HDI Impact Assessment Report on findings from Quantitative Household and Village Survey

Transcription

1 HDI Impact Assessment Report on findings from Quantitative Household and Village Survey Final Draft Report Part 1 of the HDI Impact Assessment

2 Maps showing location of sample villages (includes 170 out of 220 sample villages) 2

3 Black dots: HDI sample villages, yellow squares: Non HDI control villages. 3

4 4 Table of Content Executive summary... I 1 Introduction The baseline 2008 and 2009 surveys Objective and research questions Methodology Statistical tests of significant differences Findings from Village Profiles Village characteristics Economic activities Perceived trends in quality of life Agriculture Summary and conclusions on HDI impact assessed on the basis of Village Profiles Household Survey Perceived impact of HDI support on households in Demographic Characteristics Housing, Fuel, Water and Sanitation Monetary Poverty Indicators Food Security Labour Force and Employment Education, Literacy and Gender Health Status Ownership of Household durables/assets Agricultural land use and ownership of other agricultural resources Non-agricultural economic activities Economic Migration Indebtedness and Household Debts Summary of findings from the Household Survey Conclusions Recommendations Annex with tables and details of statistical tests in separate volume. List of tables Table 1 Main research questions for the impact assessment... 4 Table 2 Further research questions... 5 Table 3 Impact Assessment framework... 6 Table 4 Survey Sampling Design... 8 Table 5 Sample villages Table 6 Assessed living standards by HDI and Non HDI village Table 7 Housing condition over the last 3 years Table 8 Road condition of villages (last 3 years) Table 9 Improvement in quality of life over the last 3 years Table 10 Main reasons for change in the quality of life over the last 3 years Table 11 Main problems related to selling of agricultural products... 24

5 Table 12 Providers of technical assistance Table 13 Three main credit/loan providers to agricultural producers Table 14 Perceived impact of HDI by supported households support in Table 15 Perceived impact of HDI by supported households by State/Division Table 16 Scaled change in perception on the HDI project assistance (2008/ 2009 & 2012).. 32 Table 17 Perceived impact of UNDP support by types of support (% of all responses) Table 18 Average household size (2008/ 2009 & 2012) Table 19 Sex ratio (males to females) Table 20 Age dependency ratio (2008/ 2009 & 2012) Table 21 Age dependency ratio by gender of household head (2008/ 2009 & 2012) Table 22 Proportion of female headed households (2008/ 2009 & 2012) (% of households) 35 Table 23 Average household size by gender of household head (2008/ 2009 & 2012) Table 24 Economic dependency ratio Table 25 Economics dependency ratio by gender of household head (2008/ 2009 & 2012). 36 Table 26 Proportion of households using electricity for main lighting source (2008/ 2009 & 2012) (%) Table 27 Proportion of households using improved stove for domestic cooking (2008/ 2009 & 2012) (%) Table 28 Proportion of households using improved drinking water source (2008/ 2009 & 2012) (%) Table 29 Proportion of households using fly proof latrine (2008/ 2009 & 2012) (%) Table 30 Poverty lines Table 31 Food poverty headcount index (2008/ 2009 & 2012) (%) Table 32 Poverty head count index 2008/ 2009 & 2012) (%) Table 33 Poverty headcount index by gender of household head (2008/ 2009 & 2012) Table 34 Poverty gap (2008/ 2009 & 2012) Table 35 Squared poverty gap (2008/ 2009 & 2012) Table 36 Adult equivalent consumption expenditure excluding health expenditure per year (2008/ 2009 & 2012) (Kyat at current prices) Table 37 Food expenditure share in % of total expenditure excluding health expenditure (2008/ 2009 & 2012) Table 38 Food expenditure share in total expenditure excluding health expenditure by gender of household head (2008/ 2009 & 2012) (%) Table 39 Average months without borrowing for food in the past 12 months (2008/ 2009 & 2012) Table 40 Labour force participation rate (2008/ 2009 & 2012) (%) Table 41 Unemployment rate (2008/ 2009 & 2012) (%) Table 42 Labour force participation rate by gender (2008/ 2009 & 2012) Table 43 Unemployment rate by gender (2008/ 2009 & 2012) Table 44 Gross enrolment ratio in primary education (2008/ 2009 & 2012) (%) Table 45 Net enrolment ratio in primary education (2008/ 2009 & 2012) (%) Table 46 Gross enrolment ratio in secondary education (2008/ 2009 & 2012) (%) Table 47 Net enrolment ratio in secondary education (2008/ 2009 & 2012) (%) Table 48 Self reported morbidity incidence past 14 days before survey (2008/ 2009 & 2012) (%) Table 49 Proportion of population sleeping under bed net last night (2008/ 2009 & 2012) (%) Table 50 Proportion of households that own motorcycle (2008/ 2009 & 2012) (%) Table 51 Proportion of households that own bicycle (2008/ 2009 & 2012) (%) Table 52 Proportion of households that own sewing machine (2008/ 2009 & 2012) (%) Table 53 Proportion of households that own generator (2008/ 2009 & 2012) (%)

6 Table 54 Proportion of households that own radio (2008/ 2009 & 2012) (%) Table 55 Proportion of households that own TV (2008/ 2009 & 2012) Table 56 Proportion of households that own agriculture land (2008/ 2009 & 2012) (%) Table 57 Average area of agriculture land owned by landowner households (2008/ 2009 & 2012) (acre) Table 58 Proportion of agricultural households (2008/ 2009 & 2012) (%) Table 59 Proportion of agriculture households that grow second crop (2008/ 2009 & 2012) (%) Table 60 Proportion of agricultural households that own draught animal (2008/ 2009 & 2012) (%) Table 61 Proportion of agriculture households that grow rice as main crop (2008/ 2009 & 2012) (%) Table 62 Average area of rice sown as main crop (2008/ 2009 & 2012) (acre) Table 63 Average yield of rice as main crop (2008/ 2009 & 2012) (basket) Table 64 Proportion of agriculture households that grow sesame as main crop (2008/ 2009 & 2012) (%) Table 65 Proportion of agriculture households growing maize as main crop (2008/ 2009 & 2012) (%) Table 66 Proportion of households that are breeding large animals (2008/ 2009 & 2012) (%) Table 67 Average number of large animal bred (2008/ 2009 & 2012) Table 68 Proportion of households that has pig or goat (2008/ 2009 & 2012) (%) Table 69 Average number of pig or goat raised (2008/ 2009 & 2012) Table 70 Proportion of households practicing poultry breeding (2008/ 2009 & 2012) (%) Table 71 Average number of poultry bred (2008/ 2009 & 2012) Table 72 Proportion of non-agriculture household (2008/ 2009 & 2012) (%) Table 73 Proportion of trading or retail shop households among non-agriculture household (2008/2009 & 2012) (%) Table 74 Proportion of street vendor households among non-agriculture household (2008/ 2009 & 2012) (%) Table 75 Proportion of households in cottage industry among non-agriculture household (2008/ 2009 & 2012) (%) Table 76 Proportion of households with migrant workers (2008/ 2009 & 2012) (%) Table 77 Proportion of households with household head migrated for work at any time in last year among the households with migrant workers (2008/ 2009 & 2012) (%) Table 78 Proportion of households with 15 year and above female household member migrated for work at any time in last year among the households with migrant workers (2008/ 2009 & 2012) (%) Table 79 Proportion of households with 15 year and above male household member migrated for work at any time in last year among the households with migrant workers (2008/ 2009 & 2012) (%) Table 80 Proportion of indebted households (2008/ 2009 & 2012) (%) Table 81 Average loan size per indebted household (2008/ 2009 & 2012) (Kyat) Table 82 Proportion of indebted households that borrow money from relatives or friends (2008/2009 & 2012) (%) Table 83 Proportion of indebted households that borrow from moneylender (2008/2009 & 2012) (%) Table 84 Proportion of indebted households that borrow money from SRG (2008/ 2009 & 2012) (%) Table 85 Summary of double difference in Key Indicators Table 86 Summary of double difference in indicators in Household Survey

7 7 List of figures Figure 1 Composition of households in the surveys... 7 Figure 2 Overview of main support programmes by supporting organization and type of village Figure 3 Sex ratio in villages Figure 4 Main diseases of adults by State/Division Figure 5 Main diseases of children by State/Division Figure 6 Changes in job opportunities last 3 years Figure 8 Key buyers of agricultural produce Figure 9 Difference in trends in agribusiness and sales of agricultural products Figure 11 Main problem with credit related to agribusiness Figure 12 Per cent of agribusinesses running on credit and loans by State/Division Figure 13 Perceived impact of HDI by households by year of SRG establishment Figure 14 Changes in poverty head count by HDI household membership Figure 15 Poverty head count in HDI villages by membership and geographical area Figure 16 Poverty head count in HDI and Non HDI villages by geographic area Figure 17 Change in Food security by State/Division and HH type Figure 18 Differences in gross enrolment ratio in primary education Figure 19 Differences in net enrolment ratio in secondary education... 51

8 8 Acknowledgements This study was carried out on the initiative of UNDP, Myanmar, as part of the Impact Assessment of the Human Development Initiative (HDI) programme and financed through a grant from DfiD, Myanmar. The study team comprising Jens Sjørslev, international consultant, Htun Htun Oo, national consultant quantitative and Ye Thaung Htut, national consultant qualitative, designed the study on the basis of the 2008/09 Impact Assessment, analysed the data and reported on findings and conclusions. Htun Htun Oo did the analysis of the quantitative household survey data, while Jens Sjørslev did the analysis of the quantitative village profiles data and compiled the report. The fieldwork for Household Survey and Village Profiles was carried out by the company MMRD. The Team would like to thank Salai Cung Lian Thawng, who guided the study in the first stages and Minn Sann, Monitoring and Evaluation Unit in UNDP, Elisabet Dahlberg Frisk and M Shafiquer Rahman, Policy Unit of UNDP for their support and inputs, and Murshid Khan, consultant to UNDP, and Elinor Bajraktari, UNDP for their comments to the drafts report. Abbreviations and acronyms CBO DD FGD HDI IHLCA LG pp SRG UG Community Based Organization Double Difference Focus Group Discussions Human Development Initiative Integrated Household Living Conditions Assessment Livelihood Group Percentage points Self Reliance Group User Group

9 I Executive summary Background The Human Development Initiative (HDI) is UNDPs flagship programme in Myanmar. It consists of two large integrated community development projects - the Integrated Community Development Project (ICDP) and the Community Development for Remote Townships Project (CDRT). The two projects operate in different areas, and were earlier separate with different strategies, but in 2010 a common logical framework was established and they have since then been closely coordinated, but still have separate reporting mechanisms. The 2012 Impact Assessment measures progress towards the immediate objective of the HDI as described in the Project Document for Phase IV, which is to Cultivate and strengthen the capacity of village institutions and Community Based Organisations and households in project villages to plan and undertake development activities that address their basic and social and food security needs in a participatory, sustainable and transparent manner. The Impact Assessment reports on the main indicators in the programme monitoring systems as well as on a number of variables that were analysed in the 2008 Impact Assessment. These variables are similar to those applied in the nation-wide Integrated Household Living Conditions Assessment (IHLCA) of Outcome and impact assessments have been carried out in the past in 2006, 2008 and in 2009 in the Delta. In 2008 the impact assessment applied a household survey in 160 supported villages and in 60 control villages, which established a proper baseline. The 2009 Delta survey applied a similar design. The present 2012 quantitative household analysis comprises panel data of the households that were surveyed in the two baseline surveys and which were revisited in a survey carried out in February The overall 2012 HDI Impact Assessment comprises three separate but integrated studies, which are presented in three volumes: The present report is Vol. 1: Quantitative Household and Village Survey, which presents the findings from quantitative analysis of Village Profiles and the Household Survey panel data. The two other reports are: Vol. 2: Report on study of Social Capital Formation which present findings from a quantitative analysis of Village Profiles with regard to social capital and village level development outcomes, and results from 65 Focus Group Discussions in 25 villages. Vol. 3: Study of Poverty Dynamics and Causes of Poverty Based on Household Survey Panel Data from the Impact Assessment of HDI 2008/ and a qualitative study of impacts of life events on socio-economic well being in 200 households. Limitations The present impact assessment is not an evaluation and does not assess the relevance, efficiency, effectiveness, and sustainability of HDI. It does not include analysis of inputs and outputs as they relate to outcomes but focuses purely on detecting impacts.

