Quella che segue e' l'introduzione del rapporto dell'Undp (United Nations Development Programme.
Seguono i capitoli dedicati ai singoli paesi - Bulgaria Croazia Repubblica Ceca Uungheria Macedonia Romania Serbia/ Montenegro Kosovo - ai quali si puo' accedere attraverso il link:
http://vulnerability.undp.sk

Faces of poverty, faces of hope

Vulnerability profiles of Roma population in Decade of Roma Inclusion countries
Why these profiles? Alleviating poverty and overcoming exclusion is now a global challenge. All countries both developing and developed economies face poverty, although in various forms. Poverty pockets and excluded, marginalized groups exist throughout Europe, depriving whole communities of equal participation in development.
The countries of Central and Southeastern Europe face similar problems. The rate of transition there varies widely among different socio-economic groups, with some vulnerable populations, such as the Roma, in danger of being left behind.

- The Millennium Development Goals agenda. The occasion of the millennium prompted the United Nations Secretary General, as well as most of the world, to analyze past human development trends and their future directions. To address the global challenge of poverty, UN members accepted a comprehensive agenda for human development, including eight selected goals, targets with deadlines and quantitative indicators. The UN Millennium Development Goals (MDGs) originate from the Millennium Declaration signed by 189 countries, including 147 Heads of State, and adopted at the Millennium Summit at UN Headquarters, New York in September 2000. These goals were selected in an effort to tackle the world s most important development challenges. The eight UN Millennium Development Goals are intended to help governments take action to improve the situation of poor and marginalized social groups. They are as follows:
Goal 1 calls for halving absolute poverty (defined as living below PPP$1/day; PPP$4/day for developed countries such as those in Central Europe) by the year 2015.
Goal 2 envisions 100 percent primary school completion by 2015.
Goal 3 supports gender equality, empowering women and eliminating gender disparities in primary and secondary education.
Goal 4 calls for reducing child mortality by two thirds by 2015.
Goal 5 aims to reduce maternal mortality by 75 percent.
Goal 6 deals with combating HIV/AIDS, malaria, TB and other socially significant diseases.
Goal 7 addresses environmental causes of poverty.
Goal 8 calls for building global partnerships for development.
Each of these goals includes a number of specific and measurable targets (such as improving access to safe water, sanitation and increasing access to development opportunities for different groups). Monitoring the Millennium Development Goals, targets and indicators is appropriate not just for developing countries but for developed nations, too. While reaching the national targets is not a problem for developed countries, poverty pockets and excluded communities can be hidden within the national averages. The real challenge is meeting the respective MDG targets for marginalized groups.

- Roma and the MDGs.

Roma communities are an example of a group deprived of the benefits of transition. The depth and magnitude of problems Roma face require concerted action through an inclusive approach involving government, civil society and other partners working together. UNDP has consistently called for MDG disaggregation, so that the concerns of those most in need are reflected. Avoiding the Dependency Trap, the 2002 award-winning regional report on the status of Roma in five Central European countries, called for monitoring Roma MDGs as a necessary analytical tool for improving the situation of these groups. Two prerequisites were necessary, however: governments political commitment and relevant data.

- Political Commitment.

Political commitment came with the Decade of Roma Inclusion, which grew out of the June 2003 conference Roma in an Expanding Europe: Challenges for the Future, hosted by the Government of Hungary. The World Bank, Open Society Institute and the European Commission organized the conference, with support from UNDP, the Council of Europe Development Bank and the governments of Finland and Sweden. At this high-level conference, five prime ministers and high-level representatives from eight countries (Bulgaria, Croatia, the Czech Republic, Hungary, Macedonia, Romania, Serbia and Montenegro and Slovakia) made a political commitment to close the gap in welfare and living conditions between the Roma and the non-Roma and to break the vicious circle of poverty and exclusion. The governments Action Plans elaborated and implemented during the Decade of Roma Inclusion correspond to the MDGs, i.e. poverty, employment, education, health and housing. At a practical level, the Decade is an opportunity for countries to meet the MDG targets for Europe s most vulnerable group the Roma providing another link between the Decade implementation and the long-term commitments of UNDP as an MDG campaign manager and scorekeeper.

- The need for disaggregated data.

