91666 National Statistics Bureau Royal Government of Bhutan The World Bank ISBN 978-99936-28-21-7 9 789993 628217 BHUTAN Poverty Analysis 2012 National Statistics Bureau Royal Government of Bhutan The World Bank ISBN: 979-99936-28-21-7 Copyright© National Statistics Bureau, 2013 www.nsb.gov.bt Photos contributed by Nima Tshering, Tashi Dorjee, Tshering Penjor and Dorji Phuntsho Design by Loday Natshog Communications Contents Acknowledgement iv Foreword v Executive Summary vii Chapter 1: Introduction 1 1.1. Background 1 1.2. Objectives 1 1.3.  Data Source 1 Chapter 2: Updating the Poverty Lines 5 2.1.  Updated Food Poverty Line 5 2.2.  Updated Non-food Allowance and Total Poverty Line 5 2.3.  Spatial Price Index 6 Chapter 3: Patterns in Consumption Poverty 9 3.1.  Poverty Rate 9 3.2.  Depth and Severity of Poverty 10 3.3.  Poverty trend 12 3.4.  Poverty by Household Characteristics 12 Chapter 4: Basic Needs 17 4.1. Education 17 4.2. Income 17 4.3. Health 17 4.4.  Household Amenities, Assets, and Access to Services 18 4.5.  Perception and Priorities 21 Chapter 5: Inequality 23 5.1.  Consumption Quintiles 23 5.2.  Gini Index 23 Chapter 6: Conclusion 27 Annex I: Additional Statistical Tables 30 Annex II: Technical Notes 36 iii Acknowledgement This poverty analysis report is prepared by a four-member team robustness tests for comparability with the 2007 poverty from National Statistics Bureau – Mr. Cheku Dorji (Dy. Chief indicators. They also reviewed and provided comments on Statistical Officer), Mr. Tashi Namgay (Statistical Officer), the draft report. Mr. Sonam Gyeltshen (Research Officer), and Mr. Cheda Mr. Kuenga Tshering (Director General) and Mr. Phub Jamtsho (Research Officer). Experts from the World Bank Sangay (Chief Statistical Officer) through their advice and – Dr. Aphichoke Kotikula (Sr. Economist), Dr.   Srinivasan guidance that better shaped the report also deserve special Thirumalai (Sr. Economist),and Dr. Hiroki Uematsu (Junior thanks. Finally, the office wishes to acknowledge the support Professional Officer) and Dr. Nobuo Yoshida (Sr. Economist) of the World Bank for all the technical and financial support greatly contributed by verifying the results and conducting in bringing out this report. iv Foreword The National Statistics Bureau (NSB) is pleased to present the program also extends the education of the poor children and “Poverty Analysis” Report (PAR) 2012. It is based on the data the support for elderly and needy citizens of the country. from Bhutan Living Standards Survey (BLSS) 2012 conducted These initiatives could have direct impact in improving the by NSB with the support of the Asian Development Bank. living standard of the poor. A key objective of this report is to prepare updated It is our earnest hope that the report will undoubtedly cast poverty estimates that are as comparable as possible with the light on a huge range of policy issues and that the poverty estimates prepared in 2007. Both the BLSS 2007 and 2012 statistics presented in this report will be used to design policies questionnaires were nearly identical and the data from both and programs aimed at improving the living standards of the the surveys were checked for comparability by the experts poor. from the World Bank. The variables included in estimating Finally, NSB would like to acknowledge with deep poverty lines and rates were also checked for robustness to appreciation the support of the Royal Government of Bhutan. confirm the comparability. Our sincere thanks extend to the World Bank for financial This report measures poverty in Bhutan in 2012, and and technical support in bringing out the PAR 2012. The data evaluates the change in poverty compared to 2007. It is analysis team deserves all the appreciation for the hard work reassuring to learn from the report that poverty, as measured and dedication exhibited in bringing out this report. by the percentage of poor has declined from about 23.2 percent in 2007 to 12 percent in 2012. Poverty is multi-dimensional and there is no single solution. In general, people are poor because they are stuck in circumstances which don’t allow them to get ahead. One of the reasons for poverty reduction can be attributed to the noble Royal Kidu Program. Through the program, many landless households were able to get land permanently registered in their names which changed their lives forever. Generally, Kuenga Tshering landless households are more vulnerable to poverty. The Kidu Director General v Executive Summary Poverty Rate About 50 percent of the non-poor adult population (15+) has PAR 2012 established the total poverty line at Nu. 1,704.84 not attended school/institute compared to about 70 percent of per person per month. The total poverty line is obtained by the adult poor population. adding the food poverty line of Nu. 1,154.74 to non-food Around 17 percent of the surveyed population reported allowance of Nu. 550.10. An estimated 12 percent of the that they had suffered from sickness or an injury in the four population is found to be poor. Thus, poverty has declined by weeks prior to the survey, with no significant difference about half from the estimate of 23.2 percent in 2007. between the poor and non-poor. However, of this population, Poverty in rural areas (16.7%) is significantly higher than only about half (53%) of the poor visited a medical facility urban areas (1.8%). Only about three percent of the population compared to 69 percent of the non-poor. Among those is subsistence poor i.e., persons belonging to households with who gave birth during the 12 months prior to the BLSS per capita consumption below the cost of subsistence diet 2012, a smaller proportion of poor women in rural areas food. Poverty rates are observed to be high in Dagana, Samtse, received ante-natal care than non-poor women. However, a Lhuentse, Pema Gatshel, and Zhemgang. considerable proportion of women received pre-natal care in urban areas, even amongst the poor Household Characteristics Majority (98%) of the population have access to improved In both urban and rural areas, a poor household has a much water source with hardly any disparity existing between larger family size than a non-poor household. However, the poor and the non-poor households. At least 80 percent the number of households with large size is much less than households have access to improved sanitation; between poor the number of small sized households. Persons living in and non-poor households both in urban and rural areas the households where the head is currently working have higher disparity is around 15 percentage points. living standards than those living in a households whose head Nearly all (97%) households in urban areas, even the poor is either unemployed or out of the labour force. Among the use electricity for lighting purpose. In rural areas, however, employed, poverty levels are higher in households whose only 69 percent of the poor households have electricity as head works in agriculture. their primary source of lighting. Nationally, only 21 percent The poverty rate is about three percent for those below 25 among the poor households have TV compared to 59 percent years of age as compared to 14 percent for those aged 65 years in the non-poor households. and older. This may indicate a person’s inability to engage Most of the poor, especially in the rural areas, suggest that productively in economic activity with age. At least 68 percent road infrastructure and bridges, Commerce, transport and of the household heads in Bhutan are between 25 and 54 years communication and water supply should be the priorities old, while less than five percent are below 25, and about 13 for the government. In urban areas, poor households specify percent are 65 and above. housing, labour and employment creation, and land and About 44 percent of the poor live in households whose resettlement, as priority concerns head is engaged in agriculture; and at least 16 percent in households whose head is not actively participating in the Inequality labour force. On average, a person in the top 20 percent of the national population consumers 6.2 times more than a person in the Basic Needs poorest 20 percent of the population he poorest 20 percent of The analysis shows that the poor have a much lower (52%) the population. The Gini index, which measures inequality, literacy rate than the non-poor (65%). The literacy rate of the has remained almost the same at the national level (0.35 in poor in urban areas is 17 percentage points lower than the rate 2007 and 0.36 in 2012). However, it has slightly increased for for the urban non-poor while in the rural areas the rate for the the both urban (0.32 to 0.35) and the rural areas (0.32 to 0.34). poor is six percentage points lower than the rural non-poor. vii Chapter 1: Introduction 1.1. Background Chapter one describes briefly the BLSS 2012, which is The purpose of this report is to provide updated poverty the primary data source used in preparing these updated estimates for Bhutan using newly available data from the 2012 poverty estimates. Chapter two summarizes the work Bhutan Living Standards Survey (BLSS) 2012. Baseline done to update the 2012 poverty lines for inflation. Chapter poverty estimates were produced in 2003 and 2007 using the three presents patterns in consumption poverty. Chapter corresponding BLSS data. The updated poverty estimates four presents an analysis of socio-economic indicators that in this report can be used to monitor Bhutan’s success in provide an independent source of information on poverty reducing poverty during the past five years since the last reduction during the period 2007-2012. Chapter five provides poverty estimates in 2007. It is also useful for broadening and measures of income inequality (for example, estimates of Gini deepening our understanding of the changing dimensions of coefficient). Chapter six provides the report’s conclusions and Bhutan’s poverty and for designing appropriate interventions recommendations for future poverty monitoring. for poverty reduction and monitoring efforts. 1.3. Data Source 1.2. Objectives The data used for this report is from the BLSS 2012 which is the A key objective of this report is to update poverty estimates latest and third in a series of national household surveys that that are as comparable as possible with the estimates prepared have been conducted by the NSB. Like the previous rounds, for 2007. This involves the following steps: the BLSS 2012 followed the World Bank’s Living Standard • The 2012 poverty lines are updated for inflation in food Measurement Study (LSMS) Methodology. It is comparable and non-food prices during the 2007-2012 period. in size to the 2007 survey, but more than twice the size of the • New estimates of per capita household consumption survey in 2003. The BLSS 2012 surveyed 8,968 households are prepared that are as comparable as possible with across the country from a planned sample size of 10,000. It the consumption estimates prepared in 2007. provides the same level of detailed information needed to • The per capita consumption of each household in the prepare updated poverty estimates. The questionnaire that sample is compared to the updated poverty lines to was administered in both the BLSS 2007 and 2012 is similar.1 identify the poor and to calculate the relevant poverty Using the BLSS 2012 data, an aggregate of household indicators. consumption was generated and subsequently analyzed. This Although there are a few changes in the components of aggregate excludes household expenditures on durables, questionnaire in the BLSS 2012 compared to BLSS 2007, irregular expenses, health expenses (on consultations and robustness test carried out revealed that in both years the food consumption shares are roughly the same and the distribution 1 In the 2012 questionnaire, the ‘purchased’ item is broken into ‘purchased of major food items are comparable thus suggesting a high domestic’ and ‘purchased imported.’ The education expenditure in 2012 was the expenditure incurred in the last academic year unlike the 2007 survey that collected degree of comparability between the results for 2007 and 2012. estimated expenditure for the current academic year. 1 Bhutan Poverty Analysis 2012 hospitalization) from the total household consumption and rural areas were allocated across all dzongkhags and strata expenditures (found in the BLSS 2012 report), but includes in proportion to the number of households. The primary expenses on medicines. Details on the computation of this sampling units (PSUs) were blocks for urban (towns) areas consumption aggregate are provided in Technical Note 1 of and chiwogs for rural areas while the secondary sampling Annex-II. units (SSUs) were the households within the selected blocks/ The BLSS 2012 gathered data on household consumption chiwogs. expenditure, and as such, provides a means of assessing A set of household weights are needed when interpreting the level of poverty and well-being in Bhutan. Besides statistics from the BLSS 2012 household data. These weights collecting consumption expenditure data, it also collected are needed to correct for the varying area and household in data on demographic characteristics of household members, the survey design. They can be regarded as made up of three household assets, credit and income, remittances, housing, components: (a) a correction for the differing sampling rates access to public facilities and services, education, employment, of PSUs used in the strata at the area stage of sampling; (b) health of household members and prices paid for commodities. a correction for varying numbers of households selected in Also, it included an additional module on social capital and each PSU; and, (c) a correction for non-response. questions on happiness and self-rated poverty. The survey population coverage included all households in The sample households for the BLSS 2012 were selected the country except (a) diplomatic and expatriates households; on the basis of two mutually exclusive sampling frames for (b) institutional households, i.e., residents of hotels, boarding rural and urban areas. The total sample size was set to about and lodging houses, monasteries, nunneries, school hostels, 10,000 (comparable in scale to BLSS 2007) and allocated orphanages, rescue homes, and under trials in jails and indoor equally between the rural and urban areas to capture higher patients of hospitals; and, (c) barracks of military and para- variability of data in the urban areas. Sample sizes of urban military forces, including the police. 2 Chapter 2: Updating the Poverty Lines Bhutan’s poverty lines, defined in 2007, consist of a single capita expenditure distribution ensures that expensive nor national food poverty line and single non-food allowance cheap food items are heavily represented in the basket. After and refer to monthly per capita levels of food and non-food all, prices paid even of the same items could differ across consumption. Both the food poverty line and the non-food the population. Although food consumption patterns differ allowance measured in current prices must therefore be across the country, a single food basket was used to ensure updated for inflation, i.e., they need to be converted into 2012 a consistent comparison of welfare levels of people living in prices. This chapter of the report discusses the procedures different areas of Bhutan. The 2007 poverty line is updated for used to update the 2007 poverty lines. inflation to the year 2012. The methodology used to update for inflation involves (1) updating the food poverty line using 2.1. Updated Food Poverty Line the ratio of the food CPI is 2012 to the food CPI in 2007 CPI The poverty line, the minimum acceptable standard of per (2) using the food price data collected in the BLSS 2012 to capita consumption needed to assure a minimum standard estimate spatial (regional) differences in food prices in the of living, is obtained using the Cost of Basic Needs (CBN) survey year. The CPI is believed to be a reliable source of approach, a commonly used methodology for constructing information about inflation because of its rigorous collection. the poverty lines in many countries. This approach estimates Households (and their members) consuming (in real terms) the food component of the poverty line as the cost of a food less than the food poverty line, of Nu. 1,154.74 per person per bundle that provides a predetermined minimum required month are considered subsistence poor. level of food energy. The total poverty line is obtained by adding to the food component the cost of the non-food 2.2. Updated Non-food Allowance and allowance. Total Poverty Line The food poverty line is based on the estimated cost of a The 2007 baseline non-food allowance was estimated as the single national reference food bundle providing an average per capita monthly non-food consumption of households in subsistence diet of 2,124 Kcal per day (i.e., averaged over the reference population whose food spending was near the persons of all ages and both sexes).2 The reference food food poverty line. This is a conservative non-food allowance bundle was designed to reflect the actual food consumption because it represents non-food consumption that is at the patterns of Bhutanese in 2007 who consumed a diet yielding expense of food consumption that could otherwise be used approximately 2,124 Kcal per day. The food basket used in this to achieve the reference food bundle of 2,124 calories per day report is representative of the diet of a reference population, per person.3 namely population in the second, third or fourth decile based on nominal per capita consumption. The selection of households in the second to the fourth deciles of the per 3 Although persons with total per capita consumption below the food poverty line would have to sacrifice some food consumption to purchase non-food items, they would presumably substitute cheaper foods for more expensive foods within the 2 There are 53 food items in the food bundle reference food bundle. 5 Bhutan Poverty Analysis 2012 In order to update the non-food allowance for inflation Table 2.2.  Regional Price Deflator (Median of Household-level Paasche Indices), by Dzongkhag and Area in different regions, it is necessary to develop regional non- food price indices similar to the food price index. Estimates of Dzongkhag Urban Rural inflation in non-food prices developed in this report are based Bumthang 1.12 1.12 on non-food price data collected for the 2007 and 2012 CPI. Chhukha 1.10 1.02 Nationwide, the non-food allowance was estimated at Dagana 0.77 0.92 Nu. 550.10 per person per month. Adding this non-food Gasa 1.30 1.00 allowance to the food poverty line yields the total poverty line, Haa 0.97 1.01 estimated to be Ngultrum 1,704.84 per person per month, at Lhuentse 1.02 1.00 2012 prices. Monggar 0.87 0.88 Households (and their members) consuming (in real terms) Paro 1.12 0.95 less than the total poverty line, of Nu.1,704.84 per person per Pema Gatshel 0.87 1.03 month are considered poor. Punakha 1.02 0.95 Table 2.1 shows the comparison of poverty lines (food- Samdrup Jongkhar 0.80 0.82 poverty line, non-food allowance and poverty line) for 2007 and 2012 along with the inflation in the Consumer Price Samtse 0.75 0.81 Index (CPI). As mentioned, the 2012 food and non-food Sarpang 1.07 0.94 poverty lines are derived from the 2007 values by adjusting Thimphu 0.88 1.02 for inflation that occurred between 2012 and 2007. Trashigang 0.89 0.87 Trashi Yangtse 0.82 0.89 Table 2.1.  Poverty Lines of 2007 and 2012 and CPI Inflation Trongsa 1.00 1.03 Poverty lines 2007 2012 CPI inflation Tsirang 0.81 0.87 Food poverty line 688.96 1,154.74 1.68 Wangdue Phodrang 1.03 0.95 Non-food allowance 407.98 550.10 1.35 Zhemgang 0.89 0.86 Total poverty line 1,096.94 1,704.84 1.55 Bhutan 0.94 0.92 2.3.  Spatial Price Index Consequently, the average monthly household Prices differ across the country and therefore per capita consumption in 2012 for Bhutan was estimated at Nu.20,913 consumption expenditures (in nominal terms) across regions in real terms as a result of adjustments in differences in cost are not directly comparable. An important staple food like of living (and exclusion of some non-food expenditures rice is found to be much more expensive in Gasa than in on durable items and other irregular expenses). Average Wangdue Phodrang, so that a household in Gasa consumes monthly per capita consumption in real terms was estimated less with the same nominal consumption expenditure on rice at Nu. 5,493 per person per month. In 2007, average monthly than a household in Wangdue Phodrang. To make per capita household consumption was estimated at Nu.11,777 and consumption between regions comparable, values must be Average monthly per capita consumption in real terms was deflated using a cost of living index. However, no such index estimated at Nu. 2,745 per person. is available. The usual approach to controlling for spatial price differences is to use a price index that approximates the true cost-of-living index. One possible spatial price index is the Paasche index, which calculates the cost of buying a region’s basket of goods using base reference prices. A Paasche index was computed with food items using the BLSS 2012 median price data. Details on these computations are provided in Technical Note 1 (d). 6 Chapter 3: Patterns in Consumption Poverty Households with per capita real consumptions below the illustrates subsistence and poverty rates for population across poverty line are said to be poor and those with per capita urban and rural areas. These rates are poverty head counts real consumption below the food poverty line are subsistence i.e., the percentage of the poor persons. For the 2012 the poor. Subsistence poverty may be viewed as extreme poverty, total poverty rate for Bhutan is estimated to be 12 percent. i.e., those whose consumption expenditure is insufficient This means that, around one out of eight persons belong to even to meet basic food needs even if they devote their entire households whose per capita real consumption is below the consumption expenditure to food alone. total poverty line of Nu. 1,704.84 per person per month. It Consumption poverty in this report is measured at the can be observed that subsistence incidence, i.e. extreme household level since data from the BLSS 2012 does not poverty, is relatively small in the country: only about three allow intra-household analysis. Consequently, if a household percent of the population in Bhutan belongs to households is considered poor, then all its members are considered that are spending less per person than the food poverty line of poor. Similarly, if a household is non-poor, then none of its Nu. 1,154.74. Poverty in Bhutan is still a rural phenomenon members is poor. with about 17 percent of the rural population being poor as Three aspects of consumption poverty are of particular against only about two percent in the urban areas. While the interest: four percent of extremely poor persons in rural areas is quite • Poverty Incidence – the proportion of persons (or small, it is quite large in relation to that of urban area rate of households) identified as poor; 0.3 percent. • Poverty Gap (or Depth of Poverty) – the extent to which those identified as poor fall below the poverty Figure 3.1. Poverty and Subsistence Poverty in Bhutan line (in relation to the poverty line); 18 • Poverty Squared Gap (or Severity of Poverty) – a 16.7 Poverty Headcount 16 Subsistence Headcount measure of the inequality among the poor. These poverty measures are presented in this report 14 12.0 for the country as a whole, and for certain groups of the 12 population, such as for households in urban and rural areas, 10 Percent and in dzongkhags, and in by the sex of the household head, 8 among others. For more information on indices of poverty, see Technical Note 4. 6 3.9 4 3.1. Poverty Rate 1.8 2.8 2 The food poverty line and total poverty line are used to compute 0.3 subsistence and poverty incidence, respectively. Figure 3.1 0 Urban Rural Bhutan 9 Bhutan Poverty Analysis 2012 Table 3.1.  Population Poverty and Subsistence Poverty by Area Dzongkhag level estimates of poverty incidence and Poverty Subsistence Poverty subsistence poverty for the population and for households Area Population are shown in Table 3.3 (together with their standard errors). Standard Contribution Standard Contribution Share Rate Rate error to National error to National Ranks for dzongkhags are difficult to determine due to Urban 1.8 0.3 4.6 0.3 0.1 3.1 34.0 overlapping confidence intervals, but it can be observed that Rural 16.7 0.8 94.4 3.9 0.5 96.9 66.0 poverty rates are highest in Dagana, Lhuentse, Pema Gatshel, Bhutan 12.0 0.6 100.0 2.8 0.3 100.0 100.0 Samdrup Jongkhar, Samtse, and Zhemgang. However, the survey shows that Gasa and Paro have the least poverty. It is also important to observe the distribution of the poor Table 3.2.  