10 II Approach The main analytical approach is the double difference (DD), which comprise two steps: first analysis of the differences between HDI supported households and the control households in Non HDI villages in 2008/09 and in 2012 and secondly analysis of differences between the changes in these two groups from 2008/09 to The DD is reported as the difference in percentage points (pp), i.e., the absolute differences, but the double difference approach captures the relative impact of HDI in the context of general development trends in the rural areas. The DDs for the various indicators and socio-economic variables have been calculated at the aggregate level for the whole sample and also for each geographical area. Tests for statistical significance have been applied comprising mainly t-tests and Chi-square tests. Statistical significance has been set at an alpha level of 5% for the village profile analysis, and calculated for alpha levels of 10%, 5% and 1% for the household survey. Weights have been applied to both the baseline 2008/09 data and the 2012 data. This was not done in the previous impact assessments. In effect the present impact assessment compares the 2012 data to recalculated baseline data. Findings on Country Programme Indicators The HDI Impact Assessment covers five out of the seven UNDP Country Programme indicators and targets set for the end of year Two other indicators for the HDI, namely the percentage of beneficiary households moving to higher wealth rank, and the percentage of SRGs moving to higher maturity level, are part of the normal on-going monitoring system and have not been covered in the present study. The results of the analysis are shown in table below. It is important to note that none of the aggregate double differences on these indicators are statistically significant. This means that HDI impacts on these indicators cannot be ascertained, though the difference between planned targets and actual results are reported. Country Programme outcome indicator Average # of months food-secured by the beneficiary households % in Poverty Headcount Index among target beneficiary households % of beneficiary household s budget spend on food % in disease morbidity of the beneficiary households (disaggregated) % in net primary enrolment ratio (disaggregated) Baseline Target (end 2011) 7,2 months 10 months 42,7% 38,5% 64,4% 62% 13,4% 5% 89,3% 98% Impact Assessment finding based on double difference Actual 2012: 6,6 months Target: 2,8 months increase Actual: 0,5 months decrease Actual baseline: 40,4% Actual 2012: 38,8% Target decrease: 4,2 pp Actual decrease 3,7 pp Actual baseline: 66,1% Actual 2012: 66,2% Target decrease: 2,4 pp Actual increase: 2,2 pp Actual baseline: 17% Actual 2012: 11,6% Target decrease: 8,4 pp Actual increase: 1,4 pp Actual baseline: 90,3% Actual 2012: 84,7% Target: 9,7 pp increase Actual: 1,2 pp increase

11 III The targets set for end 2011 on key indicators would appear not to have been reached: the target for poverty head count has almost been achieved with only 0,5 pp below target. Food security has decreased by 0,5 months, which is 3,3 months off target. Food expenditure ratio is 4,6 pp off target, morbidity 7 pp off target and net enrolment in primary education is 8,5 pp off target. However, as mentioned, at the aggregate level the DDs on these indicators are not statistically significant. Statistically significant DDs for main indicators were found in two geographical areas: For the Poverty Head Count Index statistically significant differences were found in some geographic zones, but negative DDs indicating a positive HDI impacts were only found for the Dry Zone (at alpha 1%) and Shan State (at alpha 5%). For the percentage of beneficiary households budget expenditure on food the DD was negative with statistical significance (at alpha 5%) in Northern Rakhine State indicating a positive HDI impact. Findings from Village Profiles Data from 220 Village Profiles were compiled during the survey in 2012 based on key informants. The analysis does not apply the double difference, since the baseline data from 2008/09 were not cleaned properly and could not be used. The analysis of the Village Profiles therefore reports on the single difference between HDI supported villages and Non HDI villages. Statistically significant differences in favour of HDI impact were observed in the following village level indicators: in the HDI village group the ratio of a perceived improvement in the quality of life over the last three years was 24 pp higher than in the Non HDI village group; the ratio of HDI villages with improvements in job opportunities was 13 pp higher than in the Non HDI village group, but job opportunities have has also worsened in 6 pp more HDI villages. Trends in agribusiness over the last three years were found to be more significantly more diverse in HDI villages. A number of other minor statistically significant differences between HDI and Non HDI villages were found on the following variables some of which possibly could be considered indicators of spin-off of HDI impact. However these have very weak correlations between type of village and the variable in question. The minor differences include: the condition of road leading to the village has improved in slightly more HDI villages; there is access to small markets in more HDI villages, but still in only 8,5% of these. With regard to access to a health facility the ratio of Non HDI villages using Station Hospitals is 13 pp higher, while 7 pp more HDI villages use Township Hospitals. With regard to economic activities of women the ratio of Non HDI villages with casual work as the main activity for women is 7-10 pp higher than in the HDI village group. With regard to the following variables no statistically significant difference between HDI and Non HDI village was found: cost of housing; minimum average daily wages for men and women; the most serious children s diseases; access to nearest town; nos. of teachers in schools; nos. of retail shops; and economic activities of men, and some additional variables. As these variables are not influenced by HDI this is evidence of the basic similarity of HDI and Non HDI villages.

12 IV The conclusion based on analysis of the Village Profiles is that there are no outstanding impacts of HDI in terms of village level development. Please also refer to the findings and conclusions of the Study of Social Capital Formation (Vol. 2 of the HDI Impact Assessment) for further results from Village Profiles. Findings from Household Survey Overall perceived impacts The panel data included 2499 households that have received support from HDI in some form. These households were asked how do you perceive the impact of HDI overall activities on your household? About half of the supported households found their situation has noticeably improved and about half that is has improved slightly. Households that are members of both SRGs, Livelihood Groups, Community Based Organizations and User Groups responded more positively. There is a downward trend since 2008/09 in perceived positive impact. On a scale where 1 is noticeably negative affected and 5 is noticeably improved the average for all HDI supported households has gone from 4.7 in 2008/09 to 4.3 in The longer households have been members of HDI initiated groups, mainly SRGs, the more likely they are to answer noticeably improved. 35% of all answers refer to provision of credit as the main factor for a positive impact, however out of the 35%, 20 pp of the answers were slightly improved. Double difference on other variables Since the questionnaire for the Household Survey in 2008/09 was modelled after the IHLCA, the panel data includes information on a large number of additional variables grouped under the headings of demographic characteristics, ownership to assets, employment, housing and basic facilities, agriculture, indebtedness, non-agricultural activities and economic migration. Not all of these variables were expected to be influenced by HDI. Nevertheless the DD has been calculated for all variables to avoid missing possible spin-off or unintended impacts, and to check for comparability between the HDI and Non HDI households. The results are summarized in the following paragraphs. Demographic characteristics: The aggregate DDs with regard to changes in demographic characteristics is marginal and not statistically significant. Since demography is not impacted by HDI this is evidence of the basic similarity and comparability of the HDI and Non HDI households. However a significant negative DD in sex ratio was found in Chin; and between the survey years significant differences in age and economic dependency ratio were found in Chin and Eastern and Northern Rakhine, possibly indicating male out migration. A significant positive DD on the proportion of female headed households was found in Northern Rakhine indicating a higher proportion of HDI households where the man has left. Ownership to assets is an indicator of wealth and a positive DD value would indicate positive impact of HDI. The DDs for ownership of motorbikes, bicycles, and generators are positive whereas for radio, TV and sewing machines the DD is negative. However the aggregate DDs are not statistically significant and no conclusion of positive impact can thus be inferred from ownership to assets taken as a whole. Employment: the aggregate DD for labour force participation rate is 0,7 pp and for unemployment rate -0,1 pp. Neither of these DDs is statistically significant.

13 V Housing and basic facilities: for the proportion of households with electricity, households using improved stoves and households using improved drinking water, the aggregate DDs are negative, but not statistically significant. Only the DD of -24 pp on households using fly proof latrines is statistically significant, which is due to a large proportion of Non HDI households having started to use such latrines. The assessment includes sixteen agriculture related indicators. Due to the different conditions in the geographical areas the aggregate DD is not an adequate measure to assess whether there have been positive HDI impacts with respect to these variables. The impact assessment has not been scoped to include a detailed analysis of agricultural outcomes in the different geographical areas but statistically significant DDs for geographical areas are reported in the relevant sections of the report. The aggregate DD for the proportion of agricultural households is -2,0 pp indicating slightly fewer such households in the HDI group, but not statistically significant. The aggregate DD for the proportion of households that own agriculture land is -4,6 pp, and the aggregate DD for the average area of agriculture land among landowner households (Acres) is -0,9. Neither of these DDs is statistically significant. Only the DD for Chin for the proportion of households that own agricultural land is statistically significant (at alpha 5%), and this is because many Non HDI households have gotten land in the past three years. The aggregate DD for the proportion of agriculture households that grow rice as main crop is 0,1 pp, and for the proportion that grow sesame as main crop the DD is 1,9 pp; these DDs are not statistically significant. The aggregate DD for the proportion of households that grow second crop is -6,4 pp; for the percentage of agriculture households that grow maize as main crop -0,6 pp; for the average area of rice sown as main crop (acre) -2,1 pp; and for the average yield of rice as main crop (baskets) the DD is -6,6 pp. None of these aggregate DDs are statistically significant but the DDs for some geographical areas are. The aggregate DDs for livestock are negative but none are statistically significant. They include the proportion of households that have large animals: DD of -4,4 pp; households that own draught animal: -1,3 pp; households that have pigs or goats: -4,0 pp; and the average number of poultry raised by households: -2,4 nos. The aggregate DD for the proportion of households practicing poultry breeding is 2,3 pp and the for the average number of pigs or goats raised it is 0,1 nos; these are not statistically significant. However the aggregate DD for the average number of large animal raised, which is 0,7 nos, is statistically significant (at alpha 5%). Indebtedness The aggregate DD for the percentage of indebted households in HDI villages is 3,4 pp, which is not statistically significant. There has been a decline in households borrowing from SRGs with a DD of -10,9 pp (statistically significant at alpha 1%). Borrowing from relatives or friends has an aggregate DD of 3,3 pp but this is not statistically significant. Only the DD for Eastern Rakhine is statistically significant (at alpha 1%), which may be due to a number of households that has left the SRGs since Borrowing from moneylenders has a DD of 4,9 pp, and is not statistically significant. With regard to the average loan size the aggregate DD is not statistically significant. However, there has been a statistically significant increase in the size of outstanding loans of HDI households in Shan (at alpha 10%), Chin (at alpha 1%)

14 VI and Eastern Rakhine (at alpha 1%), while the size of loans has gone down in the Delta (at alpha 10%). Non-agricultural economic activities The proportion of non-agricultural households among the HDI households has decreased with an aggregate DD of -10,4 pp (statistically significant at alpha 5%). The decrease has occurred mainly in the Dry Zone, the Delta and in Northern Rakhine where the DDs are statistically significant. Among non-agriculture households there is a positive aggregate DD of 9,8 pp for the proportion of households that engage in trading or retail shops, however not statistically significant, though the DD for the Dry Zone is at alpha 5%. Also among the non-agricultural households the DDs for the percentages of street vendors and households engaged in cottage industry are negative with -1 pp and -4,5 pp respectively, but not statistically significant. Economic migration The aggregate DD for the proportion of households with migrant workers is 3,8 pp, and not statistically significant. However in the Dry Zone the proportion of the HDI households with migrant workers has gone up with a statistically significant DD of 14,5 pp (at alpha 1%) while it has gone down in Eastern Rakhine with a negative DD of 14,4 (at alpha 1%) In the Dry Zone it is significant (at alpha 10%) that it is female household members above 15 years of age who migrate. The aggregate DD for households with female migrate workers is 10,3 pp, which is not statistically significant, and for male household members the DD is -2,9 pp. The DD for the percentage of households where the household head has migrated is -5 pp, which is not statistically significant. Recommendations The results of the impact assessment presented in this report leads to the following general recommendations Further analysis of the survey data should be done in combination with analysis of M&E data on inputs to evaluate efficiency and effectiveness of support interventions and lessons learned. Future support to rural development should be designed with great consideration to the different geographical, socio-economic, social and agricultural areas. In view of the modest impacts detected, the spread of HDI over a large number of villages appears to be inefficient in terms of delivering development outcomes. Therefore prioritization of the most important support interventions in each township based on demand driven approach and local development plans with higher per capita investment should be considered.

15 1 1 Introduction The HDI consists of two large integrated community development projects - the Integrated Community Development Project (ICDP) and the Community Development for Remote Townships Project (CDRT), and further a microfinance project, a smaller HIV/AIDS project, and the IHLCA project, which focuses on data collection and analyses of household living conditions based on nationwide surveys. The Integrated Community Development Project (ICDP) and the Community Development for Remote Townships Project (CDRT) are the subjects of the present assessment. The Logical Framework of 2010, which the two projects have in common, states the purpose of the HDI to be 128,000 poor and vulnerable households (in the project areas) have improved food security, increased incomes and enhanced organizational capital. Output 1 was Increased food production of 19,200 poor and vulnerable households in the project area from agriculture, livestock, fisheries and forestry and output 2 Village community groups strengthened to implement village development priorities. Both projects implement a wide range of support activities including agricultural training and support to agricultural inputs, organising village community groups in women s self reliance groups and various livelihood groups, village development activities as well as health and education related support. An outcome and impact assessment of the HDI programme is conducted periodically. The first Outcome Impact Assessment was conducted in 2006 based on research and analysis of programme & project documents, reports and project data. In 2008, Impact Assessment of the HDI was conducted using sample household survey and participatory assessment tools. The survey design employed project group (treatment group) and non-project group (control group) to generate a counterfactual. The assessment focused on the impact of community development projects and to a lesser extent on HIV/AIDS project but did not include the Ayeyarwady Delta area due to effects of cyclone Nargis (May 2008). However, later in 2009 a similar assessment was conducted in the Nargis affected Delta area. The HDI Impact Assessment 2012 focuses on the results of the two community development projects. It uses 2008 assessment and 2009 assessment (for the Delta) surveys as baseline and establishes a panel of households from the baseline surveys that have been revisited in Thus comparability of the 2012 assessment and 2008 and 2009 assessments has been critical in designing the present impact assessment. The stakeholders of the Impact Assessment include UNDP management and field staff, beneficiaries/target group, development partners, government, other development organizations and the public in general, both in Myanmar and in donor countries. UNDP, both management and field staff, is the main stakeholder in the impact assessment, with a serious interest in obtaining a credible and accurate picture of how HDI in its past and current set up is doing. The beneficiaries of the Impact Assessment are stakeholders, both because their time and effort in participating in HDI, in many cases over a number of years, should show positive results, and because the assessment may point to new directions for the support HDI will provide (perhaps under a different name) in the future. UNDP s development partners will scrutinize the impact assessment as well and the results may influence the level of future support.

