of households in period 1 would have achieved the same consumption levels in period 1 or vice versa. Bhutan Poverty Assessment 2014 89 Synthetic Panel Results Table B–1  Poverty Dynamics* for Two Periods, 2007-2012 Table B–2  Poverty Dynamics* for Two Periods, 2007-2012 (Joint Probabilities, %) (Conditional Probabilities, %) First Period, Second Period Poverty Status First Period-->Second Period Poverty Status Poor, Poor 8.3 (0.1) Poor--> Poor 44.1 (0.2) Poor, non-Poor 10.5 (0.1) Poor--> non-Poor 55.9 (0.4) non-Poor, Poor 4.1 (0.0) non-Poor--> Poor 5.0 (0.0) non-Poor, non-Poor 77.2 (0.2) non-Poor-->non-Poor 95.0 (0.1) N 6,045 N 6,045 * Based on synthetic data * Based on synthetic data Note: 1. Predictions are obtained based on data in the second Note: 1. Predictions are obtained based on data in the survey round. We use 500 bootstraps in calculating standard second survey round. We use 500 bootstraps in calculating errors. standard errors. 2. All numbers are weighted by population weights. 2. All numbers are weighted by population weights 3. Household heads’ ages are restricted to 25-55 years for the first 3. Household heads’ ages are restricted to 25-55 years for survey round and adjusted accordingly with the year difference for the first survey round and adjusted accordingly with the year the second survey round. difference for the second survey round Figure B–1  Chronic Poverty and Upward Mobility, by Sex of Household Head, 2007-2012 Percent of population Percent of poor population Female-headed Male-headed Female-headed Male-headed Poor-poor Poor-nonpoor 90 Bhutan Poverty Assessment 2014 Figure B–2  Downward Mobility by Sex of Household Head, 2007-2012 Percent of population Percent of nonpoor in 2007 Female-headed Male-headed Female-headed Male-headed Nonpoor-nonpoor Nonpoor-poor Figure B–3  Chronic Poverty and Upward Mobility by residence area, 2007-2012 Percent of population Percent of poor in 2007 Poor-poor Poor-nonpoor Bhutan Poverty Assessment 2014 91 Figure B–4  Downward Mobility, by Area of Residence, 2007-2012 Percent of population Percent of nonpoor in 2007 100 80 60 40 20 0 Nonpoor-nonpoor Nonpoor-poor Figure B–5  Chronic Poverty and Upward Mobility, by Employment Status, 2007-2012 Percent of population Percent of poor in 2007 60 40 20 0 Poor-poor Poor-nonpoor 92 Bhutan Poverty Assessment 2014 Figure B–6  Downward Mobility, by Employment Status, 2007-2012 Percent of population Percent of nonpoor in 2007 100 80 60 40 20 0 Nonpoor-nonpoor Nonpoor-poor Figure B–7  Chronic Poverty and Upward Mobility, by Employment Sector, 2007-2012 Percent of population Percent of poor in 2007 60 40 20 0 Agriculture only Wage work only Agriculture Any non-wage work Agriculture only Wage work only Agriculture Any non-wage work & wage work & wage work Poor-poor Poor-nonpoor Bhutan Poverty Assessment 2014 93 Figure B–8  Downward Mobility, by Employment Sector, 2007-2012 Percent of population Percent of nonpoor in 2007 100 80 60 40 20 0 Agriculture only Wage work only Agriculture Any non-wage work Agriculture only Wage work only Agriculture Any non-wage work & wage work & wage work Nonpoor-nonpoor Nonpoor-poor Figure B–9  Chronic Poverty & Upward Mobility, by Remittance Receipt Status, 2007-2012 Percent of population Percent of poor in 2007 No remittance Remittance No remittance Remittance Poor-poor Poor-nonpoor 94 Bhutan Poverty Assessment 2014 Figure B–10  Downward Mobility, by Remittance Receipt Status, 2007-2012 Percent of population Percent of nonpoor in 2007 100 80 60 40 20 0 No remittance Remittance No remittance Remittance Nonpoor-nonpoor Nonpoor-poor Figure B–11  Chronic Poverty and Upward Mobility, by Land Ownership in Rural Areas, 2007-2012 Percent of population Percent of poor in 2007 60 40 20 0 Landless 2 acres or less 2-5 acres 5 acres or more Landless 2 acres or less 2-5 acres 5 acres or more Poor-poor Poor-nonpoor Bhutan Poverty Assessment 2014 95 Figure B–12  Downward Mobility, by Land Ownership in Rural Areas, 2007-2012 Percent of population Percent of nonpoor in 2007 100 80 60 40 20 0 Landless 2 acres or less 2-5 acres 5 acres or more Landless 2 acres or less 2-5 acres 5 acres or more Nonpoor-nonpoor Nonpoor-poor Figure B–13  Chronic Poverty and Upward Mobility, by Literacy, 2007-2012 Percent of population Percent of poor in 2007 60 40 20 0 Poor-poor Poor-nonpoor 96 Bhutan Poverty Assessment 2014 Figure B–14  Downward Mobility, by Literacy, 2007-2012 Percent of population Percent of nonpoor in 2007 Nonpoor-nonpoor Nonpoor-poor Figure B–15  Chronic Poverty and Upward Mobility, Education Achievement, 2007-2012 Percent of population Percent of poor in 2007 Poor-poor Poor-nonpoor Bhutan Poverty Assessment 2014 97 98 0 20 40 60 Pema Gatshel Trashigang Tsirang Dagana Gasa Samtse Lhuentse Trashi Yangtse Monggar Samdrup Jongkhar Zhemgang Bhutan Poverty Assessment 2014 Sarpang Haa Percent of population Trongsa Paro Punakha Percent of population Bumthang Wangdue Phodrang Chhukha Thimphu Poor-poor Nonpoor-nonpoor Pema Gatshel Gasa Trashigang Samtse Figure B–16  Downward Mobility, by Education Achievement, 2007-2012 Samdrup Jongkhar Sarpang Figure B–17  Chronic Poverty and Upward Mobility, by Dzongkhag, 2007-2012 Dagana Trashi Yangtse Poor-nonpoor Nonpoor-poor Tsirang Chhukha Monggar Haa Percent of poor in 2007 Zhemgang Lhuentse Wangdue Phodrang Bumthang Trongsa Percent of nonpoor in 2007 Punakha Thimphu Paro Percent 0 20 40 60 80 100 50 60 70 80 90 Pema Gatshel Rural Trashigang Urban Tsirang Female Dagana Male Gasa Regular paid Samtse Casual paid Lhuentse Unpaid family worker Trashi Yangtse Self-employed Monggar No education Samdrup Jongkhar Note: dashed line is national average 8 Yrs or less Zhemgang 9-12 Yrs Sarpang Haa >12 Yrs Trongsa Percent of population No remitance Paro Received remitance Punakha Landless Bumthang 2 Acres or less Wangdue Phodrang 2-5 Acres Chhukha Figure B–19  Upward Mobility, Bhutan, 2007-2012 5 Acres or more Thimphu Nonpoor-nonpoor Agriculture Manufacturing Figure B–18  Downward Mobility, by Dzongkhag, 2007-2012 Services Pema Gatshel Less than average Bumthang Population Groups Trashigang Chhukha Dagana Dagana Gasa Gasa Haa Tsirang Lhuentse Samtse Monggar Trashi Yangtse Nonpoor-poor Paro Lhuentse More than average Pema Gatshel Monggar Punakha Samdrup Jongkhar Samdrup Jongkhar Zhemgang Samtse Sarpang Sarpang Haa Percent of nonpoor in 2007 Thimphu Trongsa Trashigang Wangdue Phodrang Trashi Yangtse Paro Tsirang Bumthang Trongsa Punakha Wangdue Phodrang Chhukha Zhemgang Thimphu Bhutan Poverty Assessment 2014 99 100 Percent 0 2 4 6 8 Rural Urban Female Male Regular paid Casual paid Unpaid family worker Self-employed No education Note: dashed line is national average 8 Yrs or less Bhutan Poverty Assessment 2014 9-12 Yrs >12 Yrs No remitance Received remitance Landless 2 Acres or less 2-5 Acres Figure B–20  Downward Mobility, Bhutan, 2007-2012 5 Acres or more Agriculture Manufacturing Less than average Services Population Groups Bumthang Chhukha Dagana Gasa Haa Lhuentse More than average Monggar Paro Pema Gatshel Punakha Samdrup Jongkhar