THE IMPACT OF SOCIAL POLICIES ON SELF-RELIANCE INCENTIVES FOR ROMA IN CENTRAL AND EASTERN EUROPE

Poverty Reduction and Economic Development United Nations Development Programme Regional Centre Bratislava Gr6sslingova 35 81109 Bratislava, Slovak Republic Tel: (00421-2) 59337-179 Fax: (00421-2) 59337450 E-mail: susanne.milcher@undp.org Katarina Zigova Strategic Research and Economy Branch United Nations Industrial Development Organisation Vienna International Centre P.O.Box 400 A-1400 Vienna, Austria Tel:(0043-1) 26026-3469 Fax: (0043-1) 26026-6802 E-mail: k.zigova@unido.org


Introduction
According to many studies, the economic transition in Central and Eastern Europe has adversely affected the Roma more than any other group (UNDP, 2002 andRingold, 2000). Their socio-economic status is characterised by poverty, low educational achievement and consequently high unemployment. There are two generally accepted views why this is the case. According to the first one, the Roma face discriminatory practices on the labour market and in the educational system. The second one assumes that Roma themselves don't actively search for employment, don't educate their 58 children and don't improve their living conditions. Both views, albeit exclusive, reflect the complexity of Roma exclusion. Whatever the underlying cause, the outcomes are unemployment, unstable income flows and hence dependency on stale transfers. In this paper we try to assess to what extent this dependency aids or hinders the Roma's pursuit of their independence from social transfers. We base our investigation on the UNDP/ILO Roma survey held in 2001 in five countries of Central and Eastern Europe: Bulgaria (BG). the Czech Republic (CZ), Hungary (HU), Romania (RO) and Slovakia (SK). The choice of these five countries is not incidental; more than 50% of Roma are settled there (Vasecka, Juniskova and Nicholson, 2003).
It is clear that for a lot of households, including Roma ones, social transfers are the only way to ensure a minimum income and avoid falling into poverty. However, the social systems of these countries are sometimes inconsistent. On the one hand, the government's social contribution budget is often limited. Hence, some social transfer receivers might still live close to the subsistence level. On the other hand, the beneficiaries are rarely motivated to seek employment if the wages offered are close to their social transfers. In countries with high unemployment, this system creates tension between the part of the population that is actively working and paying taxes (and thus being net contributors to the social welfare system), and the net beneficiaries. When these cleavages coincide (or even correlate) with ethnic identities, the systemic causes of ethnic conflict are in place (UNDP, 2002). This paper contains two parts. The state social policies section contains a review of social policies in five countries. We focus here in particular on laws towards people with no regular wage income (or other income) below the national subsistence level. This section is the basis for the following section, in which we analyse the relationship between Roma dependency on social transfers and factors measuring their efforts to get a job as well as factors describing their lifestyle. The last section infers possible implications for policy makers and concludes the paper.

State social policies
Here, we review the social systems in five countries: we assess the approximate income level of households that depend on state transfers in the form of social, child or unemployment benefits. l Furthermore, we compare this approximation with the minimum wage in the respective countries. Finally, we link these two types of incomes, both expectedly low, with the minimum subsistence levels. This exercise provides us with an interesting picture of subsistence levels and poverty of dependent households as well as households earning minimum wages. The conclusion drawn from this section is that social transfers do not always offer enough incentives to look for a job. In fact, the difference between social transfers and the minimum wage is very small and almost equal for households with many children.