16 2 1.1 The baseline 2008 and 2009 surveys The 2008 Household Survey was carried out in 220 sample villages of 20 ICDP and CDRT sample townships (N=4400) in November 2008 to January The 2008 survey used two types of survey instruments for the quantitative survey Community Key Informant Survey questionnaire (Village Profile) and Household Survey questionnaire. The instruments were also applied for the 2012 survey, with additional questions relating to social capital formation. The Village Profile focused on village characteristics, access to infrastructure and programme and projects being implemented in the respective village. The household survey questionnaire focused on 1) Household Member Characteristics; 2) Housing; 3) Education and Literacy; 4) Health; 5) Consumption Expenditure; 6) Household Assets; 7) Livelihood, and 8) Food Security and Indebtedness. The 2009 assessment survey (for Delta) was fielded in 160 villages in all 5 Early Recovery (ER) project townships (N=4000) in Delta in August to September It used the same Community Key Informant Survey questionnaire and Household Survey questionnaire adopted from 2008 assessment survey with some modification to be compatible with ground situations under the ER programme The 2008 assessment survey produced a report named Outcome and Impact of the UNDP Human Development Initiative in Myanmar, 2008 that covered outcome/impact of the CDRT project and the ICDP project. It reported that the poverty headcount index in HDI project villages was 43%. In addition, the report also highlighted the findings on food security, source of debt, roof and wall material, type of stove, type of farm business, distribution of occupation, ownership and access to productive assets, and perceptions of the impacts of the HDI. The 2009 assessment survey (for Delta) produced the report Outcome/Impact Assessment of UNDP Support for Recovery of Cyclone Nargis Affected Rural Communities in the Ayeyarwady Delta, Myanmar published 30 November It also used almost the same set of survey instruments that were used in the 2008 assessment. However, there were no estimates of poverty headcount available. A qualitative assessment was made based on 300 households in 30 villages in 2 ER townships, namely Bogale and Labutta. Group discussions, key informant interviews and in-depth household interviews were used for the qualitative exercise. Contextual issues In 2012 insurgency made it impossible to conduct the household survey in Kachin State. The CDRT township offices in Moemauk, Man-Si and Myitkyina townships reported that it was not safe to access the sample villages. The Tropical Storm Two that made landfall near the Myanmar-Bangladesh border (near Maungdaw Township) on 19 October 2011 resulted in heavy rain (up to mm) and triggered flash floods in Magwe, Mandalay and Sagaing Regions. Pakokku District in Magwe Region was the worst affected by the floods and one sample township from the 2008 assessment exercise Pakokku Township was moderately affected.

17 3 Comparability The following observations may be noted with regard to comparability of data from the various rounds of surveys: In the present Impact Assessment 2012 exercise, data from 2009 (Delta) is included as part of the baseline; In 2012 Kachin State was not included in the sample due to security reasons; In 2012 a number of additional questions pertaining to social capital were included there are no baseline data for these; In 2008 weights were not applied in the statistical analysis. In the present Impact Assessment 2012 weights have been applied including to the 2008/09 survey data to ensure comparability. In 2008 the Village Profile data were not adequately cleaned preventing comparative analysis with the 2012 data. 2 Objective and research questions The overall approach to the Impact Assessment has been to apply the original outcomes in Project Document as the overall yardstick for assessment of impact. It is important to note that the Impact Assessment is not an evaluation and does not specifically assess the relevance, efficiency, effectiveness, and sustainability of HDI. Further, the impact assessment does not include analysis of inputs and outputs as they relate to outcomes. However, in addition to assessing the impact, the rich data can and should also be used for the purpose of lessons learned to feed into design of the new programme. The overall objective of the 2011 Impact Assessment is To provide an assessment of the extent to which HDI has been successful in achieving its immediate objective to: Cultivate and strengthen the capacity of village institutions and Community Based Organisations and households in project villages to plan and undertake development activities that address their basic and social and food security needs in a participatory, sustainable and transparent manner. Through the three outcomes: Village community institutions are able to prepare and implement more effective village development activities in a participatory and equitable manner. Village communities are able to take care of their and their families basic social needs (health and education) as a result of increased skills and knowledge and access to social infrastructure and services. Village communities achieve improved incomes and food security as a result of increased skills, knowledge and access to sustainable livelihood services and assets. It is recognized that the prioritization of outcomes and outputs of HDI has varied over the years, but as mentioned above it is found appropriate to apply the original outcomes as the overall yardstick for the IA. The main specific research questions for the Household and Village level HDI Impact Assessment listed below relate to various aspects of these expected outcomes and are all included as MMR indicators for 2011.

18 4 Table 1 Main research questions for the impact assessment Research Detailed research Details of method question question % of HDI households being below the poverty Poverty line in 2011 sample Headcount compared to Non-HDI Index and to 2008 HDI and Non- HDI households % of beneficiary households budget expenditure on food Average # of months foodsecured by the beneficiary households % in disease morbidity of beneficiary households (disaggregated) % in net primary enrolment ratio (disaggregated) The distribution of HDI and Non-HDI, 2008/09 and 2011 households expenditure on food. Have HDI households better food security than Non-HDI households, and has their food security improved since 2008? Have HDI households (SRG and Non-SRG) lower morbidity than Non HDI households, especially diarrhoea, and less than in Have HDI households (SRG-Non SRG) a higher enrolment ratio than Non HDI households. Has there been a development since the enrolment ratios since 2008/09? * Calculated using the formula used in IHLCAS: Poverty line definition is linked to Food Poverty Line calculation Food expenditure to total expenses as percentage is calculated for each household. The distributions of the percentages for each group are compared: mean, median, variance Distribution of responses to direct question about how many months in the last year the HH could eat meals regularly without borrowing in cash or kind Based on 2 weeks recall time for individual household members on diarrhoea/sickness events. Disaggregation in 2008 report is by SRG-Non-SRG households, not by sex. Data source Module E in HH questionnaire. Module E in HH questionnaire. Module H in HH questionnaire. Module D in HH questionnaire. Module C in HH questionnaire Data sources: Household Survey Notes: * 2008 found no significant difference between HDI and Non HDI, or SRG and Non-SRG households on enrolment; slightly higher on literacy for SRG HHs. There was no disaggregation by sex. Furthermore, the Impact Assessment includes an analysis of a number of general and contextual indicators that were included in the baseline surveys 2008/09, and where originally from the IHLCA national survey. These are listed in table 2.

19 5 Table 2 Further research questions Research Detailed research question question Changes in Demographic Characteristics Variables assessed Average Household Size Demographic Dependency Ratio (DDR) Economic Dependency Ratio (EDR) Female- headed Households (%) Changes in livelihoods Changes in Distribution of Occupation among Economically Active Population Ownership and Access Labour Force Participation and Employment Housing, Water and Sanitation Health and Nutrition Analysis of the occupational categories Average Land Area Owned (Acres) Landless Rate in Agriculture (%) Access to Credit (Agriculture) Debt (Percentage of Households) Labour Force Participation Rate Unemployment Rate (7 days) Underemployment Rate (7 days) Access to 'Quality' Roofing Access to Safe Drinking Water Access to Improved Sanitation Access to Electricity Self-Reported Morbidity Incidence Access to Health Care Health Share in Consumption Exp. 3 Methodology The HDI Impact Assessment comprises three separate but integrated studies in three volumes: Vol. 1: Quantitative Household and Village Survey (the present), presenting the findings from quantitative analysis of Village Profiles and the Household Survey panel data. Vol. 2: Report on study of Social Capital Formation which present findings from a quantitative analysis of Village Profiles with regard to social capital and village level development outcomes, and results from 65 Focus Group Discussions in 25 villages. Vol. 3: Study of Poverty Dynamics and Causes of Poverty Based on Household Survey Panel Data from the Impact Assessment of HDI 2008/ and a qualitative study of impacts of life events on socio-economic well being in 200 households. Table 3 shows the data sources and sampling for the three studies, and Figure 1 shows the composition of different types of households in the three surveys, of which the 2008 and the 2009 survey in the Delta in combination provides the baseline for the present Impact Assessment.

20 6 Table 3 Impact Assessment framework Survey/Study Quantitative Impact Assessment Social Capital Household Survey Village Profiles Formation study Poverty Dynamics and Causality Study Data type Quantitative Quantitative Mixed methods: Qualitative and Quantitative Total sample 4564 households 220 villages 20 Townships 220 villages 20 Townships Of which Non HDI control 60 villages (50 villages in 2012 some have been included in HDI since 2008) 1225 households in Village Profiles 65 Focus Group Discussions in 25 villages 20 Focus Groups 10 villages 4564 households data from questionnaire 200 households life events 80 households Baseline 2008 and 2009 (Delta) None 2008 and 2009 (Delta) Data Questionnaire, same as 2008/09 + Questionnaire, collection tools additional questions relating to Social same as 2008/09 Capital Main approach Questionnaire, same as 2008/09 + a few additional questions relating to Social Capital Quantitative analysis measuring indicators and variables in Double Difference: Same format used for all villages HDI Non HDI, SRG - Non-SRG households Baseline to 2011 (Panel Data) Social Capital Formation Tool (SOCAT) Focus Group discussions and participatory methods Mixed methods: In-depth Focus Group discussions (3 in each village) supported by quantitative analysis of data from Household Survey and Village Profiles Participatory life event timeline and cause-effect tables Mixed methods: quantitative analysis of poverty dynamics of all sample households Life event timelines capturing causes of economic mobility

21 7 Figure 1 Composition of households in the surveys It is important to be aware of the overlap between the types of membership of HDI supported households and the geographical areas. In effect the households that are members of both SRGs and Livelihood Groups (LG), community based organizations (CBOs, and/or user groups (UG), are all located in Chin and Rakhine, and the same is the case for households that are only members of LG/CBO/UGs. This is also the area where one of the two programmes of HDI, namely the Community Development in Remote Townships (CDRT) operates. In the other geographical areas HDI is implemented through the Integrated Community Development Programme (ICDP). The box below describes the features of the different kinds of HDI initiated groups that households can be members of. Self Reliance Groups - SRG The Self Reliance Group (SRG) was first introduced into the UNDP HDI program through the Community Development in Remote Townships (CDRT) HDI project in the period Initially operating in remote townships of Rakhine, Chin and Kachin Zones, those SRGs were based on the model developed by the Indian NGO, MYRADA in Southern India. The Integrated Community Development Project (ICDP) of UNDP HDI program also introduced SRGs among its new activities to support community livelihoods, especially for the poor, in mid July SRGs were formed purely with women on an affinity basis in ICDP, whereas the CDRT project had some mixed sex SRGs. At the end of 2008, the two projects included 3,774 functioning SRGs in Chin, Kachin, Mon and Kayin, Rakhine, Shan, Dry and Delta zones throughout Myanmar, of which 99% were exclusively women. The two community developments CDRT and ICDP have increasingly made efforts to mainstream gender empowerment through SRGs by providing increased access to credit, technology, and inputs.

22 8 Livelihood Groups, Community Based Organizations and User Groups LG/CBOs/UGs HDI has formed marginal farmer groups, marginal fisherman groups, and rice bank groups. Training of these groups has included community development awareness raising, training in management, agriculture, fishery, livestock, cottage industry such as weaving, tailoring, bee keeping and fish paste making. Training also included workshops to raise the community development awareness and build managerial and technical know-how/ capacity of the communities. A system with revolving in kind assistance was established so that a livelihood group could receive assistance from other livelihood groups. Inputs to the groups have included agriculture related assistance such as fertilizer provision, seed provision, and provisions of draught cattle and hand tractors for marginal farmer groups to be used as revolving funds for marginal farmer groups. Also fishing gears and small fishing boats have been provided. The HDI program has furthermore formed soil conservation and gully control groups, community initiated forest (CIF) groups, fuel-saving stove making groups, watershed management groups and mangrove forest conservation groups. It has conducted awareness-raising, managerial and technical training and workshops to raise the environmental ecosystem awareness and build managerial and technical know-how of the communities. The formation of these groups and networks have been useful and helpful to implement the activities and projects which were designed to address the issues of soil erosion and water-shed forest depletion in Shan and Dry zone; the issues of slash and burn cultivation in Chin zone; and Mangrove forest depletion in Delta and Rakhine zone. The sampling design is presented in table 4 below. Table 4 Survey Sampling Design Sampling Stage Sampling Unit Selection method Remark Design Stratified three stage sampling Stratification Based on type of project (ICDP/ CDRT) and duration of intervention in the township (Old/ New Township) First strata HDI Township PPS with replacement PPS = Probability Proportional to Size Second strata Village PPS Systematic 2 village sampling frames provided HDI villages from the ICDP/ CDRT villages: (with SRG villages frame and presence of SRG in the village) Non-HDI villages frame Non-HDI villages Third strata Household Systematic 2 household sampling frames in HDI SRG household random sampling village: SRG households frame and Non-SRG CDE households frame - Non-SRG CDE 1 household sampling frame in Non- HDI village: CDE households frame household A&B wealth ranked households are excluded from the frame in all villages Weighting Applied for each sampling frame separately for each survey The sampling was done with emphasis on the SRG households (therefore, Both SRG and LG/CBO/UG is also representative) but not specific to only LG/ CBO/ UG households. The sample allocation to no membership households is quite low compared to the sample

