Inclusive Growth Research Infrastructure Diffusion

Final Report Summary - INGRID (Inclusive Growth Research Infrastructure Diffusion)

Executive Summary:
The InGRID research infrastructure (RI) serves the social sciences aspiring to make an evidence-based contribution to the European policy challenge of inclusive growth. This research community focuses on deprivation, vulnerability-at-work and related social and labour market policies from a European comparative perspective. It is an interdisciplinary European Research Area of poverty research, labour studies, policy analysis and social statistics.
The InGRID RI supports these social scientists to access and analyse comparative data and help them translating their evidence-based knowledge to the field of European policy innovation. Key in this top-level, interdisciplinary Research Area is looking for and interpreting problematic trends in social situations and workplaces and to monitor or assess policy influences and innovations on these trends. The ‘helping hand’ of the RI is hereby providing integrated data, analytical facilities to link policy and practice, and indicator-building tools. The RI includes therefore centres that integrate data at a European level; competence centres that have expertise on key official European data sources; institutes that have invested in comparative policy databases and/or microsimulation models to investigate the impact of policies with comparative microdata; organisations providing standards of harmonisation; new and innovative data collectors; and statistical departments providing methodological know-how and equipment.
The FP7 project integrated for the first time this existing, but distributed European social sciences RIs on ‘Poverty and living conditions’ and ‘Working conditions and vulnerability’. Opening, widening and innovating the integrating RI was the key goal by improving the transnational access, organising knowledge exchange and improving the RI methods and tools.
184 scientist made a highly-valued visit to one of the RIs. 462 early-stage-researchers attended one of the 18 training events. 14 workshops were attended by in total almost 500 experts.
Besides continued inventory and assessment of existing databases and tools, joint research and development work was conducted to improve the RI: the building of an integrated poverty and living conditions indicator system (IPOLIS) for particular vulnerable groups; recommendations on the data collection of hard-to-identify and to-reach groups; high-precision poverty mapping; new indicator-methods to identify and map vulnerability-at-work; further steps in the harmonised surveying of occupations; the investigation of the added-value of web crawling for monitoring new jobs and new skills; improved tools to generate comparative social policy indicators; a well-focused enhancement of tax-benefit simulation model EUROMOD with a hypothetical household tool and the domain of child care policies; crucial statistical improvements to data analysis methods, such as solving survey sampling problems ex-post in the analysing phase, improving techniques to construct small area estimations, advancing multidimensional indicator-building and exploring the potential use of Big Data from a statistical point-of-view.
The FP7 project has thus strengthened and opened for the first time this existing and integrating InGRID RI and gained considerable recognition for these accomplishments from a wide range of stakeholders. RI gaps and an innovation agenda based on among others a users’ survey have been summarised in a future planning. The project achievements will be sustained and the RI will be further advanced through a 4-year follow-up EU Horizon 2020 project.

Project Context and Objectives:
2.1 Concept: integrating a distributed social sciences research infrastructure
The ‘Europe 2020’ Strategy is the overarching policy initiative which brings together all areas of EU competence and activity in order to prepare the EU economy for the next decade. The third priority of inclusive growth is aiming to raise labour market participation, to fight poverty and to strengthen social cohesion. Combating poverty and ensuring quality of work and employment are core elements in achieving this objective. Referring to this EU2020-ambition of Inclusive Growth, the general objectives of the InGRID project were to integrate and to innovate for the first time existing, but distributed European social sciences research infrastructures (RI) on ‘Poverty and living conditions’ and ‘Working conditions and vulnerability’ by improving the transnational data access, organising mutual knowledge exchange activities and improving methods and tools for comparative research. This integration provided the related European scientific community-of-practice with new and better opportunities to fulfil its key role in the development of evidence-based European policies on Inclusive Growth.
Figure 1 Facilities and services of InGRID research infrastructure (see attachment)

For this purpose 17 key actors of the related European Research Area were brought together in the InGRID consortium, representing data infrastructures and competence centres with EU FP5-FP6-FP7 cumulated know-how and providing the following types of social sciences services and facilities:
1. international expertise and competences on key data collections, mostly European surveys or data (EU-SILC, LFS, European Household Finance and Consumption Survey, European Working Conditions Survey, European Company Survey, ESENER survey), but also data from Member States (census data, national working conditions surveys and employer’ surveys, socio-economic panels, ...). Taking data protection rules into consideration, researchers are guided and helped in the use and analysis of these data;
2. integrated data sets: multicountry, harmonised income (LIS) and census microdata (IECM), indicator databases built-from the mentioned surveys (EUROMOD, IPOLIS); the CSB-EU Reference Budgets database;
3. innovated data collection strategies beyond these ‘official’ data: web survey initiatives like the WageIndicator databases, MEADOW/InGRID protocols for organising/improving working conditions surveys, access to national best practices like British Household panel and Understanding Society UK Household longitudinal study or French REPONSE (linked employer/employee survey);
4. tools for harmonised classification: WISCO or the world data base of occupations; translation and classification tables for national surveys and indicator-building based on textual policy databases;
5. comparative policy databases: Institutional characteristics of Trade Unions, Wage Setting, State Intervention and Social Pacts (ICTWSS); the Social Policy Indicators Database (SPIN); CSB-MIPI data set on minimum income protection;
6. microsimulation environments: tax-benefit microsimulation model for the European Union EUROMOD (UESSEX and others), LIAM2-based microsimulation models (LISER); UNI-Trier social statistics large simulation infrastructure (including AMELIA);
7. statistical quality standards: small area estimating techniques (UNIPI); S3RI centre for technical support on small area estimation, sampling and inference for big data; data linkage;
8. indicator-building techniques: high expertise on indicator-building: quality of work (KU Leuven), vulnerable groups (TARKI, UvA), social policies (SU, UA, LISER).
2.2 Detailed project objectives
The detailed project objectives were the following:
1. Transnational access to 13 infrastructures: an integrated visiting grant system to and learning network on existing European data infrastructures; target: 205 visitors and 2009 visiting days.
2. 30 networking and mutual learning activities
The objectives of the 12 expert network workshops (WP3) were among others:
▪ to identify and discuss key technical issues and possible solutions in particular areas of InGRID;
▪ to reinforce the interaction of the consortium with relevant others in the European Research Area and the European Statistical System;
▪ to open the distributed research infrastructure to a broader, transnational research community and community-of-interest;
▪ to provide a critical contribution to solve the current and future data and methodology needs of the research infrastructure.
Each workshops aimed at attracting 25 participants.
The objectives of the 18 Summer school training events (WP4) were to raise the competence level of early-stage researchers, engaged in RI-related research activities and of the broader public of users (experts coming from politics, civil society, governments or administrations) by providing training:
▪ to satisfy the learning and development needs in the research infrastructure based on a combination of short training courses and short-term transnational access;
▪ to transfer the assembled scientific excellence, cumulated in a series of EU FP SSH projects, to the advantage of the broader InGRID community;
▪ to train the next generation of scientist, engaged in the use of the RI;
▪ to organise ‘hands-on’ training events to diffuse output to the broader network of users/professionals.
Each training aimed also at attracting 25 participants.
3. Four well-focused and innovative Joint Research Activities
The overall and common objectives of the JRA were:
▪ to provide inventories of national and EU data sources of both micro- and macro-data;
▪ to produce metadata;
▪ to provide input into the series of mutual learning networking activities;
▪ to provide input to the future development planning of the research infrastructure including the e-infrastructure.
For the poverty and living conditions research (WP20) the objectives were in particular:
▪ to synthesise and make accessible data and knowledge in a structured form that backs comparative research on social policies, and improves infrastructure to study vulnerable groups at risk of poverty;
▪ to organise indicators and research results in a theoretical frame, with the aim of avoiding bottlenecks and temptations to carry on purely data driven research on poverty and living conditions;
▪ to create synergy between existing knowledge and data infrastructures and to support the development of methodologies and data infrastructure facilitating territorial analysis of poverty and living conditions.
For the working conditions and vulnerability research (WP21) the objectives were:
▪ to integrate repositories on European and national working conditions and occupational safety and health (WC&OSH) data using a web-based platform;
▪ to integrate and further develop tools and competences in the collection of cross-country comparative WC&OSH data;
▪ to provide the research community and stakeholders with classification and analytical tools for WC&OSH analysis;
▪ to develop new methods and tools to generate comparative WC&OSH data of relevance for the EU New skills new jobs strategy;
▪ to contribute in an innovative way to the development of an integrated research infrastructure on WC&OSH analysis with a focus on vulnerable groups.
For the policy analysis research (WP22) the objectives were:
▪ to develop an integrated research infrastructure on policy analysis;
▪ to develop new tools and competence in the collection of comparative and institutional policy data;
▪ to provide input for a web-portal that provides the research community and stakeholders with classification and analytical tools for on-line policy analysis;
▪ to develop new methods and tools to generate comparative and institutional policy data of relevance for poverty, living and employment conditions;
▪ to develop an institutional and comparative perspective on policy areas and programmes where good quality data are lacking;
▪ to improve the possibilities for linking institutional data to micro-level outcomes in the area of poverty and social inclusion;
▪ to create necessary incentives to assure commitment and operation of the integrated research infrastructure beyond the project period.
For the statistical quality research (WP23) the objectives were:
▪ to improve share and consolidate knowledge on theories and best practices to judge the quality and appropriateness of indicators through an empirical analysis
▪ to provide appropriate definitions of the indicators at multidimensional level.
▪ to define in a robust way their values to obtain statistically sound estimates for unplanned domains (typically small areas)
▪ to contribute to a correct usage and formulation of survey weights with attention to the impact of the weighting system and survey designs both on the definition of indicators and on the fitting of models for the analysis.
▪ to formulate and estimate measures of accuracy of the estimates, such as mean squared errors and non-sampling errors due to non-response, imputation of missing data and measurement errors arising from disclosure limitation methods
▪ to implement case studies in order to demonstrate the methodology as well as strengths and weaknesses.
▪ to use the result of the scenario analysis and simulation studies to develop best practice recommendations and to point out needs of further methodological research.
4. Innovative solutions in these RTD activities are sought by:
▪ the process flow integration of research with networking and transnational access;
▪ the strategic investigation of ‘missing infrastructure links’ to enhance the European Research Area and European Statistical System in the related fields;
▪ focused research to innovate fundamental technologies and techniques of data collection and analysis with vulnerable groups;
▪ scoping the feasibility of a pan-European approach to survey the effects of employers’ behaviour;
▪ contributing to socio-economic impact by improving instruments and indicators for the assessment of labour market and social policies.
In order to reach these objectives, the key milestones for these JRAs were methodological and data infrastructure reports, identifying the missing infrastructure links and final indicator and data infrastructure reports, providing innovative solutions to the problems identified.