Samtse Sarpang Thimphu Trashigang Trashi Yangtse Tsirang Trongsa Wangdue Phodrang Zhemgang Annex C: Qualitative Assessment of Poverty Introduction administrators, and Dzongkhag Statistical Officers (DSOs) were consulted and relevant poverty The Bhutan Poverty Analysis 2012 Report documents used to select the most suitable shows that living standards continue to improve gewogs. The selected gewogs for each dzongkhag are in Bhutan, with the percentage of people below presented in Table C–1. the official poverty line falling from 23.2 percent Separate focus group discussions were con- in 2007 to 12 percent in 2012. However, while ducted for men and women in each selected Bhutan overall has made tremendous progress community, where possible, although some in poverty reduction, especially in reducing groups comprised both genders. The reason for rural poverty, some dzongkhags continue in separating the men and women was mainly to poverty or have a harder time reducing poverty. provide the women with an enabling environ- The drivers of this rapid poverty reduction, and ment to more candidly share their opinions on the reasons why some regions remain mired in issues of poverty and wellbeing from a gender poverty, are not generally well-understood. In perspective. addition to desk-based quantitative research, a To the extent possible, participants for the qualitative approach is necessary to understand focus groups were selected from poor households the problems. and non-poor households. The poor households This annex presents the findings from 13 comprised participants with small land holdings, focus group discussions (FGDs) organized in four some were agriculturists, or derived livelihoods dzongkhags of Bhutan during January and February from nonfarm activities, and some were 2014. The four dzongkhags are Dagana, Zhemgang, women heads of households. Participants for Pema Gatshel, and Lhuentse. The findings from the better-off households were shopkeepers, these discussions complement the desk-based cash-crop owners, non-poor farmers, small quantitative research in order to deepen the businessmen, contractors, etc. In-depth insights understanding of poverty dynamics in Bhutan. into the communities were also obtained through interviews with the key informants, including Focus Group Approaches and Methods local leaders such as gewog heads (gups), sector For the purpose of the focus group discussions, heads in the community such as the Renewable the four dzongkhags were selected based using the 2012 Bhutan Poverty Analysis Report. In each of Table C–1  Gewogs and Dzonkhags Selected for Focus Group Discussions the dzongkhags two communities were identified and these were represented by as near to equal Dzongkhag Gewog numbers of men and women as possible. A gewog Dagana Kana and Drujeygang (which serves as the administrative center for a Zhemgang Nangkhor and Phangkhar group of villages, or chiwogs) was therefore taken Pema Gatshel Shumar and Khar as a “community”. The dzongkhag administration, Lhuentse Gangzur and Metsho dzongkhag planning officers, respective dzongkhag Bhutan Poverty Assessment 2014 101 Natural Resource (RNR) extension agents (of as deprivation12 of basic necessities required for a the Agriculture Ministry), health workers, and decent living. For most of them basic necessities local committees such as village women’s groups, meant having sufficient food to eat, water vegetable groups, and interviews with the elderly accessibility, and a house to live in. Deprivation from the community typically known as the “Go- of these necessities affected the wellbeing and Shey Nyen-Shey”. likelihood of a household’s vulnerability to A similar format for the questions was used in poverty. For most of the participants whose order to obtain a standardized to set of responses livelihood system was dependent on cash crops, so as to understand the various communities’ poverty was about risk and vulnerability to pests perceptions of poverty, the strengths of the and diseases, natural disasters such as drought, communities, factors hindering or helping com- irrigation constraints, and human-wildlife munity prosperity at the community level, and conflict. The situation was further constrained improvements or declines in income at household by the fact that communities lacked concrete level. The specific questions are: solutions to the problems and absence of coping mechanisms and strategies to offset the losses In your context what do you think of poverty? which directly affected their wellbeing. Community Level Primary indicators of poverty identified by Let us talk about your community. What is going well the participants: here? “Insufficient food to eat. We just work to eat”. “Not having an income earning source”. In the last five years what do you think about the “Lack of Income”. community? Is it “Children not able to eat”. • More prosperous? “No proper house to live in”. “Living in a bago • Remained the same? (bamboo hut)”. • Less prosperous? “Not having land”. Household Level Women’s Perspective on Poverty: Women generally perceived poverty as lacking money, Do you think that your income increased in the last having insufficient food to eat and drink, and five years? Yes or No. no proper house to live in. A few participants All discussions were recorded and notes considered money a key criterion to differenti- were taken simultaneously. A paired wise ate between rich and poor. For some, having no ranking matrix was used for the community endowments such as land, being a sharecropper level discussions to list down and identify the and relying on other’s land for livelihood is what top three community prosperity or declining poverty about. Similarly, for some households factors through a consensus-building approach. poverty is when members of the households, Summary findings and detailed findings from the including children, do not have sufficient food discussions are presented below. to eat and parents are not able to afford higher education for their children. Poverty is about old Summary of Key Findings age, lack of employment, with elderly left behind in the villages unable to do hard labour, thereby Community perspective on poverty: Focus group participants generally perceived poverty 12  Scarcity and not necessarily denial 102 Bhutan Poverty Assessment 2014 Table C–2  Gender Perspective on Poverty, by Dzongkhag and Community Participants Dzongkhag Community Male Female “Not having wealth and not having enough food to eat”. “Not able to work and inequality is poverty” “Loss of income to pest & disease hence Household members especially children not Drujeygang difficult to meet annual expenditure”. having enough to eat and not able to afford education for children “Having land but not able to work due to old Dagana age” “Lack of money. Having money is rich, having “Problem of sufficient food to eat, drink and no money is poor. Being rich also means clothes to wear” owning car, land, everything house” “Not enough money as compared to rich Kana “Having no