23 9 allocation to SRG households (refer table 22 in Annex for the number of sample households by membership). An important element of the present Impact Assessment has been calculation and application of weights for both the baseline 2008/09 data and the 2012 data. This was not done in the previous impact assessments. In effect the present 2012 impact assessment compares the 2012 data to recalculated baseline data. The main analytical approach is to calculate the double difference (DD) between HDI supported households and the control households in Non HDI villages, and between the 2008/09 and the 2012 surveys. The DD is calculated as the difference in percentage points (pp), i.e., the absolute difference. The DDs have been calculated at the aggregate level for the whole sample as well as for households with different types of membership and by geographical area. Tests for statistical significance have been applied comprising t-tests and Chi-square tests. The double difference captures the relative impact of HDI in the context of general development trends in the rural areas. Furthermore, the double difference approach is also applied to analysis of the households that are members of Self Reliance Group (SRGs) versus households that are not members, but live in villages supported by HDI (Non SRG households). In addition to SRGs, the HDI has initiated Livelihood Groups (LGs), Community Based Organizations (CBOs), and User Groups (UGs). Some households in the sample are members of these groups, and some are even members of both SRGs and the LG/CBO/UGs. The within HDI difference in impacts on these groups is also analysed. HDI Villages and Households Non HDI (Control) Villages and Households Difference HDI Non-HDI 2008/ Difference % of totals % of totals pp % of totals % of totals pp pp pp Double Difference - pp It should be noted that the Village Profile analysis does not apply the double difference approach since the corresponding 2008 Village Profile data proved incompatible as these were not cleaned and corrected thoroughly. Thus, the analysis was done on the 2012 Village Profile cross-section data only. Further, only 186 villages were included in both the 2008/09 and the 2012 surveys -Kachin state was not accessible for the 2012 survey and the 2009 impact survey data from Ayeyarwady was included. Another change is that 10 villages that were Non - HDI in 2008 were incorporated into HDI subsequently and considered as HDI villages for the present assessment. 3.1 Statistical tests of significant differences In statistics "significant" means probably true (not due to chance). However, it should be noted that a finding might be probably true without being important.

24 10 For the analysis of the village profiles statistical significance has been set as per custom practice at an alpha level 0,05, which means there is a 5% chance that the finding is not true, and 95% chance it is true. The report on village profiles therefore only mentions whether the differences are significant or not, i.e., at an alpha of 5%. For the analysis of the household survey data statistical significance has been calculated at alpha 10%, 5% and 1%. However, the statistical analysis presupposes that the sample is random, and this study has a stratified sample where the surveyed number of villages and households per village with different types of membership has been fixed and thus does not represent the actual population proportionally. Therefore weights have been applied to the village and household surveys to make up for the selection of strata by randomizing them, i.e., make them representative of a larger population based on the probability of them being selected if a truly random selection had been done. Analysis of categorical data or nominal responses The categorical data variables of the village profiles have been analysed through contingency analysis using the software JMP from SAS. This section describes the way JMP applies contingency analysis 1. When the response column (column assigned the Y role) has a nominal modeling type, the probabilities are fitted so that the response is one of r different response levels given by the data values. The basic model is that for each observation, Probability (Y = jth response level) = some function of the Xs and parameters The probability estimates must all be positive. For a given configuration of X s, the probability estimates must sum to 1 over the response levels. The function used to predict probabilities is a composition of a linear model and a multi-response logistic function. This is sometimes called a log-linear model because the logs of ratios of probabilities are linear models. For example, if Y can be either 1 or 2, the log-linear model is expressed as Log = function of the Xs and parameters The fitting principle is called maximum likelihood. It estimates the parameters such that the joint probability for all the responses given by the data is the greatest obtainable by the model. Rather than reporting the joint probability (likelihood) directly, it is more manageable to report the total of the negative logs of the likelihood. The uncertainty ( loglikelihood) is the sum of the negative logs of the probabilities attributed by the model to the responses that actually occurred in the sample data. For a sample of size n, it is often denoted as H and written If a probability of 1 is attributed to each event that did occur, then the sum of the negative 1 Partly cited from: SAS Institute Inc JMP 8 Statistics and Graphics Guide, Second Edition. Cary, NC: SAS Institute Inc.

25 11 logs is zero for a perfect fit. The simplest model for a nominal response is a set of constant response probabilities fitted as the occurrence rates for each response level across the whole data table. In other words, the probability that y is response level j is estimated by dividing the total sample count n into the total of each response level nj, and is written All other models are compared to this base model. The base model serves the same role for a nominal response as the sample mean does for continuous models. The coefficient of determination R-square statistic measures the portion of the uncertainty accounted for by the model, which is The R-square (U) that has been calculated is analogous to the R-square in a regression. R- square (U) is computed as follows: -log likelihood for Model -log likelihood for Corrected Total The total negative log-likelihood is found by fitting fixed response rates across the total sample. The R-square (U) has been used as a simple indicator for how successful the fit of the independent variable and the dependent variable is in explaining the variation in the data. R- square is the square of the correlation between the response values and the predicted response values. Thus an R-square of say 0,5 translates into 50% of the variation is explained by the correlation. In the social sciences an R-square of 0,3 and above is often considered an indicator for a correlation, but even lower values can be considered worthy of investigation if Chi-square tests show significance. If one categorical variable is considered as Y and the X is regarded as fixed, the Chi-square statistics test that the distribution of the Y variable is the same across each X level. In other words, Prob > Chi-square lists the probability of obtaining, by chance alone a Chi-square value greater than the one computed if no relationship exists between the response and factor. Even in cases where the Prob> Chi-square has a significant low value below 0,05 the R-square value from the statistical analysis of the categorical data can be low. Analysis of household survey data and double difference In the household survey, there are three sets of hypothesis for the statistical tests of differences between HDI households and Non HDI households. 1) Hypothesis for the difference of two time points within HDI villages (1 difference) H 0, HDI villages : H 1, HDI villages : There is no difference between the estimates of respective indicator between 2012 and 2008/09 (Indicator 2012(HDI) Indicator 2008/09(HDI) = 0) There is a difference between the estimates of respective indicator between 2012 and 2008/09 (Indicator(HDI) 2012 Indicator(HDI) 2008/09 0)

26 12 2) Hypothesis for the difference of two time points within Non-HDI villages (1 difference) H 0, Non-HDI villages : There is no difference between the estimates of respective indicator between 2012 and 2008/09 (Indicator(Non-HDI) 2012 Indicator(Non-HDI) 2008/09 = 0) H 1, Non-HDI villages : There is a difference between the estimates of respective indicator between 2012 and 2008/09 (Indicator(Non-HDI) 2012 Indicator(Non-HDI) 2008/09 0) 3) Hypothesis for the double difference the difference of the difference of two time points within HDI villages vs the difference of the difference of two time points within Non-HDI villages H 0, Difference : There is no difference of the difference of two time points (2012 vs 2008/09) between HDI villages and Non-HDI villages of the estimates of respective indicator (Difference HDI village Difference Non-HDI village = 0) H 1, Difference : There is a difference of the difference of two time points (2012 vs 2008/09) between HDI villages and Non-HDI villages of the estimates of respective indicator (Difference HDI village Difference Non-HDI village 0) The normal probability distribution is used for testing hypothesis mentioned above. The test statistics is, where SE is notation for the standard error. The mean difference is specified in the brackets of the respective hypothesis. The standard error (SE) of the respective mean difference must be calculated. There is about 90% overlap of the sample across the surveys of 2008/09 and 2012, i.e., the panel data. Sample households from HDI villages and sample households from Non-HDI villages do not overlap and each can be considered a distinct sample. According to Kish (1965), we need to apply the following two formulae to calculate SE of the mean difference from the survey data. For the SE of difference between the estimates of respective indicator between 2012 and 2008/09 (about 90% overlap sample), the following is used where R is 0.2 for HDI villages and 0.3 for Non-HDI villages based on the correlation estimates of the consumption expenditure from the 4166 panel households survey data. For the SE of difference of the difference of two time points (2012 vs 2008/09) between HDI villages (there is no overlap in the sample) the following has been used, The detailed results of the statistical tests are provided in the Annex 1.

27 13 The notations of ***/**/* means the difference is significant at 1%/5%/10% level respectively. No such notation means there is no significant difference even at 10% level. Please note that these statistical tests are not conducted for the population subgroups within HDI villages, i.e., households with different types of membership. Therefore, the lack of indication of significance for those statistics does not mean that differences could not be significant. It is worth to recall that the surveys were conducted at different times in the year, which could lead higher variation in seasonal related estimates. This is probably the reason why the correlation estimates for the consumption expenditure between the two surveys were lower than what could be expected. However, the use of the double difference method cancels out those variations. Calculation of the poverty line There are two basic methods for establishing the poverty line: 1) the Food-energy intake (FEI) method and 2) Cost-of-basic-need (CBN) method. The present impact assessment applies the CBN method, which is also the method used in the nation wide IHLCA surveys. There are two components in setting the poverty line using the CBN method: 1) A food component - the poverty line is almost universally anchored to nutritional requirements for good health, 2) A non-food component - an estimate of the budget share that goes to food. Establishing poverty lines is complex and done through a series of steps, as follows: 1) Constructing consumption aggregates and adjusting the price differentials with Paasche Price Index (PPI) 2 2) Defining reference household as defined in Impact Assessment 2008 (25% - 75% of consumption expenditure distribution). 3) Estimate the minimum calorie required for the reference group. 4) Define the food consumption basket and calculate the food poverty line (FPL). 5) Calculate budget share around +/- 10% of FPL to food consumption expenditure. 6) Add the monetary value of non-food portion to FPL to arrive at poverty line (PL). 2 The Passche index is a ratio that compares the total purchase cost of a specified bundle of current-period commodities (commodities valued at current prices) with the value of those same commodities at base-period prices; this ratio is multiplied by 100. The Paasche price index tends to understate price increases, since it already reflects some of the changes in consumption patterns that occur when consumers respond to price increases i.e., increased consumption of goods will indicate reduced relative prices. (Encycl. Britannica)

28 14 4 Findings from Village Profiles Key informants in the villages provided the data for the Village Profiles during the household survey. Thus, the data must be considered soft with a margin for errors in terms of preciseness of quantitative information. However, the data are valid as providing a general picture of the status of relevant variables. The sample villages by State/Division included in the 2012 assessment are listed below. Table 5 Sample villages 2012 TYPE OF VILLAGE State/Division HDI Non HDI Total Chin Magway Mandalay Rakhine Shan (South) Shan (North) Ayeraywaddy Total As described in the previous section sample weights were incorporated into the estimations. With weights the extrapolated number of villages are as follows: Nos. villages in Sample Extrapolated nos. villages HDI villages Non HDI villages The focus of the analysis is on impacts detected through differences between HDI supported villages and Non HDI villages ( with-without ). Types of support programmes by the supporting organization The HDI villages have received support from UNDP but some Non HDI villages have also received external support from organizations. Figure 2 gives an overview of which types of programmes the various supporting organizations have provided to the two types of villages.

29 UNDP (ICDP/ CDRT) Other UN agencies INGOs Local NGOs Community as a whole Monastery/ Religious org. Govt. Political parties/ Politicians Main programme 15 Figure 2 Overview of main support programmes by supporting organization and type of village Main programme by Supporting organization Sized by % of Type of programmes in HDI and Non HDI villages Appointing additional health professional/ trained health personal Appointing additional teacher Appointment of doctor/ nurses and health practitioner Type of village HDI Non HDI Building hospital/ clinics and related development of physical infrastructure Providing other health related training Building monestry/ rest house and related public places Building school and related physical development Seeking knowledge/info on agriculture related issues Seeking knowledge/info on livestock related issues Seeking knowledge/info on other income generating related issues Credit Employment and labor migration Land use Other environmental issues Water resources Drinking water Sanitation Electricity Roads and transportation Rice Religious contribution Supporting organization Circle Size

30 Village characteristics The differences between HDI and Non HDI villages have been analyzed for a number of variables presented in this section. In the case of many variables there was no significant difference found. For variables that are not expected to be impacted by HDI, a lack of impact confirms the comparability of the HDI and the Non HDI villages, or in other words, the propensity of the treatment group, i.e., the HDI villages, with the control group, i.e., the Non HDI villages. Some differences are statistically significant but may not be attributable to HDI. With regard to topographical variation, the HDI covers a large area of Myanmar and both the HDI and Non HDI sample villages reflect this variation. Half of both types of villages are in the plains, 28% in hills/mountains, 15-18% in the Delta, and the remaining in valleys and on plateaus (table 1 in annex). Demographic characteristics Non-HDI villages are smaller than HDI villages with a median of 76 households compared to 122 in the HDI villages (table 2 in annex). However, the variation in village size is quite large between States/Divisions with a median of 70 households in Chin and a maximum of 190 households in the Delta (table 3 in annex). The sex ratio in the sample villages in total is skewed towards females. Figure 3 Sex ratio in villages A closer look at the six outliers show that they are all HDI villages, with two villages in Chin, two in Shan (South), one in Shan (North) and one in Magway. In these villages there are about double as many women as men. This indicates high level of out migration in some areas. Even without these outliers the mean sex ratio of females-to-men is 107. The changes in total village population from 2008/09 to 2012 cannot be ascertained, as the Village Profile data from 2008/09 were not cleaned sufficiently to allow for comparison. General living standards and conditions Key Informants were asked to assess the general living standard in their village and there is a significant difference. HDI villages are assessed as generally better off than Non HDI villages with 27 pp more of the Non HDI villages assessed as poor or very poor.