Project Results:
The mission of InGRID as RI is thematically and policy driven: it is about serving a research community focusing on the major European policy strategy of ‘inclusive growth’. By definition this is an interdisciplinary community of sociologists, economists and political scientists. It is a research community of which the networked output is among others:
1. detectable in accomplished and on-going European framework projects. EU research and innovation will address social exclusion, discriminations and various forms of inequalities as one of the societal challenges of Horizon 2020;
2. appearing in a series of applied research networks of European policy actors: the European Social Policy Network; the European Employment Policy Observatory; and EurWORK. InGRID partners are important contributors to these networks.
As FP7 4-year project InGRID organised, diffused and innovated for the first time in a structured way the integration of the key existing distributed research infrastructure that serve this European and international (comparative) research in the fields of poverty, working life and living conditions. For these purposes InGRID involved 17 partners contributing to 23 different work packages. Jointly, these work packages generated an integrated and reinforcing cycle of mutual learning and knowledge-sharing, data access and technical innovation of the research infrastructures in the respective fields. In order to strengthen the integration of these activities and to secure the effectiveness of its impact, the different activities were clustered into two thematic pillars, ‘Poverty and living conditions’ and ‘Working conditions and vulnerability’, and into two horizontal pillars, ‘New tools and indicators for policy analysis’ and ‘Statistical quality management’.
As such, this INFRA project contributed to expanding and strengthening the pan-European social sciences impact on the Inclusive Growth Strategy of EU2020 by:
• providing a unified and harmonised transnational access to 13 social science installations based on a diversified grant system with open selection procedures (WP7-19);
• organising more than 30 networking mutual learning and training activities to provide services to the related research communities (expert workshops and summer schools)and to strengthen and open the research infrastructure (WP2 to 5);
• conducting 4 well-focused innovative joint research activities to enhance the technical quality of the research infrastructure (WP20-23).
The following graph presents these components of InGRID and their interdependencies and visualises the formalised work package structure of the project.
Figure 2 Work package structure of InGRID (see attachment)

In the following sections we discuss the main results of these different project activities.
2.1 Transnational access
A key activity of the InGRID project was providing transnational access to the research infrastructures. Researchers working and living within the EU member states (or associated countries) were invited to apply for free-of-charge access to and support in one of the 13 research infrastructures by a visiting grant system. Next to access to statistical competence centres with expertise in social science statistical research, access was granted to major surveys and tools such as LIS, LWS, IECM, SPIN, CSB-MIPI, WISCO, ICTWSS, EWCS, ECS, wage indicator data, EUROMOD, ... in research infrastructures with extensive experience in working with these data. Experiences of the users can be found at Previous visitors.
Transnational Access means free of charge, trans-national access to research infrastructures or installations for selected users or user groups. Central in a visiting grant is the free access to information and data, respecting and following data protection rules, held at the infrastructures. The access includes furthermore the logistical, technological and scientific support and the specific training that is usually provided to external researchers using the infrastructure. Transnational Access includes also support for travel and subsistence during the stay. Via short and long-term visiting grants (between 1 week and 1 month), researchers were able to experiment with and work on data in a context of mutual knowledge exchange and cross-fertilisation. Therefore, 10 calls for visiting grants were launched quarterly during the project. More information on the application process and terms and conditions for access to InGRID research infrastructures is available on the InGRID website. A whole logistical and administrative handling process was set-up in the organisation of this TNA, including a welcome desk, toolkit and ombudsservice. The granting of the visits was decided by a special created selection panel (including independent experts from the InGRID Advisory Board).
Table 1 Overview of visitors and visit days for each research infrastructure (see attachment)

1,930 visit days in total were used. Originally, 2213 visiting days were awarded to 200 applicants by the selection panel in the consecutive calls. However, due to planning and a lack of time to welcome visitors at especially the last contract period, some RIs could not use all the originally assigned visiting days. Visitors came from 29 different countries, representing 43 nationalities. 1 in 4 of the visitors came from newer EU Member States (see figure 3).
Figure 3 Geographical distribution of Transnational access visitors InGRID (see attachment)

The visitors evaluated on average the whole TNA visiting as an excellent opportunity that was smoothly organised (see consistent above 9.0 evaluation results in table 2).
Table 2 Questions for evaluation of the TNA process (see attachment)