Household Poverty and Subsistence Poverty by Area population. Among the dzongkhags, 17 percent of the poor Poverty Subsistence Poverty population resides in Samtse followed by Samdrup Jongkhar Population Area Rate Standard Contribution Rate Standard Contribution Share (9.1%), Chukha (8.8%) and Pema Gatshel (8.6%). error to National error to National In terms of the subsistence poverty, the rates are high Urban 1.4 0.2 5.5 0.2 0.1 4.2 34.0 in Lhuentse (11.1%) and Zhemgang (9.9%). In terms of the Rural 12.4 0.6 94.5 2.6 0.3 95.8 66.0 distribution of subsistence poor, Samtse and Zhemgang have Bhutan 8.6 0.4 100.0 1.8 0.2 100.0 100.0 the highest proportion of the subsistence poor population (Table 3.4). The poverty and subsistence poverty statistics are shown The estimated number of poor and subsistence poor in Table 3.1 together with their standard errors. Because households across dzongkhags are provided in Table 3.5 the poverty incidence figures are estimates from a sample and Table 3.6. These tables include the contribution of survey, it is important to consider their standard error when each dzongkhag to household poverty and subsistence evaluating the precision of these estimates. While the best poverty. Dagana and Lhuentse have high proportion of poor estimate of poverty rate in Bhutan in 2012 is 12 percent, this households. The largest proportion of poor and subsistence estimate has a margin of error of 1.2 percentage points. That poor households is in Samtse. is, we are 95 percent confident that the true poverty rate is between 10.9 percent and 13.2 percent. We are also confident 3.2.  Depth and Severity of Poverty that urban poverty, estimated at 1.8 percent (but could range Poverty analysis is not limited to examining poverty rates and between 1.2% to 2.3%) is much lower than rural poverty of comparing the statistics across sub groups of the population. 16.7 percent (that could range between 15.1% to 18.3%). In It is important to also look into the depth and severity of addition, we observe that about 95 percent of poor persons poverty. The poverty gap and poverty squared gap indices throughout the country reside in rural areas. Among the measure the depth and severity of poverty, respectively. For extremely poor, the proportion is even higher: 97 percent an individual, the poverty gap is the difference between the resides in rural areas. Consequently, efforts toward poverty poverty line and actual per capita expenditure (the gap is zero reduction ought to continue to focus on ural development. for all non-poor individuals). The poverty gap index measures The poverty estimates of 2012 (Table 3.1) are comparable the average extent to which individuals in a population fall with previous estimates of 23.2 percent poor and 5.9 percent below the poverty line and expresses it as a percentage of the subsistence poor in 2007. poverty line. The poverty squared gap index gives more weight Table 3.2 presents poverty incidence and subsistence to the very poor than those who are less poor. It is the average incidence as a percent of households. About nine percent of value of the square of depth of poverty for each individual households are poor, and about two percent are subsistence measured relative to the poverty line. More explanation on poor households. Hence, of the estimated 127,942 households, these indices is available in Technical Note 4. 11,049 are poor, and 2,322 are extremely poor. For both the poverty gap and poverty squared gap, as well A comparison of the poverty statistics in Table 3.1 and as for poverty rate, the larger the value of the index, the greater Table 3.2 indicates that poverty measures based on population the degree of poverty. These poverty measures are important are larger than those based on the number of households for planning poverty reduction programs. All things being because poor households, on average, have more household equal, sub-groups of the population with higher measures members should receive priority for poverty reduction programs. 10 Bhutan Poverty Analysis 2012 Figure 3.2.  Depth and Severity of Poverty in Bhutan Figure 3.3  Population Poverty Rates for 2007 and 2012 4.0 35 Poverty Gap 3.6 Poverty Squared Gap 30.9 3.5 30 3.0 2.6 25 23.2 2.5 20 Percent Percent 2.0 16.7 15 1.5 1.2 12.0 1.0 0.9 10 0.5 0.3 5 0.1 1.7 1.8 0.0 0 Urban Rural Bhutan 2007 2012 Urban Rural Bhutan Figure 3.2 shows that poverty is deeper and more severe in rural areas than in urban areas. Figure 3.4.  Population Subsistence Poverty Rates for 2007 and 2012 The poverty gap and poverty squared gap (with their 9 standard errors) across dzongkhags are listed in Annex I 8.0 (Table A-1). The table also includes the contribution of 8 the dzongkhags to the national poverty measures. Some 7 dzongkhags such as Lhuentse, Pema Gatshel and Zhemgang 5.9 6 have very high poverty measures (whether in terms of poverty 5 Percent rate, gap or severity). However, Samtse has a very high share 3.9 of the contribution to the national poverty measures, partly 4 because of its high population share. 3 2.8 2 3.3.  Poverty trend Figure 3.3 shows that the over poverty reduced from 23 1 0.2 0.3 percent in 2007 to 12 percent in 2012 and the reduction rural 0 2007 2012 poverty from 31 to about 17 percent. However, the proportion Urban Rural Bhutan of poor in urban areas remained practically unchanged at about two percent. As shown in Figure 3.4 subsistence poverty decreased is larger in rural areas than in urban areas. from six percent (2007) to about three percent (2012). In the As shown in the figure 3.5 the poverty rates and rural areas, the rate was reduced from eight percent in 2007 subsistence poverty rates also increase with the size of the to four percent in 2012. In the urban areas, the subsistence household. The increase in the poverty rate is faster than the poverty rate is quite low and remains unchanged (around subsistence rate as the household size increases. The share three in 100 persons). of households increases rapidly reaching a maximum of 40 percent for households containing four or five members. 3.4.  Poverty by Household Characteristics However, the share then decreases and reaches a minimum Households differ in their demographic composition and with four percent of the households containing nine or more characteristics. Household sizes in Bhutan are, on average, members This indicates that, although the poverty rates are larger in rural than urban areas. Table 3.7 shows that, across higher among the larger household size, especially those with the country, a poor household typically has much larger an average size of more than five, the corresponding share of family (6.3) than a non-poor household (4.4). The difference total households is much less. 12 Chapter 3: Patterns in Consumption Poverty Table 3.7.  Average Household Size by Area, Poverty Status and Sex of Figure 3.5.  Household Poverty and Subsistence Poverty Rates by Head Household Size Household head 45 Area/Poverty status Total 40.1 Male Female 40 Urban 4.2 3.7 4.1 35 33.1 Poor 5.4 4.6 5.3 Non-poor 4.2 3.7 4.1 30 26.6 Rural 4.8 4.6 4.7 25 Percent 24.3 Poor 6.5 6.3 6.4 20 Non-poor 4.6 4.4 4.5 16.6 15 Bhutan 4.6 4.4 4.5 11.3 Poor 6.4 6.2 6.3 10 6.2 5.1 Non-poor 4.4 4.2 4.4 2.9 3.6 5 3.9 1.6 1.0 0.7 0.3 0 2-3 4-5 6-8 9+ Table 3.8.  Household Poverty and Subsistence Poverty Rates, by Area Poverty Rate Share of total households and Sex of Household Head Subsistence Rate Poverty Rate Subsistence Rate Area/ Share of Household Contribution Contribution Total Heads Head Index Index Figure 3.6.  Population Poverty Rate by Economic Activity of the to National to National Household Head Urban 1.4 5.5 0.2 4.2 34.0 Male 1.5 4.9 0.2 3.3 27.4 100.0 Bhutan Female 0.8 0.6 0.3 1.0 6.6 12.0 Rural 12.4 94.5 2.6 95.8 66.0 Economically 16.1 Inactive 15.8 Contribution to national Male 12.9 64.7 2.4 57.8 43.3 poverty rate 0.6 Poverty incidence Female 11.3 29.8 3.0 37.9 22.7 Unemployed 6.5 Bhutan 8.6 100.0 1.8 100.0 100.0 27.3 Services Male 8.5 69.5 1.6 61.1 70.7 3.1 Female 9.0 30.5 2.4 38.9 29.3 12.1 Industry 4.1 43.9 Agriculture 18.5 Typically, welfare and household demographic composition are observed to have a nexus with the characteristics of the household head. Male headed households are observed to be, on average, less poor than female headed households. This Figure 3.7 displays poverty rates by the highest level of difference is more pronounced for subsistence poverty (Table educational attainment of the household head. The education 3.8). levels in Bhutan are generally quite low, especially among Figure 3.6 combines information on poverty, participation household heads. As expected, the higher the level of learning in the labour force, and main sector of employment of the completed by the household head, the lower the poverty rate. household head. Persons living in households where the head The returns to education increase considerably if the head is currently working have higher living standards than those has attended secondary levels irrespective of whether the whose head is either unemployed or out of the labour force. household is in an urban or rural area. Among the employed, poverty rates are higher in households Poverty rates increase with the age of the household head whose head works in agriculture. About 44 percent of the (Table 3.9). The poverty rate is about three percent for those poor live in households whose head is engaged in agriculture; below 25 years as compared to 14 percent for those aged 65 and 16 percent in households whose head is not actively years and older. This may indicate an inability to actively participating in the labour force. engage in economic activity with age. It is noticed that most 13 Bhutan Poverty Analysis 2012 Figure 3.7.  Household Poverty Rate by Educational Attainment of Figure 3.9.  Household Distribution and Poverty in Rural Areas Household Head by Area 30 16 15.0 24.5 14 25 13.0 12 Percent of households 20 18.3 16.3 10 15.0 Poverty rate 15 16.5 12.9 15.4 8 7.0 13.8 13.6 11.5 12.1 6 10 8.1 4 3.1 4.2 6.9 5 5.4 2 0.4 0 1.2 0.0 Landless Upto 1 acre 1 - 2 acres 2 -3 acres 3 - 4 acres 4 -5 acres 5 + acres 0 None At most grade VIII XI to XII Beyond XII Poverty incidence Distribution of households Educational attainment Urban Rural Bhutan Figure 3.8.  Distribution of (a) Type of Floor and (b) Type of External Walls by Household Poverty Status 60 60 58.5 (a) (b) Poor Non-poor Poor 50 50 Non-poor Total 45.1 43.9 40.9 Total 40 38.1 38.3 40 Percent Percent 30.2 30 28.5 30 28.4 26.4 23.2 21.0 20 18.0 18.2 20 17.7 10.9 10.1 10.8 11.4 10.5 10 8.8 10 8.3 7.7 6.1 6.2 5.1 5.6 4.4 4.3 2.8 3.6 3.8 1.1 0.60 .6 1.0 0 0 Wood Cement/tile Concrete Clay/earthen Plank/shingles Other Mud-bonded Cement-bonded Concrete Mud Wood/branches Other Table 3.9.  Household Poverty and Subsistence Poverty Rates by Age of Table 3.10.  Household Land ownership by Area and Poverty Status Household Heads Area Poor Non-poor Total Poverty Subsistence Poverty Age of Urban 20.8 32.4 32.3 Share of Household Contribution Contribution Rate Rate Total Heads Rural 92.8 82.3 83.6 Head to National to National Bhutan 88.8 64.0 66.1 < 25 2.8 1.4 0.7 1.6 4.4 25-34 3.3 9.1 0.5 7.1 23.9 35-44 7.5 20.0 1.8 22.4 22.9 45-54 10.5 25.5 1.9 22.3 20.9 55-64 13.3 23.4 2.9 24.2 15.2 65 + 14.0 20.6 3.2 22.5 12.7 All ages 8.6 100.0 1.8 100.0 100.0 14 Chapter 3: Patterns in Consumption Poverty household heads (68%) in Bhutan are aged 25 to 54 years, Table 3.10 shows land ownership in urban and rural while less than five percent are below age 25, and about 13 areas by poverty status. Across the country, 66 percent of percent are 65 and above. households own land with a higher proportion owned by Figure 3.8 shows the distribution of floor and external poor households. The proportion of households owning land wall types by household poverty status. There is not much in rural areas is at least two and half times that of urban areas. difference between poor and non-poor households in the Figure 3.9 illustrates the distribution of the number of use of materials, except for cement/tile and plank/shingles. households and the poverty rate by size of land holdings in Just over 10 percent of the poor households has cement/tile rural areas. The largest proportion of households own up to compared to 30 percent for the non-poor. At least 23 percent one acre of land but the proportion decreases with the size of the poor have clay/earthen floor while less than 10 percent of land holding. The poverty rate is the lowest for landless of non-poor have such flooring. Regarding the main materials households. The incidence of poverty is slightly lower for of the wall, only seven percent of the poor households have households who own two to three acres compared to those cement-bonded or concrete compared to 37 percent for the who own one to two acres, but is almost similar to those non-poor households. households who own more than three acres. 