31 17 Table 6 Assessed living standards by HDI and Non HDI village % of villages Well-to-do Average Poor Very poor HDI village 8,27 47,68 35,19 8,86 Non HDI village 0 28,53 50,09 21,37 Difference pp 8,27 19,15-14,9-12,51 An analysis of the data on cost of housing and house rents shows no significant difference between HDI and Non HDI villages. There is however quite a variation between State/Divisions with Shan South and Rakhine as the most expensive places in terms of housing. With regard to the housing conditions in general it has improved in 9 pp more HDI villages compared to Non HDI villages, and has worsened in 2 pp less. The difference is significant but may not be attributable to HDI (table 7). Table 7 Housing condition over the last 3 years % of villages Improved Remained the same Worsened HDI village Non HDI village Difference pp The minimum average daily wage rate is not significantly different between HDI and Non HDI villages whereas it is significantly different between State/Divisions both for men and women. Shan, Chin and Mandalay have higher daily wage rates for both men, median of Kyat 2000, and women, median of Kyat 1500, than the other state/divisions. The lowest rates for women are in Rakhine with a median of Kyat 1000, and in the Delta for men with a median of Kyat 1500 (table 4, figure 1 in annex). Health Figure 4 and 5 shows the results of analysis of the information provided by the key informants regarding the most important diseases for adults and children. The graphs show HDI and Non HDI villages that have reported various diseases as the most important, with the density shadows covering the diseases in the areas from which the highest frequencies were reported. There is a significant relationship both between location of village in State/Division and type of village, and morbidity among adults, e.g. from malaria, jaundice, and diarrhea/ dysentery, with morbidity reportedly higher in Non HDI villages. On the other hand, paralytic stroke has a higher reported frequency in HDI villages (table 5 in annex, figure 1), a finding that requires further exploration. The differences between State/Divisions are shown in the Contour Plot graph. Intervention targeting would aim hygiene promotion at Shan South and Chin (for seasonal illness) and Rakhine (for dysentery), and nutritional education at Rakhine, the Delta and Chin (for high blood pressure). It is noteworthy that TB appears to be a serious problem in Rakhine, and malaria in Shan, Rakhine and Chin. With respect to children s diseases there is no significant difference between HDI and Non HDI villages. The difference between State/Divisions is however seen in Figure 5. This would point to possible interventions such as hand washing campaigns, sanitation and health and hygiene promotion in Rakhine and Chin, South Shan and the Delta, and mosquito nets also in Shan (North).

32 18 Figure 4 Main diseases of adults by State/Division The most serious disease of adults the last 12 months vs. State/Division by Type of village Chin Rakhine Type of village HDI Non HDI HDI Non HDI Magway Mandalay Shan (North) Shan (South) Ayerwaddy Diarrhea/ Dysentery Typhoid Jaundice Gastritis TB Malaria HIV/ AIDS Seasonal illness Diabetes/high blood pressure Paralytic stroke Sore eyes Itch Asthma Figure Graph 5 Builder Main diseases of children by State/Division The most serious disease of children the last 12 months vs. State/Division by Type of village Chin Rakhine Type of village HDI Non HDI HDI Non HDI Magway Mandalay Shan (North) Shan (South) Ayerwaddy Dengue fever Malaria Polio Measles Seasonal illness Diarrhea/ Dysentery ARI Cough Where(65 rows excluded)

33 19 Access to markets and services Over the last 3 years the ratio of HDI villages where the road conditions has improved is 5 pp higher and where it has worsened 9 pp less than for the non-hdi villages - a significant difference that may or may not be attributed to HDI. Table 8 Road condition of villages (last 3 years) % of villages Improved Remained the same Worsened HDI village Non HDI village Difference HDI Non HDI pp With regard to access to town the difference between HDI and non-hdi villages were not significant. The mean distance is 15 and 16,5 miles for HDI and Non HDI villages respectively, and distance to nearest main road was around 2 miles for both. Slightly more non-hdi villages have town access through earthen roads (table 7 in annex). In terms of transport the difference is also insignificant. Public transport is available in 62% of HDI villages and 57% of Non HDI villages (table 9 in annex). With regard to access to markets there are significantly more HDI villages which have a small market, but still it is only 8,5% of all HDI villages. There is no significant difference with regard to the number of retail shops. Access to schools In this respect there are slight differences with the ratio of HDI villages having a primary school within the village 11 pp higher, and the ratio having a school for 5-8 graders and 9-10 graders within the village 7 pp higher than Non HDI villages. On average there are slightly more students enrolled in schools in HDI villages (table 11 in annex). Access to Health Facilities The ratio of HDI villages that have a health facility within the village is 20 pp higher than for HDI villages, and 9 pp higher with regard to having a drug store or drug selling shop. The ratio of Non HDI villages that have a Station Hospital as the main type of health facility is 13 pp higher than for Non HDI villages (Table 12 in annex. Table 13 shows the average distance to health facilities). Access to telephones Non-HDI villages have poorer access to telephones (17 %) compared to HDI villages - 11% more HDI villages can connect on mobile telephones and 6% more have a landline connection (Table 14 in annex) Economic activities With regard to economic activities of males, no significant difference was seen between HDI and Non HDI villages. In around 56% of both types of villages, the main occupation is agriculture, and in around 17% it is casual work. In 5-6% of both types of villages the main occupation is fisheries (table 15 in annex). There is a significant difference in women s main occupation agriculture/cultivation is more prevalent in HDI villages (14% more) whereas casual work is more popular in non-hdi villages (7% more) see table 16 in annex.

34 20 There is also a significant difference in the trend in job opportunities. In 13% more HDI villages it has improved in the last 3 years, but it has also worsened in 6% more HDI villages (Figure 6, table 17 in annex). Figure 6 Changes in job opportunities last 3 years Perceived trends in quality of life Over the last 3 years there is a significant trend in that 21% more HDI villages reported an improvement in the quality of life. Sixty-five per cent of the HDI villages found that quality of life has improved. This is of course a soft indicator by itself, but still a very positive sign. Table 9 Improvement in quality of life over the last 3 years. Row % Improved Remained the same Worsen HDI village 64,65 22,71 12,64 Non HDI village 43,96 43,33 12,71 Difference HDI - Non HDI 20,69-20,62-0,07 The reason for the changes both positive and negative - in both types of villages - are related to agriculture, but in Non HDI villages employment conditions played a bigger role in determining the overall quality of life (Table 10, Refer to Tables 18 and 19 in annex for details). UNDP support is the biggest reason in HDI villages for improvements in quality of life according to key informants.

35 21 Table 10 Main reasons for change in the quality of life over the last 3 years HDI and of Non HDI villages Improved Remained the same Worsened Row % HDI villages UNDP support 23,16 0, ,61 Agriculture 17,3 16,9 8,3 42,55 Improved knowledge 7, ,92 Transportation 7, ,37 Employment 2,5 2,9 1,4 6,76 All HDI all reasons 64,51 22,8 12, Non HDI villages Agriculture 18,4 23,85 2,96 45,2 Improved knowledge 9, ,16 Employment 4,9 4, ,67 Transportation 2,57 0 2,05 4,62 All Non HDI all reasons 39,76 46,59 13, Note: Only main reasons included therefore cells do not add up to 100%. Seasonal well-being Key informants were asked whether each of the months in the last year had been better, the same or worse than the previous. Figure 8 shows the seasonal well being as graphs for HDI and Non HDI villages respectively for each of the State/Divisions. There is a general agreement between the two types of village, but in Mandalay and Shan (South) the types of villages appears to be out of sync in this respect. However, in those State/Divisions the sample of villages was very small which may account for this. In general well-being is at a peak during the post harvest time by the end of the calendar year, and the most difficult time is during the monsoon season. Interestingly there are deeper and higher movements and variations, e.g. in Chin, Magway (Dry Zone) and Shan (North), which is probably due to cropping patterns in those areas.

36 Figure 7 Seasonal well being by type of village and State/Division Agriculture Access to agricultural markets Access to markets for selling agricultural produce is not very different between HDI and Non HDI villages. For 60% in HDI and 50% in Non HDI villages the market is in the nearest town while in 22% to 26% this is in another village in the township, and in 12% of both types of villages the market is within the same village (Table 10 in annex). Also with respect to the key buyers of agricultural produce the two types of villages are similar. In around 60% of both types of villages products are sold to a trader, broker or trading centre and only in 40 % of the villages are products sold directly at local markets or regional markets.

37 23 Figure 8 Key buyers of agricultural produce Figure 9 Difference in trends in agribusiness and sales of agricultural products

38 24 However, there is a significant difference in the trends in agribusiness and sales of agricultural products in the last 3 years between HDI and Non HDI villages, with 27% more HDI villages responding agribusiness has improved and 18% more HDI villages responding that sales of agricultural products has improved (figure 8). The reason for this difference is not clear, but could be interventions by HDI. The main problems that HDI and Non HDI villages face with regard to selling of agricultural products are almost the same and mainly relate to access and low prices (table. 11). Fortytwo per cent of both HDI and Non HDI villages lists problems with access to markets as the main problem. Thirty per cent of HDI villages mention low prices as a main problem and 24% of the Non HDI villages do the same. Table 11 Main problems related to selling of agricultural products Type of problem % of HDI village % of Non HDI village Difference pp Access: sell at collector because of difficult transportation 26,78 22,11 4,67 Access: sent to trader because of no market access 6,24 7,01-0,77 Access: transport expensive 5,45 9,22-3,77 Access: could not rent car and so cannot get high price 3,94 3,86 0,08 Access problems total 42,41 42,2 0,21 Low price: Could not get current price because of advanced money 16,3 14,62 1,68 Low price: Can't know the current price 8,01 4,96 3,05 Low price: 5,01 4,63 0,38 Low price: Could not get high price and don t take from sugar industry 1,02 0 1,02 Low price problems total 30,34 24,21 6,13 Could not use inputs because of advanced money 1,61 0 1,61 Had to pay border security agent 4,25 9,25-5 Cannot store products long time 2,02 2,61-0,59 Low quality products due to natural disaster 2,68 1,25 1,43 No customer 0,86 0 0,86 Nothing 12,55 12,05 0,5 Stole weight 0,94 8,45-7,51 Testimonial from government 1,99 0 1,99 Other problems total 26,9 33,61-6,71 Agricultural calendar Village key informants from the 220 villages were asked about the agricultural activities during the calendar year. Based on this information it was been possible to make an agricultural calendar for the different State/Divisions. Since the calendars from each village within states differs somewhat the graph is based on the quintiles of responses, thus showing the most common calendar for each State/Division. The graph shows the frequency

39 25 of agricultural activities as the dark areas, with the lines extending showing the range of points in time of the activities as reported from the villages. The white line in the middle of each bar is the mean point in time of all the responses. The agricultural calendar provides basic information which can be used as a general tool for understanding fluctuations in socio-economic well being as well as planning of timing of various technical assistance interventions. The agricultural calendar can be compared to the analysis of poverty dynamics and causality presented in volume 3 of the HDI Impact Assessment.