The output following from these InGRID visits is very diverse and still growing steadily. Next to academic articles in journals (of which many are still in the review process), books and book chapters followed from the visits. Many users consolidate the work from their visit in a working paper or a conference presentation or presentation at the end of the visit or at the home institution as a first step in valorising the work which is done. Further some visits have led to other types of output, such as a database, a data collection module, a statistical package, a webtool, a documentation tool, etc. An overview of the output per visitor can be found at: https://inclusivegrowth.be/visiting-grants/selection-results.
2.2 Training and networking
2.2.1 Training programme
The training programme of InGRID focused on raising the competence level of early-stage researchers engaged in the RI-related research activities. A specific part was dedicated to the broader public of RI end users (experts or policy innovators from politics, civil society, governments and administrations). A normal programme lasted 5 days, but also shorter 2.5 days courses were organised. In line with InGRID’s project scope, the winter, summer and autumn schools offered the possibility to combine training activities with the opportunities to network with leading international experts and other colleagues involved in these topics. They mix traditional lectures with hands-on and practice oriented sessions. The schools were student-centred, providing participants opportunities to discuss and develop their own projects, either as doctoral students or as postdocs/early-career researchers. The participants in these training events included academics, graduate students, policy makers, civil servants and other stakeholders in the area of quality of life, work and labour markets, and social policy. The sessions included in general keynote lectures on the core themes by experts (visiting teachers), hands-on exercises and participants had the possibility to present their own work. The following trainings were organised:
• Advanced poverty research: Poverty and social exclusion in three dimensions: multidimensional, longitudinal and small area estimations (21-25/07/2014, UNIPI-DEM); Intergenerational and life-course transmissions of poverty (UBremen, 10-15/09/2014); Poverty and material deprivation dynamics (6-10/07/2015, LISER); Reaching out to hard-to-survey groups among the poor (30/05-3/06/2016, KULeuven);
• Advanced labour studies: The gender pay gap revisited: causes and consequences of horizontal and vertical gender inequalities in the labour market (7-11/07/2014, UvA); Quality of working life and vulnerabilities (1-5/06/2015 & 9-13/05/2016, CEE); Migrant women on the labour market (24-28/10/2016, UvA);
• EUROMOD and microsimulation training: six 2.5 days training (organised by UESSEX or UA); another 2,5 days training on LIAM2-based microsimulation models (LISER, 16-18/11/2015);
• Policy experts training: New skills and occupations in Europe: challenges and possibilities (25-27/11/2013, CEPS), Intergenerational inequalities (03-05/11/2014, CEPS), From Uber to Amazon mechanical Turk: non-traditional labour market driven by technological and organizational change (23-25/11/2015), Theory and practice of programme evaluation (LISER, 9-10/11/2016).
A total of 8 five-days schools and 11 2.5 days training events were organised. We received in total 1186 applications to attend these training events. 462 got the opportunity to attend the training events and networked with 169 speakers/experts. The original target was to have 450 attendees. This target has thus clearly been attained. Participants (trainees and speakers) came from 47 different countries. 54 participating scientists came from outside the EU.
2.2.2 Expert workshops
Fourteen expert workshops were organised (two more than planned). The innovative activities of the joint research activities provided the key input for these workshops, which were an opportunity to identify and discuss key technical issues and possible solutions in particular areas of the InGRID infrastructure with a broader group of stakeholder experts. The following topics were dealt with:
• Poverty & living conditions: Indicators specific vulnerable groups, survey protocols hard-to-reach and hard-to-identify groups; high-precision census maps, visualisation;
• Working conditions and vulnerability: New tools measuring occupations; web crawling data, linked employer/employee data, job quality indicators;
• Social policy: Euromod improvements (child care policies, hypothetical household tool); local statistics; new types of indicators; minimum income protection research; big data.
Input from the high-performance statistical quality management JRA were streamed at several places into the workshops.
498 experts attended in total the different workshops throughout the project period. The original target of 312 was thus largely achieved. Participants came from 36 different countries. 5% of the experts were from non-EU countries.
2.3 Joint research activities
As already explained, the joint research activities (JRA) to improve and expand the InGRID research infrastructure have been tackled in four different work packages.
2.3.1 Poverty and living conditions
Poverty trends in the European Union Member States are in many instances still disappointing and it is doubtful whether the European countries by the end of this decade will reach the poverty reduction target set in the Europe 2020 Growth Strategy. Poverty and social inclusion are complex, multidimensional issues, something that clearly is highlighted in the Social OMC, which is still part of the soft governance of the EU. Here, a wide set of indicators are used to monitor and define targets in relation to several vulnerable groups, such as children, youth, elderly, migrants and Roma. In order to keep up with the sometimes rapid re-organisation of the European welfare states, for example, following the 2008 global financial crisis, it is important that monitoring and target setting in European policymaking are based on solid, flexible, standardised, and up-to-date data infrastructures and analytical instruments. The JRA activities of the InGRID poverty pillar worked in this regard on two innovations: a new integrating database and a mapping method.
2.3.1.1 Integrated poverty and living conditions indicator system (IPOLIS).
While they are invaluable for constructing material well-being indicators, both EU-SILC and EU-LFS, two prominent Eurostat data sources, fail to produce appropriate data to answer some non-standard questions. They are particularly not well-suited to map comparatively the situation of a series of vulnerable groups. To improve the RI for analyzing and monitoring the situation of these most vulnerable groups InGRID JRA started to build an integrated poverty and living conditions indicator system (IPOLIS) (Deliverable 20.1: Gabos & Kopasz, 2014; Deliverable 5.10: Gabos & Toth, 2017).
This IPOLIS (ipolis.tarku.hu) is the core outcome of the work package on innovative tools and protocols for poverty and living conditions research of the InGRID project. IPOLIS fits within the frame defined by the overall objectives of the project in many respects.
• Material living conditions and poverty and social exclusion in particular (also as defined by the Europe 2020 strategy target), stays at the core of the integrated indicator system.
• IPOLIS is conceived as an innovative tool by including interactive data visualization.
• It allows not only researchers, but also the broader stakeholder community to follow the situation of most vulnerable groups.
• It builds mainly on the European Statistical System, while other data sources are also considered as inputs.
Nine specific vulnerable groups are identified from the IPOLIS perspective:
• easy-to-reach groups: (a) children (0-17 years), (b) youth (15-30 years) and (c) older people (65+ years);
• hard-to-identify groups: (d) migrants, (e) Roma, (f) travellers;
• hard-to-reach groups: (g) institutionalised people, (h) undocumented immigrants and (i) homeless people.
As a start, a full and integrated monitoring database (with data visualizations) has been produced for the easy-to-reach groups (children, youth and elderly).
IPOLIS is built on the quality of life concept, which includes the concept of poverty and living conditions, but does not fully cover what we understand as (subjective) well-being. The multidimensional nature of IPOLIS is represented by six domains of Quality of Life: material living conditions, labour market attachment and work-life balance, education and training, health and risk behaviours, social connectedness and participation, physical environment and safety. These six domains are completed by policy and context indicators. IPOLIS ensure a coherence across modules of the indicator system by providing direct linkages between modules through common indicators at the level of domains, components and sub-components. These indicators allow for a comparative assessment of the relative position of vulnerable groups – primarily according to the dimensions of poverty and material living conditions. A set of indicators, referred to as cross-module indicators, characterizes all three groups. These measures have the same definition and preferably should be produced on the same data source. Household level indicators, like household income and material living conditions, meet these criteria.
A code book and guidelines have been compiled to facilitate the use.
3.2.1.2 Methodological and data infrastructure reports
Related to IPOLIS, methodological and data infrastructure reports have been produced for easy-to-identify and easy-to-reach vulnerable groups: children (Gabos & Kopasz, 2015), the youth (Schäfer, Zentarra & Groh-Samberg, 2015) and the elderly (Kopasz, 2015). In addition, a similar report has been carried out for policy and context indicators in IPOLIS (Limani, 2017). These papers reflected each time on the feasibility of including these vulnerable groups in IPOLIS. Working papers on the data infrastructure available for the Roma (Bernat & Messing, 2016) and the (undocumented) migrants (hard-to-identify) (Wets, 2017) have been also produced. A final paper investigated reaching out to hard-to-reach groups from the perspective of survey protocols, statistical issues and research design (Schepers, Juchtmans & Nicaise, 2015). We summarise here in brief the main results of the Roma and hard-to-reach groups papers.
Evidence-based policy making aimed at Roma inclusion faces serious limitations, because basic information is lacking about Roma people’s social and economic situation (Bernat & Messing, 2016). The paper addresses issues related to measuring Roma inclusion and the outcomes of EU wide and national policies aiming at this population. In this effort the paper first outlines the political and conceptual framework and describes availability and limitations of data on Roma populations. In a next section it comparatively overviews indicators of Roma inclusion applied by National Roma Inclusion Strategies in 16 EU Member States and critically assesses them. Finally, it proposes alternative indicators that reflect challenges of Roma inclusion in a comprehensive, multidimensional way and discusses the possible data sources as well as gaps in available data.
Schepers, Juchtmans & Nicaise (2015) show which difficulties may arise when surveying hard-to-reach groups, in particular populations living in poverty. In general, the reasons that explain why these difficulties appear can be subsumed in three broad, but interrelated categories:
• as a result of specific characteristics of the target group (for instance mobility, linguistic barriers);
• as a result of the research design (for instance the use of ambivalent concepts, culturally inappropriate questionnaires) or;
• as a result of the interaction between researcher and (potential) respondent (for instance mistrust, communication problems).
In order to overcome these difficulties and to reach a hard-to-reach group living in poverty, researchers should carefully consider the aim and the design of their research and anticipate some ethical issues (for instance gaining consent of vulnerable groups), even before taking any steps in contacting the target group. Furthermore, the complex and multilayered nature of the difficulties discussed requires multiple, multidimensional and innovative strategies or approaches. In this regard the paper proposes five research strategies to overcome trust barriers between researchers and respondents in poverty research: doing (community-based) participatory research, recruiting peer researchers, data collectors or interviewers, providing training, support and supervision for interviewers, developing culturally appropriate questionnaires and innovative data collection methods, such as non-verbal methods and methods using digital technology. When applied thoroughly, the strategies look promising in enhancing communicative success with hard-to-reach groups living in poverty and thus increasing the validity of research results. However, it has to be noted that successful implementation of these strategies may require more time, commitment and thus budget.
2.3.1.3 Experimental high-precision poverty mapping
Estimating economic indicators is crucial for achieving a targeted implementation of welfare policies. However, for such policies to be effective, policy makers must have access to a detailed picture of deprivation that goes beyond aggregate estimates at the country (national) level, extending to finer geographical levels and to other domains of interest, such as specific groups of individuals. One possible solution for obtaining accurate indicators at finer spatial scales is by using small area estimation methodologies. The term ‘small areas’ is typically used to describe domains (e.g. geographic areas) whose sample sizes are not large enough to allow sufficiently precise direct estimation, i.e. estimation that is based only on the sample data from a domain. In such cases, model-based estimation procedures can be considered for improving the precision of the direct estimates. More recently, some of the research effort has been shifted towards methods for estimating poverty (deprivation) indicators at the small area level, also known as poverty mapping. Poverty mapping can offer a detailed description of the spatial distribution of poverty and inequality within a country. It combines individual and household survey data with Census/administrative data with the objective of estimating welfare indicators for geographic sub-areas. In recent years, a range of alternative model based small area methodologies for poverty mapping have been proposed. In the InGRID paper the so-called Empirical Best Prediction (EBP) model suggested by Molina and Rao (2010) is adopted. Using the most recently available data from European censuses and the EU-SILC household surveys, the main goal of this exercise has been to generate poverty maps (via the EBP method) with the highest geographical precision that the available data permitted, in as many countries as possible (see: Esteve et al., 2017; deliverable 20.3).
2.3.2 Working conditions and vulnerability
Employment growth in Europe has been a policy aim since the establishment of the European Employment Strategy (EES) at the Luxemburg Employment Council 1997. The aim of improving the quality of jobs was introduced into the EES at the Lisbon summit in 2000, summarised in the widely cited objectives of ‘more and better jobs’. In the Europe 2020 agenda these objectives are still prominently on the agenda. The Employment strategy has been recently re-worked to stress the challenge of inclusive labour markets (2012).
2.3.2.1 Monitoring new jobs and skills needs
A specific concern of EU employment policies is expressed in the New Skills for New Jobs initiative and the future European Skills, Competences and Occupations taxonomy, ESCO, that will serve as a multilingual framework. For policy measures, particularly in the field of education, these forecasts use to aggregate occupational categories, which one has tried to counterbalance by sector-based qualitative studies. The challenge to identify which new jobs are emerging and what the related skill requirements are, nevertheless remains. The lack of tools to identify new jobs and the related new skills at various levels of aggregation is the greatest data challenge for the years to come. InGRID JRA in the working conditions pillar contributed to tackle this challenge by on the one hand investigating how occupations can or should be measured in a harmonised way especially in (web)surveys and on the other hand explored how web crawling data could be used to monitor trends in jobs and skills.