Next was the data. Without a means to measure both the implementation of policies and monitoring, MDGs risk becoming merely hollow slogans. Governments, donors and implementing partners require quantitative data to outline priorities and measure progress. Disaggregated quantitative data is a precondition for relevant national policies for sustainable inclusion of vulnerable groups, the Roma in particular. Therefore, UNDP decided to conduct a comprehensive survey of Roma and other vulnerable groups in Central and South-Eastern Europe. The survey was designed to map the Roma vulnerability levels compared to other groups. Its purpose was to provide both important analytical inputs and a baseline for the Decade implementation and its progress monitoring.
No less important were the capacity development aspects. The very process behind the survey was as important as the results. The survey was not conventional data collection because a set of crucial questions regarding disaggregated data collection, MDG monitoring and reporting in Central and East European countries is still open:
How to distinguish communities?
How to define vulnerable groups (in the case of the Roma, who are the Roma)?
What sources of data can and should be used?
How to obtain it?
How to deal with privacy, with multiple identities?
Another set of issues emerged related to the mechanics of monitoring and reporting MDG-related indicators at subnational levels:
Should international targets and benchmarks be used to measure progress?
What should be considered a country s commitment? Meeting the targets at the national level or for all distinct groups?
Can a country be considered meeting its MDG targets if certain marginalized groups (like the Roma) fall well behind? These methodological challenges added new dimensions to the survey and turned it into a rather long-term process. UNDP established an Experts Group on data and measurements whose task was the elaboration of consistent and comparable approaches to the issue of quantitative socio-economic data disaggregated for major vulnerable groups.
The group s purpose was to suggest specific (and feasible) ways of overcoming existing barriers in the area of ethnically disaggregated data collection so that in a few years the capacity for disaggregated data collection is in place at the country level. By 2006 2007, the whole responsibility for data collection should be transferred to the relevant bodies in the individual countries.

- Outline of the survey methodology.

The objective of the survey was to provide quantitative and comparable data on development problems and challenges of vulnerable groups and Roma in particular in Central and Southeastern Europe. Given the launch of the Decade of Roma Inclusion in 2005, the survey was supposed to provide baseline data that will enable measuring the progress in achieving MDG indicators for Roma. Of course, the survey is sample-based research and cannot be as representative as a household budget survey would be. But it still provides quantitative data enabling the rough calculation of poverty lines, poverty depth, employment/unemployment rates, educational levels, educational attainment and housing conditions. Based on this data, a set of indicators can be calculated that is consistent with the individual-oriented indicators envisaged by the Millennium Development Goals monitoring.

- The survey instrument and main assumptions.

The survey questionnaire was designed accordingly to: Reflect the logic of MDG goals and provide necessary data for computation at the individual and household level Monitor MDG indicators. It followed the philosophy of an integrated household survey with separate components containing both household and individual modules. Within the individual module, each household member s profile was registered (demographic characteristics, economic status, education, health). The household module addressed issues related to the household in general (dwelling type, access to and type of use of basic infrastructure, household items and possessions, etc.). Questions related to incomes and expenditures were addressed in both modules making it possible to crosscheck the results. The primary universe under study was the whole population of areas with present or over-represented proportion of Roma community. These were defined as administrative units/settlements where the share of the Roma population equals or is higher than the national share of Roma population in the given country as reflected in census data. Of course, in most countries Roma are underreported in censuses. Officially registered figures on Roma population traditionally differ from the experts estimates. Hence the first assumption of the survey: Censuses understate the absolute number of the Roma population, but provide a reasonably adequate picture of its structure and territorial distribution mainly for those who identify themselves as Roma. The second assumption was that major disparities in socio-economic status of the populations are most obvious (and can be explored best) at the municipal level (or other relevant territorial units). Since at this level vulnerability factors exist that affect both Roma and the majority populations, vulnerability profiles of the two groups (Roma and the majority) in the same territorial unit would make possible the identification of those vulnerability factors that particularly affect Roma.

- General principles of the sample design.