Since the early 1990s, the five countries under review have had to transform their social protection systems in order to make them compatible with market economies. Reforms varied across the countries according to their speed and performance in the transition process. Although social spending stabilised at 15-20% of GDp, social protection systems have remained a necessary safety net. Cash transfers formed an important part of household income. In Romania and Bulgaria, over 80% of households received at least one benefit in 2001(Fox, 2003. However, benefits were usually poorly targeted and support was mainly given through categorical benefits, such as child allowances. Nevertheless, reforms of the social assistance system are underway in most of the five countries. This is mainly due to the EU accession process, which put a priority on combating social exclusion. By now, all countries have introduced a means-tested social assistance system. The need for social assistance is unquestionable, as poverty rates before transfers are still very high (cf. Figure 1). In four countries, poverty would be more than three times higher without transfers than after them. The impact of pensions is higher for the three new EU member states. Further, the risk offalling into poverty is higher for a family with more than three children (European Commission, 2004). However, the coverage and efficiency of the social systems for non-pensioners in Central and Eastern Europe vary considerably among the five countries. For example, in 1993For example, in -1995 in Hungary the social welfare system covered 43% while in Bulgaria only 10% of the poor. This low number in Bulgaria can be explained by the country's targeting of the non-working population, even though there is a high proportion of "working poor". The "exclusion error", the share of the poor not receiving social assistance, was 90% in Bulgaria versus 57% in Hungary. On the other hand, targeting can also 60 go to non-poor (inclusion error). In Hungary, the share of the non-poor receiving social assistance was 86%, and in Bulgaria it was 92% (Braithwaite, Grootaert and Milanovic. 2000). Both inclusion and exclusion errors within targeted programmes have been high in Central and Eastern Europe. Inflation also affected the ability of the social system to reduce poverty, as few countries have indexed the value of cash benefits. Similarly, the purchasing power of the social assistance eligibility threshold eroded in real terms.
During the first half of the decade, several factors influenced the evolution of social welfare systems. Targeting benefits according to income was a new concept, and many preferred to continue to support categorical benefits. Due to initial reforms, a high proportion of people considered themselves poor, and they were too many to be covered by the social system. In addition, transient poverty was common, and households tended to cycle in and out of poverty due to large income shocks. Overall, social assistance delivery systems were weak. In many cases, decentralisation meant delegating responsibility to local governments that had limited capabilities and resources. Finally, loose eligibility criteria encouraged people to continue receiving benefits instead of looking for a job (Andrews and Ringold, 1999). With EU membership, the governments of the Czech Republic, Hungary and Slovakia are obliged to reduce public spending in order to fulfil the Maastricht criteria. 3 The greatest move was done by the Slovak Republic in changing its very generous social welfare system to a "make work pay" system in November 2003. The following part will describe in detail the structure of social assistance, child and unemployment benefits in the five countries in 2000-2002 (Table 1). Subsequently, a model of Slovakia's new social assistance system will be presented.

Second-wave countries
In Bulgaria, the Ministry of Labour and Social Policy (MLSP) supervises unemployment and social assistance, among other areas. At the same time, decentralisation has shifted the responsibilities to local governments and social assistance is financed by the municipal budget, subsidies and earmarked transfers from the central government. Since 2002, the Unemployment Fund has been separated from the state budget. The period of eligibility for unemployment benefits is nine months. The nine months must fall within 15 months of the last contribution to the social insurance sys-tem. The unemployment benefit is 60% of the average monthly wage during the last nine months of insurance but should neither be lower than 35 euros nor higher than 67 euros (2001). The length of benefit payments ranges between four months and 12 months depending on years of insurance. In December 2001, only 24.2% of all registered unemployed received unemployment benefits. The low coverage is a result of the restrictive qualifying period, limited access for seasonal workers and rising long-term unemployment. Bulgaria's social assistance programme is a differentiated minimum income guarantee scheme, which provides eligible households whose income falls below the Guaranteed Minimum Income (GMI) level with a benefit equal to the difference between their income and the GM!. In addition, larger households are entitled to a higher income level based on an equivalent scale. The GMI is calculated according to a consumer basket of 22 food items and energy expenditures. However, the Institute for Trade Union and Social Studies separately calculates a "basic needs consumer basket" reflecting the standards of the Food and Agriculture Organisation and the World Health Organisation. The difference between these two subsistence levels is striking: 20 versus 57 euros in 2001. Besides, the purchasing power of the minimum wage is more than twice that of the GM!. In 2001, the minimum wage could buy 130 kg of bread while the GMI could only buy 62 kg. Child benefits (8 euros) are income-tested and accrue to families whose income does not exceed 75 euros per person.