a) Improved (survey) measurement of occupations
Deliverable 21.1 (Tijdens & Vesintin, 2017) contributes to this measurement issue of occupations and is based on a series of papers/milestones (Tijdens, 2014; Tijdens, 2015; Tijdens & Vesintin, 2016). It provides first a review of the measurement of occupations in surveys in Europe. It specifies how occupations are measured in web surveys, and outlines the methodology currently used to test the comparability of the job content and skill requirements in occupational titles. The results of the validation efforts are presented, including the design of a project to measure occupations on a global scale. Occupation is a key variable in socio-economic research, used in a wide variety of studies. Where such studies use quantitative approaches, they usually rely on survey data. In this paper an inventory of 33 surveys is analysed with respect to the phrasing of the question. The vast majority uses an open text format for the occupation question, but the phrasing of the question is different across almost all surveys. In an additional question, half of the surveys ask for a job description, and again the phrasing varies largely across the surveys. Coding of the open format question is usually (semi-) automatic, survey agencies applying dictionary approaches for automatic occupational coding. In web surveys closed survey questions can be asked using text string matching and search trees for navigating. Recently, machine learning algorithms appear to be a promising development, requiring a substantial amount of manually coded occupations to be used as training data for the automatic classification. A huge training set is required for an auto-coder to apply machine learning algorithms. The deliverable details a design to develop such a training set in a multi-country approach.
Occupation is a key variable in socio-economic research, used in a wide variety of studies, but the measurement of occupations is a major challenge. In case of open-ended survey questions, web surveys pose extra challenges, as respondents are more likely to key in odd text compared to other survey modes, particularly those with an interviewer. Web surveys however offer new opportunities for closed survey questions with self-identification of occupations, particularly when using semantic matching and a database with large numbers of coded occupational titles. A second part of InGRID deliverable 21.1 details the design requirements for such a database to facilitate surveys in multiple countries. The basis is formed by the WISCO Database of Occupations that is related to the ISCO occupational classification. It also reviews how occupations can be measured in web surveys in open-ended or closed survey questions. For the closed question three approaches are detailed, notably scrolling, search trees and semantic matching. Given that any national labour market easily cover 10,000 or more occupational titles, the semantic matching is considered the best approach provided that the look-up tables include 5,000 or more occupational titles for each country. Finally, the Section details the design principles in the database used for semantic matching. It discusses the size of the source list in relation to the ISCO coding. It details the issue of male and female titles and the use of start or end years for occupations It specifies how the database copes with highly aggregated occupational titles such as clerk or manager, with abbreviations and organisation specific job titles, with synonyms, with skill levels within and across occupations, with occupations in the corporate hierarchy, with composite jobs, with handicraft workers, and with subsistence farmers and hunters. Next, it details how to cope with respondents indicating that their job title is not in the database. The Section ends with a future outlook for a program to test the validity and reliability of occupational measurement.
A last part of the deliverable 21.1 addresses the question whether occupations refer to the same work activities, as assumed in occupational classifications such as ISCO-08. Up to now, no large-scale empirical testing of this assumption has been conducted, whereas occupations are a core variable in socio-economic research. Using the task descriptions provided for all ISCO-08 4-digit occupations, the frequency of task implementation was tested using respondents in the multi-country, multilingual WageIndicator web survey on work and wages in 13 countries. The web survey targets individuals in the labour force. Depending on their self-selected occupation, the relevant task list was shown and respondents were asked to tick on a 5-point scale how often they performed each task. For 427 occupations (ISCO-08 4-digits) in total 3,237 occupation-specific tasks were available. Between November 2013 and August 2015 33,678 respondents had completed the tasks questions for their respective occupations. The results show among others that task measurement is feasible because it can generate sufficient observations to allow for analysis for a range of detailed, 4-digit occupations and how it can be related to wage data.
b) Web-based monitoring of new jobs and skills
New jobs and skills have been high on the agenda of policymakers for several years now, as evidenced by the Europe 2020 strategy, the ‘New Skills for New Jobs Agenda’ and other policy documents. At the same time, current approaches used to capture and understand these dynamics may not fully meet their goal. Deliverable 21.2 (Beblavý et al., 2017) of the InGRID project brings in this regard together four chapters that CEPS prepared. Each of these chapters aspired to shed light on the identification of new occupations and skills, from different angles. The report explores methods of identifying the dynamics of new or emerging occupations, jobs, tasks and skills and especially investigates the potential of web data in this area.
The first chapter of the report reveals that while there is an extensive literature on new jobs and skills, which extends beyond academic research, many issues remain obscure. The conceptualisation of new occupations and new skills often is unclear, imprecise or narrow in scope (yet, concepts like occupation, job, skill and task are clearly defined, but are not always easy to disentangle in practice). Only a handful of contributions explicitly study these concepts, their meaning and ways to identify them. The first chapter further reveals that generally new jobs (and skills) are identified through surveys, interviews, classifications, forecasts, trade literature or other data sources.
The second chapter of the report builds on the findings of the first chapter. It evaluates the data sources and methodologies that are commonly used to capture new occupations and skills. It starts with a thorough discussion of the strengths and limitations of these methods and data sources, pointing to issues related to timing, precision and scope. The report then explores what alternative methodologies and data sources could be used to overcome these limitations. More specifically, we turn to web data and carefully explain how these sources could be used and what advantages they could bring. Web data are a very promising source but their use is still limited. In addition, a lot of methodological research is needed to further support the development of this research field.
The third chapter presents the results of pilots carried out for InGRID. It shows what information can be derived from online job boards and their vacancies with regards to language skills, IT skills and other education, skills or related requirements in labour markets in Europe and the US. The chapter also presents the findings obtained from an occupations observatory prototype, which builds on metadata extracted from online job boards, and their occupational classification in particular. One of the pilots is highlighted in the fourth chapter of the report, linking educational requirements in vacancies to the educational attainment of jobholders to assess skill mismatches in the Czech Republic.
2.3.2.2 European measurement of working conditions and vulnerability-at-work
In relation to the comparative study of job quality, labour market inclusiveness and vulnerability, the InGRID JRA focused on two main issues.
a) Inventory of the data landscape
On the one hand, and as a first step towards further data integration in the targeted and serviced comparative research area of working conditions, inventory and assessment studies were conducted. Deliverable 21.4 (Smits et al., 2017) integrates these studies.
In an analytical approach of cross-national and cross-sectoral investigating a series of job dimensions and connecting these to the policies and behaviours of a series of actors, survey material is very important. A first chapter/paper (Szeker & Van Gyes, 2015) focuses as a consequence on the available working conditions surveys in Europe. Seventeen active national and transnational surveys are inventoried on their scope, methodology and quality control procedures. This evaluation shows than only 10 have already a longer tradition and even less have a sort of cohort panel approach. A lot of variety reigns this landscape of national working conditions surveys in Europe. The chapter concludes with the observation that only a small quality group of established surveys can be distinguished in Europe. This small group is situated in North-western Europe and includes Denmark, Sweden and Finland as the Nordic countries accompanied with the Netherlands, Germany and France. In other words only 6 EU-countries have a well-established and high-quality national tradition of working conditions surveys
In the survey field of comparative research on working conditions in Europe, the European Working conditions Survey (EWCS) is – especially when taking this ‘poor’ national landscape into account - a key resource. Since its launch in 1990 the European Working Conditions Survey has provided an overview of working conditions in Europe. Acknowledging this highly valued central position of the EWCS data in the InGRID community, the second chapter/ paper of this InGRID report (Vercruyssen & Van Gyes, 2016) implies as an exercise – positive, but critical – the Total Survey Error evaluation framework to the EWCS data collection. The question whether there are (maybe) methodological elements of improvement to detect, is tackled from this methodological ideal type quality framework. This evaluation confirms once again that the EWCS is currently the most complete source of information on working conditions and job quality in Europe. Nevertheless, some methodological points of attention are stressed (e.g. sample sizes, response remote areas, stability in variables).
Linked employer-employee data can fill an important gap in the set of data used to study working conditions, shedding light on questions that cannot be addressed using firm- or individual-level data alone. The inventory by Greenan & Seghir. (2016) illustrates the two forms of linked surveys (linked employer-employee and linked employee-employer) and their advantages and drawbacks. Fifteen national surveys from eight countries are identified in the inventory. All-in-all it is again concluded that the concept of linked data is still not widespread, despite its richness as tool to survey and analyse working conditions in its organizational and managerial context.
The inventory of comparative policy databases (Smits et al., 2017) – presented in the fourth chapter of this report – shows that a series of comparative text bases exist on especially OSH policies. Legal textual databases are provided in particular by the International Labour Organisation (ILO). Company surveys of European agencies complement this picture with information on company management policies. However, distilling from this information, comparative indicators on policies is largely in its infancy.
When looking to the field of comparative data to measure working conditions and related policies in Europe, this report inventoried thus the available resources. It stressed the advantages of linked employer-employee data; the need of large, longitudinal, cohort samples; and the required construction of comparative policy indicator databases. The conducted inventories and assessments learned in this regard that the current reality is one of a ‘worrying’ patchwork of (national) data sources. Harmonisation is very low and comparability between countries and over time is seriously hampered. A majority of the EU Member States have no organized and structured high-quality survey tradition in relation to working conditions, job quality and vulnerability-at-work. Existing initiatives struggle to continue. The current, first and only stronghold is especially the European Working Conditions Survey, organized by the European Foundation for the Improvement of Living and Working Conditions. Other interesting base resources are the text databases of ILO and others on regulations and policies and academic efforts to harmonise classification schemes and to compile indicators on social dialogue. Nevertheless, the conducted inventories learn that there is an urgent need for a coordinated, transnational initiative on ‘modernisation’ of European working conditions data.
b) Indicators of vulnerability-at-work: methodological innovations
Deliverable 21.3 is based on the joint work done by HIVA-KU Leuven and the CNAM-CEET on indicator methodology regarding the issue of vulnerability to poor working conditions and OSH issues in European countries. In this study, we propose two different but complementary approaches to define and to identify vulnerable groups to poor working conditions and OSH issues across European countries. The first approach sets out a typology of jobs relying on various relevant aspects of working conditions. The outcome is an improvement guide for each job type. Using a methodology developed to study vulnerability to multiple deprivations, the second approach proposes a measurement frame for assessing vulnerability to adverse working conditions. Such a framework would be part of a warning system to identify employees at risk in workplaces. Both approaches relies on the most reliant source of information regarding working conditions in European countries, namely the European Working condition Survey (EWCS) performed by Eurofound.
The first chapter (Szeker, Smits & Van Gyes, 2017) starts from the job types’ methodology proposed by Holman (2012) to capture and better understand the complex and multidimensional nature of job quality. A typology of seven job types is made relying on the latest data of the EWCS (2010 and 2015) and using a broad set of job characteristics such as complexity, autonomy, voice and wage. In the second part, the relation between the job types and subjective workers’ outcomes (such as job satisfaction, perceived work related health problems, feeling of job insecurity, etc.) is examined and trade-offs between the job characteristics impacting workers’ outcomes are discussed.
The second chapter (Greenan & Seghir, 2017) applies a methodology originally developed to study vulnerability to poverty to measure vulnerability at the workplace. Vulnerable workers are defined as carrying the burden of working under the threat of adverse physical and psychosocial working conditions. Vulnerability is thus a forward-looking concept that allows identifying workers that are the most exposed to work resource deprivations and more generally to ill-being at the workplace. The proposed methodology allows tracking cohorts of employees from 1995 until 2015, comparing the level of vulnerability across European countries and identifying the characteristics of vulnerable groups.
The third and final chapter (Vercruyssen & Van Gyes, 2016) assesses the possibility of an international standard for scale construction with the fifth European Working Conditions Survey (EWCS) data. As both previous chapters rely on this survey to tackle issues related to job quality and quality of working lives, the reliability of used data and scales are of primary importance. This chapter takes it a step further in discussing methodological issues related to the construction of one internationally applied scale for measuring job quality and its various aspects. Harmonisation of data on item level and on scale level are key to this.
2.2.3 Policy analysis tools and instruments