land, being a sharecropper, who households who have money to finance has to depend on other’s land” education of their children”. “No road, no drinking water, people with no “Money is not alone sufficient. We need both education” hands, both feet” “Not having road connection. Poverty is about “Poverty is having very small amount of land, poor family who cannot afford to send their not sufficient even for making a vegetable children to school” garden”. “Not being equal with others, no access to “Having no land, depending on other’s land for Nangkhor drinking water and a house to live in”. cultivation” “Poverty is about remote people who face “Having not enough land, having no income. problem of not having enough land, and Poor background cannot provide good condition of the house is not good” education to children because of less income”. “Poverty means no road, long distance to “Poverty is lack of opportunity and no road, marketing, wastage of agricultural products, Zhemgang housing problem, shortage of meals, and lack no proper living and housing condition and not of facilities to the people” enough to eat and drink”. “Not proper housing and not sufficient food “Having no electricity and no property” to eat” Phangkhar “No basic necessities like food, clothing and “People who could not earn income and have shelter and problem in sending children to never earn income by themselves” school”. “Being in a poor family. Poor family means people who cannot express to community and   to the government” “ Having no income, no cash, not sufficient to “Low income, doing hard work in the field but eat and not having an income earning source” earning no income” Shumar “Inability to deal and protect crops from pest “Not having sufficient food to eat, lack of and disease, wild animals” facilities like schools and hospitals” Pema Gatshel “Loss of produce to wild animals” “For me poverty is the wild boar”. Khar “Insufficient food to eat. We just work to eat”   “Insufficient labour to work due to old age”   “ Poverty is women without husband, without   road, working hard but not earning any income” Gangzur “Poverty is due to small landholding, having   Lhuentse labour force to work but limited land” “Poverty is due to land defragmentation”.   “When households do not have land but they Metsho   lease in the land of their relatives.” Source: Poverty Qualitative Assessment, 2012 Bhutan Poverty Assessment 2014 103 age directly harming their livelihood. Women principal food crop in addition to community’s also perceived poverty as isolation caused by re- popularity for oranges. In Pema Gatshel maize moteness, lack of accessibility to markets, and is widely grown but orange is the main cash lack of health and educational opportunities. crop. The community is also very rich in mineral Men’s Perspective on Poverty: Men’s resources such as gypsum. Lhuentse is charac- perception of poverty tend to converge with terized by rugged terrain and land endowments those of women in terms of limitation of income, are restricted to small holdings and subsistence deprivation of basic necessities for a decent living agriculture is predominant. Most of the commu- in addition to remoteness, lack of road network, nities under study have access to basic facilities lack of electricity, health and educational such as road network, power supply, mobile con- facilities. More precisely men identified poverty nectivity, a Basic Health Unit (BHU), a school as being in a poor family and constrained by the and access to Renewable Natural Resources household’s inability to express their plight to the (RNR) and livestock extension services. community or the government. Poverty is about Factors Affecting Community Prosperity: people who could not earn income and never have Despite the community’s richness in natural earned income by themselves. When discussing resources and endowments majority of the poverty, men spoke of it as the destruction of FGD participants identified factors which the principal income-earning source, such as they considered was hindering the community oranges and cardamom by pests and diseases and prosperity. These factors included lack of lack of knowledge to cope with such disasters. irrigation, vulnerability of crops to pest Some participants identified poverty as not and diseases, market in-accessibility, small having wealth, having only small land holdings, landholding, human wild life conflict, absence and female-headed households with no male of road networks, lack of access to rural credit, members to work. lack of labour force, declining social capital Livelihood Resources: The livelihood portfolio and cohesion, impact of mining industry, lack in the study area is diverse and most commu- of school, and rural to urban migration. These nities identify subsistence agriculture, land, factors have significantly affected the community livestock, cash crops, industries, organic farm- in different ways. During each of the focus group ing and employment in nonfarm sectors as their discussion factors identified by the participants livelihood resources. Subsistence agriculture were listed and a paired wise ranking matrix was characterized by labour-intensive traditional used through a consensus building approach methods of farming is common in all the com- to identify the top three factors specific to and munities. Livelihood for community in Dagana affecting the community. is predominantly dependent on cash crops such In Dagana vulnerability of cash crops to pest as oranges and cardamom which are exported and diseases have affected the annual income to India. In addition communities also grow of the community. Market inaccessibility due to maize and vegetables such as beans, cabbages, longer distance to the market, poor quality of farm cauliflower for self-consumption. Communities road and inaccessibility throughout the year was in Nangkhor in Zhemgang grow rice although limiting opportunities for farmers to exploit the the area is also suitable for cardamom and sup- benefit of organic farming. Drought and drying of port organic farming. In lower regions of the natural streams and human wild life conflict have dzongkhag such as in Phangkhar maize is the also significantly affected this community. 