The most difficult problem was Who are the Roma? and determining how to identify the respondents. The primary objective of the survey was to map vulnerability of groups with common socio-economic, cultural and linguistic patterns not the exact way they refer to themselves. Given the fact that Roma identity is often associated with underclass status and/or discrimination, avoiding self-identification as 'Roma' is a logical pattern. Simply asking, 'Are you Roma?' does not work. Another challenge was the lack of clarity on identification criteria and the multiple identities people tend to have (particularly Roma). A question 'Are you Roma'? implicitly suggests its opposite wording ('You aren t Bulgarian, Romanian, Macedonian, etc'). There can be much confusion regarding a person s ethnicity, nationality and citizenship.
This is why relying solely on self-identification would not produce a representative sample. On the other hand, forcing people into certain categories, applying external identification only, is not acceptable either. Given these considerations, a compromise between the two, self-identification and external identification, was reached within the 'implicit endorsement of identification'. Sample design took place in three stages. First, the universe was defined using an average and above share of Roma in each administrative unit/settlement. Second, sampling clusters were determined taking into consideration estimations of Roma organizations (suggesting, for example, that in municipality X Roma dominate, but for various reasons tend to be reported or declare themselves as Y or Z), the distribution of the settlements and population size. Third, respondents were identified using random route selection.
At different stages, internal (self-identification) or external (outsider's) identification prevailed: self-identification (reported during the census) at the first stage, external (assessment of local people, NGOs, experts) at the second. At the final, third stage (respondents selection), the two identification methods were confirmed or rejected by implicit endorsement of identification . This means that having identified the sample clusters and the households to be interviewed, the introductory sentence at the beginning of the interview was 'Good morning/day, we are conducting a survey among the Roma population. Would you mind being interviewed?' In case of explicit denial ( I am not Roma, why should you interview me? ) the interview was cancelled. Participation was interpreted as the household member s implicit endorsement of belonging to the universe under study.
In some cases (particularly in big cities and capitals) a numerically large group of Roma still constitutes a proportionally low share in the total. In such cases, the sample model followed the administrative subdivisions: Usually the capital municipality is divided into smaller municipalities and/or lower levels of self-government. These smaller units were chosen as the sampling units. Such cases were also corrected typologically introducing additional sampling points.

- Majority boosters.

Apart from Roma respondents, majority booster samples were constructed using similar procedures (representative for the majority population living in settlements with Roma population 'average and above', not for the total majority in a country). The idea was to have a sample for the majority living in close proximity to Roma populations and facing similar socio-economic challenges often associated with regional disparities.
Applying majority boosters gave the survey a benchmark, allowing judgments as to the depth of poverty and vulnerability among Roma vs. non-Roma populations living in a similar socio-economic environment. This approach, despite all technical difficulties in sample design, enables distinguishing various vulnerability factors, in particular those that are related to minority status (and hence can be attributed to various forms of discrimination) from those due to regional disparities or depressed local economies (i.e. due to the fact that populations studied live in less developed territories).
In cases of municipalities with a high share of Roma not having substantive number of majority population for a majority booster (for example, in cases of isolated settlements or segregated neighbourhoods), the majority booster was based on the population from a typologically similar settlement in the same (or adjacent) district (administrative unit), residing in the nearest proximity to the surveyed Roma target population. The criterion for choosing an administrative unit/settlement was the closest one accessible by road connection. It is important to bear in mind that the approach would not guarantee national representativeness for the majority population, and the surveyed universe of the majority in each country is actually composed of those who live in closer proximity to Roma populations.
The advantages and impediments of the adopted approach A sample based on municipalities with an average and above share of Roma is not fully representative of the whole Roma population, but roughly covers about 85 percent of Roma in each country and provides a good basis for quantitative socio-economic indicators for Roma (quality of life, life expectancy, access to services, income, etc.). The resulting sample is also representative not just for segregated Roma but also for the majority of Roma.
The data acquired will be acceptably consistent with census information, since the data is based on relative numbers (structure and regional distribution) instead of absolute numbers of Roma population registered by the censuses. The data provides comprehensive snapshots of the regions with concentrated Roma population based on a clear approach to the questions: Which are these regions ? and How concentrated is the Roma population there ? The approach gives some standardized criterion for majority booster selection. The majority boosters are based on populations in direct proximity to Roma (in the same municipality or region) despite all limitations of such a definition.
The major impediment of the adopted sampling model relates to municipalities with the share of Roma population below the national average: They fall out of the scope of the sample. They could be either concentrated in mini poverty pockets or are dispersed (presumably integrated with the majority). However, both groups are represented in the sample:
In the first case (concentrated mini poverty pockets), because most of the 85 percent covered are living in similar poverty pockets (which were representatively sampled).
In the second case (the dispersed and integrated group), because part of that 85 percent of Roma is functionally integrated (employed, maintaining contacts with the majority and institutions etc.) and thus typologically similar to dispersed (presumably integrated) Roma from the remaining 15 percent.
Those of the 15 percent who are dispersed and integrated and identify themselves as Roma are typologically close to the integrated Roma from the 85 percent. Those who have been assimilated and do not identify themselves as Roma fall out of the scope of the research either because they do not meet the criterion of 'being Roma', whatever that means, or because they do not meet the vulnerability criterion. Overall, the suggested approach is based on the assumption that existing demographic information on the size and structure of the Roma population can be analysed and compacted in a reliable enough picture: total and territorial distribution. This is quite difficult to achieve, and will inevitably be partly based on estimates and experts assumptions, but is a prerequisite for any representative sampling procedure. All effort has to be made in this direction. An alternative is very large national samples in each country, ensuring a statistical minimum of Roma sub-samples. Levels of comparability The combination of two samples (Roma and majority) with the format of the survey instrument following the philosophy of an integrated household survey provides the unique opportunity for three levels of comparability:
Between Roma and the local majority living in depressed areas
Between Roma and the status of the average population of the country (reflected in national household and labour force surveys)
Between majority populations living in depressed areas and the national averages.
In addition, applying common methodology in all countries covered by the survey allows for cross-country comparisons. Given the major constraint uncertainty of the absolute number of the Roma population (due to the unclear identification criteria) the data (and all possible comparisons) have certain limitations. The survey does not provide the answer to questions like How many Roma live in poverty ? or How many Roma have completed secondary education ? It gives the answers to questions like What share of Roma is living in poverty ? and What share of Roma has completed secondary education ? Such answers are comprehensive enough for policy purposes because they outline the distance between various groups and provide clues to the reasons why disparities exist. However, they may be inadequate for resource allocation (usually based on headcount) until some national-level consensus is reached on the number of people referred to as Roma . This issue goes beyond the scope of the current survey.