In Romania, the provision of social services suffered due to reduced public revenues and tight monetary and fiscal policies. Poverty increased dramatically due to this austerity programme, and for families with 3-4 children the poverty rate reached 84% in 1998. The Ministry of Lcbour and Social Solidarity (MLSS) supervises social insurance, unemployment insurance, family and child benefits. The social assistance system is financed from state and local budgets. However, the state provides the total funds for local authorities without specifying their use. Therefore, local authorities have to set their own priorities, which do not necessarily focus on social assistance. The unemployment insurance system is based primarily on contributions. Only half of the active population contributes to the insurance system since it charges among the highest contribution rates (60% of wages). Unemployment benefits represent 75% of the minimum gross wage at the date of establishing the benefits (41 euros in 2002). The length of benefits depends on the contribution period but is no more than 12 months. Minimum guaranteed income is ensured by a monthly benefit that reflects the difference between the minimum guaranteed income (specified by the minimum Guaranteed Income Act of 2001) 62 and the monthly net income ofthe household. The minimum income threshold is based upon family composition with higher levels for larger families (24 euros for a single person). Child benefits (5 euros) are universal for children up to 16 years of age or a maximum of 18 years if they are still in education. In Bulgaria and Romania social transfers do not provide disincentives to work, since benefit levels are very low compared to the minimum wage. In these countries, the question is whether benefit levels are high enough to provide a subsistence income. This is also a result from the survey data on Roma. Table 6 shows that Romania and Bulgaria have the highest share of households that are independent of state benefits, i.e. many households cannot live on benefits but need to look for employment.

EU member states
The Ministry of Labour and Social Affairs (MoLSA) is responsible for state social support in the Czech Republic. Nevertheless, re-gional and municipal self-governing authorities have implemented social activities with money from their own budgets, which stem mainly from transfers from the central government. Unemployment benefits are financed through contributions and equal 50% of the assessment base for the first three months and 40% of the assessment base for the following three months.4 People are entitled to receive benefits if they have been insured for at least 12 months in the last three years and if they have not refused employment without a valid reason. In the Czech Republic, also a minimum-subsistence-income-based social assistance scheme is in place. The minimum Subsistence Amount Act (2001) sets the eligibility threshold for benefits at 52 euros for a single person. Total benefits increase with the number of household members. Child benefits are income-tested and correspond to a range of 15 to 25 euros depending on the age of the child.
In Hungary, the social protection system does not guarantee a standard minimum income level. The social assistance system is fragmented by eligibility criteria and by territorial units. It is very decentralised and the Ministry of Health, Social and Family Affairs has only limited responsibility and control. The social assistance is financed 75% by the central budget and 25% by the local government. The minimum income threshold defined by legislation is 83 euros in 2002. As a reference, the Central Statistical Office calculates a minimum basket of goods containing food, housing and non-food expenses (394 euros in 2002). However, both levels are not used in order to target social assistance; instead, the minimum oldage pension is used as a benchmark. Social assistance is distributed in various forms, of which the benefits for families with children make up the largest part. Per capita income of households with three children is only half that of single-child households, and family and child benefits have a strong role in reducing poverty among families with children. In 2000, the poverty rate was 9.1 % and would have been 13% without family allowances. Child benefits are universal, and the benefits are differentiated according to family composition (16 euros for a couple with one child). Child support, on the other hand, is a benefit for families raising three or more children and the youngest child is between age three and eight. The monthly amount is equal to the minimum income threshold (83 euros). Regular child protection support is a benefit for families where per capita income is equal to or lower than the minimum income threshold. The monthly amount is 17 euros per child.