104 Bhutan Poverty Assessment 2014 In Nangkhor community of Zhemgang provided little or no benefits to the community. irrigation constraints, market inaccessibility Employment opportunities for the local and small land holding were the top three community were limited, constrained by the factors affecting the community experience with lack of skills of the local people in operating prosperity. Women had a somewhat different heavy equipment. The community observed view. For them it was the small land holding, lack that the dust, pollution and frequent explosions of irrigation and poor transport infrastructure in the mining site was damaging their crops, (road) affecting the community. Similar problems and affecting water scarcity. In Lhuentse both of drying streams were reported forcing farmers the communities identified small land holding, to increasingly depend on monsoon hence market inaccessibility and lack of irrigation labour intensive farming was impossible for although there were differences in the ranking of households who lacked enough labour force. these factors. Market inaccessibility was as a result of poor Across the four dzongkhags and the road network (farm roads), longer distance to the communities most of these factors frequently domestic market and often problem in economies featured during the discussion in both the groups of scale in production due to smaller land holding. disaggregated by gender. The most common Community from Phangkhar in lower factors identified were: Zhemgang identified absence of road network, 1. Lack of Irrigation limited higher educational facilities and 2. Pest and diseases affecting the cash crops vulnerability to wild life as the principal factors. (oranges and cardamom) Women identified lack of credit facilities limiting 3. Market inaccessibility and Small Land entrepreneurial opportunities in addition to Holding absence of road network and educational facilities. 4. Human wild life conflict In Pema Gatshel both the communities 5. No road network &no education facilities from Shumar and Khar identified similar and limited access to credit factors affecting the community experience 6. Impact of mining & labour shortage & with prosperity. These included vulnerability poor infrastructure (road) of high valued cash crops to pest and diseases, Poverty and Household Wellbeing: At the lack of irrigation, and human wild life conflict. household level participants both men and The difference was, for the Khar community women generally believed that their income labour shortage was impacting them the most improved now as compared in the past. This as compared to wild animals attacking crops was possible because household members could for the Shumar community. However women now diversify their sources of income through participants from Shumar community identified nonfarm activities by working as daily wage two new factors including the negative impact of workers, as small contractors and selling livestock mining and limited access to credit in addition to products and vegetables in small quantities to pest and disease affecting the cash crops as the public servants at the gewog centers. top three factors. Majority of the participants also observed In Pema Gatshel gypsum is one of the main that daily wages for both skilled and unskilled natural resources where mineral is extracted and workers have increased with increase in demand exported to India. Majority of the participants for such workers. Some of the emerging issues both men and women observed that mining affecting the households were use and abuse Bhutan Poverty Assessment 2014 105 of alcohol, limited social cohesion, lack of self- community prosperity were identified which is help groups and increasing trends in divorce. presented in Table1. Such emerging issues at times shifted the entire Common Factors Affecting Community burden of raising family including children on women. Female headed households with no male Prosperity members also found it challenging to exchange Some common factors derived from Table C–1 labour in the neighborhood because of preference affecting the community prosperity were identi- and expectations of reciprocal arrangements of fied based on the number of times it occurred as labour contributed by a male workforce. shown in Table C–2. Across the 13 focus groups, Factors Affecting Community Prosperity: The lack of irrigation significantly affected communi- focus group participants identified the strengths ty prosperity cutting across men and women. Pest of their own community in terms of factor and diseases affecting the cash crops came as the endowments such as land, fertile soil, favorable next important factor followed by market inac- climatic conditions, production and trade in high cessibility & small land holding. While all factors value cash crops like oranges and cardamom, were common across men and women, limited and developments in physical infrastructure. access to credit was one factor women raised as Participants recognized that investment in a hindrance to community prosperity. Human physical infrastructure by the government like wild life conflict is of course a pervasive problem development of road networks including farm across all the communities but it did not strongly roads construction, provision of electricity, feature in the list of top three factors across all mobile connectivity, and access to drinking communities, men and women. water have improved their living conditions as The dzongkhag and community-wise well helped them diversify their income earning presentation of top three factors affecting opportunities/portfolios. community experience with prosperity showed These developments over the last five years male participants in Dagana identifying pest enhanced improvements in people’s lives as and diseases affecting their cash crops, market households observed a rise in income, facilitated inaccessibility, and irrigation constraints. Women easier access to public services, increased access participants identified human wild life conflict to health and educational facilities, access to as the third factor affecting the community essential goods and services, RNR support prosperity in addition to irrigation constraints, services amongst others. However, despite the pest and diseases affecting the cash crops. The community richness in the factor endowments preference for market inaccessibility was the least and the developments in physical infrastructure for the women group. and improved accessibility participants For the male communities of Nangkhor expressed their inability in having fully exploited gewog under Zhemgang dzongkhag, irrigation the endowments which otherwise would have constraints, market inaccessibility and having helped them enhance their livelihood. The small land holding were the top three factors focus group participants identified range of affecting the community mobility. Women had factors which according to them was hindering a somewhat different view. For them it was community prosperity. By using a paired wise the small land holding, lack of irrigation and ranking matrix through consensus building poor transport infrastructure (road) affecting approach a list of top three factors hindering the the community wellbeing. Similarly male 106 Bhutan Poverty Assessment 2014 Table C–3  Top Three Factors Affecting Community Prosperity, by Dzongkhag and Selected Community Participants Dzongkhag Community Male Female Pest and disease affecting cash Lack of irrigation crops Dagana Drujeygang Pest and disease affecting cash Market inaccessibility crops Lack of irrigation Human wild life conflict Lack of irrigation Small land holding Nangkhor Market Inaccessibility Lack of irrigation Small land holding Poor infrastructure (Road) Zhemgang No road access No road access Phangkhar No