15 Chapter 4: Basic Needs Other non-monetary dimensions of welfare, such as Figure 4.1. Literacy Rate by Area and Poverty Status health and education status that pertain to basic needs, are 90 Poor complementary to consumption poverty. The health status 79.5 79.2 Non-poor 80 of an individual undoubtedly determines her/his quality Total 70 of life. Literacy and education status are widely recognized 62.3 64.6 63.0 to be important for improving the living standards of the 60 56.9 55.9 51.1 51.6 population. People with little or no education are likely to be 50 Percent unemployed, or if they do get employed, they often have low- 40 paying labour-intensive occupations. Such occupations often put them at risk of staying poor. More education provides 30 individuals with the basic knowledge, skills and competence 20 required for economic productivity, which, in turn, will 10 provide her/him assets and other capabilities for further improving her/his living standards and consequently some 0 Urban Rural Bhutan degree of social mobility. secondary qualification while adult population among the 4.1. Education non-poor has seven percent (Figure 4.2). According to Figure 4.1, poor persons in Bhutan have a much lower literacy rate than non-poor persons: 52 percent against 4.2. Income 65 percent, respectively. Disparities persist in literacy rates The BLSS 2012 collected information on the household income. between poor and non-poor both in the urban and rural areas. Figure 4.3 illustrates that the average monthly income in urban The literacy rate of the poor in urban areas is 17 percent lower areas is higher than in rural areas by about Nu. 10,000. The than the non-poor while in the rural areas the literacy rate of disparities between the poor and non-poor households in the poor is just 6 percent lower than the non-poor. terms of average income exist both in urban and rural areas. In At least 70 percent of the poor population 15 years and urban areas, the average income of the non-poor households is older have never attended school/institute while just over half Nu. 23,784: more than three times that of the poor households. of the non-poor have not attended. Although there is almost In the rural areas, the average income of the non-poor is Nu. equal proportions of the poor and non-poor adult population 9,348: more than twice that of the poor households. that have some educational attainment up to at most class eight, the proportion who have XI and XII qualifications 4.3. Health among the poor is just half that of the non-poor population. The BLSS 2012 collected information about the health Just about one percent of the poor population has beyond conditions and access to health services. Around 17 percent of 17 Bhutan Poverty Analysis 2012 Figure 4.2.  Distribution of Adult (15+) Educational Attainment by Figure 4.4.  Health Seeking Behaviour by Area and Poverty Status Poverty Status 80 75.9 73.2 None 70.1 68.9 Poor 62.3 18.8 9.9 0.9 At most grade VIII 70 XI to XII Non-poor 52.2 20.0 20.8 7.0 60 Beyond XII 50 53.7 52.8 Total 54.3 19.9 19.5 6.3 40 02 04 06 08 0 100 30 Figure 4.3.  Average Monthly Household Income by Poverty Status and 19.2 20 15.3 17.4 Area 12.5 13.9 15.2 10 25,000 23,784 23,784 0 Poor Non-Poor Poor Non-Poor Poor Non-Poor Average monthly household income 20,000 Urban Rural Bhutan Percentage of Persons Who Reported Sick During the Four Weeks Prior to the Enumeration Date 14,647 Proportion of Persons Who Reported Sick and Consulted Health Provider 15,000 13,736 9,348 Figure 4.5.  Proportion of Women Who Received Ante-Natal Care by 10,000 8,674 Areas and Poverty Status 7,387 Poor 5,000 4,096 Non-poor 100 3,905 100 94.3 94.3 Total 89.2 87.3 90 86.0 0 83.6 Poor Non-poor Total 80 74.7 73.8 Urban Rural Bhutan 70 60 the population reported that they had suffered from sickness Percent 50 or an injury in the four weeks prior to the survey, with no 40 significant difference between the poor and non-poor (Figure 4.4). However, of the population that reported some illness 30 only over half (53%) of the poor visited a medical facility 20 compared to 69 percent of the non-poor. 10 Table 4.1 illustrates a disparity between the percentage 0 of the poor (36%) and the non-poor (58%) who first visited Urban Rural Bhutan JDWNRH, or a regional referral or district hospital when they suffered from sickness or injury four weeks before the 4.4.  Household Amenities, Assets, and interview. At least 60 percent of the poor visited BHU/ORC compared against only 38 percent of non-poor. The disparity Access to Services is more common in rural areas. The living conditions of a household are often highly correlated When examining women who gave birth during the 12 months with its amenities, assets and access to services. Household prior to the BLSS 2012 interview, there is no difference between amenities, including suitable sanitation facilities, and access poor and non-poor women. However, in rural areas, a smaller to safe water sources, are not only wealth indicators, but also proportion of poor women received ante-natal care than non-poor improve welfare conditions of the household. Lack of safe water women. A considerable proportion of women received pre-natal or basic sanitation affects an individual’s health by increasing care in urban areas, especially among the poor (Figure 4.5). her/his chances of contracting diseases that are transmitted in 18 Chapter 4: Basic Needs Table 4.1.  Distribution of Persons who Suffered from Sickness/Injury four weeks prior to the survey with Health Seeking Behaviour by Area and Poverty Status Urban Rural Bhutan Health Service Provider Consulted Poor Non-Poor Total Poor Non-Poor Total Poor Non-Poor Total JDWNRH 6.9 33.2 32.8 3.1 9.2 8.5 3.3 16.1 15.0 Govt. regional referral hospital 8.8 14.4 14.3 6.9 10.4 10.0 7.0 11.5 11.1 Govt. district hospital 51.6 31.2 31.5 23.9 30.4 29.7 25.3 30.6 30.2 Govt. BHU/ORC 25.2 14.8 15.0 62.2 46.9 48.6 60.2 37.7 39.6 Indigenous centres 0.0 1.1 1.0 0.7 0.4 0.4 0.7 0.6 0.6 Traditional practitioner 0.0 0.7 0.6 0.0 0.0 0.0 0.0 0.0 0.0 Others 7.5 4.7 4.8 3.3 2.8 2.8 3.5 3.5 3.5 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Figure 4.6.  Proportion of Households with Access to Improved Water Figure 4.7.  Proportion of Households with Access to Improved Sanitation Source by Poverty Status and Area by Poverty Status and Area Poor Poor Non-poor Non-poor Total 100 95.7 95.5 Total 99.4 99.3 100 96.7 97.2 97.5 97.5 97.2 98.2 98.1 90 81.4 82.7 81.0 90 80 75.2 73.5 80 70 61.5 62.6 70 60 Percent 60 50 Percent 50 40 40 30 30 20 20 10 10 0 Urban Rural Bhutan 0 Urban Rural Bhutan Table 4.2.  Household Distribution of Subjective Poverty by Area and Poverty Status unsanitary environments. Some assets may allow households to cope with the risks brought about by seasonal variations Area/ Neither poor Very Poverty Not poor Poor Don’t know nor non-poor poor in incomes from farming, or other sources of vulnerability. If Status the head of the household suddenly becomes unemployed, or Urban 20.1 64.1 13.4 0.5 2.0 dies, or if a natural disaster occurs, the household could use its Poor 0.0 44.4 50.9 4.7 0.0 assets to smooth consumption. Consequently, it is important Non-poor 20.4 64.4 12.9 0.4 2.0 to look at the amenities and assets of a household as well as Rural 5.9 61.7 28.1 3.6 0.6 their access to basic social services to get a comprehensive Poor 1.9 49.9 39.3 8.6 0.3 assessment of their welfare conditions. Non-poor 6.5 63.4 26.6 2.9 0.6 The BLSS 2012 shows that across the country, about all Bhutan 10.7 62.5 23.1 2.6 1.1 (98%) households have access to an improved water source, Poor 1.8 49.6 39.9 8.4 0.3 i.e., piped water, public tap, protected wells/spring, bottled Non-poor 11.6 63.8 21.5 2.0 1.1 water and rain water collection. There is hardly any disparity in access to improved water source between poor and non- 19 Bhutan Poverty Analysis 2012 Table 4.3.  Household Distribution of Subjective Happiness by Area and Poverty Status Area/Poverty Status Very happy Moderately happy Neither happy unhappy Moderately unhappy Very unhappy Urban 33.7 54.2 10.4 1.2 0.5 Poor 34.0 54.3 10.0 1.1 0.5 Non-poor 14.4 47.8 34.4 3.4 0.0 Rural 32.2 50.8 12.7 3.0 1.3 Poor 33.7 50.4 12.2 2.7 1.1 Non-poor 21.9 54.0 16.6 5.1 2.4 Bhutan 32.7 52.0 11.9 2.4 1.0 Poor 33.8 51.8 11.4 2.1 0.9 Non-poor 21.5 53.7 17.6 5.0 2.3 Figure 4.8.  Proportion of Households Fuel Use for Lighting by Poverty Figure 4.10.  Rate of Characteristics for Perceived Poor and Poor Status and Area Households Poor 96.9 3.1 Electricity 100 Urban 86.2 Kerosene 90 Non-poor 97.6 1.9 78.4 Candles 80 Poor 69.3 21.6 70.2 Rural Others 70 74.6 Non-poor 85.0 10.3 60 69.1 64.4 Percent Poor 70.8 20.5 50 Bhutan 35.5 Non-poor 89.6 7.2 40 30.9 30 01 02 03 04 05 06 07 08 09 0 100 20 23.6 21.6 10 Figure 4.9.  Proportion of the Population with Ownership of Mobiles, TV 0 and Internet by Poverty Status Literacy rate of Land ownership TV ownership Antenatal care Safe sanitation household head 100 Poor Perceived Poor Poor 93.8 92.8 90 Non-poor 81.7 Total 80 Regarding access to improved sanitation (Figure 4.7), 70 at least 80 percent households have access to improved 58.6 60 55.3 sanitation (sewers or septic tanks, flush-latrines, pit with Percent 50 slab, or ventilated improved pit latrines). The disparity of around 15 percentage point is observed between poor and 40 non-poor households both in urban and rural areas. In urban 30 20.9 areas, 81 percent of poor households have access to improved 20 12.6 11.6 sanitation compared to only 62 percent in the rural areas. 10 The BLSS 2012 found that the main source of energy for 1.5 0 lighting throughout the country is electricity (88%) which is Mobile TV Internet proportionally higher in urban areas (98%) than in rural areas (83%). Figure 4.8 illustrates that nearly all (97%) of poor and poor, both in urban and rural areas (Figure 4.6). However, non-poor households in urban areas depend on electricity for further observation reveals that piped water into dwelling lighting in rural areas, however, only 69 percent of the poor is less common among poor households forcing about one- households have electricity as their main source of lighting. third of the poor households to depend on a neighbour’s pipe Figure 4.9 shows that 82 percent of poor households own or public outdoor tap. mobile phone, but the corresponding proportion for the non- 20 Chapter 4: Basic Needs poor is much higher (94%). The disparity between the poor true that the subjective poverty line is much higher than and the non-poor is also evident for ownership of TV and the poverty line. The mean per-capita expenditure of the internet connection in their homes. Only 21 percent among perceived poor is Nu. 3,201 compared to Nu. 1,353 for the the poor households have TV compared to 59 percent in the poor. Figure 4.10 shows the difference for some indicators. non-poor households. The internet connection at home is low The literacy rate of the household heads, TV ownership, at 12 percent. Among the poor households, barely about two access to ante-natal care and safe sanitation are lower for the percent have a connection. poor than for the perceived poor. The only exception is for the rate of land ownership. 4.5.  