INTEGRATED SAFEGUARDS DATA SHEET IDENTIFICATION / CONCEPT STAGE Public Disclosure Copy Report No.: ISDSC16871 Date ISDS Prepared/Updated: 04-Oct-2016 I. BASIC INFORMATION A. Basic Project Data Country: Armenia Project ID: P159054 Project Name: Strengthening Armenia's Integrated Living Conditions Survey Team Leader(s): Moritz Meyer,Nistha Sinha Estimated Date 01-Dec-2016 of Approval: Managing Unit: GPV03 Lending IPF Instrument: Financing (in USD Million) Total Project Cost: 0.2 Total Bank Financing: 0 Financing Gap: 0 Financing Source Amount Trust Fund for Statistical Capacity Building 0.2 Environment C - Not Required Category: Public Disclosure Copy B. Project Development Objective(s) The development objective of this activity is to strengthen poverty monitoring in Armenia via improvements to the quality of microdata (ILCS) collected by the National Statistical Service (NSSRA). C. Project Description This grant project aims to improve the quality of the ILCS. First, the ILCS survey last underwent improvements in 2004, more than 10 years ago. A review of the sampling design and survey procedures would be important to undertake so that the survey design can be updated. Second, over the last decade the country has experienced population shifts and changes in structure of economic activity which should be reflected in the ILCS survey design. Non-response has emerged as a concern with the highest rate of refusal (18 percent) in the capital city, Yerevan. Up until the 2013 survey, ILCS was using a sampling frame (database of addresses) based on the 2001 population census which contributed to the non-response rate. From 2014 round, the NSSRA has begun to use the list of addresses (sampling frame) derived from the 2011 population census. Given the population dynamics in the country, it is important to understand the extent to which this list of addresses is still current 6 years after the census was conducted. Because any selected household that refuses to be interviewed is replaced from a list of reserve addresses, use of an outdated sampling frame will exacerbate non-response because the reserve address might no longer exist and new Public Disclosure Copy housing units created after 2011 will not be represented. Procedures to address this issue will have to be developed. Third, there appears to be underreporting of consumption expenditures in the ILCS perhaps in some part due to better off households refusing to participate in the survey. Consumption estimated from the ILCS is only about a third of private consumption calculated from the national accounts. Such discrepancies between household and national accounts based consumption estimates are common across the world but the discrepancy tends to be much smaller than what we observe in the ILCS. This underreporting in the survey could arise from richer households refusing to participate and/or from responding households not fully reporting income/consumption. Training of field teams will help address underreporting or concealing of consumption and income by responding households given that they visit households 5 times during the month of interview and recording of consumption and income. Analysis of the characteristics of those who refuse to participate in the survey will help understand to what extent selective refusal by the better off is affecting the survey. Finally, working with the NSSRA, the World Bank?s poverty team in Armenia has identified the precision of poverty estimates produced by the ILCS as a key concern. The ILCS is designed to produce consumption-based poverty estimates at the national, rural-urban, and 11 regional (or marz) levels. Having precise and reliable estimates of poverty at all these levels is important for informing government?s development policies especially at the regional level. Statistically, precision or degree of reliability of poverty estimates is represented by the standard error of the point estimate of poverty (the number that is usually reported) and the associated 95 percent confidence intervals which indicate the range of poverty estimates within which the actual poverty rates must lie. One rule of convention used by statisticians is that this confidence range should not be too wide: if the point estimate of poverty is z then the confidence range should not exceed 2*0.05z. Although poverty estimates at the national and rural-urban level are estimated with some precision (that is, low Public Disclosure Copy standard errors and narrow 95 percent confidence intervals), marz-level poverty estimates tend to be less precisely estimated. In 2014, the national poverty estimate was 30 percent with a confidence interval of 27.9 percent and 32 percent implying that the actual poverty rate was somewhere within this range with 95 percent certainty. At the marz level this confidence interval varies widely making it difficult to rank regions by their poverty rates or estimating the number of poor. For example, both Ararat and Armavir marzes are estimated to have a poverty rate of 29 percent but the confidence interval for the latter ranges from 19 percent to 38 percent while that of the former ranges from 23.5 percent to 34.6 percent. Expanding the sample size is important for improving reliability of poverty estimates. The accuracy or precision of the poverty estimates depends on both the sampling error, which can be measured through variance estimates, and the nonsampling error, which results from all other sources of error, including response and measurement errors as well as coding, keying and processing errors. The sampling error is inversely proportional to the square root of the sample size and hence expandi ng sample size is one way in which precision of poverty estimates can be improved. Expansion of sample size would be particularly important for the regions were the ILCS currently reports very wide confidence intervals. The grant will provide additional resources to NSSRA to enhance the quality of 2017 ILCS data by addressing the issues identified above. This grant will be complemented with a Bank-executed technical assistance supporting a review of survey design and processes, identifying the updates/ changes needed, and providing training to NSSRA staff on these topics. The recipient executed grant will support a) implementation of updates to the survey design and training for field teams; and b) Public Disclosure Copy expansion of the sample size. For the given budget and costs, the grant can support the addition of 2,600 households, taking the total sample to 7,784 households. The expansion of the ILCS sample will come from addition of enumeration areas (primary sampling units) rather than increases in the number of households interviewed within each enumeration area. The proposed activities will build on the World Bank?s ongoing engagement with NSSRA and will be complemented through a Bank-executed technical assistance for reviewing and updating sampling design and survey implementation. These enhancements will be implemented with the 2017 ILCS round. The grant has two components: 1) Component 1 ($190,000, expanded ILCS sample) will implement a sampling design with expanded the sample size. The revised sampling design will be supported through a separated Bank- executed TA to the NSSRA. Based on early discussions with NSSRA, the grant could finance additional field personnel needed (consultant services), training of the additional personnel and printing of additional interview material (questionnaire, field manuals etc). 2) Component 2 ($10,000, Project Management) will finance the operating costs of the Grant. D. Project location and salient physical characteristics relevant to the safeguard analysis (if known) E. Borrower’s Institutional Capacity for Safeguard Policies F. Environmental and Social Safeguards Specialists on the Team Public Disclosure Copy II. SAFEGUARD POLICIES THAT MIGHT APPLY Safeguard Policies Triggered? Explanation (Optional) Environmental Assessment OP/ No BP 4.01 Natural Habitats OP/BP 4.04 No Forests OP/BP 4.36 No Pest Management OP 4.09 No Physical Cultural Resources OP/ No BP 4.11 Indigenous Peoples OP/BP 4.10 No Involuntary Resettlement OP/BP No 4.12 Safety of Dams OP/BP 4.37 No Projects on International No Waterways OP/BP 7.50 Public Disclosure Copy Projects in Disputed Areas OP/BP No 7.60 III. SAFEGUARD PREPARATION PLAN Appraisal stage ISDS required?: No IV. APPROVALS Team Leader(s): Name: Moritz Meyer,Nistha Sinha Approved By: Safeguards Advisor: Name: Nina Chee (SA) Date: 27-Sep-2016 Practice Manager/ Name: Carolina Sanchez (PMGR) Date: 29-Sep-2016 Manager: 1 Reminder: The Bank's Disclosure Policy requires that safeguard-related documents be disclosed before appraisal (i) at the InfoShop and (ii) in country, at publicly accessible locations and in a form and language that are accessible to potentially affected persons. Public Disclosure Copy