The unemployment insurance is financed through contributions. To be eligible for the benefit, one must have paid contributions for at least 200 days during the previous four years. The amount of the benefit is 65% of the previous average earnings but should not be less than 90% of the minimum income threshold and not more than twice the minimum income threshold. The maximum length is nine months.
At the beginning of 2004 Slovakia changed its social welfare system. Until then the level of benefits had been based on the number of children in a family. Now it depends on what the unemployed person does -how he/she seeks a job, increases his/her education, or works for the community. The social sector is within the competence of the Ministry of Labour and Social Affairs and Family (MOLSAF) and is also completely centralised. The Subsistence Minimum Act (1998) adopts two lev-els of material distress under which individuals are entitled to social benefits. Individuals get one level or the other depending on whether they are in need because of subjective (i.e. their own fault) or objective reasons. 5 The subsistence minimum was set by law at 92 euros in 2002 for a single person. Social assistance transfers were allocated according to family composition with no upper ceiling on the amount a household can receive. Child benefits (6 euros) are universal, with the exception that they are tied to the children's school attendance. In addition, households whose income does not exceed 1.37 times the subsistence level receive a bonus to the child allowance based on the child's age. Unemployment benefits are financed through statutory insurance premiums administered by the Social Insurance Agency and the National Labour Office. Unemployed individuals are entitled to benefits lasting at most nine months if the person has contributed to unemployment insurance for at least 24 months in the last three years. The amount of benefits equals 50% of the average monthly assessment base in the first three months and 40% of the assessment base for the remaining time. This amount should not exceed 1.5 times the minimum wage (191 euros in 2002).
The reform of social assistance, effective from January 2004, aims to increase the difference between employment-generated income and income from social benefits in order to encourage the unemployed to seek work. Before, the system created a culture of dependence upon benefits by not rewarding people 5 People unable to earn an income through their own abilities are considered to be in a state of need due to objective reasons. People who do not seek employment and refuse to cooperate with the labour office are consid· ered to be in a state of need due to subjective reasons. 64 when they find employment. The new system introduces an activity-tested income support that provides 24 euros to an individual who takes up smaller jobs or participates in retraining courses. Already in 2003, the law was changed to curb unlimited benefits for large families by establishing a maximum ceiling of 253 euros for the total amount of benefits. The new system reflects the "making work pay" principle by increasing the difference between welfare payments and the minimu-m wage ( Table 2). In the pre-reform system, it was not worth having one additional income at minimum wage for families with more than one child. The total benefits with income from employment were the same as total benefits without income from employment. In the post-reform system, this disincentive has been removed. The difference between the value of social assistance benefits and the minimum wage increases steadily up to almost 120 euros (4977 SKK).

Impacts on Roma
The new system should increase incentives to look for a job, even if paid at minimum wage. But certain groups like the Roma may be affected negatively. Roma leaders argue that Roma cannot find employment because of discrimination on the labour market, their low education level and general unfavourable economic conditions in the regions where Roma predominantly live. Unrest in Eastern Slovakia's Roma communities in early 2004 demonstrated these fears. In addition, these concerns are also proven by the fact that many Roma are long-term unemployed and accounted for the majority of social assistance beneficiaries during 1996-2001. In Bulgaria, the Roma are six times more likely to be eligible for social assistance than non-Roma; they  1  7510  7510  0  3860  6196  2336  8190  9914  1724  2  9090  9090  0  4410  7126  2716  8740  10872  2132  3  10 670  10 670  0  4960  8056  3096  9290  11829  2539  4  12250  12250  0  5510  8987  3477  9840  12787  2947  5  13 830  13830  0  7060  9917  2857  11390  14744  3354  6  15410  15410  0  7610  10848  3238  11 940  15702  3762  7  16990  16990  0  8160  11778  3618  12490  16660  4170  8  18570  18570  0  8710  12708  3998  13040  17617  4577  9  20150  20150  0  9260  13608  4348  13590  18567  4977 are four times more likely in Hungary, and three times more likely in Romania (Revenga, Ringold and Tracy, 2002). Since Roma have often more than three children, they are affected by the cuts in benefits based on the number of children. Generally, in determining the amount of social assistance, all family members necessitating higher social benefits are taken into consideration (Table 1). For example, a family in the Slovak Republic with five children receives only 170 euros of social benefits plus child support in the post-reform system instead of 324 euros before the reform (Ministry of Labour and Social Affairs of the Slovak Republic, 2004). This is a cut by almost half. Therefore, in the pre-reform system having more children (common for Roma households) further decreases the difference in income between households who are living on social transfers and those receiving the minimum wage (very likely Roma households). Hence, there is no sufficient incentive for those households to move from receiving benefits to working at the minimum wage. In the postreform system, these disincentives have been removed but it remains to be seen if households are able to cope with the cuts in benefits given limited job opportunities in disadvantaged regions.