higher education facilities Limited access to credit Human wild life conflict No higher education facilities Pest and disease affecting cash Impact of mining crops Shumar Lack of irrigation Limited access to credit Pest and disease affecting cash Human wild life conflict Pema Gatshel crops Lack of irrigation   Khar (Male Female: Labour shortage   Combined) Pest and disease affecting cash   crops Small land holding   Gangzur (Male Female: Market inaccessibility   Combined) Lack of irrigation   Lhuentse Market inaccessibility   Metsho (male Female: Lack of irrigation   Combined) Small land holding   Source: Poverty Qualitative Assessment, 2014 Table C–4  Common Factors Hindering Community participants of Phangkhar community identified Mobility* lack of transport (farm road network), no higher Frequency of education facilities and human wild life conflict as Sl. No Factors occurrence the principal factors. The women group identified 1 Lack of irrigation 11 and ranked lack of transport (farm road), access Pest and disease affecting cash 2 5 to credit and no higher education facilities as the crops. Market inaccessibility & small top three factors. 3 4 land holding In Pema Gatshel both the communities 4 Human wild life conflict 3 identified similar factors affecting the community. No road network &no education 5 facilities and limited access to 2 These included pest and diseases affecting the credit cash crops, lack of irrigation, and human wild Impact of mining & labour 6 1 shortage & poor infrastructure life conflict. The difference was, for the Khar community the group identified labour shortage * Derived from community responses in Table C–2 Bhutan Poverty Assessment 2014 107 impacting them the most as much as wild animals basic needs, buying of essential items, meeting attacking crops for the Shumar community. the educational expenses of their children, and However women from Shumar community even financing of higher education of their identified two new factors including the negative children amongst others. impact of mining and limited access to credit in In Dagana participants from both the addition to pest and disease affecting the cash communities reported loss of the cash crops to crops as the top three factors. In Lhuentse both pest and diseases. This has reduced the total yield the communities identified similar factors such and subsequently affected the total annual income as small land holding, market inaccessibility and of the households. The communities also lacked lack of irrigation although there were differences coping strategies to offset the loss of income in the ranking of these factors. from pest and disease although some households Besides the top three factors hindering the reported having shifted to pulses (dal) farming community prosperity the most, there were factors which they export to buyers in India. which emerged strongly during the discussion but In Zhemgang the pest and disease severely did not feature in list of three factors during the affecting the cash crops was however not consensus building process. Some of these factors reported. In Pema Gatshel both the communities include, rural to urban migration, declining mentioned pest and disease having affected their conditions of social capital and cohesion, lack of cash crop and hence the income. Men ranked it as self-help groups, use and abuse of alcohol, natural the top factor while women ranked it as the third disasters, increasing trend in divorce cases etc. factor affecting their livelihood. Both male and The impact of top three factors and the emerging female participants believe that dust from the issues affecting the community prosperity are gypsum mine responsible for causing damage to discussed in detail below under economic, social the cash crop. and environmental factors. Market Inaccessibility: Participants perceive increasing and potential opportunities in organic Economic Factors farming to supplement cash income when their Loss of Income to Pests and Diseases: The principal income sources have been affected dzongkhags of Dagana, Pema Gatshel and by pest and diseases and irrigation constraints Zhemgang are located in the southern belts of have affected agriculture. They also immediately the country. The sub-tropical climatic conditions recognize the absence of a market generally and soil quality make the dzongkhags suitable for characterized by limited buyers. The community’s growing cash crops like oranges and cardamom. understanding of a market also remotely extend Most commonly the lower regions of the beyond the local market such as the nearest town, dzongkhags are popular for producing oranges the dzongkhag headquarters and to the extent while in the higher altitudes of Dagana cardamom possible the national market such as Thimphu is also grown. For the people of Dagana cardamom and border towns of Phuentsholing, Gelephu, and and oranges are therefore the two main sources Samdrup Jongkhar. Communities have limited of income, compared to mainly oranges for the knowledge of export markets such as India communities of Pema Gatshel and Zhemgang. although they recognize lack of competitiveness The livelihood of these communities is dependent of their products as a result of cheaper on the annual income from the sale of cash crops. alternatives available from India. Participants Households use the annual income in meeting the recognized both external and internal factors 108 Bhutan Poverty Assessment 2014 “We even tried producing and selling local vegetables however potential buyers lack interest in our vegetables because they say that our vegetables lack quality even though we think that our prices are reasonable selling a Kg of Cabbage at Nu20 and a bundle of broccoli at Nu 50. For example I and my friend hired a Bolero pickup truck paying Nu 4,000 as transport charges in delivering our vegetables including broccoli and cabbage to Punakha but could not sell because buyers and customers were not interested in our products” – A male FGD participant from Drujeygang gewog. “The income from the oranges has gone down from average Nu 50,000 to Nu 20,000 so we do not know what to do next. May be we should plant mountain hazelnut as an alternative. Heard that a weather condition of our area is similar to that of Lingmithang in Monggar and it might work here” – A male FGD participant, Shumar Pema Gatshel “Many orange orchards are damaged by disease these days. Now oranges are not even available to eat. First it affected the trees in Denchi village and shifted upwards. Many people of our locality believe that dust from Gypsum powder factory leads to dying of the crops as well as orange” – A female FGD participant, Shumar Pema Gatshel “I didn’t see the insect but the root of the orange tree has been damaged”– A female FGD participant, Shumar, Pema Gatshel. “According to the agriculture sector the solution to the disease, is after many rounds of discussion, we have been advised to cut down all the orange trees even if all the trees in the orchard are not affected. If