Policy evaluations require micro-level data of high quality, but need also reliable and systematic information concerning the institutional arrangements. In order to account for individual level outcomes in terms of poverty, living and working conditions, it is necessary to incorporate these arrangements of social and labour market policies into the analytic framework. A particular pillar of InGRID invested time and resources in improving the comparative evaluation tools in the field of social policies. In the area of social policy, it is common to use large-scale comparative research designs and various forms of quantitative analytical techniques to analyse data. Typically, researchers try to include as many countries as reasonably possible, and to use as many time points as readily available. However, lack of relevant and reliable comparative social policy data has hitherto constrained research, thus seriously restricting possibilities for comparative welfare state analysis. InGRID WP22 on innovative solutions for comparative policy indicators and analysis worked in this regard, besides conducting some inventory work, on the extension and improvement of a key InGRID infrastructure, namely EUROMOD. In addition, work was conducted on the integration of minimum wage databases and the innovation of indicators on out-of-work benefits.
2.2.3.1 Inventory of social policy databases
Also in these JR activities, the first goal was to provide inventory. Deliverable 22.1 (Doctrinal et al., 2017) provides an inventory of core social policy databases and indicators for comparative research. 26 databases and infrastructures are mapped that fruitfully can be used in comparative research to analyse the causes and consequences of social policy. Each database is compared according to a set of characteristics, including type of data (expenditures, institutional indicators, beneficiary statistics, socio-economic/income surveys, micro-simulation), policy areas included (cash benefits: family benefits, unemployment benefits, sickness benefits, pensions, work-accidents, social assistance, and disability/invalidity/survivors benefits; public services: child care, health care, elder care, and active labour market policy), countries and years covered, as well as interval for updating of data.
2.2.3.2 Improving and extending EUROMOD
EUROMOD provides the baseline for the InGRID contributions in this pillar. EUROMOD is a state-of-the-art policy analysis tool that allows studies on the functioning of and interplays between different types of tax and benefit programmes, thereby making in depth distributional policy analysis possible. EUROMOD links micro-data from household surveys and policy regulation (codified into analysable units) in a single user interface. EUROMOD can be used to answer a wide range of counterfactual questions related, for example, to the effects of policy reforms in terms of poverty, inequality, incentives and government budgets. EUROMOD can also be used to study, for example, how demographic and labour market changes affect the functioning of policy. EUROMOD is now expanded to cover all EU Member States.
a) Hypothetical household tool
In order to evaluate the outcomes of policies and to provide relevant information for evidence-based policy making, a proper understanding of the functioning and interaction of policies is required. Taxbenefit hypothetical household simulations are an important tool for better understanding the interactions between tax-benefit policies and for detecting trends in tax and benefit levels. These simulations are based on a set of well-defined hypothetical households (or so-called ‘model families’). The hypothetical households are defined on the basis of a clear set of assumptions with regard to the relevant parameters to examine tax liabilities and benefit entitlements (e.g. household composition, place of residence, health status, employment status, etc.). On the basis of these sets of characteristics, a microsimulation model calculates tax liabilities, benefit entitlements and disposable income for the specified hypothetical households. Subsequently, net disposable incomes and the level of relevant income components can be used in further analyses. The primary advantage of tax-benefit hypothetical household simulations is their illustrative function: they are easy to understand and show how tax-benefit policies work and interact. In addition, hypothetical household simulations can be used to construct indicators for evaluating the adequacy, fairness and work incentives of tax-benefit policies. They are also a very convenient tool for crosscountry comparisons, on the condition that assumptions are equivalent across countries. Also, when microdata are lacking for specific vulnerable groups (due to small sample sizes), hypothetical household simulations can provide further insights.
Currently, hypothetical household simulations are brought together in databases such as CSB-MIPI and SaMip. The main disadvantage of these databases is that they contain information only on a restricted number of hypothetical households, and that it is not possible to change the underlying characteristics, or policy rules (e.g. to evaluate the effects of a policy reform). In contrast, the OECD has developed a tax-benefit calculator, which allows for specifying more flexibly the characteristics of hypothetical households. However, for complex household types, the model is not very user-friendly. In addition, if one wants to complement the results with microsimulations, one needs to fall back on other tax-benefit models which may not provide consistent results. To innovate this approach, a new Hypothetical Household Tool (HHoT) has been developed as part of the InGRID project. The unique features of HHoT are: (1) its flexibility and user-friendliness for specifying hypothetical households; (2) its integration in a regularly updated, validated and comparative microsimulation framework, namely EUROMOD.
A beta version of HHoT (Deliverable 22.2: Hufkens et al., 2016) is currently available to researchers in UA and UEssex and the participants of the two UA InGRID Winter Schools for testing; and will also be made available to experienced EUROMOD users on request. Continuing and future work will not only extend the number of checks but also, based on these, produce a manual explaining the use of HHoT in more detail. Due to the flexibility of the HHOT to generate households with a wide range of characteristics, it has to be extensively explained why the user obtains results that are different from those generated by the existing tools. A public release will follow soon afterwards.
b) Child care policies
In a second task, the teams at the University of Essex and the University of Antwerp collaborated to extend the policy sheets of EUROMOD to include childcare arrangements (Deliverable 22.4: Hufkens & Verbist, 2017). The childcare policies are included in a version of EUROMOD for the selected countries (DE, FI, SE, NL, LT and PT). For LT and HU the simulation is included in the policy year 2014, for the other countries policies are modelled in 2015. EUROMOD is extended by calculating the parental fees for formal child care and other possibly related measures such as the tax treatment of these fees. The policies are added to EUROMOD G3.0+ using EU-SILC 2012 as input data. Description of the country policies and the assumptions to model these policies in EUROMOD are reported in D22.4 (Hufkens & Verbist, 2017). The childcare policies are currently included in a EUROMOD version at the UA. This version can be used by visitors to UA. The UEssex and UA are discussing how best to include this EUROMOD extension in the Essex master version of EUROMOD.
2.2.3.3 Empirical demonstration comprehensive indicators out-of-work benefits
The purpose of deliverable 22.3 (Doctrinal, Nelson & Siren, 2015) is to discuss new and innovative ways of measuring income replacement in out-of-work benefits that help us to formulate and test more precise hypotheses about the causes and consequences of welfare states and of social policy. Replacement rates are commonly used in research to evaluate the performance of cash benefits, for example, in the areas of unemployment and social assistance. However, replacement rates are often based on calculations that involve too simplified assumptions, thus seriously restricting the validity of results. Comparative social policy research requires more detailed and accurate indicators on the institutional design of public programs for income redistribution.
Based on OECD data we present a new comparative dataset that includes a variety of synthesised measures on income replacement in outofwork benefits. All of these synthesised measures on income replacement are published in the Out-of-Work Benefits (OUTWB) dataset, which is part of the Social Policy Indicators (SPIN) database at the Swedish Institute for Social Research (SOFI), Stockholm University (http://www.sofi.su.se/spin/). The OUTWB dataset includes 39 countries for the years 20012011. Focus is on measures that capture the overall replacement rate of outofwork benefits as well as measures that describe the distribution of income replacement across earnings-levels. Whereas the former dimension is often used in analyses of social policy and income distributions, albeit measured in more simplified ways, the latter dimension is often neglected in comparative research.
The deliverable is organised as follows. First possibilities and pitfalls associated with various types of social policy data are discussed. Thereafter it is explained how income replacement in out-of-work benefits can be more accurately measured. In the following section some early descriptive results are presented using the new data on overall replacement rates and progressiveness of income replacement in out-of-work benefits. The deliverable ends by offering some conclusions and suggestions for further infrastructure research on this matter.
2.2.3.4 Integration of minimum wage databases
In social policies the pre-distribution instead of the re-distribution agenda is getting rising attention. In this regard work was conducted by the University of Amsterdam in deliverable 22.5 (Tijdens, 2017). Minimum wages are in this regard of high relevance for combating poverty. Twelve databases with information for 2011-2015 about MW fixing mechanisms and their coverage were merged, resulting in a database for 195 countries, of which 97 with data covering all five years. The latter showed that between 2011 and 2015 the percentage of countries with a SMW policy increased from 92% to 94%. The results are documented in D22.5 (Tijdens, 2017).
2.3.4 High-performance statistical quality management
In the world of the InGRID research community statistics plays an important role. Analysing quantitative data and building comparative indicators involves often a series of statistical operations and treatments. High-tech statistical techniques can in these research activities be of particular help to overcome methodological quality issues and problems (e.g. related to sampling, non-response, mathematical basis of used analytical technique, multidimensionality of indicators, etc.). Based among others on knowledge exchanges and discussions with other InGRID partners (and beyond) specific areas were identified in InGRID WP23 to investigate and experiment with new high-performing statistical techniques as a ‘helping hand’ to tackle methodological questions raised in the InGRID field of study. D23.1 (Case studies; Berger et al., 2016) and D23.2 (Future needs in statistics; Articus et al., 2017) focusses in particular on samples beyond classical sampling theory (missing values, non-probability samples, and big data), multidimensional indicators as well as regional indicators and advances in small area applications.
2.3.4.1 Multidimensional Indicators
a) Principal Components Analysis and Complex Survey Designs
The work (Smith & Shlomo, 2016) identifies common social science approaches for dimensionality reduction and latent class analysis using principle component and factor analysis. These approaches work well under simple random sampling survey designs but may have serious bias under complex survey designs. In particular, disproportionate stratification with unequal inclusion probabilities and clustering are the main causes of bias when using these approaches. Using simulation techniques an in-depth exploration of the underlying causes of bias was carried out. The conclusions from the research were that survey weights arising from disproportionate stratification must be accounted for in the estimation of the underlying population correlation matrix which underpins these approaches but the impact from clustering had little impact on potential bias.
b) A multidimensional approach to measure children's living conditions
In this study (Giusti, 2016) the UNIPI-DEM team gained insights into Italian children’s living conditions and deprivation of capabilities using the Capability Approach, an alternative normative framework for the evaluation of human development, well-being and freedom by thinking in terms of human functionings and capabilities. The idea of the study was born upon the InGRID TNA visit of Dr Antoanneta Potsi at the Department of Economics and Management of the University of Pisa. From a methodological point of view, an approach based on a fuzzy methodology is applied to data from the EU-SILC 2009 ad-hoc module on children. The use of this methodology makes it possible to preserve the richness of the data available from the EU-SILC survey that include both monetary and non-monetary aspects of children deprivation. The fuzzy methodology is combined with the capability approach at a disaggregated level of analysis by three social economic factors (single parent household, household educational level, macro-region of residence). Besides the well-known Italian North/South disparity of financial indicators confirmed also for households with children the findings suggest a new duality for Italian children quality of life, given by the multidimensional domains of deprivation internal or external to children’s households.
The work was then extended by including into the analysis other mainly Southern European Countries (PT, IT, IE, EL and ES) (Guisti, D’Agostino & Potsi, 2017). The “common framework” that these five countries shared in time of crisis stimulates interesting analyses aiming to measure and compare the living standards of children in different households across these five countries as well as a more general discussion on the social impact that the economic crisis had on this particular vulnerable sub-group of the society. In this analysis, on a methodological point of view, the UNIPI team proposed a simplified Jackknife variance estimates for fuzzy measures of multidimensional poverty and extended fuzzy measures of poverty to the longitudinal perspective (Betti, Gagliardi & Verma, 2017).
2.3.4.2 Non-response and Imputation
a) Compensating for missing data in cross-sectional and longitudinal data
An in-depth, extensive and comprehensive literature review (Cernat, Byrne & Shlomo, 2016) with examples of how missing data should be compensated for in both cross-sectional and longitudinal data, taking into account the missing data mechanisms of missing at random (MAR) and not missing at random (NMAR) was carried out in the first deliverable (D23.1). Different solutions for correcting for missing data were presented and compared. In the second deliverable (D23.2) (Byrne, Shlomo & Chandola, 2017), compensating for missing data in longitudinal data using a standard approach of multiple imputation is compared with a new approach which preserves the hierarchical nature of longitudinal data. For example, imputation across waves should come before imputation across data subjects. It is concluded that more research and development is needed before determining the optimal method of imputation for longitudinal data. It is nevertheless suggested that this further research and development should focus on imputation methods that preserve the hierarchy inherent in the data as well as on methods compensating for informative missingness.
Berger (2017) extends an empirical likelihood design-based approach for multiple parameters to a case of unit non-response in a multi-stage sample design. Empirical likelihood has been mostly developed under the assumption of independent and identically distributed observations. This assumption is violated under multi-stage sampling. The proposed inference approach takes as a consequence into account the design, the response mechanism, the randomness of the estimated response propensities and the population level information. This confidence interval does not rely directly on the normality of the point estimator and does not require variance estimation, linearisation, re-sampling and estimation of a design effect.