In addition, employment does not guarantee protection against poverty. Even though the principle of "making work pay" encourages employment, it does not reduce the number of the so-called "working poor". This group comprises mainly unskilled labour of which Roma are a majority. Accepting a lowpaid job often leads to an increase in other expenses connected to the job (e.g., transport, meals, etc.). As a result, reducing benefits for families with many children could possibly in-66 crease the number of "working poor". Rather, an activity-targeted social benefits system could be a more efficient option to provide incentives to look for a job. Child benefits should be administered separately to maintain support for households with many children.
Furthermore, it is important that additional earnings are not taxed. For example, Bulgaria disregards 30% of wage income in calculating eligibility (Andrews and Ringold, 1999). Hungary provides an income tax relief for parents with children. The Slovak Republic links income support to employment. Romania, apart from linking income support to employment, awards 15% additional assistance to a household if a member gets a job (Ministry of Labour, Social Solidarity and Family of Romania, 2004). Other approaches to increasing people's incentives to work include requiring registration in labour offices and limiting the eligibility period. The three new EU member states have introduced incentives more extensively than in the secondwave countries, as can be seen by higher unemployment payments and longer duration of benefits in Bulgaria and Romania (Figure 2 and Figure 3).
The review of social systems revealed that the new EU member states have a more generous social safety net and are able to lift more households out of poverty. However, because of this generosity, members of large households in need are less motivated to look for employment, since the added value of employment in terms of income is rather small. In the second-wave countries it is the opposite case. The social safety net is often insufficient to alleviate poverty, and households in need are forced to look for employment in order to cover subsistence expenses.

Roma households and state social transfers
The UNDP/ILO survey revealed that most of the Roma are to some extent dependent on social transfers. On average, more than 30% of Roma households' major source of income consists of some kind of social transfers, either in the form of unemployment benefits, social assistance or child support. In Slovakia, this rate reaches up to 60% (cf. Table 3), which is, as discussed above, due to the very generous social welfare system that Slovakia had in place until end of 2003. This picture is extended by a similar table based on the percentage of households that are receiving social transfers, without stating that those form most of their incomes (cf. Table 4). Here, the numbers are obviously higher in comparison with Table 3. A high percentage of the households are receiving child benefits. This is in line with the previous assumptions that Roma do have a higher number of children. A surprising result is that around 50% of households are receiving social assistance, what means that many members of the household are long-term unemployed who already lost eligibility for unemployment benefits. Looking at the aggregate numbers we get the following picture. There are 7.7% households receiving all three types of transfers, and almost ten times more are receiving at least one of the three. In Slovakia, this number is just below 100%. In the subsequent parts, we compare the Roma households, which are self-sustainable or independent of the social transfers, with households dependent on social transfers. In particular, we assess, for both groups, their efforts in terms of job search, in promoting education of their own children, in civic engagement, allocation of their household income and health care. Our results reveal that the dependent households show lower efforts in some of the criteria examined. Hence, based on the survey sample, this implies that social transfers do not always offer sufficient incentives to be active and become self-sustainable. In light of our empirical analysis, the countries studied might wish to explore the option of increasing the minimum wage, rather than decreasing the amount of social transfers.