Perception and Priorities Regarding the perception on happiness (Table 4.3), the In the BLSS 2012, questions about the perception of poverty majority of the household heads reported they are moderately and happiness were included. The household head was asked happy (52%) or very happy (33%). Although there is hardly if he considered the household to be poor. This can be thought any difference between the poor and non-poor who reported of as a measure of perceived poverty. Across the country, at on moderate happiness, the proportion of household heads least a quarter (26%) of the household heads considers their who reported being very happy is much higher among the poor households to be either poor or very poor. In the urban areas, households; around 20 percentage point and 10 percentage the perceived poverty rate is 14 percent which is mostly point higher in urban area rural areas, respectively. It is more driven by the poor households (56%). There is at least 10 likely that the non-poor households report being neither percent of the household heads, who do not consider their happy nor unhappy compared to poor households especially households poor yet the analysis of survey data shows they are in the urban areas. actually poor. About a quarter (23%) of the household heads The BLSS 2012 respondents were asked to identify an belonging to non-poor households consider their households action agenda for the government that would improve their to be poor and the proportion is more than double in urban welfare. Most of the poor, especially in the rural areas, suggest areas (30%) compared to rural areas (13%). Table 4.2 further that road infrastructure and bridges, commerce, transport shows that in the urban areas there is no household head that and communication and water supply should be the priorities belongs to a poor household that considers itself to be non- of government. In urban areas, poor households specified poor. housing, labour and employment creation, and land and It could be useful to develop a different poverty profile resettlement, as priority concerns. based on the perceived (subjective) poverty. It is generally 21 Chapter 5: Inequality While poverty indicators focus on the population or households 5.2. Gini Index at the bottom of the per capita consumption distribution, it Consumption inequality can also be examined using graphical is also important to look at the spread of consumption over tools, such as the Lorenz curve, which maps the cumulative the entire population using inequality indicators. There is consumption share on the vertical axis against the distribution much interest in measuring inequality since high levels of of the population on the horizontal axis. If each household inequality may contribute to, if not exacerbate, poverty. had the same consumption, the resulting curve would be a Growth is known to be important for poverty reduction. 45-degree line known as the line of perfect equality. Figure 5.2 High inequality may result to lower subsequent economic illustrates the Lorenz curve of total household consumption growth and, consequently, in less poverty reduction. A high in Bhutan. The further away is the Lorenz curve from the level of inequality may make it difficult for the poor to have line of perfect equality, the higher is the level of inequality. a substantial share of the benefits of subsequent economic The Lorenz curve indicates that inequality in urban and rural growth. Inequality indicators attempt to measure the areas is very pronounced. The degree of inequality is similar deviation of a given consumption distribution from the ideal in urban and rural areas. This similarity may be the result of distribution, called perfect equality. within country remittances, or households residing in “rural” areas that have some members who are earning in “urban” 5.1. Consumption Quintiles areas. In addition, it may suggest the need to examine the Consider the distribution of real per capita consumption. current definition of urban and rural areas. Typically the population is ranked by ascending order of per Figure 5.1. Per Capita Consumption Quintiles capita consumption and the distribution is divided into fifths, i.e., 20 percent of the population, or equivalently quintiles. 120 In Bhutan, the share (7.1%) of national consumption of the 100.0 poorest quintile is only one sixth that of the share of the 100 richest quintile of the population (Figure 5.1). Table 5.1 shows that a person belonging to the richest 80 20 percent of the national population consumes on average Percent 6.7 times more than a person belonging to the poorest 60 20 percent of the population. This difference represents 43.7 40 a decrease compared to the estimates in the BLSS 2007 suggesting improvements in consumption inequality. As 22.4 20 15.5 is to be expected from Engel’s Law, the proportion of total 11.3 7.1 consumption allocated to food tends to decrease as the level 0 of per capita real consumption increases. Lowest Lower Middle Middle Upper Middle Upper Bhutan 23 Bhutan Poverty Analysis 2012 Table 5.1.  Average Monthly Real Per Capita Consumption (Nu), Share in National Consumption, Average Share of Food to Total Consumption, Average Household Size by Consumption Quintile Indicator Lowest Lower Middle Middle Upper Middle Upper Overall Average Per Capita Consumption 1,881.6 2,972.4 4,102.5 5,901.5 11,525.5 4,603.2 Share of National Consumption 7.1 11.3 15.5 22.4 43.7 100.0 Average Share of Food Consumption to Total Consumption 71.7 67.7 62.2 57.8 52.0 62.3 Average Household Size 6.1 5.1 4.5 3.9 3.2 4.5 Figure 5.2.  Lorenz Curve of Per Capita Household Consumption by Area The Gini coefficient, measured by the ratio of the area between the line of perfect equality to the Lorenz curve, to the area (of the triangle) under the line of perfect equality, Cumulative proportion of real per capita consumption is a commonly used indicator of inequality. The Gini index 81 ranges between 0 to 1 (with zero meaning perfect equality and one meaning perfect inequality). The typical values of the 6. Gini coefficient is between 0.2 to 0.5. While comparisons with previous estimates and international comparisons may be 4. done, such comparisons should be done with much caution. 2. Cumulative Percent of Population Comparisons are more meaningful across groups within the country. Figure 5.3 provides the Gini index at the national 0. 0. 2. 4. 6. 81 level and within urban and rural areas. The Gini at the Lorenz curve for Bhutan Line of Perfect Equality national level (0.36) is observed to almost equal that of urban Urban Areas Rural Areas (0.35) and rural areas (0.34). Figure 5.3.  Gini Coefficient by Area 0.4 0.36 0.35 0.34 0.3 0.2 0.1 0.0 Urban Rural Bhutan 24 Chapter 6: Conclusion The Royal Government of Bhutan has been, over the past current analysis are updated from 2007 using the ratio of years, implementing developmental activities with the focus the Consumer Price Indices of 2012 and 2007. The total of increasing the living standards of its citizens. The 10th FYP poverty line is obtained at Nu. 1,704.84 per person per month in particular aimed to alleviate poverty under the theme of and subsistence poverty line of Nu. 1,154.74 per person per “Poverty Reduction.” Such themes reflect the Government’s month. Consequently, the poverty rate is observed to be at commitment towards improving the welfare of the people 12 percent and subsistence poverty at 2.8 percent. Compared especially those who live in poverty. The Millennium to the poverty rate in 2007, the result in 2012 represents a Declaration, signed by the global community in 2000 at the reduction of about 50 percent. However, in urban areas there United Nations, was a commitment to ensuring that poverty is considerably smaller (i.e., from 0.2% to 0.3%) increase in is reduced to half its 1990 by 2015. the estimated total poverty index relative to the food poverty The PAR 2004 was the first attempt to measure the index, reflecting the more rapid inflation in food prices than poverty situation in Bhutan based on consumption data from in nonfood prices during 2012 and 2007. the BLSS 2003. It highlighted the poverty conditions in the Besides providing comparable and updated poverty country and unquestionably contributed to putting poverty profiles, the PAR 2012 also presents a spatial distribution on the development policy agenda, even though the FYPs of poverty in Bhutan down to the dzongkhag level. Updated have always had a pro-poor focus. information about the conditions of the poor presented The second analysis was carried out in 2007 using the BLSS in this report conveys information necessary to guide the 2007 data, which was designed to provide a portrait of the implementing plans and programs needed to eradicate poverty conditions down to the dzongkhag level. The report poverty and improve the living standards of the poor in examined an enriched set of information from the BLSS 2007, Bhutan. coming up with two poverty lines: a food poverty line of Nu. This report shows that poverty is still very much a rural 688.96 per person per month for measuring subsistence (or phenomenon in Bhutan, and that living standards vary extreme) poverty, and a total poverty line of Nu. 1,096.94 per considerably across the dzongkhags. person per month for measuring absolute poverty. Using In terms of demographic characteristics and educational these poverty lines, 23.2 percent of the population was living in attainment, the analysis shows the households that are poor poverty in 2007. The rate of subsistence (or extreme) poverty tend to be larger in size with more children, and to have heads was estimated at 5.9 percent, - one in sixteen Bhutanese did with no education and whose employment is concentrated in not have enough income to purchase even their food needs. the agricultural sector. The current analysis report using BLSS 2012 data is Estimated literacy rates and inequality measures in this based on a similar questionnaire, sample size and sampling report appear to be improvements from the 2007 levels, thus methodology as BLSS 2007 that allows valid comparisons giving a sense that public investments in basic social services, of poverty indicators over time. The poverty lines for the especially in education, have been successful in the rural areas. 27 Bhutan Poverty Analysis 2012 Such efforts must be continued and intensified. source of income by selling surplus dairy products. Another Like previous poverty analyses, this report confirms that option is to create market opportunities to enable the rural poverty is still a very much a rural phenomenon where the population sell their goods at favourable prices. majority (66%) of the population resides. While building a Poverty is bad not only for those who are poor but also causal role for public policy in poverty reduction in Bhutan represents a social problem that entail a joint responsibility by is beyond the scope of this report, some tentative conclusions the government, private sector and the development partners can be suggested. Efforts in rural and regional development in addressing this issue. Development plans should promote will thus have to be continued, and even expanded and inclusive growth, speeding up growth in lagging regions, accelerated. Improving access to credit in rural areas, assisting and reduce poverty in more deprived population groups. farmers in bringing their produce to vegetable markets in the There is a need to look into the successes and failures in towns, training farmers as entrepreneurs to transform their poverty reduction in other countries, and customize plans for rural products should enable farmers to better reap the fruits Bhutan. It is hoped that this report will help all development of their labour. stakeholders to understand the living conditions of the poor, . Livestock development is an excellent way to reduce and to listen to their often unheard voices, thereby leading to poverty in the poorest of rural areas as it increases the standard informed discussion and policy action. of living for the recipients. Families get better nutrition and a 28 Annex I: Additional Statistical Tables Table A-1.  