even one tree is affected rest of the trees also has to be cut down and burnt. The government is providing free orange saplings. Now farmers are apprehensive to the advice because the question is how they would manage without income until the new trees start bearing fruits. It takes at least five years to start bearing fruits”– A key informant, Shumar, Pema Gatshel rendering market inaccessible. Factors such the control of the community. The community as longer distance to a potential market, poor also lacked production, marketing, packaging and quality of the road, high cost of transportation, handling skills as internal constraints resulting in and absence of marketing infrastructure, and non-competitiveness of the products. cheaper alternatives were external and beyond Small Land Holding: Participants from the Bhutan Poverty Assessment 2014 109 formal financial institutions require immovable “We do not have sufficient land. property such as a land or a house as collaterals Small landholding is a problem. Why while extending credit facilities in absence of because of distributing the land to micro finance institutions. The participants feel the children resulting in land getting that small land holding and inflexible credit smaller and smaller due to division/ requirements limit innovations, entrepreneurial defragmentation” A male FGD opportunities with lack of economies of scale in production. Small land holding is also the participant, Gangzur community. outcome of land defragmentation as a result of “Getting loan is also a problem land division among family members. because collaterals are required. Limited credit facilities: Due to the seasonality of agriculture production and downturn in the Suppose if we think we can get some production of cash crops communities greatly loan to buy livestock (pig) and earn recognize the importance of working capital and some income it is not possible. We consumption credit. The difficulty in accessing are asked if we have land or not. credit facilities in the rural areas limit business When we say no we do not have land, opportunities for young entrepreneurs, small then we are told that we would not be eligible for the loan. In place like Thimphu it would be easy to get loan “Our vegetables are not competitive simply by mortgaging a building against the one imported from and can make profit from it” – A border town of India because it is male FGD participant, Gangzur said that our cabbages contains lot community. of water inside, the cabbages are not green, the leaves are yellowish and Nangkhor community in Zhemgang pointed due to poor quality of road vegetables out their constraints to economies of scale in get damaged while transporting production due to small land holding. Otherwise them to longer distance” – A male they acknowledge the favorable climatic FGD participant, Drujeygang gewog. conditions, soil fertility for growing different types of vegetables, and cash crops like cardamom. “Although we have now, electricity, Small land holding is also the consequence of road but the prices in the market land defragmentation because of divisions among have increased. Things are very family members. Female participants defined expensive. We do not have a market poverty in relation to small landholding not even to sell our products like vegetables. sufficient for a vegetable garden and depend on People are far off from the market in others land for cultivation. In Lhuentse both the the cliffs. We do not have even place communities pointed out the implications of small to keep tourist. They all return to landholdings limiting opportunities to access Monggar” – A male FGD participant, credit from formal financial institutions like the Gangzur community, Lhuentse Bhutan Development Bank Ltd. (BDBL). The 110 Bhutan Poverty Assessment 2014 contractors in diversifying their economic activities. Formal financial institutions have “We also request for farming collateral requirements which make it not machineries because of water feasible for the farmers to access loan in absence scarcity we depend on monsoon of specific collaterals demanded by the bank. It rain to transplant paddy. When is not lack of collaterals. In fact farmers have there is monsoon everyone in the specific collaterals such as small land holding, village start transplanting paddy labour, but which are not acceptable to the bank. Limited access to credit was one of the reasons and therefore we cannot exchange cited by women group of Phangkhar community labour and sometimes we have to in Zhemgang. According to the participants keep our land fallow” – A male FGD men earn income from different sources such participant, Nangkhor Community as construction works, as daily wage earners working in the road side and engage in other “During summer we face problems nonfarm activity. In contrast women sources of of paddy transplantation as the income is limited to daily wages from carrying water on irrigation channel dries oranges. Since the community do not have access on the way to our fields and we to road woman carry oranges from villages until have to depend on the monsoon the highway and is the only source for them once and every households start planting a year during the harvest season. when it starts raining so we face In the remaining part of the year women labour shortage” – A male FGD mend their fields. Since the orange yield has been participant, Nangkhor Community affected by pest and disease their income has also been affected. Women see potential opportunities through self-employment. Young women also households and exchange of labour is impossible. have been trained in tailoring, hair dressing, Sometime households with fewer labour forces beauticians, etc. through government’s rural cannot carry out the transplantation leaving the poverty reduction strategies but credit facilities is land fallow. a big constraint for them. Social Factors Labour shortage (Labour as input & number of labourers): Community understanding of labour The community understanding of poverty was shortage differed according to the situation not limited to income alone. Some of the factors affecting the labour intensive agriculture sector. that emerged during the discussion were social For some community it is the decreasing input as in nature. A factor such as lack of accessibility to labour because of old age since young people have market was due to poor road infrastructure and migrated to urban centers leaving the elderly absence of road network in some community. behind. In comparison Nangkhor community A community in Zhemgang pointed out lack of in Zhemgang believed limited work force in the higher educational facilities in the community households affecting the labour as input affecting affecting their livelihood because a huge amount their livelihood because the community is entirely of recurring expenditure was incurred on children dependent on monsoon for transplanting rice. in arranging education facilities outside the During monsoon transplanting coincides across