Data and variables of use
Throughout the empirical analysis we are using the UNDP/ILO survey conducted in 2001 on a representative sample of 5000 Roma households in five countries: Bulgaria (BG), the Czech Republic (CZ), Hungary (HU), Romania (RO) and Slovakia (SK). The survey is both individual and household-specific and is representative in terms of the Roma population structure in the respective country (UNDP, 2002). The issue of main concern is the variable that expresses the financial dependency of the household on the state. Throughout the rest of the paper we will refer to this variable as 'state dependency'. We create it in two consecutive rounds. In the first round, we raise two questions (used for Table 3 and Table 4): Q1: 'From which of the following sources does the household usually receive money during the last 6 months? ' and Q2: 'Which of the above sources provides most money for the household?'. Note that the Q1 allows multiple answers, while the Q2 requires a single answer. These questions helped us to express not only the state budget dependency as such but also the level. The procedure is as follows: we took into account the respondents' answers only with respect to three types of state transfers: unemployment benefits, child support and social transfers. We categorised households into three groups according to their answers to these two questions.
A totally independent household is the one that does not receive any money from the government, i.e. answers' No' to both Q1 and Q2. The mid-level of dependency comprises households that are receiving at least one of the three types of state transfers, but transfers do not form the majority of the households' incomes (Answer 'Yes' to Q1, but 'No' to Q2). The highest level of dependency comprises households stating that the highest share of their income stems from one ofthe three state transfers, i.e. answer 'Yes' to both Q1 and Q2.
In the course of our work we revealed that there is a significant number of Roma households who state that their source of income is the pension benefit of a household member. This situation as such cannot be judged as state dependency if the household consists of retired people. However, in the case of Roma households we have found that such households, consisting of members in active age apart from the retired members, receive and even live on pensions. In fact, this is very typical of Roma who commonly live in three-generation households. Hence, the second revision was needed. In this round, we searched for a selection criterion to tell us whether a household is a 'deserved' receiver of a pension or whether the household is 'feeding' members in productive age, partially or mainly, from pensions. We used again the Q1, Q2 answers with respect to pensions, and the selection criterion was the absence of an unemployed member in the household. Hence, we abstracted from the pension receivers the households that had an unemployed member. This was a sufficient condition that the household had a member in active age who lived on pension benefits. This additional 'revision' increased the share of strongly dependent households about 4%, while it decreased the share of the independent households about 5%. The share of households across our dependency groups and countries is presented in Table 5. The distribution on the total level is apparently skewed in 'favour' of the dependent households. In the Czech Republic the distribution is almost symmetric, with small values on the tails and high values for the weakly dependent group, while in Slovakia the highest percentage is among the 'strong' category. In Romania the situation is reversed. There are several explanations why this is the case. The five countries can be divided into two subgroups: the wealthier ones (CZ, HU, SK, at present fresh EU members) and the less wealthy ones (Ra and BG).6 The three wealthier countries have a relatively small 'no' category. This is caused by the larger state budget devoted to the social net, which is able to lift more Roma households out of poverty. Slovakia, additionally, has the highest long-term unemployment rate and hence, more Roma need to be covered by the social safety net. On the contrary, the two poorer counterparts have weaker social safety nets; hence more Roma might have to rely on themselves.
In the next three parts, we will test the variable of state dependency against other household characteristics and discuss the differences among the groups. These differences reflect the impact of state transfers on Roma's behavb This division is supported by a brief look at the GDP per capita values. In the year 2001, the numbers reach 5,576 (CZ), 5,200 (HU), 3,793 (SK), 1,790 (RO) and 1,693 (BG) US dollars (Eurostat, New Cronos). iour. The analytical tool will be cross-tabulation of the constructed 'state dependency' variable against one other factor. We present only those results where significant differences among the groups existed. 7 Three aspects of household's lives are reviewed: education and employment, life style and civic activity. In the following parts, when we refer to some activity made by households we mean at least one member of the household.

