1 Technical Design Issues in Constructing Migration Event HistoriesProfessor Tilman Brück ISDC – International Security and Development Center Workshop on “Civil Registration in Refugee Settings: Advancing Assessment of CRVS in Unsettled Environments” 19-20 December 2016, Beirut

2 Disclaimers CRVS is typically a government activityI am an academic researcher Key issues in CRVS focus on statistics I am a development economist CRVS focuses on administrative data My expertise is in collecting and analyzing survey data Most CRVS take place in countries at peace My research focuses on conflict and fragile states → Outside-in perspective

3 Overview Conceptual issues Case studies Angola Kyrgyzstan Syria

4 1: Conceptual Issues

5 Peace birth death Challenges statistical capacity weak base datacomplex regulations undocumented citizens

6 Migration Syria Lebanon Additional challengesavailability and validity of foreign documents illegal migration

7 War birth war death Additional challengesdestruction of capacity and of records multiplicity of actors in humanitarian setting scale of displacement security of access political agendas even more constraining what are “vital statistics” for war-affected people?

8 War + Migration birth in Syria war death in LebanonAdditional challenges paucity of complementary data fluidity of household boundaries and household dispersion strong behavioral responses to war and displacement

9 Objective of CRVS @ War Recording Organizing Analyzing Discussionidentifying residents and citizens basic demographics Organizing planning eligibility and needs Analyzing understanding trends and causality understanding key “transitions” Discussion overlapping concepts fluid boundaries between CRVS and surveys learn from household rosters in survey work clarify objectives of CRVS before considering validation strategy

10 Levels of Aggregation Individual level Household level National levelrecording: vital event, e.g. own birth organizing: verification for government services, eg health care analyzing: relationships between events and outcomes per person Household level recording: someone else’s vital event, eg birth of child organizing: verification for government services, eg housing analyzing: relationships between events and outcomes, within HH National level recording: vital event within or across groups organizing: e.g. spatial planning analyzing: population dynamics (by group)

11 Requirements Completeness Quality Timeliness Relevanceeven for formal records: what do you want to record and why? that is, what is an “event”? e.g.: live birth, stillborn birth, miscarriage, abortion, pregnancy? what about trauma, experience of violence, loss of home etc? risk of omitted variable bias? Quality beyond accuracy: know your error patterns and errors will vary widely across sub-groups Timeliness this makes sense in domestic peace-time but also record events ex-post, from time spent abroad? Relevance check if CRVS is relevant for the context

12 Which Events to Register?In war, more than birth and death matters location/jurisdiction is unclear date of border crossing (what about repeated crossing?) events concerning an individual vs. an individuals family family vs. household (role of guardian, unaccompanied minors etc) legality of status (who determines this?)

13 Who to Register? Also look at host populationwar in the region and being a host community has implications for capacity to collect data for political discussions and priorities for service delivery Maybe re-define standard protocol for CRVS

14 2: Case Studies

15 2: Case Studies Study of Ex-Combatants in Angolacollecting and analyzing sensitive survey data in an orphan post-war country with Wolfgang Stojetz (ISDC) work-in-progress data collection based on: Brück, T. et. al. (2016): “Measuring Violent Conflict in Micro-Level Surveys: Current Practices and Methodological Challenges”. World Bank Research Observer, 31(1):

16 The Project Study focus Overall hypothesisparticipation of Angolan demobilized soldiers in their local community, economy, and family in the long run Overall hypothesis military service and exposure to war shape post-war behavior in economy, society and family Relevance for our discussion highlight heterogeneity in outcomes rooted in war exposure indicate feasibility of enumerating war experiences gain analytic strength from multi-disciplinary cooperation stress importance of involving multiple stakeholders

17 The Survey Fieldwork Interviews and data Huambo province, Angola760 veterans and partners from 34 clusters complementary ethnographic study (Spall 2015) Interviews and data from veteran: pre-service, service and post-service (including wartime violence, e.g. sexual abuse) from spouse: family and relationship outcomes (including domestic violence) from couples: household outcomes

18 Challenge: Sensitive TopicsRespondents’ perspective: intimate and traumatic experiences incentives to under-report e.g. social pressures, fears of retaliation or legal sanctions incentives to over-report e.g. hopes of monetary benefits or health treatments Government perspective ex-combatants a ”high-risk group” inefficient post-war assistance to ex-combatants

19 Strategies: Sensitive TopicEngage with relevant/all authorities for consent and support Multi-disciplinary survey instruments psychology module on traumatic experiences and mental health ethnographic companion study (Spall 2015) Select interviewers carefully align language, ethnic group, social group (no ‘outsiders’) Questionnaires pre-test, pre-test, pre-test! Interviewers train, train, train! (key: atmosphere, trust) Interviews practice, practice, practice! (key: privacy)

20 Challenge: Capacity Post-war setting ’Orphan’ setting (JL Arcand)formal institutions are weak informal institutions are strong and diffuse existing government data likely not reliable/biased ’Orphan’ setting (JL Arcand) few NGOs few projects few existing findings

21 Strategies: Capacity Local partners Engage with authoritiesNGO Local networks/visits Engage with authorities at all levels Try to get hands on any secondary data available e.g. ongoing census (design) Build population lists at the cluster level (design)