community. Bhutan Poverty Assessment 2014 111 Some of the emerging social issues identified by the participants were use and abuse of alcohol, “One of the main problems is that increasing trend in divorce, limited number we are not able to protect our crop of self-help groups, lack of social capital and from wild animals such as wild social cohesion and rural to urban migration. boars and monkeys. On top of that The abuse of alcohol is mostly associated with important cash crops like oranges men. According to some women participants a and cardamom have been affected by considerable income is spent on buying alcohol diseases and their yield have declined which is easily available in the market. It not only over the years. These problems have impacts the income but also affects the household also resulted some of the households wellbeing because participants think that alcohol is also associated with increasing trend in divorce to migrate to urban areas. When one and other social problems. Divorce and alcohol household migrates other households related challenges are pervasive in nature and also think of migrating and the present in most of the communities. There is a area and the farm of the migrated general agreement that social capital and cohesion household when not maintained it among communities is slowly degenerating. The turns into thick jungle making it existence of informal network is very helpful but easier for wild animals to attack it demands reciprocal treatment. Women headed crops in the locality” – A male FGD households are vulnerable to poverty because participant, Kana Community, reciprocal arrangements are expected for male Dagana labour contribution. When discussing rural to urban migration “I also feel that we do not have participants pointed out absence of permanent educated people remaining in the migration in large numbers. Of course few community because they all are in households in the community have migrated to the urban centers employed in some urban centers accompanying their children. Rural sectors or doing some business”. – to urban migration according to the community A male FGD participant, Kana is mostly common among educated youths Community, Dagana who move to urban centers in search of better opportunities. It is therefore not on account of surplus labour generated by the agriculture sector. The agriculture sector is unattractive to youths “In my opinion if we have access to and labour shortage continues to be a constraint better extension services (marketing) for the sector with mostly elderly labour force staying behind. access to credit, provision of seeds, hybrid cattle, provision of cattle Environmental Factors feed, pesticides, enough water for irrigating our vegetable garden Irrigation Shortage: Lack of irrigation is would solve most of the problem” – A discussed under this section because most female FGD participant, Kana gewog communities attributed the problem to 112 Bhutan Poverty Assessment 2014 environmental destruction, climate change, and drought. Irrigation emerged as principal factor “We have only a primary school that affecting all most all the communities in the was established some 30 years ago. four dzongkhags. In Lhuentse and Pema Gatshel Upgrading of the schools will have irrigation shortage has affected both domestic benefits such as we can sell some consumptions as well as affected farming. While of our local produce and we do not for the Nangkhor community in Zhemgang the have to send our children to school shortage has mostly affected farming. which is very far. It incurs huge Irrigation shortage has also linkages with additional cost in transport, living the dying of the cash crops in Dagana and Pema arrangements, frequent buying of Gatshel because firstly these cash crops require irrigation in absence of rain. Secondly organic school uniforms, shoes when we have farming is the only alternative for communities to send our children to other schools. to offset the loss of income to pest and dis- If we do not have to change schools eases. In Nangkhor, the drying of the streams so frequently school uniforms, shoes has left farmers to depend on rain. In absence last for many years and children can of timely rainfall and lack of labour force wet stay with us and attend classes” – A lands have remained fallow. In Pema Gatshel male FGD participant, Phangkhar too, drying of streams have rendered some 30 community, Zhemgang. acres of paddy field fallow. The unique prob- lem in Pema Gatshel is that the streams are “There is no up gradation of the frequently shifting their locations downwards school in the Gewog due to which we every year and households located in higher al- have to send our children to other titude have water scarcity. far off school. We face financial Vulnerability and Risks: Crop losses due to problems. The BHU also has male pest and diseases, and wild life and natural di- health assistant. We women face sasters like storms, earthquake and drought problem in discussing our health make the community vulnerable to poverty. The issues” – A female FGD participant, principal risks in agriculture across all the com- Phangkhar community, Zhemgang. munity was identified as wild life attacking both food and cash crops. Farmers are left to harvest sometimes only the remnants of the crops. The Concluding Observations community believes that increasing conflict is as a result of human encroachment due to de- The findings of the focus group discussion forestation, construction of roads, erecting of present similarities in patterns of factors electricity poles and other developmental activi- hindering the community prosperity. Lack of ties. Farmers have no access to compensation irrigation, vulnerability of principal crops to for the damage given the challenges in assessing pest and diseases, market inaccessibility and loss the extent of the damage caused and in absence of crops to wild animals amongst others were of crop insurance. perceived as important conditions of community wise experience with decline. The findings suggest that these common factors are often external in Bhutan Poverty Assessment 2014 113 “For this community maize has been principal food from time immemorial. People use to grow maize and eat as a special diet to work in the fields but due to drought we could not harvest like before which has affected our food security” – A male FGD participant, Drujeygang, Dagana “We have lot of wet land for paddy cultivation but now the water sources have started drying up and there is limited volume of water left for sharing among households. Lack of irrigation channel is a problem on top of that due to which wet land remains fallow” – A female FGD participant, Drujeygang, Dagana “Nowadays we have been experiencing hot weather with rise in temperature and may be this is because of lot of constructions works going on and building of factories elsewhere which causes pollution. Water sources have been drying up because may be we are using excessive wood for construction of houses and blasting. Even