Employment and education
Having quantified the variable of our main interest, the level of state dependency, we assess the differences of the three groups in terms of job search, as well as supporting the education of children in their household. Each of the five countries has its own strategy how to include Roma into the life of society. With respect to adults, there are many employment programmes or training activities organised by various NGOs as well as municipalities. The natural question arises whether Roma participate in them. The participation can be a sign for the government as it expresses the desire of Roma to work and to be active. A rather more challenging task is to start a small business. This requires more effort and self-confidence and hence is more valued. In Figure 4 we plot the shares of the Roma households that joined some organised training or employment programme and the shares of the Roma households that ever tried to start their own business. The two lines have reversed tendencies within the state dependency grouping. The -0-line, participation in the training or employment activities, is growing with a higher state dependency. On the one hand, approximately 15% of strongly dependent households attended some programme targeted at them, while only 7% of households participated which are independent. This is only obvious that the households that are independent of state transfers are those that are employed and do not need to participate in public works type of programmes. On the other hand, the households that are strongly dependent on state transfers show only little interest in own market activity (-eline); only 5% of them ever tried to start their own business, while the share of independent Roma households reached 20%, implying that one fifth of households was engaging in some kind of their own market activity. This is a good sign, particularly for the government, that they should pursue the development of small businesses for Roma, as this might be the way for them to escape the dependency trap from the state. This relationship can be supported by an example from Bulgaria. A small microfinance project aiming at around 100 Roma households helped them to become self-reliant on the basis of private small agricultural businesses (Ivanov, 2002). The definitive leader in organising employment programmes for Roma is Slovakia. The government created a net of public work programmes, like street cleaning or maintaining city parks, which were organised by municipalities. The Roma who attended were valued with an additional premium to their social transfers. As a result, in Slovakia more than one quarter of dependent households (27%) participated in some employment programme. Unfortunately, these types of jobs did not bring permanent employment, only 7% of those who attended those programmes admitted that they 'helped them substantially' to find a regular job.
With respect to young Roma, the main interest of the government should be good education aiming at future successful employment and independency of all citizens. The attitude towards education is established first of all in households. Hence, as a next step we discuss the Roma attitude towards school attendance of their children. Part of the reasons for Roma exclusion is their insufficient education. The survey asked what could be a justifiable reason for a boy/girl from the household not to attend school. The response to the statement 'I would not stop my child from going to school under any condition' was of particular interest to us. The results, for 'a boy', are depicted in Figure 5. 8 Interestingly, the mid-level group gained the highest share. The reason seems to be rather simple, as in some countries social assistance is bound to school attendance of children (e.g., in Slovakia). This result, however, might reveal another gap, namely that Roma do not believe that education matters. In Slovakia, the motivation rule in terms of eligibility for social assistance seems to be effective. Here, more than 70% of households, H The shares for 'a girl' were almost identical.

72
independently of the category, agree with the above-mentioned statement.
One of the reasons why Roma are not integrated into society is the perception of the majority population that they are inactive and not sufficiently educated. This obviously never holds for a whole nation or a whole group. The reason why the majority judges the Roma in this way is because they observe specific characteristics, such as their long-term unemployment and their children that are often repeating classes in school. The survey in this respect focused on the reasons for Roma children to attend special schools. 9 Table 6 presents the shares of households having a child in a special school. The dependent households (weak as well as strong) have a 4 times higher percentage than the independent ones. The 16% is an alanning number, and it is very unlikely that there are so many Roma children who are in need of special education. When searching for the parents' reasons, more than 60% ad-  mitted that the reason is only that the school programme is easier there. Here the government needs to improve its effort in integrating Roma children into mainstream education in order to assure adequate education to as many Roma children as possible.