40 Figure 10 Agricultural calendars by State/Division. 26

41 27 Access to Technical assistance to farmers In 71% of the Non HDI villages there is no access to technical assistance for farmers according to the key informants and only 20% received technical assistance from INGOs. In 81% of the HDI villages, technical assistance was provided by UNDP. Technical assistance from the government is minimal and only around 6% of both types of villages has received this. This shows the great need and importance of making technical assistance available. Table 12 Providers of technical assistance Technical Assistance provider % of HDI village % of Non HDI village Difference HDI - Non HDI - pp Not applicable 7,5 70,8-63,2 INGO 3,3 19,5-16,3 Unknown 0,0 3,5-3,5 Government 5,4 6,2-0,8 Agricultural company 0,4 0,0 0,4 Other UN org 0,9 0,0 0,9 Agriculturist 1,5 0,0 1,5 UNDP (HDI) 80,9 0,0 80,9 Sorted ascending by % difference Loans and credits for agricultural purposes With regard to differences in problems with credit for agricultural purposes the differences between HDI and Non HDI villages are however significant. Figure 11 Main problem with credit related to agribusiness

42 28 Figure 11 shows the difference in percentages of HDI and Non HDI villages that report different types of main problems. Eight per cent more Non HDI villages rate the lack of credit facility as a main problem and 7% more lack of security deposits. On the other hand 10% more HDI villages mentions that loans cannot be given without guarantee as the main problem. This indicates that Non HDI villages have less access to credit, whereas in HDI villages with access to a credit facility, the conditions for the credit such as short payback period, timing and availability mean that not all farmers can benefit from it (Table 20 in annex). When we look at the three main providers of credits and loans to agricultural producers in the villages (table 13) it emerges that UNDP/HDI is main provider in 28% of the HDI villages and SRG/LG/CBO the main provider in 15% of the HDI villages. Further, buyers, i.e., middlemen and traders and moneylenders are the main providers in 17% and 14% respectively in HDI villages. In 19% of the Non HDI villages the main credit providers are buyers, i.e., middlemen and traders, and in 17% it is moneylenders. In between 10% to 14% of both types of villages the main credit providers are agricultural/development banks. Notably in 19% of the Non HDI villages there is no credit provider. Thus the main difference between HDI and Non HDI villages is the credit from HDI and SRGs/LG/CBOs and the lack of credit providers in Non HDI villages. Buyers and moneylenders play a major role for credit in both types of villages, with agricultural development banks in third position. Table 13 Three main credit/loan providers to agricultural producers Column % Main Second Third Credit/Loan provider HDI Non Differ HDI Non Differ HDI Non Differ HDI ence HDI ence HDI ence Relatives or friends 2,23 3,85-1,6 8,89 10,66-1,8 8,89 10,66-1,8 Land lord/ employer 6,76 5,43 1,3 6,19 2,08 4,1 6,19 2,08 4,1 Money lender 14,39 16,65-2,3 12,13 10,94 1,2 12,13 10,94 1,2 Pond shop 1,75 1,26 0,5 0,36 3,51-3,2 0,36 3,51-3,2 Buyers (middleman/ traders/ 17,44 18,64-1,2 3,58 1,64 1,9 3,58 1,64 1,9 exporters) Private company 0,3 5,27-5,0 2,68 1,44 1,2 2,68 1,44 1,2 National banks 1,73 4,09-2,4 2,83 5,35-2,5 2,83 5,35-2,5 Agricultural/ development 10,02 13,9-3,9 12,99 15,93-2,9 12,99 15,93-2,9 banks SRG/ LG/ CBO 15,27 2,31 13,0 16, ,2 16, ,2 Credit from UNDP HDI project 27,59 4,5 23,1 2,4 0 2,4 2,4 0 2,4 Credit unions or cooperatives 1,11 5,35-4,2 0,81 0 0,8 0,81 0 0,8 Private bank 0,78 0 0,8 0,78 0 0,8 USDP/ WAA 0,65 0 0,7 0,65 0 0,7 Not applicable 1,42 18,76-17,3 29,55 48,44-18,9 29,55 48,44-18,9 Total % ,99 0, ,99 0,02 Percentage of agribusiness with credits and loans Key informants were asked what percentages of agricultural businesses in their village use loans and credits to support their activities. The difference between HDI and Non HDI

43 29 villages in this respect is insignificant, but the difference between the State/Division is however significant (Figure 12, Table 21 in annex for statistical details). Figure 12 Per cent of agribusinesses running on credit and loans by State/Division Note: blue line: median of HDI villages, red line: median of Non HDI villages Figure 12 shows the quintiles and medians of the distribution of sample villages with regard to the percentage of agribusinesses that use loans and credits. Chin, Mandalay and Rakhine have the lowest medians, but especially Rakhine also a large variation between villages. Shan (South) has the highest percentages and apart from two outliers with zero percent, between 70% and 100% of business in all villages use credit and loans to support the agribusinesses. It may be noted that this type of information can provide inputs to future programming and enable development interventions to be better targeted, by State/Division Summary and conclusions on HDI impact assessed on the basis of Village Profiles This section presents the findings and analysis of a number of indicator variables for detecting HDI impact, but also for assessing the degree of comparability of HDI and Non HDI villages. The significant differences between HDI and Non HDI villages in favour of a positive impact by HDI are listed below:

44 30 Perceived quality of life has improved in 24% more HDI villages compared to Non HDI villages Job opportunities improved in 13% more HDI villages, but also worsened in 6% more Trends in agribusiness are more diverse in HDI villages Other minor differences (weak correlations) Condition of road leading to village has improved in slightly more HDI villages Access to small market in village in more HDI villages, but still in only 8,5% of all HDI villages Access to health facility: 13% more Non HDI village use Station Hospitals, 7% more HDI villages use Township Hospitals Economic activities of women: 7-10% more Non HDI villages with casual work as main activity Indicators where no significant difference between HDI and Non HDI villages were found are indicated below: Cost of housing Minimum average daily wages for men and women Most serious children s diseases Access to nearest town Nos. of teachers in schools Nos. of retail shops Economic activities of men The data from the Village Profiles can be useful for future programming purposes. Further analysis could be done to screen for more significant relationships between the variables in order to tweak out causal links. Not least, the data could be combined with data on inputs and duration of support to provide a basis for an assessment of efficiency and effectiveness of the various support interventions.

45 Household Survey The first sub-section of this section reports findings from a soft (qualitative) indicator: the perceived impact of UNDP support on supported households. Thereafter, the sub-sections report on findings and analysis on the various indicators and variables in the quantitative data. The tests for statistical significance of the double differences are found in the tables in annex Perceived impact of HDI support on households in 2012 The 2012 household questionnaire included a question of how the household respondent perceived the impact of UNDP HDI overall activities on the household. Of the 2499 households in the sample that have received support from UNDP 48% responded that the support has noticeable improved their conditions. Another 48% responded the support has slightly improved their conditions, while 4% responded it had had no effect. Less than 10 households expressed HDI has had a negative impact. Table 14 Perceived impact of HDI by supported households support in 2012 Membership Noticeably Slightly No effect Nos improved improved SRG 50% 47% 3% Both SRG and LG/CBO/UG 63% 36% 1% 739 Only LG/CBO/UG 40% 55% 5% 304 The highest proportion of supported households that finds that their condition has noticeably improved are in Chin with 79% and in Shan (North) with 46%, whereas the lowest proportion is found in Rakhine with 29% (table 14). In Mandalay, Magway, Shan (South) and the Delta, around 35% of the households noted a noticeable improvement, and in these areas around 56% reported slight improvements. No effect was reported by around 8% of the households in Magway, Shan (South), the Delta and in Rakhine. Table 15 Perceived impact of HDI by supported households by State/Division State/Division Noticeably improved Row % Slightly improved Row % No effect Row % Chin 78,6 19 2,4 Shan (North) 46,3 49,8 3,9 Mandalay 37,5 58,7 3,8 Magway 36,2 54,9 8,9 Shan (South) 35,6 56,4 8,1 Ayeyarwady 32,3 59,9 7,8 Rakhine 28,5 63 8,5 Sorted descending by responses in State/Divisions Since the 2008/09 surveys the overall perceived impact has decreased. On a five-point scale from noticeably negatively affected with a value of 1, to noticeably improved with a value of 5, the change looks as shown in table 16.

46 32 Table 16 Scaled change in perception on the HDI project assistance (2008/ 2009 & 2012) All HDI Villages 4,7 4,3-0,4 SRG 4,8 4,5-0,3 supported Both SRG and LG/CBO/UG 4,8 4,7-0,2 HHs Only LG/CBO/UG 4,7 4,3-0,4 estimates No membership 4,6 4,1-0,4 Thus perceived impact has fallen by 0,4 points, which means that more households now consider that their condition has slightly improved rather than noticeably improved (table 23 in annex). There is a significant relationship between duration of membership in a SRG and the perceived (positive) impact in 2012 as shown in figure 13. However, the R-square is weak at 0,01. Figure 13 Perceived impact of HDI by households by year of SRG establishment Note: blue lines are probability curves, If the response is completely predicted by the value of the factor, then the logistic curves are effectively vertical. Red lines are rate curves connecting each year. In terms of which support interventions that have had most impact, credit is by far the most important (Table 17 - does not show types of support with less than 1% of responses). Knowledge of income generation, drinking water and sanitation and building of schools follows on the list. These include school meals, knowledge on agricultural issues, other environmental issues, knowledge on other health related issues, appointment of health professionals, health related infrastructure, other health related training, etc. The responses

47 33 indicate which support interventions are the most effective and relevant. Table 17 Perceived impact of UNDP support by types of support (% of all responses) Type of support Noticeably Slightly improved improved No effect Credit 15,1 20,7 0,5 Seeking knowledge/info on other income generating related issues 4,3 3,6 0,1 Drinking water 4,1 6,5 0,7 Sanitation 3,2 7,8 0,7 Building school and related physical development 2,9 5,3 0,4 Roads and transportation 1,1 1,2 0,1 Mosquito net 1,1 1,0 0,0 Appointing additional teacher 1,0 2,4 0,2 Employment and labor migration 0,7 0,5 0,0 Seeking knowledge/info on livestock related issues 0,6 0,5 0,0 Water resources 0,5 1,2 0, Demographic Characteristics Comparison of HDI supported households and Non HDI households with regard to demographics shows no statistically significant differences. This is an indicator of the degree of comparability between the treatment group and the control group. Table 18 Average household size (2008/ 2009 & 2012) HDI Villages 5,0 5,1 0,1 SRG 5,0 4,9-0,1 supported Both SRG and LG/CBO/UG 6,1 6,3 0,2 HHs Only LG/CBO/UG 6,2 5,9-0,3 estimates No membership 4,8 4,8 0,0 Non HDI Villages 4,9 5,0 0,1 Difference (HDI - Non HDI) 0,1 0,1 0,0 The age dependency ratio is calculated as the number persons in a household below 15 years of age and above 59 years divided by the number of person between 15 and 59 years of age, i.e., persons considered in the productive age group. Thus a family with 2 children, one elderly person and four adults below 59 years of age will have a dependency ratio of 0,75. A lower dependency ratio indicates a better chance of having a higher material living standard. Sex ratio There are more females than males in both HDI and Non HDI households and the sex ratio has even decreased since 2008/09. The small difference of 2 points between HDI and Non HDI households in 2008/09 has shrunk to 1 point in 2012 so that the sex ratio is now 95 in HDI villages and 94 in Non HDI villages (refer table 29 in annex). The aggregate DD is not

48 34 statistically significant, however in Chin the DD is -10,4 pp, significant at alpha 10%, indicating outmigration of males in HDI households. Table 19 Sex ratio (males to females) HDI Villages 97,7 95,0-2,7 SRG 97,2 96,2-1,0 supported Both SRG and LG/CBO/UG 93,1 84,7-8,4 HHs Only LG/CBO/UG 84,7 92,6 7,9 estimates No membership 100,1 97,5-2,6 Non HDI Villages 95,7 94,0-1,7 Difference (HDI - Non HDI) 2,0 1,0-0,9 Dependency ratio 3 Neither the aggregate DD nor DDs for geographical areas are statistically significant (table 30 in annex). Both in 2008/09 and in 2012 the Non HDI households had a lower dependency ratio than HDI supported households, but the gap has closed by 0,03 in This could speculatively indicate that Non HDI households either use family planning slightly more, or on average have experienced a comparative decrease in elderly people. It could also mean that HDI households have more children, or household members are becoming comparatively older. Table 20 Age dependency ratio (2008/ 2009 & 2012) HDI Villages 0,70 0,73 0,02 SRG 0,70 0,68-0,02 supported Both SRG and LG/CBO/UG 0,77 0,80 0,02 HHs Only LG/CBO/UG 0,54 0,81 0,26 estimates No membership 0,70 0,71 0,01 Non HDI Villages 0,76 0,75-0,01 Difference (HDI - Non HDI) -0,06-0,03 0,03 Within the HDI supported households there are differences according to membership in groups. In 2012 SRG member households have the lowest dependency ratio of all, which could indicate more use of family planning. Households that are only members of LG/CBO/UG are households in Rakhine and Chin and they have much higher increase. However this sample is very small and no great importance can be given to this. A differentiation of age dependency ratio by household head shows an equal increase in both female-headed HDI households and male headed HDI households. 3 The age dependency ratio provides information on the number of dependents (i.e. children aged less than 15 and people aged 60 year-olds and above), compared to the number of persons aged 15 to 59 years.