Population Poverty Gap and Poverty Squared Gap by Dzongkhag Poverty Gap Poverty Squared Gap Dzongkhag Share of Population Index Standard error Contribution to Total Index Standard error Contribution to Total Bumthang 0.3 0.1 0.2 <0.2 0.0 <0.2 2.2 Chhukha 2.3 0.5 8.2 0.7 0.2 8.1 9.4 Dagana 5.8 1.3 7.4 2.0 0.7 7.5 3.3 Gasa <0.2 0.0 <0.2 <0.2 0.0 <0.2 0.5 Haa 1.4 1.3 0.8 0.5 0.4 0.8 1.5 Lhuentse 8.4 2.3 7.9 3.2 1.0 8.9 2.5 Monggar 1.8 0.3 4.4 0.6 0.2 4.2 6.6 Paro <0.2 0.0 <0.2 <0.2 0.0 <0.2 5.4 Pema Gatshel 5.5 1.1 8.1 1.7 0.5 7.4 3.8 Punakha 2.5 1.0 3.6 1.0 0.5 4.3 3.8 Samdrup Jongkhar 4.6 1.1 9.2 1.5 0.4 8.8 5.2 Samtse 4.7 0.8 17.0 1.4 0.3 15.6 9.5 Sarpang 0.7 0.3 1.5 0.2 0.1 1.3 5.9 Thimphu <0.2 0.0 0.2 <0.2 0.0 0.2 15.4 Trashigang 2.7 0.5 7.9 0.9 0.2 8.0 7.5 Trashi Yangtse 2.8 1.9 3.0 1.0 0.8 3.1 2.8 Trongsa 3.5 1.0 3.1 1.1 0.4 3.0 2.3 Tsirang 2.5 0.7 3.2 0.7 0.3 2.7 3.3 Wangdue Phodrang 2.3 0.7 5.2 0.8 0.3 5.3 5.8 Zhemgang 7.2 2.0 9.0 2.9 1.1 10.8 3.3 Bhutan 2.6 0.2 100.0 0.9 0.1 100.0 100.0 Table A-2.  Population Subsistence Poverty Gap and Subsistence Poverty Squared Gap by Dzongkhag Subsistence Poverty Gap Subsistence Poverty Squared Gap Dzongkhag Share of Population Index Standard error Contribution to Total Index Standard error Contribution to Total Bumthang <0.2 0.0 <0.2 <0.2 0.0 <0.2 2.2 Chhukha 0.4 0.2 6.7 0.1 0.1 6.5 9.4 Dagana 1.3 0.7 8.0 0.3 0.2 7.4 3.3 Gasa <0.2 0.0 <0.2 <0.2 0.0 <0.2 0.5 Haa 0.3 0.3 0.8 0.0 0.0 0.3 1.5 Lhuentse 2.4 1.0 11.0 0.7 0.3 10.6 2.5 Monggar 0.2 0.2 3.1 0.1 0.1 5.4 6.6 Paro <0.2 0.0 <0.2 <0.2 0.0 <0.2 5.4 Pema Gatshel 0.9 0.6 6.3 0.3 0.2 7.0 3.8 Punakha 0.8 0.5 5.7 0.3 0.2 6.2 3.8 Samdrup Jongkhar 0.9 0.4 9.3 0.2 0.1 8.3 5.2 Samtse 0.8 0.3 13.4 0.2 0.1 13.0 9.5 Sarpang <0.2 0.0 0.3 <0.2 0.0 <0.2 5.9 Thimphu <0.2 0.0 <0.2 <0.2 0.0 <0.2 15.4 Trashigang 0.5 0.2 7.3 0.1 0.1 6.4 7.5 Trashi Yangtse 0.7 0.7 3.7 0.1 0.1 2.6 2.8 Trongsa 0.7 0.4 3.1 0.1 0.1 2.2 2.3 Tsirang 0.3 0.2 2.1 0.1 0.0 1.1 3.3 Wangdue Phodrang 0.5 0.3 5.4 0.2 0.1 7.1 5.8 Zhemgang 2.3 1.1 13.9 0.8 0.4 15.8 3.3 Bhutan 0.5 0.1 100.0 0.2 0.0 100.0 100.0 30 Annex I: Additional Statistical Tables Table A-3.  Population Poverty Rate by Dzongkhag and Area Urban Rural Dzongkhag Share of Population Index Standard error Distribution of the poor Index Standard error Distribution of the poor Bumthang 2.5 1.6 2.4 3.8 2.3 2.4 2.2 Chhukha 5.2 1.4 42.9 16.8 3.5 42.9 9.4 Dagana 2.8 2.6 2.1 28.3 4.1 2.1 3.3 Gasa <0.5 0.0 <0.5 <0.5 0.0 <0.5 0.5 Haa <0.5 0.0 <0.5 8.5 5.7 <0.5 1.5 Lhuentse 4.9 5.8 2.0 34.6 6.3 2.0 2.5 Monggar <0.5 0.0 <0.5 12.7 2.1 <0.5 6.6 Paro <0.5 0.0 <0.5 <0.5 0.0 <0.5 5.4 Pema Gatshel 4.0 3.9 3.5 30.1 3.6 3.5 3.8 Punakha 1.5 1.5 1.8 11.9 3.1 1.8 3.8 Samdrup Jongkhar 2.7 1.6 7.1 28.1 5.6 7.1 5.2 Samtse 1.8 0.8 5.4 26.5 3.5 5.4 9.5 Sarpang 3.8 1.4 12.0 4.3 1.5 12.0 5.9 Thimphu <0.5 0.1 6.5 2.0 1.2 6.5 15.4 Trashigang 2.9 1.6 4.8 12.7 1.7 4.8 7.5 Trashi Yangtse <0.5 0.0 <0.5 15.7 5.7 0.0 2.8 Trongsa 2.0 1.2 1.4 17.5 4.5 1.4 2.3 Tsirang <0.5 0.0 <0.5 16.1 3.0 <0.5 3.3 Wangdue Phodrang 1.7 1.4 4.9 14.3 2.8 4.9 5.8 Zhemgang 3.6 2.0 3.1 30.0 5.0 3.1 3.3 Bhutan 1.8 0.3 100.0 16.7 0.8 100.0 100.0 Table A-4.  Household Poverty Rate by Dzongkhag and Area Urban Rural Distribution of Dzongkhag Index Standard error Distribution of the poor Index Standard error Distribution of the poor households Bumthang 2.7 1.8 3.2 1.7 1.1 <0.5 2.2 Chhukha 4.2 1.1 48.7 11.5 2.8 6.4 9.4 Dagana 1.8 1.7 1.9 21.4 3.3 7.9 3.3 Gasa <0.5 0.0 <0.5 <0.5 0.0 <0.5 0.5 Haa <0.5 0.0 <0.5 4.0 2.8 0.5 1.5 Lhuentse 2.6 3.1 1.2 27.1 5.1 7.2 2.5 Monggar <0.5 0.0 <0.5 9.2 1.5 5.3 6.6 Paro <0.5 0.0 <0.5 <0.5 0.0 <0.5 5.4 Pema Gatshel 3.2 3.1 3.1 23.7 2.9 9.3 3.8 Punakha 1.0 1.0 1.6 9.2 2.0 3.1 3.8 Samdrup Jongkhar 2.3 1.4 7.5 20.0 4.2 10.0 5.2 Samtse 1.1 0.5 4.3 19.9 2.7 17.8 9.5 Sarpang 1.9 0.7 8.6 2.6 0.9 1.3 5.9 Thimphu <0.5 0.1 5.9 2.4 1.2 0.6 15.4 Trashigang 2.0 1.0 4.2 10.0 1.4 8.5 7.5 Trashi Yangtse <0.5 0.0 <0.5 10.2 3.0 3.1 2.8 Trongsa 1.6 0.9 1.5 12.9 3.6 2.8 2.3 Tsirang <0.5 0.0 <0.5 13.4 2.4 4.8 3.3 Wangdue Phodrang 1.5 1.2 5.6 10.7 1.9 4.9 5.8 Zhemgang 2.7 1.5 2.9 22.8 4.2 6.2 3.3 Bhutan 1.4 0.2 100.0 12.4 0.6 100.0 100.0 31 Bhutan Poverty Analysis 2012 Table A-5.  Population Poverty Gap and Poverty Squared Gap by area Poverty Gap Poverty Squared Gap Area Share of Population Index Standard error Contribution to Total Index Standard error Contribution to Total Urban 0.3 0.1 4.1 <0.2 0.0 3.6 31.0 Rural 3.6 0.3 95.9 1.2 0.1 96.4 69.0 Bhutan 2.6 0.2 100.0 0.9 0.1 100.0 100.0 Table A-6.  Population Subsistence Poverty Gap and Poverty Squared Gap by area Poverty Gap Poverty Squared Gap Area Share of Population Index Standard error Contribution to Total Index Standard error Contribution to Total Urban <0.2 0.1 2.9 <0.2 0.0 2.8 31.0 Rural 0.8 0.2 97.1 0.2 0.1 97.2 69.0 Bhutan 0.5 0.1 100.0 0.2 0.1 100.0 100.0 Table A-7.  Household Poverty Rate, Poverty Gap and Poverty Squared Gap by Area and Sex of Household Heads Poverty Rate Poverty Gap Poverty Squared Gap Area/Sex of HH Head Share of Population Index Contribution to Total Index Contribution to Total Index Contribution to Total Urban 1.4 11.4 0.3 100.0 <0.2 100.0 100.0 Male 1.5 88.6 0.3 84.4 <0.2 81.4 80.6 Female 0.8 11.4 0.2 15.6 <0.2 18.6 19.4 Rural 12.4 100.0 2.6 100.0 0.8 100.0 100.0 Male 12.9 68.4 2.6 65.4 0.8 63.2 65.6 Female 11.3 31.6 2.6 34.6 0.9 36.8 34.4 Bhutan 8.6 100.0 1.8 100.0 0.6 100.0 100.0 Male 8.5 69.5 1.7 66.4 0.5 64.1 70.7 Female 9.0 30.5 2.0 33.6 0.7 35.9 29.32 Table A-8.  Household Poverty Rate Poverty Gap and Poverty Squared Gap by Area and Age of Household Heads Poverty Rate Poverty Gap Poverty Squared Gap Area/Age of HH Head Share of Population Index Contribution to Total Index Contribution to Total Index Contribution to Total Urban 1.4 100.0 0.3 100.0 <0.2 100.0 100.0 < 25 0.6 2.9 <0.2 2.0 <0.2 1.3 6.6 25-34 1.1 30.2 0.2 25.0 <0.2 24.0 37.4 35-44 1.5 31.2 0.2 23.5 <0.2 16.1 28.1 45-54 1.4 17.3 0.3 17.8 <0.2 15.4 17.5 55-64 1.9 8.8 0.3 8.2 <0.2 5.5 6.5 65 + 3.3 9.6 1.6 23.5 0.8 37.7 4.0 Rural 12.4 100.0 2.6 100.0 0.8 100.0 100.0 < 25 5.2 1.4 1.3 1.7 0.5 1.8 3.3 25-34 5.7 7.9 1.2 7.6 0.3 6.9 16.9 35-44 11.8 19.3 2.6 20.4 0.9 20.8 20.2 45-54 14.2 26.0 2.9 25.4 0.9 25.2 22.6 55-64 15.2 24.2 3.1 23.7 1.0 22.5 19.7 65 + 15.3 21.3 3.2 21.3 1.1 22.7 17.2 Bhutan 8.6 100.0 1.8 100.0 0.6 100.0 100.0 < 25 2.8 1.4 0.7 1.7 0.2 1.8 4.4 25-34 3.3 9.1 0.6 8.5 0.2 7.7 23.9 35-44 7.5 20.0 1.6 20.6 0.5 20.5 22.9 45-54 10.5 25.5 2.1 25.0 0.7 24.8 20.9 55-64 13.3 23.4 2.7 22.9 0.8 21.7 15.2 65 + 14.0 20.6 3.0 21.4 1.1 23.4 12.7 32 Annex I: Additional Statistical Tables Table A-9.  Household Poverty and Subsistence Poverty Rate by area and household size Poverty Rate Subsistence Rate Area/Household size Share of total households Index Contribution to National Index Contribution to National Urban 1.4 5.5 0.2 4.2 34.0 1 <0.5 0.1 0.4 0.5 2.2 2-3 0.7 0.8 0.2 0.9 9.6 4-5 1.2 2.1 <0.2 0.4 15.7 6-8 3.2 2.3 0.6 2.0 6.1 9+ 5.0 0.2 2.0 0.4 0.4 Rural 12.4 94.5 2.6 95.8 66.0 1 2.5 0.9 1.0 1.7 2.9 2-3 4.1 8.1 0.4 3.5 17.0 4-5 9.4 26.5 1.6 21.0 24.4 6-8 21.1 44.5 4.6 46.0 18.2 9+ 36.3 14.6 12.3 23.5 3.5 Bhutan 8.6 100.0 1.8 100.0 100.0 1 1.6 0.9 0.7 2.1 5.1 2-3 2.9 8.9 0.3 4.4 26.6 4-5 6.2 28.6 1.0 21.4 40.1 6-8 16.6 46.8 3.6 48.1 24.3 9+ 33.1 14.8 11.3 24.0 3.9 Table A-10.  Population Literacy Rate for Aged Six Years and Above by Table A-11.  Proportion of Women (15-49 years) Who Received Ante- Dzongkhag and Poverty Status natal Care by Dzongkhag and Poverty Status Dzongkhag Poor Non-poor Total Dzongkhag Poor Non-poor Total Bumthang 62.9 67.8 67.6 Bumthang * 84.6 84.6 Chhukha 58.3 72.0 70.4 Chhukha 69.8 96.5 93.9 Dagana 52.6 63.7 61.0 Dagana 100.0 100.0 100.0 Gasa * 49.4 49.4 Gasa * 43.9 43.9 Haa 69.8 68.5 68.6 Haa * 92.1 92.1 Lhuentse 47.2 58.5 54.9 Lhuentse 53.4 72.8 62.0 Monggar 56.2 59.9 59.5 Monggar 88.0 85.6 86.0 Paro * 67.3 67.3 Paro * 85.4 85.4 Pema Gatshel 54.5 56.6 56.1 Pema Gatshel 59.8 78.8 74.8 Punakha 47.9 55.9 55.2 Punakha 100.0 100.0 100.0 Samdrup Jongkhar 54.7 63.6 61.7 Samdrup Jongkhar 83.3 85.4 84.9 Samtse 42.3 52.0 49.8 Samtse 76.3 85.6 83.4 Sarpang 67.5 59.6 59.9 Sarpang 100.0 85.7 87.4 Thimphu 60.4 80.2 80.0 Thimphu * 98.6 98.6 Trashigang 47.8 61.7 60.1 Trashigang 49.4 72.7 70.5 Trashi Yangtse 54.4 61.4 60.4 Trashi Yangtse 100.0 100.0 100.0 Trongsa 56.4 67.4 65.7 Trongsa * 88.7 88.7 Tsirang 51.9 60.9 59.6 Tsirang 76.4 77.9 77.5 Wangdue Phodrang 42.8 52.3 51.3 Wangdue Phodrang 100.0 94.0 94.7 Zhemgang 58.3 63.7 62.3 Zhemgang 44.7 84.8 66.5 Bhutan 51.6 64.6 63.0 Bhutan 74.6 89.2 87.2 * * Figure not shown due to few cases Figure not shown due to few cases 33 Bhutan Poverty Analysis 2012 Table A-12.  Proportion of Population Who Reported Sick/Injured Four Table A-14.  Proportion of Population with Access to Improved Weeks Prior to the Survey by Dzongkhag and Poverty Status Sanitation by Dzongkhag and Poverty Status Dzongkhag Poor Non-poor Total Dzongkhag Poor Non-poor Total Bumthang 11.0 15.7 15.5 Bumthang 100.0 76.6 77.0 Chhukha 13.6 16.9 16.5 Chhukha 66.8 86.3 84.8 Dagana 17.5 25.1 23.2 Dagana 71.3 84.2 81.8 Gasa * 29.8 29.8 Gasa • 54.9 54.9 Haa 2.8 5.2 5.0 Haa • 67.6 65.6 Lhuentse 15.6 20.2 18.7 Lhuentse 88.2 88.1 88.1 Monggar 11.0 14.9 14.5 Monggar 43.7 71.8 69.7 Paro * 27.4 27.4 Paro • 93.4 93.4 Pema Gatshel 10.9 11.4 11.2 Pema Gatshel 73.9 83.0 81.1 Punakha 17.9 21.4 21.0 Punakha 60.5 69.2 68.6 Samdrup Jongkhar 20.8 22.6 22.2 Samdrup Jongkhar 53.5 84.6 79.9 Samtse 6.2 7.1 6.9 Samtse 49.8 71.9 68.3 Sarpang 18.3 7.7 8.2 Sarpang 88.4 91.2 91.1 Thimphu 47.2 14.0 14.1 Thimphu 75.8 95.5 95.4 Trashigang 19.5 25.6 24.9 Trashigang 75.0 82.0 81.4 Trashi Yangtse 8.5 22.5 20.6 Trashi Yangtse 79.3 72.4 73.0 Trongsa 23.2 22.4 22.5 Trongsa 12.4 56.8 52.1 Tsirang 22.7 32.0 30.6 Tsirang 69.1 87.7 85.5 Wangdue Phodrang 10.5 14.8 14.3 Wangdue Phodrang 38.4 70.5 68.0 Zhemgang 28.9 24.1 25.3 Zhemgang 71.8 81.6 79.8 Bhutan 15.2 17.4 17.1 Bhutan 62.6 82.7 81.0 * * Figure not shown due to few cases Figure not shown due to few cases Table A-13.  Proportion of Population with Access to Improved Water Table A-15.  Proportion of Population using Solid Fuels by Dzongkhag Source by Dzongkhag and Poverty Status and Poverty Status Dzongkhag Poor Non-poor Total Dzongkhag Poor Non-poor Total Bumthang 100.0 100.0 100.0 Bumthang 100.0 91.0 91.2 Chhukha 100.0 98.7 98.8 Chhukha 20.8 24.4 24.1 Dagana 95.5 95.8 95.8 Dagana 18.6 25.0 23.8 Gasa * 100.0 100.0 Gasa • 95.3 95.3 Haa 65.3 99.1 98.1 Haa 69.4 84.8 84.3 Lhuentse 100.0 98.7 99.0 Lhuentse 82.2 53.9 60.9 Monggar 100.0 95.8 96.1 Monggar 27.8 27.1 27.2 Paro * 98.9 98.9 Paro • 59.8 59.8 Pema Gatshel 100.0 98.9 99.1 Pema Gatshel 0.9 5.6 4.6 Punakha 93.5 96.6 96.4 Punakha 71.9 37.9 40.4 Samdrup Jongkhar 100.0 98.5 98.7 Samdrup Jongkhar 35.1 11.3 14.9 Samtse 98.2 97.4 97.5 Samtse 12.1 3.4 4.8 Sarpang 89.2 98.1 97.9 Sarpang • 3.9 3.8 Thimphu 75.8 99.6 99.5 Thimphu 33.1 16.4 16.5 Trashigang 100.0 98.8 98.9 Trashigang 15.6 30.2 28.9 Trashi Yangtse 94.7 94.9 94.8 Trashi Yangtse 77.3 54.6 56.6 Trongsa 100.0 97.9 98.1 Trongsa 68.2 61.4 62.1 Tsirang 88.1 97.3 96.1 Tsirang 11.8 30.1 27.9 Wangdue Phodrang 85.4 97.9 96.9 Wangdue Phodrang 57.1 47.0 47.8 Zhemgang 100.0 97.8 98.2 Zhemgang 34.7 23.6 25.7 Bhutan 97.2 98.2 98.1 Bhutan 30.0 28.4 28.6 * * Figure not shown due to few cases Figure not shown due to few cases 34 Annex I: Additional Statistical Tables Table A-16.  