b) Compensating for measurement errors arising from disclosure limitation methods
Perturbation of data is increasingly being used as a disclosure limitation method to enable open access data (Goldstein & Shlomo, 2017). The most common disclosure limitation method is additive random noise as well as the production of synthetic data. For simple linear regression, one can correlate additive random noise to the original data and ensure unbiased estimation of the regression parameters. However, this is not so simple under generalized linear models and multilevel models or when data has been synthesized using a range of imputation techniques. D23.2 chapter three (Goldstein & Shlomo, 2017) assess methods for integrating out the noise under a Bayesian framework to enable unbiased estimation of model coefficients for the case that the parameters used to generate the noise or synthesize the data are known.
2.3.4.3 Small Area Estimation
Several research topics were pursued under the broad heading Small Area Estimation.
At SOTON, work was done on the use of multivariate kernel density estimation methods for estimating population with specific characteristics and all for the presence of measurement error. One key advantage of this methodology is that it allows for a continuous type of geography as opposed to a discrete pre-specified geography which is the convention in many small area applications. Support was also provided to the work from CED-UAB on high-precision poverty mapping (see supra) (Tzavidis, Permanyer & Koksel, 2016).
SOTON further worked on identifying future challenges (Tzavidis, 2017). To start: access to micro-data becomes more difficult. Hence, more work on area-level models especially for estimating non-linear indicators is needed. The use of model diagnostics and approaches to amending the model using transformations, alternative parametric assumptions and robust methods are also discussed. Finally, the use of new forms of data for example, big data and the impact of changes of survey designs on small area estimates are the other two areas of future challenges that have been identified.
At UNIPI-DEM the work on Small Area Estimation in this period followed two main research streams: (1) the robust estimation of income-based inequality estimators for unplanned domains; (2) the use of Big Data in small area estimation.
As concerns the first aim, the focus of the UNIPI-DEM research team was on extending the M-quantile methodology for estimating at the small area level two specific Laeken indicators, namely the income quintile share ratio and the Gini coefficient (Marchetti, 2016). The M-quantile-based small area methodology for estimating the two inequality measures and the related Mean Squared Error estimators was developed and evaluated using Monte-Carlo simulations and by applying the methodology to real data. In the application data from the EU-SILC survey are used, and the small areas are represented by the provinces in the Tuscany region (Italy) cross-classified by the gender of the head of the household. The UNIPI team developed in addition a simulation study using the Amelia Data to compare the performance of alternative estimators of three indicators of monetary poverty defined at the small area level: the mean household equivalised income, the Head Count Ratio and the Poverty Gap (Gagliardi, 2017).
The second focus of the UNIPI-DEM research team was the development of a framework for the use of the so-called Big Data the huge amounts of digital information about human activities produced by a wide range of high-throughput tools and technologies together with small area methods. We suggested three ways to use Big Data together with small area estimation techniques to study poverty and living conditions at the local level: the first opportunity is to use Big Data sources to create local indicators and compare them to those obtained with small area estimation methods; the second possibility is to use Big Data sources to generate new covariates for small area models; the last opportunity is to use survey data to check and remove the self-selection bias of the values of the indicators obtained using Big Data. In D23.1 (Guisti et al., 2016) two real data applications based on EU-SILC data and on GPS data on mobility show how Big Data has the potential to mirror aspects of well-being and other socio-economic phenomena. The work of the UNIPI team continued following the second approach on the use of big data in SAE models, namely the use of Big Data as covariates in SAE models (Marchetti, Giusti & Pratesi, 2017). In particular, the focus was on the use of data coming from the social network Twitter to investigate their potential in predicting the share of food consumption expenditure of Italian households at local level. The work showed that the iHappy indicator derived from Twitter has a good predictive power for the share of expenditure that Italian households devote to the consumption of food and beverages - an indicator that can be used as a proxy to measure households’ living conditions.
Policy support in trans-border metropolitan areas urges the needs of a carefully selected and comparable database. Since, in general, data are provided on national level, immediately the question arises about comparability of data and concepts and their harmonization. This becomes even more evident, when small regions in these metropolitan areas have to be compared. To provide a high-quality basis for policy support in these cases, a so-called cross-border statistics concept has to be built. The needs for this were identified in the Futuring Briefing Note 2, and presented within the second deliverable of WP23 as small area border-statistics problem from Trier University (Articus, Burgard & Münnich, 2017).
2.3.4.4 Measuring level and change
a) Measuring level
Not all indicators on poverty and social exclusion in Europe are based on the EU-SILC. The prerequisite of a continuous measurement of income is not necessarily met by those other surveys. One example is the German Microcensus, which is used regularly to estimate those indicators and provides an income variable, which is classified into 24 groups. Different parametric and nonparametric approaches are considered (Lenau & Münnich, 2016).
To determine the most suitable, the classification structure has to be considered. In general, there is little point in estimating poverty from few equally spaced classes, since bias and variance are simply too high. A higher number of classes and/or higher differentiation for lower incomes are necessary to assess poverty in a suitable precision. Point estimates from parametric income distributions perform better for few or equidistant classes, while nonparametric approaches seem to be superior when more classes of ascending width are used. Moreover, inference seems in general more appropriate for the spline-based estimates, unless their bias is too high.
b) Measuring change
Measuring change over time is a central problem for many users of social, economic and demographic data. These cross-sectional estimates often include imputed values to compensate for item non-response. The estimation of the variance of an estimator of change is useful to judge whether the observed change is statistically significant.
Berger (2016) proposes to use a multivariate linear regression approach to estimate these covariances. The proposed estimator is not a model-based estimator, as it is valid even if the underlying model does not fit the data. It is showed how this approach can be used to accommodate the effect of imputation. The regression approach gives design-consistent estimation of the variance of change when the sampling fraction is small. The proposed approach is illustrated using random hot-deck imputation, although the proposed estimator can be implemented with any other imputation techniques.
2.3.4.5 Use of new data & non-probability sampling
Non-probability samples, such as Big Data or web surveys are increasingly used to overcome problems of classical survey data, e.g. when it comes to small sample sizes for subgroups. Since those data sources might by highly selective and hence biased, inference based on non-probability samples raises methodological challenges. Lenau & Münnich (2017) assess different methods for compensating the possible selectivity of such data. The results indicate that using non-probability samples without adjusting for their selectivity yields severely biased results. Further, using multivariate characteristics of auxiliary data seems in general more promising to reduce the bias than only relying on univariate information.
2.3.4.6 Monte Carlo simulation lab
This task included the testing of the proposed statistical methodology, especially the extension of the synthetic universe AMELIA (Merkle, Burgard & Münnich, 2016), which is used as a research infrastructure for statistical simulation and experimenting. As in practice we have only one sample, Monte Carlo simulations allow to learn about the applicability and performance of different statistical methods. Careful simulation set-ups are crucial since they may lead to interesting behaviour or to discovery of peculiarities which hardly can be found with mathematical proofs and sometimes the observations may lead to proofs. However, there is a problem regarding the availability of appropriate data especially for design-based simulations. This type of simulations requires a given universe from which samples can be drawn. For this reason, an artificial dataset has to be generated making use of sample data. The artificial population should have similar properties as the sample data. The AMELIA dataset was extended and used for testing of the proposed statistical methodology.
2.4 Futuring the research infrastructure
The FP7 project supported the integration, innovation and opening by improving the transnational access, organising mutual knowledge exchange and investigating improvements of the InGRID RI. As part of these activities a strategic thinking on the future of the InGRID RI was developed. This strategic thinking included a users’ needs survey (Szeker & Van Gyes, 2015) and two scoping exercises. Lenau et al. (2016) investigated future needs and challenges from the point of statistical methodology. The note links these challenges on the one hand to the ongoing ‘Modernisation of European social statistics’ and on the other hand to particular issues like ‘non-probability sampling and new types of data; new forms of (dynamic) micro-simulation; and small area estimation and cross-border statistics. Hamon-Cholet et al. (2017) scanned based on first, available material the relationship and interaction the InGRID RI has with the Central and East-European countries, which joined only in the last decade the EU and the European Research Area. In users and available data, these countries are certainly not a missing link in the InGRID activities. However, the core group of the consortium includes only one CEE partner. In addition part of work package 5 examined the e-infrastructure needs of the RI (Lenau et al., 2017). Three research infrastructure forums and a final conference titled ‘Better science infrastructure for evidence-based policies on inclusive growth in Europe’ accompanied this process.
Based on this futuring a short-term innovation agenda was outlined to advance the integrating InGRID research infrastructures:
• InGRID needs to keep investing significantly in the improvement of integrated European databases and analytical technologies that help researchers to understand the comparative structures and patterns of data on poverty, living conditions and vulnerability-at-work;
• Assessing the impact or effect of policies not only requires integrated and comparable microlevel data of high quality on ‘problematic’ situations, but also systematic information concerning the institutional organisation. In order to account for individual level outcomes in terms of poverty, living conditions and working conditions, it is necessary to incorporate the factor of social and labour policies into the analytical framework. InGRID has to invest more on this integration and incorporation of policy data into the RI system (new policy extensions in IPOLIS and EUROMOD, indicators on welfare services, OSHA policies, international indicators on social dialogue institutions and minimum wages);
• Policy uncertainties are growing due to the increased complexity of social phenomena (for example poverty in new household types or new casualisation of work due to technical changes), but also due to the growing value divergence (for example on how to get out of the socio-economic crisis). This necessitates the development of more complex analytic and simulation models to make ‘your point’ as researcher. InGRID needs to provide the related research communities with these necessary tools and instruments and ought to contribute also to the innovation of these tools (in different ways);
• At the European policy-level, benchmarking and assessing indicators is a key practice to stimulate policy debate and to advance policy changes. Helping the related research community by experimenting and innovating this indicator-building is consequently defined as an important aspect of the future InGRID activities. Joint research activities have to be conducted on norm-based job quality indicators, indicators for particular (socio-demographic) groups and statistical standards of multidimensionality or lower-level indicators. Promoting visualisation techniques is another domain that has to be picked up in this regard.
Structural and organisational strategic priorities are in this advancement of the RI integration:
• enhancing the integration and synergies between the ‘Poverty an living conditions’ and ‘Working conditions and vulnerability’ pillars;
• widening access and promoting the infrastructure in newer Member States, where on the one hand the inclusive growth challenge remains very strong, but on the other hand infrastructures engaged in integrating European comparative data and/or indicators on inclusive growth are very scarce;
• facilitating the visibility and access to the infrastructure by developing the current project website into a lightweight research portal;
• looking for the development of this e-infrastructure to other well-established research infrastructures at the European level;
• keep paying attention on the necessary scientific standards of statistical quality and thus including expert activities on this issue in the different activities of the project;
• making detectable progress in the sustainability of the distributed, but integrating InGRID infrastructure.
• maintaining and extending the stakeholdering of the RI with end users (policy innovators) and especially pre users (data providers). Increasing the networking with organisations like OECD, ILO and European agencies are important in these stakeholder activities.
• guaranteeing and promoting ‘open science’ and research ethics.
2.5 Conclusion
The FP7 project has thus integrated, strengthened and opened for the first time the existing, but distributed InGRID RIs on ‘Poverty and living conditions’ and ‘Working conditions and vulnerability’. Opening, widening and innovating the integrating RI, available in 17 different scientific centres, departments and institutes, was the key goal by improving the transnational access, organising knowledge exchange and improving the RI methods and tools.
Highlights of the R&D achievements include:
• the building of an integrated poverty and living conditions indicator system (IPOLIS) for particular vulnerable groups and accompanying visualisation tool;
• recommendations on the data collection of hard-to-identify and to-reach groups;
• the experimentation with high-precision poverty mapping;
• new indicator-methods to identify and map vulnerability-at-work;
• further steps in the harmonised surveying of occupations;
• the investigation of the added-value of web crawling for monitoring new jobs and new skills;
• the worrying assessment of a growing data gap in the European measurement of working conditions;
• improved tools to generate comparative social policy indicators;
• a well-focused enhancement of tax-benefit simulation model EUROMOD with a hypothetical household tool and the domain of child care policies;
• the growing recognition of the necessity of other simulation techniques (dynamic micro-simulation and nowcasting);
• statistical solutions to solve survey sampling problems ex-post in the analysing phase;
• improving techniques to construct small area estimations;
• advancing multidimensional indicator-building;
• exploring the potential use of Big Data from a statistical point-of-view
• showing the added-value of synthetic data sets in statistical simulation.