49 35 Table 21 Age dependency ratio by gender of household head (2008/ 2009 & 2012) Male headed Female headed Male headed Female headed HDI Villages 0,73 0,56 0,75 0,58 SRG 0,72 0,54 0,71 0,49 supported Both SRG and LG/CBO/UG 0,78 0,70 0,82 0,69 HHs Only LG/CBO/UG 0,54 0,72 0,87 0,57 estimates No membership 0,73 0,52 0,73 0,63 Non HDI Villages 0,79 0,62 0,77 0,62 Proportion of female-headed households In 2008/09 there were 1,3 pp more female headed households in the Non HDI areas but in 2012 this has reversed so that there are now 2,5 pp more female headed households in the HDI areas yielding a double difference of 3,8 pp. The aggregate DD is not statistically significant, but for Northern Rakhine the DD is at alpha 1% (table 28 in annex). As mentioned above households that are only members of LG/CBO/UGs are in Chin and Rakhine and constitute a very small sample. Table 22 Proportion of female headed households (2008/ 2009 & 2012) (% of households) Diff ( / 2009) 2008/ HDI Villages 14,5 17,3 2,9 supported HHs estimates SRG 13,9 16,6 2,7 Both SRG and LG/CBO/UG 17,6 21,3 3,6 Only LG/CBO/UG 6,2 21,9 15,7 No membership 14,4 15,8 1,4 Non HDI Villages 15,8 14,8-1,0 Difference (HDI - Non HDI) -1,3 2,5 3,8 Table 23 Average household size by gender of household head (2008/ 2009 & 2012) Male headed HHD Female headed HHD Male headed HHD Female headed HHD HDI Villages 5,2 3,9 5,3 4,0 SRG 5,1 4,1 5,1 3,7 Both SRG and LG/CBO/UG 6,4 4,5 6,8 4,5 supported Only LG/CBO/UG 6,3 6,2 4,8 HHs estimates No membership 4,9 3,7 5,0 3,8 Non HDI Villages 5,1 4,0 5,1 3,9

50 36 Economic dependency ratio 4 Economic dependency ratio is calculated as the proportion of working to non-working household members so that a family with 3 working members and 4 non-working members would have an economic dependency ratio of 0,75. The lower the ratio the more persons to support for each working member. The DDs are not statistically significant and there has been no noteworthy change in the time between the surveys (table 31 in annex). Table 24 Economic dependency ratio Diff ( / 2009) HDI Villages 1,10 1,11 0,01 supported HHs estimates SRG 0,94 0,87-0,07 Both SRG and LG/CBO/UG 1,34 1,36 0,03 Only LG/CBO/UG 1,04 1,50 0,46 No membership 1,13 1,09-0,04 Non HDI Villages 1,15 1,17 0,02 Difference (HDI - Non HDI) -0,05-0,06-0,01 The economic dependency ratio by household head shows clearly that in female-headed households there are fewer breadwinners. However, the possibility of female-headed households receiving remittances from migrated men should be considered. The survey did not include data on remittances. Table 25 Economics dependency ratio by gender of household head (2008/ 2009 & 2012) Impact Assessment Male headed HHD Female headed HHD Male headed HHD Female headed HHD HDI Villages 1,13 0,88 1,15 0,90 SRG 0,98 0,69 0,93 0,56 Both SRG and LG/CBO/UG 1,37 1,16 1,42 1,11 supported HHs estimates Only LG/CBO/UG 1,00 2,13 1,55 1,29 No membership 1,17 0,89 1,12 0,92 Non HDI Villages 1,16 1,08 1,17 1, Housing, Fuel, Water and Sanitation The percentage of households that have acquired CI roofs has increased by about 8 pp in both types of villages. The aggregate DD is not statistically significant. Only in Northern Rakhine is the DD of -1,2 pp statistically significant (table 32 in annex). 4 The economic dependency ratio provides information on the number of economic dependents compared to the number of economically active persons in the household. It is measured by dividing the number of non-working members in the household by the number of working members in the household

51 37 With regard to the use of electricity the lead by 15 pp HDI households had in 2008 has shrunk to 8 pp in 2012, which indicates that the rate of electrification in HDI villages has been slower than in Non HDI villages, but with a higher coverage at the time of the baseline survey. The aggregate DD is not statistically significant, but for Eastern Rakhine the DD of 14,3 pp is significant at alpha 1%, indicating that more HDI households than Non HDI households have been provided with electricity during the three-year period (table 33 in annex). Table 26 Proportion of households using electricity for main lighting source (2008/ 2009 & 2012) (%) HDI Villages 24,6 23,2-1,4 SRG 26,3 29,2 2,9 Both SRG and LG/CBO/UG 25,6 35,9 10,3 supported HHs Only LG/CBO/UG 9,5 estimates No membership 23,7 22,2-1,5 Non HDI Villages 10,2 15,3 5,1 Difference (HDI - Non HDI) 14,5 7,9-6,5 Only about 1%-2% of all sample households, both HDI and Non HDI, use non-solid fuel for domestic use so the marginal differences between HDI and Non HDI and between the different types of membership is of no consequence. There are no significant DDs on this variable (table 34 in annex). The use of improved stoves shows a worrying decline in most types of households, except for members of both SRG and LG/CBO/UGs, which has seen a dramatic increase of 24%. The aggregate DD is not statistically significant, but in the Dry Zone and Shan the decrease among HDI households is statistically significant at alpha 1% and 5% with DDs of -46 pp and - 20 pp respectively. The reason for this could be sought perhaps in low quality or stoves with inappropriate technology for the areas in which they are to be used (table 35 in annex). Table 27 Proportion of households using improved stove for domestic cooking (2008/ 2009 & 2012) (%) HDI Villages 33,3 20,1-13,2 supported HHs estimates SRG 37,7 23,5-14,2 Both SRG and LG/CBO/UG 17,0 40,7 23,7 Only LG/CBO/UG 6,0 No membership 36,2 19,3-16,9 Non HDI Villages 17,9 13,7-4,2 Difference (HDI - Non HDI) 15,4 6,4-9,0 Neither the aggregate DD nor the DD for geographical areas for the percentage of households with access to improved drinking water sources are statistically significant. Coverage has increased overall in both HDI and Non HDI villages but only by 5 pp in all HDI villages compared to 12 pp in Non HDI villages. In the HDI villages access for SRG members has increased also by 12 pp and thus kept pace with the general trend (table 36 in annex).

52 38 Table 28 Proportion of households using improved drinking water source (2008/ 2009 & 2012) (%) HDI Villages 54,2 59,2 5,0 SRG 51,1 63,6 12,5 Both SRG and LG/CBO/UG 58,3 65,9 7,6 supported HHs Only LG/CBO/UG 36,8 estimates No membership 54,8 63,4 8,7 Non HDI Villages 45,3 57,3 12,0 Difference (HDI - Non HDI) 8,8 1,9-7,0 In 2008/09 there were 27% more HDI supported households that used fly proof latrines compared to Non HDI households, but in 2012 the difference has decreased to only 3% more HDI households. The aggregate DD of -24 pp is statistically significant at alpha 10%. In the Dry Zone and in Eastern Rakhine state the DDs are statistically significant at 1%, and in Northern Rakhine state at 5%. This indicates activities by other providers of fly proof latrines in those areas (table 37 in annex). Table 29 Proportion of households using fly proof latrine (2008/ 2009 & 2012) (%) HDI Villages 67,9 67,2-0,7 SRG 91,4 82,0-9,4 Both SRG and LG/CBO/UG 55,2 64,9 9,7 supported HHs Only LG/CBO/UG 30,5 estimates No membership 61,4 72,7 11,3 Non HDI Villages 40,7 64,3 23,6 Difference (HDI - Non HDI) 27,1 2,8-24, Monetary Poverty Indicators The following two indicators are UNDP Country Programme outcome Indicators for 2011: 1) Poverty Headcount Index: % of HDI households below the poverty line in 2011 compared to Non-HDI and to 2008 HDI and Non-HDI households; 2) Budget expenditure on food (%) by households for HDI and Non-HDI samples. Poverty lines Notably the poverty line has not changed much between 2008/09 and The 2012 poverty line is for all areas including the Delta, which had a lower poverty line in 2009 than the survey areas of the 2008 survey.

53 39 Table 30 Poverty lines Surveys HDI Impact Assessment 2008 ICDP Delta Survey 2009 HDI Impact Assessment 2012 Minimum calorie required per day Food poverty line (Kyat) Food budget share (%) Non food portion (Kyat) Poverty line (Kyat) 2, ,212 68% 104, ,977 2, ,330 67% 105, ,736 2, ,772 67% 108, ,417 Food poverty head count 5 The food poverty head count index is the proportion of the population that cannot afford enough food to fulfil the basic calorie intake norm of 2200 per day. Table 31 Food poverty headcount index (2008/ 2009 & 2012) (%) HDI Villages 10,2 8,4-1,8 SRG 2,5 4,0 1,4 Both SRG and LG/CBO/UG 20,9 20,6-0,3 supported HHs Only LG/CBO/UG 6,5 8,6 2,1 estimates No membership 11,3 8,3-3,0 Non HDI Villages 10,9 10,7-0,1 Difference (HDI - Non HDI) -0,7-2,3-1,7 Overall there are around 10% food poor households in both types of villages. In HDI there has been a slight improvement over Non HDI villages such that the double difference is -1,7 pp, however none of the DDs on this indicator are statistically significant (table 38 in annex). Poverty headcount 6 The poverty headcount index is the proportion of population below the poverty line. The double difference is -3.7 pp, which however is not statistically significant. Only in Shan with a DD of -15 pp, significant at alpha 5%, and in the Dry Zone with a DD of -11 pp, significant at alpha 10%, has HDI had a measurable impact on this indicator (table 39 in annex). The within group differences in HDI supported households are significant. Notably, and a cause of concern, is that SRG members have increased the poverty head count. This should be investigated further. 5 Food poverty headcount index is the proportion of individuals whose consumption expenditure is lower than the food poverty line. 6 Poverty headcount ratio 6 is the proportion of the population whose consumption expenditure is below the poverty line

54 40 Table 32 Poverty head count index 2008/ 2009 & 2012) (%) HDI Villages 40,4 38,8-1,6 SRG 26,2 32,8 6,7 Both SRG and LG/CBO/UG 66,5 50,9-15,6 supported HHs Only LG/CBO/UG 33,6 42,5 8,8 estimates No membership 40,7 38,3-2,4 Non HDI Villages 42,1 44,2 2,1 Difference (HDI - Non HDI) -1,8-5,4-3,7 The breakdown of the change in poverty by HDI household s membership in different types of organizations is shown in Figure 14. Figure 14 Changes in poverty head count by HDI household membership Households that are members of both SRG and LG/CBO/UG could reduce poverty by 16 pp during the 3-year period. On the other hand, an increase in poverty of 6.7 pp was found among the SRG only member households. Another increase in poverty was found among the households that are only LG/CBO/UG members. It should be noted that the numbers of households in these sub-groups differs significantly. The reduction of 16 pp in 3 years is found only in the CDRT area. But the double differences suggest that it was not only due to HDI. Similar high reductions are also found in no membership households in the same area of Chin, ERS and NRS as seen in Figure 15.

55 41 A further breakdown of poverty head count in HDI villages by membership and geographical region shows interesting differences. In Figure 15 the green triangles show areas where poverty head count has gone down, the yellow triangles where it has gone up. Figure 15 Poverty head count in HDI villages by membership and geographical area Poverty head count in HDI villages by membership Dry Zone -1.4% +14% +15% SRG No membership SRG Shan +11% +3% +8% No membership SRG Delta -12% -3% +10% No membership Chin +12% -9% -20% SRG and LG/CBO/UG No membership ERS -5% -36% -32% SRG and LG/CBO/UG No membership SRG and LG/CBO/UG NRS -26% -2% -24% No membership Note: % should be read as pp. It is clear that the differences between geographical areas in poverty head count are larger than between HDI supported households and Non-HDI households. This points to the importance of existing conditions and factors external to HDI impacting on the socioeconomic status of all households. Figure 16 shows a comparison between poverty head count of households by HDI and Non HDI village and geographical area, which illustrates that overall poverty among target households has increased in the Dry Zone, Shan and in the Delta. However, the poverty head counts in those states was at a comparatively lower base in between 20% and 30%. Poverty among both HDI and Non HDI supported households has decreased in Chin and Eastern and Northern Rakhine states, but from a very high starting position, especially in Chin and Northern Rakhine with around 80% poor households in 2008 and slightly lower in Eastern Rakhine. The pattern shows clearly that both HDI and Non HDI households are affected by larger and state wide socio-economic conditions and trends. The impact of HDI is mainly seen in how households are able to respond to such macro or meso level factors. Please refer to the Poverty Dynamics and Causality Study for further analysis.

56 42 Figure 16 Poverty head count in HDI and Non HDI villages by geographic area 90,0 80,0 70,0 60,0 50,0 40,0 30,0 20,0 10,0 0,0 Double difference Poverty headcount in HDI villages by geographic areas + 16% + 11% + 5% - 14% - 30% Dry zone Shan Delta Chin ERS NRS - 11% - 15% + 1% - 5% - 16% + 5% - 24% 90,0 80,0 70,0 Poverty headcount in Non HDI villages by geographic areas - 9% 60,0 50,0 40,0 30,0 20,0 + 26% + 26% + 5% - 14% - 30% 10,0 0,0 Dry zone Shan Delta Chin ERS NRS Note: % should be read as pp. Poverty head count by household head There are more poor female headed households (FHH) than male headed households (MHH) overall. For FHH there has been a reduction of 7 pp in poverty head count in HDI compared to a reduction of 6 pp in Non HDI villages. For MHH, there has been a reduction of 1 pp in the poverty rate in HDI but an increase of 3 pp in Non HDI areas. Thus it appears that female-headed households are better able to cope with poverty. The reasons for the differential performance of FHH and MHH are not clear. However, it is a weakness of the present survey that it does not have data on remittances from migrant family members, which could be a factor in explaining this. The overall stronger position of female headed households on this indicator is however tainted by the fact that the poverty head count in SRG female headed households has gone up by 8.3 pp. In households in HDI villages that are not members of any group the poverty head count has gone down by 17 pp. This finding beckons further investigation. Table 33 Poverty headcount index by gender of household head (2008/ 2009 & 2012) Impact Assessments Difference Diffe- 2008/ / rence Male Female Male headed HHD Female headed HHD HDI Villages 38,9 37,8-1,1 52,1 45,1-7,0 Non HDI Villages 40,9 44,2 3,3 50,6 44,5-6,1 Difference HDI - Non HDI -2,0-6,4-4,4 1,5 0,6-0,9

57 43 Poverty gap 7 There has been a slight decrease in the poverty gap among HDI supported households whereas there has been an increase among the control group. The aggregate DD is not statistically significant. For Shan the DD is 0,037, which is significant at alpha 10% (table 40 in annex). Table 34 Poverty gap (2008/ 2009 & 2012) HDI Villages 0,087 0,081-0,006 SRG 0,040 0,060 0,021 Both SRG and LG/CBO/UG 0,160 0,142-0,018 supported HHs Only LG/CBO/UG 0,072 0,079 0,007 estimates No membership 0,091 0,081-0,010 Non HDI Villages 0,096 0,097 0,001 Difference (HDI - Non HDI) -0,009-0,016-0,007 Squared poverty gap ratio The Squared poverty gap ratio 1 is an indicator of the severity of poverty. It differs from the poverty gap ratio in that it gives more weight to the poorest households (i.e. those furthest from the poverty line). The pattern is similar to the poverty gap ratio with an overall decrease in double difference, but with a slight increase among SRG households. None of the DDs are statistically significant on this indicator. Table 35 Squared poverty gap (2008/ 2009 & 2012) HDI Villages 0,028 0,025-0,003 SRG 0,009 0,017 0,007 Both SRG and LG/CBO/UG 0,053 0,052-0,001 supported HHs Only LG/CBO/UG 0,023 0,023 0,000 estimates No membership 0,031 0,025-0,006 Non HDI Villages 0,033 0,032 0,000 Difference (HDI - Non HDI) -0,004-0,007-0,003 The aggregate DD for average adult equivalent expenditure, excluding health expenditure, is not statistically significant. However, for Eastern Rakhine state there has been a statistically significant increase among HDI households at alpha 1% and likewise, but to a lesser degree 7 Poverty gap ratio 7 is the mean distance separating the population from the poverty line (with the non-poor being given a distance of zero). The indicator measures the poverty deficit of the entire population, where the poverty deficit is the per capita amount of resources that would be needed to bring all poor people above the poverty line through perfectly targeted cash transfers. In other words, it measures the intensity of poverty, i.e. the average shortfall from the poverty line (depth of poverty) of the poor multiplied by the poverty headcount. It can be used to provide an estimate of the sums required to raise the consumption level of all poor families to the poverty line.