Proportion of Households Who Have TV by Dzongkhag and Poverty Status Dzongkhag Poor Non-poor Total Bumthang 82.6 76.3 76.4 Chhukha 28.8 65.1 62.4 Dagana 16.2 28.4 26.1 Gasa * 35.8 35.8 Haa * 79.5 77.2 Lhuentse 16.7 33.4 29.2 Monggar 12.0 38.3 36.4 Paro * 78.7 78.7 Pema Gatshel 25.6 53.1 47.3 Punakha 34.5 62.6 60.5 Samdrup Jongkhar 12.9 45.6 40.6 Samtse 14.8 48.8 43.4 Sarpang 51.4 61.6 61.4 Thimphu 66.9 90.0 89.9 Trashigang 22.3 41.2 39.5 Trashi Yangtse 29.3 27.8 27.9 Trongsa 13.2 48.3 44.6 Tsirang 15.3 34.9 32.5 Wangdue Phodrang 27.8 50.8 49.0 Zhemgang 21.8 44.7 40.3 Bhutan 20.9 58.6 55.3 * Figure not shown due to few cases 35 Annex II: Technical Notes Technical Note 1 (Measuring Aggregate (b) purchased domestic and (c) non-purchased food items. Consumption) The BLSS food purchases module contains questions on purchases of a fairly comprehensive list of food items (a) Aggregations of consumption and expenditure data were during a relatively short reference period, i.e. the last seven made following the recommendations by A. Deaton and S. days, the last 30 days, and the last 12 months in which such Zaidi (2002). Most of the information below is quoted from purchases were made. Data are collected on the total amount their paper. spent on purchasing each food item, and also on the quantities purchased, during the specified recall period. a)  Income versus consumption Calculating the food purchases sub-aggregate involved In most industrialized countries living standards and poverty converting all reported expenditures on food items to a are assessed with reference to income, not consumption. uniform reference period—one month—and then aggregating The empirical literature on the relationship between income these expenditures across all food items purchased by the and consumption has established, for both rich and poor household. countries, that consumption is smoother and less-variable The “last 30 days” data measure over the “last 7 days” than income. Observing consumption over a relatively short or the “last 12 months” has the advantage of being closer to period, even a week or two, will tell us a great deal more about the concept that we want—usual consumption — over what annual—or even longer period—living standards than will actually happened in the last 7 days, which could have been a similar observation on income. Although consumption unusual for any number of reasons—and reduces problems has seasonal components they are of smaller amplitude than of seasonality, but suffers from measurement error if seasonal fluctuations in income in agricultural societies. respondents find it difficult to calculate a reasonable answer. There are several other reasons why it is more practical The last “12 months” may be too long a recall period to reveal to gather consumption rather than income data. Where accurate data. Thus, we prefer the “last 30 days” data. If there self-employment, including small business and agriculture, are no available “30 days” data, we use the “last 7 days” data is common, it is notoriously difficult to gather accurate and rescale the results. If there are no available “30 days” or income data, or indeed to separate business transactions from “last 7 days,” we use the “last 12 months” data and rescale the consumption transactions. results. The BLSS 2012 questionnaire also asked explicitly b)  Food consumption about the total value of meals taken outside the home by all Households consume food obtained from a variety of household members; this amount was also included in the food different sources, and so in computing a measure of total consumption aggregate as part of purchased consumption. food consumption to include as part of an aggregate welfare The questionnaire contains a separate set of questions measure, it is important to include food consumed by the on consumption of home-produced food items. Data were household from all possible sources. In particular, this collected on both the value and quantity of consumption of measure should include not just (i) food purchased in the each home-produced food item. The home-production food market place, including meals purchased away from home sub-aggregate can thus be calculated by adding the reported for consumption at or away from home, but also (ii) food value of consumption of each of the home-produced food that is home-produced, (iii) food items received as gifts items in a manner analogous to that followed in the case of or remittances from other households, as well as (iv) food food purchases. received from employers as payment in-kind for services Consumption of food derived from payment in-kind, as rendered. well as in the form of gifts, remittances, etc., was added to the The BLSS 2012 food consumption module questionnaire overall food aggregate. contains separate sets of questions on (a) purchased imported All quantities were reported in standard units. Analysis was 36 Annex II: Technical Notes performed on the quantities and unit prices to treat missing In most cases, however, households own the dwelling in data and identify inconsistent data. Cases were noted where a which they reside and do not pay rent as such. Others are household had declared consuming a non-zero quantity of a provided with housing free of charge (or at subsidized rates) particular item, or households reported consumption values, by their employer, a friend, a relative, government, or other but no corresponding information on quantities. Others had such entities. Non-renter households were asked how much inconsistent data on quantities, or values (yielding outliers it would cost them if they had to rent the dwelling in which of unit prices). In such instances, median regional unit they reside, and this “implicit rental value” was used in place prices were used to make imputations. Median prices were of actual rent. preferred to mean prices, as they are less sensitive to outliers. 2) Taxes When median price was not available at the lowest geographic Expenditures on taxes and levies are not part of consumption, level, we used prices reported by other households in the and were not included in the consumption total. same Dzongkhag, depending on whichever is the next higher level of aggregation for which price information is available. 3) Repayment of debt and interest payments Medians of unit price are computed and used separately for All purchases of financial assets, as well as repayments of debt, purchased and home-produced items. and interest payments were excluded from the consumption aggregate. c)  Non-food consumption 4) Education Unlike many homogeneous food items, most non-food goods Education expenditure paid by the households was included are too heterogeneous to permit the collection of information in households’ consumption. on quantities consumed, so that BLSS 2012 collected data only on the value of non-foods purchased over the reference 5) Health period. Data on purchases of non-food items were collected Expenditure on health is to a large extent a lumpy expenditure. for two different recall periods, i.e. over the 12 months, or One argument for exclusion is that such expenditure reflects the last 1 month, depending on how frequently the items a regrettable necessity that does nothing to increase welfare. concerned are typically purchased. Constructing the non- By including health expenditures for someone who has fallen food aggregate thus entails converting all these reported sick, we register an increase in welfare when, in fact, the amounts to a uniform reference period—one year, and then opposite has occurred. The fundamental problem here is our aggregating across the various items. inability to measure the loss of welfare associated with being Not all non-food expenditures were included in the sick, and which is (presumably) ameliorated to some extent consumption aggregates. Also, some “expenditures” required by health expenditures. imputations. Including the latter without allowing for the former is clearly incorrect, though excluding health expenditures 1) Housing altogether means that we miss the difference between two What is required is a measure in monetary terms of the people, both of whom are sick, but only one of which pays for flow of services that the household receives from occupying treatment. It is also true that some health expenditures—for its dwelling. Because house purchase is such a large and example cosmetic expenditures—are discretionary and welfare relatively rare expenditure, under no circumstances should enhancing, and that it is difficult to separate “necessary” from expenditures for a housing purchase be included in the “unnecessary” expenditures, even if we could agree on which is consumption aggregate. which. It is also difficult without special health questionnaires Expenditure on house repairs and improvements were to get at the whole picture of health financing. Some people also excluded from the consumption aggregates. have insurance, so that expenditures are only “out of pocket” In the hypothetical case where rental markets function expenditures which may be only a small fraction of the total, perfectly and all households rent their dwellings, the rent paid while others have none, and may bear the whole cost. Simply is the obvious choice to include in the consumption aggregate. adding up expenditures will not give the right answer. Whenever such rental data are available, they were used for Expenditure on hospitalizations, consultations, and constructing the housing sub-aggregate and the consumption analyses were excluded from the household consumption. total. Purchase of medicine was however included. 37 Bhutan Poverty Analysis 2012 6) Remittances the regional price deflators. The Paasche price index for Another group of expenditures are charitable contributions, household h is given by: and remittances to other households. Their inclusion in the consumption aggregate would involve double-counting if, as Pph = (�wkh ( pk 0 / pk h ))-1   one would expect, the transfers show up in the consumption of other households. We therefore excluded them from 0 h where pk is the reference unit price for good k, pk is the household consumption. h unit price paid for good k by household h, and wk is the 7) Other lumpy expenditures share of household h’s budget devoted to good k. The weights While almost all households incur relatively large expenditures used for the price index are the quantities consumed by the on relatively infrequent expenditures such as marriages and household itself and therefore differ from one household to dowries, births, and funerals at some stage, only a relatively another. In other words, these indexes involve, not only the small proportion of households are likely to make such prices faced by household h in relation to the reference prices, expenditures during the reference period typically covered by but also household h’s expenditure pattern, something that is the survey. Ideally, we would want to “smooth” these lumpy not true of a Laspeyres index. 0 expenditures, spreading them over several years, but lacking The reference price vector p was inevitably selected as the information to do so—which might come, for example, by a matter of convenience. To ensure that the vector is not very incorporating multi-year reference periods for such items— different from prices actually observed, we chose to take the we left them out of the consumption aggregate. median of the prices observed from individual households as reference. The use of the national median price vector ensures 8) Durable Goods that the money metric measures conform as closely as possible Another important group of items to consider are items such to national income accounting practice, as well as eliminating as consumer durables whose useful life typically spans a time- results that might depend on a price relative that occurs only period greater than the interval for which the consumption rarely or in some particular area. aggregate is being constructed. From the point of view of Quantities and unit values were available at the household household welfare, rather than using expenditure on the level only for foods items. For non-foods, data is not available purchase of durable goods during the recall period, the at the household level. The Paachse price indices were thus appropriate measure of consumption of durable goods is computed for food items only. the value of services that the household receives from all the durable goods in its possession over the relevant time period. Technical Note 2 (Food Poverty Line) To assess the value of services, one would need data on the cost of purchase and year of purchase. Such information The Food Poverty line for 2012 is updated from 2007 using is not available in BLSS 2007. Consumption of durable goods the ratio of the Food Consumer Price Indices. The BLSS 2007 was thus not included in the overall consumption aggregate. collected data on 118 different food items. Consumption data was available in standard quantity units for all these items. d)  Computing regional price deflators For 94 of them, calories intake data was available, and of Before our measure of consumption could be used to compare these items, 53 items were used to create a reference food standards of living of individuals residing in different parts of basket. These items were used to compute the food poverty the country, it is necessary to take into account differences in line since the most frequently consumed food items by the cost of living. To convert total expenditure into money metric reference population (i.e., the second to the fourth deciles of utility, the price index must be tailored to the household’s the nominal per capita consumption distribution). These 53 own demand pattern, a demand pattern that varies with the goods accounted for 80 percent of the food consumption by household’s income, demographic composition, location, and the reference population. The quantities of each item in the other characteristics. The calculation of money metric utility food basket were established by considering the consumption thus requires that the nominal values be deflated by a Paasche pattern of the reference population. The quantities were price index, in which the weights vary from household to scaled up in such a way that the resulting basket provides a household. total of 2,124 Kcal. The cost of the basket was calculated using Data collected by the BLSS 2007 were used to construct the national median unit prices for each item. 38 Annex II: Technical Notes Table A-17.  