Potential Impact:
The strategic impact of the EU FP7 project on integrating, improving and opening the InGRID distributed research infrastructure can be identified at five different levels:
• integrated and coordination of research infrastructures of established European Social Sciences Research Areas;
• improved access to and use of the pool of research infrastructures;
• enhanced European added value of research infrastructure capabilities;
• integration and innovation of the research infrastructure based on a two-way impact process with the wider scientific community and the community-of-interest;
• contribution to a more coordinated evidence-based European policy to combat poverty and vulnerability and enhancing coordination and impact of European policies.
3.1 First integration and coordination of research infrastructures developed in previous EU research and existing ERAs
Since decades, the establishment of European Research Areas is a key goal of the European research policy. On the domains central in the call which was addressed by InGRID – ‘Poverty and living conditions’, ‘Working conditions and vulnerability’, European Social policies - longstanding collaboration between the expert institutes throughout Europe is already established. These expert institutes and researchers have now been brought together in a new consortium, making a qualitative leap to a more integrated European Research Infrastructure and to open these data, tools and services more to the broader scientific community. The leading academic community involved in InGRID indeed shares a history of cooperation enabled by a range of previous framework programme projects. Almost all partners involved have been coordinators of one or more FP projects. The added value of these experiences and European orientation is evident from the fact that it was possible to clearly identify the most urgent and pertinent needs for an effectively integrated and coordinated EU-wide research infrastructure. This is not only the result of a shared understanding of the research area, but also a proof of familiarity of doing research at the European, rather than at the national level, in an effective and efficient way.
The FP7 project – the first financing from the EU in the RI programme - played thus a stabilising role in sustaining these data facilities and expert resources. Importantly, the integration included also some additional and highly-valued technical resources. On the one hand: data centres specialising in integrating national data in comparative datasets of high relevance for the InGRID research community, namely LIS on income data and IECM/IPUMS on census data. On the other hand: some key national ‘best practices’ (French working conditions surveys and British household panels).
Hence, a first clear impact of InGRID is that European Research Infrastructure efforts have been focused to precisely those issues that are most urgent and pertinent in the given domain from a statistical, knowledge and policy perspective. A first key impact of InGRID is its contribution to inventorising and addressing the strategic gaps and missing infrastructure links in knowledge, data, methods, indicators and tools which are necessary to enhance the coherence and integration of the defined European Research Area (linked to the third major EU policy strategy, namely Inclusive Growth). Key issues addressed in the ‘Poverty and living conditions’ pillar include the infrastructure focus on the multidimensional character of poverty, the problem of hard-to-identify and hard-to-reach vulnerable groups and the improvement of longitudinal and regional poverty mapping. In the field of ‘Working conditions and vulnerability’, examples are new indicator-methods to identify and map vulnerability-at-work, further steps in the harmonised surveying of occupations, and illustrating the added-value of web crawling for monitoring new jobs and new skills. In the ‘New tools and indicators for policy analysis’ pillar, efforts have been targeted at improving data and tools to generate comparative and institutional policy data and indicators and at the linking of those institutional data to micro-level outcomes in the area of poverty and social inclusion. Further, the well-focused enhancement of EUROMOD with a hypothetical household tool and the domain of child care policies will enable to optimise the dissemination and access of this key data infrastructure. In addition to these domain-related impacts, the statistical partners of InGRID have identified crucial improvements to data analysis methods which enable more reliable, accurate and qualitative data and measures, such as solving survey sampling problems ex-post in the analysing phase, improving techniques to construct small area estimations, advancing multidimensional indicator-building and exploring the potential use of Big Data from a statistical point-of-view.
However, InGRID impacts in another and not less important way on the European Research Areas which address the topics of the call. InGRID has brought together in an unmatched way the distinctive research areas in one project with common goals. InGRID has not only improved and facilitated the better co-ordination, structuration and integration of the respective research infrastructures but has also provided links between the different related research communities and their infrastructure. By a series of networking events and especially three research infrastructure forums and a final conference, a systematic knowledge exchange and integrated learning-from-each-other has been organised. A better measurement and understanding of the needs of vulnerable groups is as a result a common and shared expected impact of the integrating research infrastructure. Examples? The RI community of labour studies learned about the IPOLIS indicator framework targeting specific vulnerable groups, the EUROMOD simulation approach and realised more than ever before the weaknesses of the European measurement data of working conditions. The poverty and social policy RI community got more acquainted with the data and indicator resources that exist in the pre-distribution field (e.g. minimum wages; industrial relations data) and got inspired by the innovative work that was conducted in the working conditions pillar with web(crawling) data. The statistical RI more strongly showed to the others how particular mathematical techniques can help to overcome particular analytical problems (linked to data inefficiencies and missingness) and especially showed how the advancements in small area estimations can be used. Common working areas were further discovered in relation to non-probability sampling (e.g. in web surveys), multidimensional indicator-building, linking different data (types) and cross-border statistics. In general: contacts and networking with the pre-users community of the RI (data providers from official national and international agencies) were strengthened by this common together under one EU umbrella (e.g. a joint EU-SILC access was obtained from Eurostat).
Furthermore, a better integration and cross-fertilisation between the different research areas has been achieved by setting up a system of research visits by which both junior and senior researchers have been encouraged to interconnect with and learn from closely related research areas. For instance, InGRID offered the opportunity to applied poverty researchers to learn directly from researchers in statistical aspects of poverty indicators, which improved improve the quality of applied research and made researchers in statistics more aware of the statistical problems and questions applied researchers are confronted with.
3.1.2 Improved access to and use of the pool of research infrastructures
InGRID’s acronym is not chosen by accident. The reference to the grid-terminology expresses rightly the essence of InGRID. In computer science ‘grid’ is a term referring to the combination of computer resources from multiple administrative domains to reach a common goal. What distinguishes grid-computing from conventional high performance computing systems is that grids tend to be more loosely coupled, and geographically dispersed. InGRID precisely structures and strengthens from the same grid conceptualisation the existing social sciences infrastructures to reach the common goal of improved access to and use of data. In this respect, new users to the infrastructures have been attracted through the systematic combination of (a) open and diversified networking events and (b) the system of transnational visit grants. Further, (c) the development of the website, which also enhanced the visibility of the research infrastructure, formed an important backbone of these access impacts.
More precisely, four tangible impacts were identified:
• 462 members of a new generation of researchers had the opportunity to attend the training events and networked with 169 speakers/experts. Participants (trainees and speakers) came from 47 different countries. 54 participating scientists came from outside the EU.
• leveraging and structuring on-site access through a diversified system of visiting grants to a range of data centres and data competence centres and by linking these visits to training and mutual learning expert events. This way strong synergies and virtuous circles were possible between learning and application of the new knowledge in lab environments; 45% of the InGRID visitors were post-graduate researchers, 30% post-doc and 25% experienced researchers.
• access has not only been organised to specific data infrastructures, but also to expert infrastructures that have an in-depth know-how on an integrated use of existing official databases within the European Statistical System (for example SILC, LFS at the European level). Important in this regard is, in addition, the inclusion of access to high-performance statistical facilities using simulation technology;
• the scheduled transnational on-site access complemented and improved the already available remote access, which most of the involved data infrastructures currently provide to the research community (by themselves or by a data archive).
Importantly, the RI is deliberately not starting from a perspective of infrastructure integration by archiving, nor starting new data collections, but wants to help and serve these research communities also in the next important research steps of analysis tools and knowledge transfer practices, besides the access to data (as very often the case in European social science infrastructure initiatives).
3.1.3 Enhanced European added value of research infrastructure capabilities
The very meaning and strategic orientation of evolving towards the European Research Area has another dimension: creating European added value.
In InGRID European added value is first and foremost achieved through the strengthening of advanced comparative research and this on one of the key EU strategic policy arenas (inclusive growth). Servicing and helping this kind of research has been the key mission of the integrating RI project.
In addition, looking for further strategic harmonisation has been here the first key word. This includes harmonisation of protocols to identify hard–to-reach and hard-to-identify groups in living conditions surveys in the ‘Poverty and living conditions’ pillar. In the ‘Working condition and vulnerability’ pillar, harmonisation relates to validated methods for comparability of the job content of occupational titles and of the skill requirements for occupations across EU Member States. In this pillar, developmental work has also been conducted on the harmonised quality of working conditions measurement tools, such as scales and indicators. Among others a thorough evaluation has been made of the measurement possibilities the European Working Conditions Surveys gives. In the ‘New tools and indicators for policy analysis’ pillar, the harmonisation of institutional indicators organised by the new EUROMOD-based model family tool will enable to set significant steps forward in the understanding how different tax and benefit systems affect the income situation of otherwise similar household types across countries. In addition, with the development of this new tool, the model family approach is fully integrated with the micro-simulation approach offered by EUROMOD, which makes results of both approaches comparable and enable researchers to critically enhance insights into European tax-benefit systems and their relation to poverty and living conditions.
However, in addition to harmonisation, a second impact with clear European added value is the construction of integrated indicator sets and integrated survey and data repositories. These sets and repositories equally strengthen future comparative analysis in the research infrastructure. In the ‘Poverty and living conditions’ pillar a common indicator set for the measurement of deprivation and an integrated poverty and living conditions indicator system (IPOLIS) is developed. In the ‘Working conditions and vulnerability’ pillar a major achievement is an integrated Working Conditions and Occupational Health and Safety inventory of surveys and policy databases. In the ‘New tools and indicators for policy analysis’ pillar a comparable inventory of available comparative databases has been made.
Thirdly and at the same time, the ‘Statistical quality management’ pillar tackled specific problems which now seriously hinder accurate and sophisticated EU comparability, for instance by developing measures of accuracy of the estimates (such as mean squared errors and non-sampling errors due to non response, imputation of missing data and measurement errors arising from disclosure limitation methods) or by improving (sub)regional data exploitation with small area estimations.
Finally, throughout the project activities, linkages and connections have been investigated with other existing European RI projects. In particular exchange-of-practices were organised with the Data without Boundaries project (which has links with the CESSDA RI). InGRID experiences with obtaining transnational access to (national) data were used as DwB case material. The IPOLIS indicator system looked also for the integration of data coming from the European Social Survey and the SHARE survey. The e-infrastructure facilities of among others EUDAT were explored in the context of future e-infrastructure developments of the own InGRID RI. In addition and within a broader context, the ‘modernisation of European social statistics’ process (started in 2009) has been closely observed. Eurostat participated in the Advisory Board of InGRID, a strategic briefing note used this modernisation as a starting point (Lenau et al., 2016) and InGRID was throughout the project an active participant of the New Techniques and Technologies for Statistics conferences (by two satellite events).
3.4 Integration and innovation of the RI based on a two-way impact process with the wider scientific community and community-of-interest
The intensive programme of mutual learning, both summer school training events and expert learning workshops, opened the RI to the interested community of practice. The series of training and workshop events created for wider user-community opportunities to access state-of-the-art research infrastructure on ‘Poverty and living conditions’, ‘Working conditions and vulnerability’ and ‘Social policies’ while at the same time contributing to its further development through active participation and input. Further, these events facilitated access to a range of well-thought and targeted learning opportunities on the methods and statistical techniques needed. Finally, the design of focused and targeted joint research activities which directly addressed major methodological problems in the respective research fields, offered to the EU scientific community the state-of-the-art research infrastructure, including the best practice statistical methods for survey design and modelling.
In addition, the development and maintenance of the website offers an information gateway for the integration of the research infrastructure within the related community of scientific practice and for reaching out towards the community-of-interest.
The key advance inherent in this integrated programme of activities is the ability to extend the capability also to users who are not advanced in using and applying such data bases, methods, indicator sets and protocols. Analysis shows that the InGRID RI certainly made promising connections with research communities of the newer Member States (Hamon-Cholet et al., 2017). The ‘new’ users include early-stage researchers, but also the scientific community of institutes and Member States which are currently not involved in the research infrastructure and which may benefit of its improved and better integrated scale. Furthermore, the active promotion and attraction of representatives from government bodies, civil society and politics opens up the research infrastructure and enhance opportunities for evidence-based policies at different levels. The events offer ample opportunities for science-related discussion between practitioners in the data collection (official statistical agencies) and policy field (experts at state administrations and social interest groups). In addition, some specific tools and outcomes are designed to directly impact on the access of the different target audiences to the research infrastructure, such as the visualisation tools which present basic data structure explorations and bivariate relationships and trends of indicators in a visually attractive manner. In this regard the experiences of the InGRID project made also clear that stronger interaction and linkages are needed with international organisations like OECD and ILO, which are important pre- and post-users of the InGRID RIs. First networking steps have been taken in this regard.
In this way, the design of the InGRID infrastructure and activities enabled a continuous development and improvement based on a range of two-way interaction processes.
3.5 Contribution to the evidence-based European policy to combat poverty and vulnerability and enhancing coordination and impact of European policy
As already stated, the rationale of the InGRID project is related to one of the major policy challenges of the EU2020-strategy, namely striving for Inclusive Growth. The keyword in this regard is evidence-based policies. The project ambitions were thematically linked to the two ‘flagship’ policy streams of this Inclusive Growth strategy.
The InGRID project contributed in an indirect or second level of impact to these evidence-based European policies with the following concrete output:
• by providing a rich amount of transnational access, embedded in a series of networking and training activities, the project enhanced the ‘critical mass’ that can be involved in the defined policy-oriented research activities;
• an important output of the InGRID project is a more integrated set of indicators for the monitoring of poverty and living conditions, for the population as a whole and for several vulnerable groups (IPOLIS). These improvements in indicators of course facilitate policy discussion on these matters at the European level. The indicators will be easy to consult thanks to the development of a visualisation interface (ipolis.tarki.hu). Innovations in indicators have also been presented in the other pillars: out-of-work benefits, job quality and vulnerability-at-work;
• a next output is a response to changing needs for data and analytical models. The failure of the Lisbon Strategy, the changes in the social OMC and its integration into the European semester cycle, and the adoption of the European Platform against Poverty (EPAP) may well cause a radical shift in the poverty debate at European level, from a ‘purely’ social to a more economic debate. Issues such as fiscal policies, labour market reforms, liberalisation of services of general interest, etc. need to be assessed from a social inclusion perspective in the context of the ‘horizontal social clause’ (art. 9 of the new Lisbon Treaty). This necessitates the development of more complex analytic and simulation models, reflecting links between economic mechanisms and social outcomes, and requiring additional data and methods. InGRID contributes to the dissemination of such data and statistical tools for social impact assessment;
• at European level, in the framework of the New Skills for New Jobs initiative, a group of independent experts recommended the development of ‘a common language between education/training and the world of work’. DG Employment and DG Education and Culture have been entrusted with joint leadership of ESCO. The intention is to gradually develop a multilingual classification of European Skills/Competences, qualifications and Occupations (ESCO). The InGRID project has investigated tools for harmonising the measurement of occupations EU-wide and benchmark how current, relevant national employee and employers’ surveys generate data concerning new emerging jobs and skills needs in European labour markets. The project also explored how new forms of Big Data can play a role in this monitoring;
• policy evaluations require micro-level data of high quality, but necessitate also reliable and systematic information concerning institutional organisation. In order to account for individual level outcomes in terms of poverty, living and working conditions it is necessary to incorporate the organisation of social and labour market policies into the analytic framework. The InGRID project innovated classification methods, indicator sets and analytical tools of policy incidence and organisation;
• there is an increasing need to improve the research infrastructure in order to keep-up with the progress and social innovation that takes place in the policy arena of the Inclusive Growth strategy. The InGRID project identified in dialogue with the stakeholder community-of-interest priorities for future data collection and comparative analysis, with the aim to address the increasing needs for a broader, more multidimensional and forward-looking analysis of social exclusion and vulnerability-at-work at national and EU level;
• last but not least, the official European and national statistical systems invest considerable resources in household panel surveys on income and living conditions (e.g. EU-SILC and its national counterparts, national censuses). European policy agencies (Eurofound, EU-OSH, ECB) complement these efforts with particular survey efforts. The InGRID project improved the academic and policy valorisation of these governmental efforts by the provided transnational access to related data competence centers and by creating mutual learning processes about these data infrastructures.
The Stakeholder Advisory Board plaid an important intermediating role in reaching these impact results.
In a video message at the final conference Marianne Thyssen, European Commissioner for Employment, Social Affairs, Skills and Labour Mobility, but also Commissioner responsible for Eurostat, acknowledged this InGRID impact: “InGRID has made a major contribution to strengthening our capacity for evidence-based policy making particularly in the social and employment fields which are crucial for the future of the European project. (...) The InGRID project plaid here an important role by building a research community which we policy makers rely on for solid evidence that we can use to trigger reform in the Member States.” (see video). At this European policy level, indicator building is an important process. European social and economic governance is framed in scoreboards, prepared by indicators’ sub-groups of the Social Protection Committee (SPC) and the Employment Committee (EMCO). Presidents of these Committees, though speaking in their personnel capacity also acknowledged in a support letters the InGRID impact and made a plea for continuation. In the same vein, related departments from ILO and OECD expressed, based on past experiences and future plans, to increase their collaboration with the InGRID RI.
3.6 Main dissemination activities and exploitation of results
3.6.1 Main dissemination activities
To ensure the dissemination of the project results and the transfer of knowledge outside the partnership, InGRID operated a number of dissemination activities and channels. Two work packages were directly dedicated to dissemination and outreach: WP5 to the broader audience and WP6 to promote the RI to the scientific users. In addition, the very nature of the transnational access to data centres and data competence centres as well as the intensive mutual exchange learning programmes of WP3 and WP4 also attracted a broad range of participants, both scientists and stakeholders, beyond the consortium.
The activities and outcome of the project were tailored to the anticipated needs and interests of the different project audiences. These include:
• the scientific community directly interested to participate in the research infrastructure;
• the broader scientific community within and beyond the EU;
• EU and national policy-making actors and bodies.
The strategy consisted of four different and complementing types of communication and dissemination channels.
3.6.1.1 Electronic means and tools
A project website (D5.1) has been launched and is running since April 2013 and includes all documentation of project publications, presentations and other outputs, such as proceedings, newsletters and newsflashes. A specific space is dedicated to TNA. Electronic application tools have been developed and are used for TNA and networking events. On average the website receives between 1000 and 1200 (unique) visitors per month.
In addition, eight newsletter have been distributed by electronic mail to the InGRID audiences (cf. contact database infra). A typical newsletter contained the following items: InGRID in brief, editorial, InGRID news (past events, output), data in the picture (in the spotlight, in brief), community news (books, events, related projects, ...) calls for new InGRID events, info about the TNA visiting grants (selection results, experiences), and a calendar. Calls for visiting grants and events have been promoted by 27 electronic newsflashes.
3.6.1.2 Existing and new networks of scientists and the community-of-interest
All partners used their extensive networks within and beyond the European Research Area and their networks with policy makers, social partners and NGOs to involve experts and stakeholders in the research infrastructure activities. In view of systemising interaction with these networks a contact database has been maintained. People can subscribe at the website. The mailing list contains more than 4,605 contacts. 14% belongs to CEE countries, which are EU member, 5% to non-EU CEE countries.
These contacts have been invited to participate in a users’ needs websurvey. This broad survey gave the InGRID community-of-interest the opportunity to deliberate about the needs, gaps and weaknesses of the RI.
Links with 15 other FP7 projects have been established at the website.
3.6.1.3 Events and exchanges
The summer school training events of WP4 and the expert learning workshops of WP3 were targeted to the wider related scientific community and the interested community-of-practice. Specific initiatives were taken to promote and facilitate the participation of these communities to the networking events and even more so, to invite them to participate in the transnational access visiting programmes that are linked to these networking events.
Partners presented at different conferences work (partly) based on InGRID activities and made at the same time promotion on the RI. At the end of the project period, InGRID work was presented or promoted at 62 scientific events and 27 events to a broader public (of mainly policy makers, civil society and policy practitioners).
3.6.1.4 Project materials and publications
The following promotion materials were available and used to disseminate at events (InGRID events and external events): leaflets, copies of newsletters, flag, pens, notepads and paper clips. Partners were mobilised to distribute these materials at conferences. An updated version of leaflet has been re-printed. More than 3,800 leaflets have been distributed to different audiences.
In collaboration with and financed by the RICH project, a video has been developed that portraits the InGRID RI (www.rich2020.eu/rich-videos). It has been posted on the web since February 2016.
3.6.1.5 Dissemination impacts
The 10 transnational access calls received in total 633 applications. 200 were granted a visit, 184 made the visit.
Table 3 Communication channel TNA call picked up by visitors (see attachment)