58 44 in the Dry Zone at alpha 10% (table 42 in annex). It appears that there has been a noticeable decline among SRGs, and among members of both SRGs and LG/CB/UGs especially there has been a very significant increase. These findings need further investigations. Table 36 Adult equivalent consumption expenditure excluding health expenditure per year (2008/ 2009 & 2012) (Kyat at current prices) HDI Villages supported HHs estimates SRG Both SRG and LG/CBO/UG Only LG/CBO/UG No membership Non HDI Villages Difference (HDI - Non HDI) The food expenditure share in % of total expenditure, excluding health expenditure, is one of the key indicators of the HDI impact assessment. The aggregate DD is not statistically significant. In HDI supported households, the share has gone up very slightly, but it has gone down among the control group households, leading to a double difference of 2,2 pp. However in Northern Rakhine the DD is -2,3 pp, which is a decrease significant at alpha 5% (table 47 in annex). Table 37 Food expenditure share in % of total expenditure excluding health expenditure (2008/ 2009 & 2012) HDI Villages 66,1 66,2 0,1 supported HHs estimates SRG 64,2 63,6-0,6 Both SRG and LG/CBO/UG 70,9 65,1-5,8 Only LG/CBO/UG 73,9 71,7-2,2 No membership 66,0 66,2 0,2 Non HDI Villages 68,0 65,9-2,1 Difference (HDI - Non HDI) -1,9 0,3 2,2

59 45 Table 38 Food expenditure share in total expenditure excluding health expenditure by gender of household head (2008/ 2009 & 2012) (%) Male headed HHD Female headed HHD Male headed HHD Female headed HHD HDI Villages 65,6 69,2 65,6 69,2 SRG 63,8 66,2 62,9 67,1 Both SRG and LG/CBO/UG 70,1 74,2 64,5 67,4 supported HHs estimates Only LG/CBO/UG 73,5 70,9 74,8 No membership 65,4 69,4 65,7 68,6 Non HDI Villages 67,8 69,2 65,5 68,5 The food expenditure share is still higher among female headed households overall, but among SRG members it has gone up by 1,1% whereas it has gone down for all other female headed households Food Security Another key HDI outcome Indicator for 2011 was the Average # of months of food-security of beneficiary households. The question is if HDI households have better food security than Non-HDI households, has their food security improved since 2008? The aggregate DD is not statistically significant. Overall there has been a decrease in the number of months in which households can have their meals regularly without borrowing, i.e. a decline in food security (table 48 in annex). Table 39 Average months without borrowing for food in the past 12 months (2008/ 2009 & 2012) 2008/ / ) HDI Villages 7,2 6,6-0,6 supported HHs estimates SRG 7,2 7,3 0,1 Both SRG and LG/CBO/UG 8,6 7,3-1,3 Only LG/CBO/UG 5,0 4,9-0,1 No membership 7,0 6,6-0,4 Non HDI Villages 6,8 6,7-0,1 Difference (HDI - Non HDI) 0,3-0,2-0,5 Figure 17 shows the differences between State/Divisions and for each of the household types. It is noteworthy that the difference between State/Divisions is greater than the differences between HDI and Non HDI households.

60 Figure 17 Change in Food security by State/Division and HH type 46

61 Labour Force and Employment 8 None of the DDs for labour force participation and for the unemployment rate are statistically significant. The aggregate DD for labour force participation is 0,7 pp and for unemployment rate -0,1 pp (tables 50 and 52 in annex). Table 40 Labour force participation rate (2008/ 2009 & 2012) (%) 9 HDI Villages 72,5 71,8-0,7 SRG 77,6 79,4 1,7 Both SRG and supported HHs LG/CBO/UG 69,5 66,5-3,0 estimates Only LG/CBO/UG 70,5 63,7-6,8 No membership 71,0 71,5 0,5 Non HDI Villages 71,6 70,2-1,4 Difference (HDI - Non HDI) 0,9 1,6 0,7 Table 41 Unemployment rate (2008/ 2009 & 2012) (%) 10 HDI Villages 1,2 0,5-0,7 SRG 0,4 0,4 0,0 Both SRG and supported HHs LG/CBO/UG 1,3 0,5-0,9 estimates Only LG/CBO/UG 6,0 0,7-5,3 No membership 1,2 0,5-0,7 Non HDI Villages 1,1 0,6-0,6 Difference (HDI - Non HDI) 0,0-0,1-0,1 Labour force participation by gender shows significant differences with fewer women than men in the labour market (table 51 in annex). In 2008/09 the ratio of SRG women in the 8 The labour force consists of those who are employed plus those who are unemployed during the relevant reference period. It is the economically active portion of the population. Employment refers to being engaged in an economic activity during a specified reference period or being temporarily absent from such an activity, while economic activity refers to the production of goods and services for pay or profit or for use by own household. (United Nations Development Group. (2003). p. 86) 9 The labour force participation rate of the population aged 15 years and over is defined as the proportion of the population aged 15 years and over that are in the labour force, i.e., working or available for work. The labour force excludes individuals who were absent due to health or other reasons, individuals doing housework fulltime, individuals studying fulltime (or other training), fulltime religious personnel, the disabled or developmentally delayed, individuals living on pension or retired, and individuals who stopped looking for work. 10 Unemployed people are all those who are not employed during a specified reference period but are available for work and have taken concrete steps to seek paid employment or self-employment. (United Nations Development Group. (2003). p. 86) The unemployment rate of the population aged 15 years and over in the last 7 days provides information on recent or short term unemployment. It is defined as the proportion of the labour force participants aged 15 years and over that did not work at any point in the 7 days preceding the survey.

62 48 labour market was 11 pp higher compared to the Non HDI households and in 2011 this advantage had increased to 14 pp. The unemployment rate of SRG women has gone down, but it has gone further down among the Non HDI households (table 53 in annex). Table 42 Labour force participation rate by gender (2008/ 2009 & 2012) Male Female Male Female HDI Villages 85,5 60,4 87,6 57,7 SRG 88,6 67,5 89,8 69,8 Both SRG and supported 81,4 59,2 82,4 53,0 LG/CBO/UG HHs Only LG/CBO/UG 86,3 58,0 87,5 44,9 estimates No membership 84,9 57,4 87,4 56,7 Non HDI Villages 87,6 56,4 86,5 55,6 Table 43 Unemployment rate by gender (2008/ 2009 & 2012) Male Female Male Female HDI Villages 1,3 1,0 0,5 0,5 SRG 0,3 0,6 0,6 0,1 Both SRG and supported 1,3 1,4 0,5 0,5 LG/CBO/UG HHs Only LG/CBO/UG 5,4 6,7 0,5 1,1 estimates No membership 1,5 0,8 0,5 0,6 Non HDI Villages 0,9 1,5 0,3 0, Education, Literacy and Gender The percent in net primary enrolment ratio (disaggregated) is a Key Outcome Indicator for The question is whether HDI households (SRG-Non SRG) have a higher enrolment ratio than Non HDI households. Have there been any developments since the enrolment ratios of 2008/09? The 2008 survey found no significant difference between neither HDI and Non HDI, nor SRG and Non-SRG households with regard to enrolment, but slightly higher literacy among SRG households. There was no disaggregation by gender done in the 2008 Impact Assessment. For both gross and net enrolment ratio in primary education neither the aggregate DD, nor DDs for the geographical areas are statistically significant (tables 54 and 56 in annex). Enrolment in primary education has decreased in all sampled households, except in SRG households for there has been an increase of 7,3 pp in gross and 2,4 pp in net enrolment. This could be encouraging and one can speculate if the tendency of SRG members to become poorer could be partly explained by the higher priority to education. However this beckons further investigation.

63 49 Table 44 Gross enrolment ratio in primary education (2008/ 2009 & 2012) (%) 11 HDI Villages 127,8 125,1-2,7 SRG 126,8 134,1 7,3 Both SRG and LG/CBO/UG 133,6 123,5-10,1 supported HHs Only LG/CBO/UG 139,5 127,3-12,2 estimates No membership 126,0 119,8-6,2 Non HDI Villages 134,7 127,4-7,4 Difference (HDI - Non HDI) -7,0-2,2 4,7 Figure 18 Differences in gross enrolment ratio in primary education 11 The gross enrolment ratio in primary education is the ratio of children of any age enrolled in primary education (Grade 1-5: KG to 4 th standard) over the total population of children of official primary school age (5 to 9 years).

64 50 Table 45 Net enrolment ratio in primary education (2008/ 2009 & 2012) (%) 12 HDI Villages 90,3 84,7-5,6 SRG 92,7 95,1 2,4 Both SRG and LG/CBO/UG 89,0 84,4-4,6 supported HHs Only LG/CBO/UG 95,2 72,4-22,9 estimates No membership 89,2 84,5-4,7 Non HDI Villages 90,2 83,5-6,7 Difference (HDI - Non HDI) 0,1 1,3 1,2 Though enrolment in secondary education has seen a decrease among HDI households with a DD of -4 pp in gross enrolment and -6 pp in net enrolment ratio, the DDs are not statistically significant (table 58 and 60 in annex). Table 46 Gross enrolment ratio in secondary education (2008/ 2009 & 2012) (%) / 2009) HDI Villages 46,6 48,7 2,1 SRG 46,6 53,5 6,9 Both SRG and LG/CBO/UG 62,0 68,1 6,1 supported HHs Only LG/CBO/UG 25,8 23,7-2,1 estimates No membership 42,4 51,2 8,8 Non HDI Villages 31,0 37,1 6,1 Difference (HDI - Non HDI) 15,6 11,6-4,0 Within the HDI supported households it is again the SRG households that have the highest increase in enrolment. Enrolment among only LG/CBO/UG members has even decreased (Figure 19). Again, these differences need further investigation. Tables to 61 to 68 in annex gives detailed results on education and literacy rates by gender. There are no noteworthy statistically significant DDs on these variables. 12 The net enrolment ratio in primary education 12 is the ratio of the number of children of official school age (as defined by the national education system) who are enrolled in primary school to the total population of children of official school age. The indicator is used to monitor progress towards the goal of achieving universal primary education, identified in both the Millennium Development Goals and the Education for All initiative. (United Nations Development Group. (2003). p. 16) 13 The gross enrolment ratio in secondary education is the ratio of children of any age enrolled in secondary education (Grade 6-11: 5 th -10 th standard) over the total population of children of official secondary school age (10 to 15 years).

65 51 Table 47 Net enrolment ratio in secondary education (2008/ 2009 & 2012) (%) 14 HDI Villages 36,4 37,1 0,7 SRG 37,9 45,5 7,6 Both SRG and LG/CBO/UG 46,7 48,5 1,8 supported HHs Only LG/CBO/UG 16,4 15,9-0,5 estimates No membership 33,1 37,9 4,8 Non HDI Villages 23,1 29,5 6,3 Difference (HDI - Non HDI) 13,2 7,6-5,6 Figure 19 Differences in net enrolment ratio in secondary education Health Status One of the UNDP Country Programme Outcome Indicators for 2011 is the incidence of morbidity experienced by beneficiaries (percent). The question is: Do HDI households (SRG and Non-SRG) experience lower morbidity rates, especially from diarrhoea, than Non HDI households, and has this changed since 2008? The data are based on 2 weeks recall time for individual household members on diarrhoea/ sickness events. As is well known this is a too long recall period to ensure precise answers and the results should be taken with that in mind. 14 The net enrolment ratio in secondary education is the ratio of students of official primary school age (10 to 15 years) over the total population of official secondary school age. The indicator is a measure of the coverage and efficiency of the school system.