Lifestyle
In this part, we assess whether there are some differences between the level of dependency categories and household expenditures and their attitude to health care. We assessed the households' shares of budget devoted to food and non-alcoholic beverages as well as the share spent on alcohol and cigarettes. Here, we did not find significant differences among the three categories. However, we were able to compare the shares of Roma with the national ones.
Roma spend a higher share of the household budget on alcohol and tobacco (8%) than the national number shows, which on average is 4% (unweighted average of the 5 countries). This is a little bit misleading as Roma also earn less compared to the majority of population, and therefore the shares might be the same or even less in nominal terms. According to the survey, Roma spend most of the income (more than 50% ) on food and non-alcoholic beverages. The national numbers are around 30%, which also supports the idea that this is mainly due to lower incomes among the Roma. III Remarkable differences were found in a completely unexpected area such as medical insurance and medical treatment. The results

Civic activity
The last issue that we cover is the civic activity, engagement or knowledge of societal events. To close the circle of a full and integrated life of Roma in society, they should not only be adequately educated, satisfactorily employed, in good health, but also they should participate in civic life. The survey posed the question of attendance of the last election. Table 7 reviews the results. Expectedly, the independent households had more than 10% higher shares than the dependent ones. The knowledge of Roma political representatives implies some level of interest and trust. In Figure 7, the -0-line indicates the share of Roma who have responded that there exists a Roma political party which they can trust. It seems that independent Roma have more trust in their representatives, what implies also a higher interest in the political scene. The -.-line in Figure 7 describes the positive answer to the question whether the household knows any programme targeted at Roma. Here the tendency is reversed: the highest share is among the weak dependent group. This is mainly due to the high attendance to training and employment programmes, which was already mentioned above.
To conclude this section, dependent Roma households are less interested in politics and their own representatives than independent households. However, they are more familiar with some programmes targeted at them, which is mainly due to their higher participation in active labour market programmes.

Conclusion
In this paper we reviewed the social systems in five Central and Eastern European countries: Bulgaria, the Czech Republic, Hungary, Romania and Slovakia. We used the UNDP/ILO survey data on Roma households to link the social policies and their impact on Roma behaviour. In the review, the focus was on the social protection system for people with insufficient income. We paid special attention to whether the system is efficient in terms of providing incentives to look for a job. Furthermore, we discussed whether the system covers the people's needs if they need to rely completely on the system.
The social system of these five countries varies according to the country's status. In the new EU member states, the Czech Republic, Hungary and Slovakia, the system is more generous and helps to provide households in need with subsistence income. However, the system does not always motivate people covered by the social net to search for a job. Particularly, if the household has more than three children it is not favourable to work for the minimum wage. The recent reform in Slovakia aimed at reducing these disincentives and making the system more efficient. Nevertheless, this efficiency only holds in presence of employment opportunities, what is not necessarily the case, especially for the Roma minority. The systems that are in place in Bulgaria and Romania do not necessarily provide sufficient income to households in need. The levels of social assistance are very close to minimum subsistence levels, forcing households to look for additional employment, which cannot be considered as an incentive effect but rather as a motivation for survival.
In the empirical part, we reveal some of our initial hypotheses. In the wealthier countries there is a significantly higher percentage of Roma who live partly or exclusively on the transfers. In the two less wealthy countries, social benefits are not necessarily sufficient for subsistence, their ratio is remarkably lower (Table 5). This does not mean that the way to motivate Roma to look for employment should be low benefits, because this can rather lead to counter-effects, such as criminality and participation in the informal economy. What seems to be more effective is to help people who are interested in starting their own small businesses, e.g., in agriculture. The dependent households are less active in terms of self-reliance, what is supported by the small percentage of their own established businesses ( Figure 4). Simultaneously, there is a pertaining negative tendency of misplacing their chil-dren in special schools, which causes even less chances for them to escape the dependency trap. They are also less active in civic events. On the contrary, it is optimistic that they use external offers to join some employment or training programmes in a larger proportion than the independent households. Therefore, it may be worth considering establishing active labour market policies that provide em-