1186 researchers made an application to attend a training event, 462 could attend. The expert workshops counted in total almost 500 participants.
Newsletter and newsflashes have been sent almost every 3 months to a growing group of more than 4600 InGRID contacts.
The InGRID RI is meanwhile recognised by a series of key end users (policy innovators) (see point 3.5). In total, 132 data providers, policy experts, scientists, academics and former visitors within the project attended the final conference, coming from 30 different countries.
3.6.5 Exploitation of results
As the project is on the one hand related to a research infrastructure and on the other hand focusing on social sciences and policy innovation, foreground that might be exploited for commercial or industrial capability is only very limitedly an aspect of the project results. Patent issues are an issue.
Nevertheless, two outputs have been defined as exploitable foreground:
• the construction of a new database (out-of-work benefits dataset) that will be made available freely to interested researchers by SU
• the Hypothetical Household tool which has been developed by UAntwerp and UEssex and will be made available to the users of EUROMOD. As use of Euromod includes restricted access to EU-SILC micro-data, the use of EUROMOD is bound by a license/users’ agreement. In a first period these users can on request use and test the tool. In a second step, the tool will be made available to all Euromod users (within the next two years).
Both applications are or will be freely available for (non-commercial) research and no confidentiality issues are at play.
The InGRID achievements will be sustained and further advanced through a 4-year follow-up EU Horizon 2020 project.

List of Websites:
Website: www.inclusivegrowth.be
Contact details:
Guy Van Gyes, InGRID project coordinator
p/a HIVA – Research Institute for Work and Society
KU Leuven
Parkstraat 47 box 5300
BE-3000 Leuven, Belgium
guy.vangyes@kuleuven.be

Final Report Summary - INGRID (Inclusive Growth Research Infrastructure Diffusion)

Share this page

Download