the taste of oranges is not that sweet like before may be because of the heat”– A female FGD participant, Drujeygang, Dagana “The dust from the mining might have affected the oranges. Drinking water was really not a problem but may be because of the dust and the bombings the source of drinking water is shifting downwards. Dust has affected the trees, animal fodder and because of no rain dust does not settle”– A male FGD participant, Shumar community Pema Gatshel nature and beyond the control of the community immediate solutions may be, sharing of best who are less endowed with technical knowledge practices, lesson learning experiences and and expertise in immediately solving the problems exploring and diversifying alternative livelihood by themselves. strategies. Problems of irrigation and drought show There are also opportunities for improving the community’s vulnerability and limited resilience community livelihood. Most of the community to the forces of climate change although the have the required infrastructure in place such country has abundant fast flowing rivers but as road, electricity, mobile services access to beyond the rich of these specific communities. health and educational facilities. The agriculture The cash crops vulnerability to pest and diseases extension services provided by the RNR sector in demands a better understanding of the causes provision of seeds, fertilizers, and technical skills and requires a long-term solution that is also in marketing of organic products are significant. acceptable to the community. Possible and 114 Bhutan Poverty Assessment 2014 References Alkire, S. 2011. “A National MPI for Bhutan: Journal of Agricultural Economics, Three Trial Measures.” Oxford Poverty 91 (4), 1007–1021. and Human Development Initiatives, Duclos, Jean-Yves. 2009. What is “Pro- Oxford. Poor”?Social Choice and Welfare, Atkinson, A.B. 1987. “On the Measurement of 32:37-58. Poverty.”Econometrica, Vol. 55, No.4: Essama-Nssah, B. 2012. “Identification of 749-764. Sources of Variation in Poverty Benjamin, Dwayne, Loren Brandt, and John Outcomes.” World Bank Policy Research Giles. 2005.“The Evolution of Income Working Paper No. 5954. Washington, Inequality in Rural China.”Economic D.C.: The World Bank. Development and Cultural Change, Vol. Essama-Nssah., B. and Peter J. Lambert. 53, No.4: 769-824. 2009.“Measuring Pro-Poorness: Bhutan Multiple Indicator Survey. 2010. A Unifying Approach with New National Statistics Bureau, Royal Results.”Review of Income and Wealth, Government of Bhutan. Series 55, No. 3: 752-778. Bourguignon, François, Francisco H. G. Ferreira, FAO (Food and Agriculture Organization). 2011. and Nora Lustig. 2005. Introduction in Guidelines for Measuring Household François Bourguignon, Francisco H.G. and Individual Dietary Diversity, FAO. Ferreira and Nora Lustig (eds), “The Nutrition and Consumer Protection Microeconomics of Income Distribution Division, Food and Agriculture Dynamics in East Asia and Latin Organization of the United Nations, America,” pp. 17-46, Washington, D.C.: Rome, Italy. Available online at Based on Cross-Sections.” World Bank Firpo, S., N. Fortin, and T. Lemieux. Policy Research Paper number 6504, 2009. “Unconditional Quantile Washington, DC.: World Bank. Regressions,”Econometrica, 77, 953-973. Datt, Gaurav, and Martin Ravallion. Foster, James E., J. Greer, and E. Thorbecke. 1992.“Growth and Redistribution 2010.“The Foster-Greer-Thorbecke Components of Changes in Poverty (FGT) Poverty Measures: 25 Years Measures: A Decomposition with Later.” Journal of Economic Inequality, Applications to Brazil and India in 8: 491-524. the 1980s.”Journal of Development Gross National Happiness Report. 2010. Gross Economics, 38: 275-95. National Happiness Survey results Dercon, S., D.O. Gilligan, J. Hoddinott, and 2010, Available online at Poverty and Consumption Growth in Hettige, H. 2006. “When do Rural Roads Fifteen Ethiopian Villages,” American Benefit the Poor and How? An In-depth Bhutan Poverty Assessment 2014 115 Analysis Based on Case Studies,” Asian Development Research Group of the Development Bank, Manila, Philippines. World Bank: http://iresearch.worldbank. IMF (International Monetary Fund). 2010a. org/PovcalNet/index.htm?0 Bhutan: Poverty Reduction Strategy Ravallion, Martin. 1994.“Poverty Comparisons.” Paper. IMF Country Report No. 10/180. Chur (Switzerland): Harwood Academic Washington, D.C.: IMF. Publishers. IMF (International Monetary Fund). 2010b. Ravallion, Martin., and Shaohua Chen. Bhutan: Poverty Reduction Strategy 2003.“Measuring Pro-Poor Growth.” Paper – Joint Staff Advisory Note. Economics Letters 78: 93-99. IMF Country Report No. 10/181. RGoB (Royal Government of Bhutan). 2013. Washington, D.C.: IMF. Poverty Analysis Report 2012. Thimphu: Jenkins, S. and Peter J.Lambert.1997.“Three National Statistics Bureau. ‘I’s of Poverty Curves, with Analysis of ———2007. Poverty Analysis Report 2007. UK Poverty Trends.” Oxford Economic Thimphu: National Statistics Bureau. Papers, 49: 317-327. Shorrocks, A.F. 1999. “Decomposition Procedure Kennedy, G., T.Ballard, and M-C. Dop. 2011. for Distribution Analysis: A Unified “Guidelines for Measuring Household Framework Based on Shapely Value”. and Individual Dietary Diversity.” University of Essex and Institute for Nutrition and Consumer Protection Fiscal studies, Colchester, England. Division, Food and Agriculture Tiwari, S., E. Skoufias, and M. Sherpa. 2013. Organization (FAO) of the United “Shorter, Cheaper, Quicker, Better: Nations, Rome, Italy. Available online Linking Measures of HouseholdFood at and Tanzania.”Policy Research Working Khandker, S.R., Z. Bakht, and G. B. Koolwal. Paper 6584. Washington, DC.: World 2009. “The Poverty Impact of Rural Bank. Roads: Evidence from Bangladesh”, Ura, K., S.Alkire, T.Zangmo, and W. Wangdi. Economic Development and Cultural 2012. “An Extensive Analysis of GNH Change, 57, 685–722. Index”, Center for Bhutan Studies: Kray, Aart. 2006.“When is Growth Pro-Poor? Thimphu, Bhutan. Available online at, Evidence from a Panel of Countries.” Journal of Development Economics, Warr, P. 2010. “Roads and Poverty in Rural 80: 198-227. Laos: an Econometric Analysis.”Pacific National Statistics Bureau. 2013. Economic Review, 15(1), 152-169. Available gress on Sanitation and Drinking Water. Osmani, S. 2005.“Defining Pro-Poor Growth.” Water Supply and Sanitation Monitor- One Pager Number 9, International ing, Joint Monitoring Programme for Poverty Center, Brazil. Water Supply and Sanitation, 2013 PovcalNet: the on-line tool for poverty World Bank. 2010. Sustainable Transport measurement developed by the for All: Helping People to Help 116 Bhutan Poverty Assessment 2014 Themselves, Rural Transport, IDA at Work in Transportation –2010 Report. Available online at, Yamauchi, F., M. Muto, S. Chowdhury, R. Dewina, and S. Sumaryanto. 2009. “Spatial Networks, Labour Supply, and Income Dynamics: Evidence from Indonesian Villages.” International Food Policy Research Institute, Discussion Paper No. 897. Bhutan Poverty Assessment 2014 117