22 Summary of Data CollectionHousehold survey fieldwork: Jun-Aug 2012, Mar-May 2013, Aug 2013-Jan 2014 population: ex-combatants in Huambo province stratification: 2/3 MPLA, 1/3 UNITA; 60% rural, 40% urban interviews: household, veteran, spouse questionnaires: electronic (tablets, ODK) data storage: electronic (online, ODK) Interviews and data from veteran: pre-service, service and post-service/-war from spouse: family and relationship outcomes from couple: household outcomes

23 War-time Violence ExposureDid veteran experience situations where... Mean SD Min Max ...fellow soldiers lost their lives? 0.49 0.500 1 ...enemies lost their lives? 0.44 0.497 ...civilians lost their lives? 0.41 0.491 ...civilians were attacked (no massacre)? 0.45 0.498 ...civilians were killed (massacre)? 0.39 0.488 ...civilian woman was sexually abused? 0.22 0.414 Note: 0=No, 1=Yes, N=759

24 2: Case Studies Life in Kyrgyzstan Studywith Damir Esenaliev (ISDC and SIPRI) and many other colleagues Brück, T., D. Esenaliev, A. Kroeger, A. Kudebayeva, B. Mirkasimov & S. Steiner (2014): “Household Survey Data for Research on Well-Being and Behavior in Central Asia”. Journal of Comparative Economics, 42 (3):

25 Life in Kyrgyzstan StudyTracking individuals in households in 120 communities Yielding over 30 million data points so far Stratified 2-stage random sampling from census Representative: national, rural/urban, North/South Data collection: around October–December So far, 4 waves from 2010, 2011, 2012 and 2013 Currently, we are collecting wave 5 Data and questionnaires are available online Research-led, open-access, independent study

26 National Coverage

27 LiK is a Multi-Topic SurveyFarming and agricultural markets Gender and intra-household bargaining Assets, income, remittances and expenditure Employment, entrepreneur-ship and migration Demographics, education and child health Public service delivery Shocks, fears and security perceptions Social capital, networks and subjective well-being Huge benefits from connecting these themes!

28 Questionnaires Individuals Households family and householdsubjective well-being education and health labor market movements social life child anthropometrics worries security and violence Households household composition child education and health housing and assets agricultural markets consumption and expenditures income migration shocks

29 6 Conflict and DisplacementBrück, Hennicke, Kröger and Vothknecht: work in progress using LiK 2010

30 3 Indicators of FragilityBrück, Hennicke, Kröger and Vothknecht: work in progress using LiK 2010

31 Impact of Conflict Aladysheva and Brück: work in progress using LiK 2010

32 Implications We regularly track for exposure to insecurity and fragility This questions are less intrusive in the field than feared by many ex ante Combining various topics is key for analysis Conflict and fragility impact on people significantly Hence we build up understanding of key life events over time

33 2: Case Studies ‘Crowd-seeding’ Conflict and Peace Data in Syriawith Ghassan Baliki (ISDC) work-in-progress

34 Design Trained reporters in 11 geographic areas in Syria They report:detailed information on violence, peace, kidnapping, and looting information at the local level, mainly within their geographic proximity information which they either observed personally or through their key local network informants which is directly coded into a database (real-time) via a secure platform

35 Codebook and Platform Codebook Online platformdeveloped in line with existing conflict event databases (with some minor adjustments) It includes detailed description of the information and definitions of the main events and their attribute data Online platform entry sheets are based on the structure of the codebook access and submission of information (events) is anonymous and highly secure data is immediately stored and updated into a shared database, and readily accessible for basic analysis

36 Generic Events Geo-location of an eventseparately divided between large cities and other geo-areas constitutes of 6 precision levels (down to city neighborhood level) reporters click an embedded map and specify the precision level (coordinates are generated automatically) Date of submission and date of event and temporal precision (exact date, within 2 days, etc) Sources of information trust and reliability for second degree observations from their local networks Sensitivity of publishing certain incidents to reduce risk Open list of actors to choose from and to add to

37 Example of Database

38 Violent Events Forms of Violence Tools of Violenceviolence between armed actors (battles) indiscriminate violence against civilians one-sided violence (with no clear target) Tools of Violence Aerial and Anti-aerial; Ground-to-ground (Heavy); Small and medium firearms; Stationary explosives; Suicide bombings; etc Outcomes of Violence death of armed actors and/or civilians per events injury destruction of dwellings or Infrastructure (very detailed) displacement as a result of a violent event territory gain or loss (with it spatial coordinates and actors involved)

39 Peace Events Mainly around agreements at the local level between local actors Forms of peace events include invitations, talks, reaching agreements rejections of talks, violations, and breaking-off agreements Types of agreements are divided into 2x2 matrix permanent or temporary involves suspension of violent or other non-violent related issues (aid access)

40 Example of Database

41 Example of Analysis

42 Comparison of Methods Crowd-Seeding Crowd-Sourcing TraditionalNon-retrospective ✔ Elimination of media bias Elimination of reporting fatigue More detailed and micro-level More precise coordinates Codebook-oriented Less costly Less risky

43 Implication Tools like our crowd-seeding database of conflict and peace events may be critical for the validation of official records and survey data

44 Conclusions Validate information across data sources Current project design asks if system works and aims to strengthen system But also ask what we do not know and what would CRVS need to do to support critical learning (i.e. what are knowledge gaps)