Food Bundle and Costs of Nutritionally Adequate Food Bundle Per Person Per Day, 2007 Items Unit Calories per units (kcals) Daily quantity consumed (units) Daily calories provided (kcals) Price per unit Cost Cereals and Pulses 101 Rice Bhutanese Gram 3.46 92.29 319.34 0.03 2.31 102 Rice fine Gram 3.49 59.83 208.79 0.01 0.79 103 Rice FCB Gram 3.46 110.24 381.41 0.01 1.47 104 Processed rice (zaw, sip) Gram 3.25 9.60 31.19 0.03 0.29 105 Maize (kharang) Gram 3.42 92.97 317.97 0.01 1.02 106 Ata, Maida Gram 3.41 9.75 33.23 0.02 0.18 107 Noodles Gram 3.47 12.13 42.09 0.04 0.49 108 Confectionery Gram 2.45 0.20 0.49 0.30 0.06 109 Biscuits Gram 3.64 4.67 17.01 0.09 0.42 110 Pulses Gram 3.43 11.47 39.34 0.03 0.34 Dairy Products 201 Liquid milk Ml 0.67 19.11 12.80 0.03 0.51 202 Milk powder Gram 4.96 6.51 32.29 0.17 1.07 203 Local butter Gram 7.29 10.44 76.09 0.15 1.57 204 Local cheese Gram 4.73 12.35 58.47 0.11 1.37 205 Egg Gram 1.73 3.68 6.37 0.08 0.32 Meat 301 Fresh fish Gram 0.97 2.25 2.18 0.08 0.23 302 Dried fish Gram 2.55 11.20 28.57 0.07 0.78 303 Fresh beef Gram 1.14 7.22 8.23 0.06 0.43 304 Dried beef Gram 2.00 1.77 3.53 0.20 0.35 305 Fresh pork Gram 1.14 4.09 4.67 0.10 0.41 306 Chicken Gram 1.09 2.91 3.17 0.10 0.29 Fruits 401 Apple Gram 0.59 0.69 0.41 0.04 0.03 402 Orange Gram 0.48 21.24 10.19 0.01 0.32 403 Mango Gram 0.74 0.52 0.38 0.03 0.02 404 Banana Gram 1.16 18.06 20.95 0.01 0.14 405 Cucumber Gram 0.13 5.95 0.77 0.01 0.06 406 Sugarcane Gram 3.98 2.70 10.73 0.02 0.05 407 Guava Gram 0.51 2.44 1.25 0.01 0.02 408 Walnut Gram 6.87 3.92 26.94 0.01 0.04 409 Other fruits Gram 0.48 0.63 0.30 0.02 0.02 Vegetables 501 Fresh beans Gram 1.58 17.36 27.42 0.02 0.35 502 Tomato Gram 0.23 17.77 4.09 0.02 0.36 503 Spinach Gram 0.26 32.93 8.56 0.01 0.40 504 Cabbage Gram 0.27 20.40 5.51 0.01 0.20 505 Potato Gram 0.97 60.56 58.75 0.01 0.71 506 Pumpkin Gram 0.25 4.42 1.10 0.01 0.04 507 Radish Gram 0.17 26.46 4.50 0.01 0.26 508 Cauliflower Gram 0.30 8.11 2.43 0.02 0.16 509 Brinjal Gram 0.24 5.54 1.33 0.02 0.08 510 Gourd Gram 0.12 2.67 0.32 0.02 0.04 511 Fresh mushroom Gram 0.25 1.95 0.49 0.20 0.39 512 Fern (damru) Gram 0.34 6.25 2.13 0.02 0.12 513 Mustard oil Ml 9.00 14.11 127.03 0.06 0.85 514 Dalda oil Ml 9.00 3.07 27.64 0.05 0.15 515 Refined oil Ml 9.00 6.61 59.53 0.06 0.40 Spices, Seasonings and Pastes 601 Fresh chili Gram 0.29 21.31 6.18 0.03 0.64 602 Dried chili Gram 2.46 6.16 15.15 0.10 0.62 603 Haldi, Jeera Gram 3.49 0.82 2.87 0.10 0.08 604 Coriander leaves Gram 0.44 6.18 2.72 0.03 0.19 605 Salt Gram 0.00 8.80 0.00 0.01 0.09 607 Sugar/gur Gram 3.98 16.00 63.69 0.03 0.48 Beverages 701 Beer Ml 0.35 3.93 1.36 0.06 0.24 702 Juice Ml 0.47 4.36 2.03 0.05 0.22 TOTAL PER DAY 2,124 kcal Nu. 22.49 39 Bhutan Poverty Analysis 2012 Technical Note 3 (Non Food Adjustment to line may also be computed (since poor households usually the Poverty Line) have a smaller size, the proportion of poor households is usually lower than the proportion of poor population). Having set the food poverty line, a non-food component must P0 = q / n be added to obtain an overall poverty line that incorporates overall needs. As M. Ravallion and Bidani (1992, 1999), where P0 is the proportion of population deemed to be suggested, the total poverty line is obtained by scaling up the poor (poverty headcount), q is the number of poor people food poverty line to allow for the purchase of some essential (below the poverty line), and n is the total population. non-food items to reach a final poverty line. The non-food The fact that poverty calculations are based on a sample of needs must be consistent with the consumption behavior of households, or a subset of the population, carries implications. households who can just afford basic food needs. Samples are designed to reproduce the whole population, A number of methodologies have been proposed for but they can never be as exact as information that covers making this non-food adjustment, including the use of another everybody in the country. They carry a margin of error, as basket of non-food items. The best solution is to measure what do poverty rates calculated from these sample surveys. When is the typical value of non-food spending by a household that monitoring the incidence of poverty over time, it is crucial to is just able to reach its food requirements. This will equal the remember that the figures are based on samples. Instead of lowest level of non-food spending for households that are able considering one figure is better to use confidence intervals. to acquire the basic food bundle. It can thus be considered a minimal allowance for non-food goods. Poverty Gap Index ( P1 ) and Income Gap Ratio What we use here is a non-parametric estimate of the The poverty incidence alone will not provide a complete non-food consumption of households in the reference picture of poverty. It is also important to look into the depth population whose food consumption is close to the food of poverty. For one individual, the depth of poverty is the poverty line. First, we calculate the mean per capita non- proportion by which that individual is below the poverty line food expenditures of households in the reference population (it has a value of 0 for all individuals above the poverty line). whose food spending lies within a plus or minus 1 percent The poverty gap index is the average depth of poverty bandwidth of the household whose food consumption is for the population. This is the sum of the depth of poverty of nearest the food poverty line. We increase the bandwidth each individual, divided by the total number of individuals to 2 percent and recalculate the average non-food per capita in the population. This gives a good indication of the depth expenses, and keep iterating up to a plus or minus 10 percent of poverty, in that it depends on the distances of the poor bandwidth. Then we take an average of all these mean per below the poverty line. Also, this index multiplied by total capita non-food expenditures and use this as our non-food population may be thought of representing the total cost of adjustment. In effect, the resulting non-food adjustment is a poverty reduction assuming perfect poverty targeting. weighted average of non-food expenses of households whose The poverty gap index can also be written as food expenses are near the food poverty line, with the highest weight on the households whose food spending are closest to P1 = H * ( z – yp  ) / z   the food poverty line (and with weights that decline as the food spending goes farther from the food poverty line). where ( z – yp  ) / z is referred to as the “income gap ratio” Similar to the Food Poverty line, the Non-food Poverty (mean depth of poverty as a proportion of the poverty line). Line for 2012 is updated from 2007 using the ratio of the Non- The income gap ratio is not a good poverty measure. To food Consumer Price Indices. see why, suppose that someone just below the poverty line is made sufficiently better off to escape poverty. The mean of Technical Note 4 (Poverty Measures) the remaining poor will fall, and so the income gap ratio will increase. And yet one of the poor has become better off, and Incidence of Poverty ( P0 ) none are worse off; one would be loathe to say that there is not The incidence of poverty is the proportion of the population less poverty, and yet that is what the income gap ratio would that is poor (percentage of the total population below the suggest. This problem doesn’t arise if the income gap ratio is poverty line). The percentage of households below the poverty multiplied by the head count index to yield P 1   . 40 Annex II: Technical Notes The poverty gap index doesn’t tell us how the poverty graph the cumulative percentage of households (from poor to is distributed among individuals; it may not convincingly rich) on the horizontal axis and the cumulative percentage of capture differences in the severity of poverty. The poverty consumption-expenditure on the vertical axis. This gives the gap will be unaffected by a transfer from a poor person to Lorenz curve as shown below. The diagonal line represents someone who is less poor. However, when the poverty gap perfect equality. The Gini coefficient is calculated as the area index is multiplied by the total population and the result A divided by the sum of areas A and B, where A and B are further multiplied to the poverty line, we obtain the aggregate as shown on the graph. If A=0 the Gini coefficient becomes gap. This represents the cost of eliminating poverty assuming 0 which means perfect equality, whereas if B=0 the Gini perfecting targeting and no targeting costs. coefficient becomes 1 which means complete inequality. Poverty Squared Gap Index ( P2    ) The Poverty Severity Index ( P 2    ) gives a weight to the poverty gap (more weight to very poor than to less poor). It is the average value of the square of depth of poverty for each individual. Poorest people contribute relatively more to the index. While this measure has clear advantages for some purposes, such as comparing policies which are aiming to reach the poorest, it is not easy to interpret. For poverty comparisons, however, the key point is that a ranking of dates, places or policies in terms of  P 2    should reflect well their ranking in terms of the severity of poverty. It is the ability of the measure to order distributions in a better way than the alternatives that makes it useful, not the precise numbers obtained. The poverty incidence, poverty gap and poverty squared gap measures all belong to a family of measures Formally, let xi be a point on the X-axis, and yi a point on proposed by Foster, Greer, and Thorbecke (1984). the Y-axis. Then Gini  = 1 – �( xi – xi–1 ) ( yi + yi–1). N α z – yi     = (1/n) � q P   i=1 α i=1 z When there are N equal intervals on the X-axis this simplifies to where α is some non-negative parameter, z is the poverty Gini  = 1 – 1 �( y  + y N line, y denotes per capita consumption, i represents individuals ). (or households), n is the total number of individuals (or N i=1 i i–1 households) in the population (or household population), and q is the number of individuals (or households) with per The Gini coefficient of inequality varies between 0, or capita consumptions below the poverty line. complete equality of expenditures, to 1, or complete inequality (one person has all the expenditure, all others have none). Technical Note 5 (Inequality Measures) b)  Quintile Dispersion Ratio a) Gini A simple measure of inequality is the quintile dispersion Graphically, the Gini coefficient can be easily represented by ratio, which represents the ratio of the average consumption different areas of the Lorenz curve, a cumulative frequency of the richest 20 percent of the population divided by the curve that compares the distribution of a specific variable average consumption of the bottom 20 percent. This ratio such as per capita expenditure with the uniform distribution can also be calculated for other percentiles (for instance, that represents equality. To construct the Gini coefficient, dividing the average consumption of the richest 5 percent 41 Bhutan Poverty Analysis 2012 – the 95th percentile – by that of the poorest 5 percent – the 5th percentile). The quintile dispersion ratio is readily interpretable, by expressing the consumption of the top 20% as a multiple of that of those in the poorest quintile (the “poor”). However, it ignores information about consumptions in the middle of the consumption distribution, and does not even use information about the distribution of consumption within the top and bottom quintiles. 42