- Fieldwork and partnerships.

As a specific and unique minority group, the Roma in some countries show distrust towards other ethnic groups. In order to overcome the possible distrust of pollsters, Roma interviewers were used for fieldwork where possible (in countries where sufficient numbers of trained Roma were available). In other cases, Roma intermediaries were used (following the pattern of Roma assistant teachers ). These were either Roma assistant interviewers (that is, a Roma representative accompanying the experienced pollster) or local social workers or Roma NGO representatives. In all cases, the intermediaries were trained (on the contents of the questionnaire, on general rules and procedures of an interview etc.) before beginning the fieldwork The general rule, however, was to approach the communities carefully, with respect and avoiding any suspicion about the purpose of the data collection.
The Council of Europe as a part of its Roma under the Stability Pact joint project with the European Commission expressed a deep interest in contributing financially to this survey, in particular for costs related to assistant interviewers or other intermediaries, in view of updating with statistics its own study on Roma Access to Employment in Southeastern Europe.
The survey was executed by the following agencies: For countries of Southeastern Europe: by each GALLUP International affiliated agency, coordinated by the GALLUP International regional office that managed the execution of the whole survey For the Czech Republic: by FOCUS Agency For Hungary: by TARKI.
After the fieldwork was completed, a field control was run on 10 to 15 percent of the sample, depending on the country. All completed questionnaires were subject to quality control for proper administration and, where it was deemed necessary, some interviews were discarded and the respective target sampling points were re-visited. For countries of Southeastern Europe, data entry was conducted locally by each GALLUP International affiliated agency. The GALLUP International regional office collected and assembled the final regional data set. For the Czech Republic and Hungary, data entry and control were conducted by FOCUS and TARKI respectively.
From the outset, all agencies involved worked in coordination under the methodological and conceptual guidance of the UNDP Bratislava Regional Centre. The methodology of the survey, sampling and fieldwork were broadly discussed with colleagues from the World Bank and members of the Data Experts Group. Three consultants (Gabor Kezdy, Valerie Evans and Dragana Radevic) were particularly instrumental in the final design of the methodology and sampling models.
The nature and purpose of this publication The data generated within the project are extremely rich and multidimensional. Their analyses and interpretation are still forthcoming. The primary purpose of this publication is to give food for thought by presenting the major socioeconomic indicators of the Roma population in the Decade of Roma Inclusion countries. The data are presented in graphs with the value of every indicator. The accompanying notes are intended to provide additional information on the indicators and explain how they were calculated but not to interprete the messages of the data. This is why the accompanying notes are the same for all countries.
One final note on comparability. The values of the indicators for the Roma are presented in relation to the values for the majorities living in close proximity to the Roma. These are not national averages. The majorities in close proximity tend to be in a disadvantaged position as well (for example, because they live in economically depressed areas). This is why the difference between Roma and national-level indicators may be higher than reflected in the graphs. This assumption can be checked by analyzing the data against the national averages of similar indicators reflected in integrated household surveys and/or labour force surveys something that will be the subject of further analysis.
Andrey Ivanov, PhD
Project coordinator
Bratislava, January 2005

PER GENTILE CONCESSIONE DELL'UNDP
19/03/2005 16:02