SOCIAL WELFARE POLICY PREFERENCES IN LATVIA: EVIDENCE FROM ISSP SURVEYS

This study seeks to find out how social welfare policy preferences have changed over time and what factors account for those preferences in Latvia. The author analyses ISSP survey data gathered in 1996, 2007 and 2016. The data analysis shows that most Latvians still support government interventions in providing social welfare. However, economic factors like material wellbeing and self-interest have decreased the overall support for social welfare policies during the last 20 years. The article provides a long-term perspective missing in previous studies on social welfare policy preferences in Eastern Europe. UDC Classification: 316.4; DOI: http://dx.doi.org/10.12955/cbup.v6.1232


Introduction
The focus of this article is individual level preferences towards various social welfare policies in Latvia. Public attitudes towards welfare policies are a frequent subject in scientific literature. Many surveys have been carried out to test various hypotheses why certain societies or social groups are more or less likely to support particular welfare policies (See Blekesaune & Quadagno, 2003;Butkeviciene, 2012;Deeming, 2016;Lipsmeyer, 2003;Morelock, 2016;Svallfors, 2004). Blekesaune and Quadagno (2003) analyse ISSP (International Social Survey Programme) survey data from 24 nations gathered from 1995 to 1997. They argue that both national and individual level variables shape people's attitudes towards welfare policies. Individuals may be influenced by selfinterest or ideological preferences (Blekesaune & Quadagno, 2003, p. 415). The unemployed may view government interventions more favourably because they are the recipients of welfare benefits. Women are more inclined to support welfare programs because they have certain values and expectations towards state, i.e. ideological preferences (Blekesaune & Quadagno, 2003, p. 415). When measuring the impact of national level variables, Blekesaune and Quadagno find out that nations with a more distinct egalitarian ideology and high unemployment tend to be more positive towards welfare policies (Blekesaune & Quadagno, 2003, p. 424). Svallfors (2004) studied data sets obtained in ISSP surveys in Sweden, Germany, Britain and the US in 1996. He argues that one's class status determines one's attitudes towards welfare policies. Those belonging to the so called service classes are less supportive of government intervention in providing welfare than, for example, unskilled labour. In other words, more well-off classes are less inclined to support social welfare measures. Deeming (2016) studied ISSP survey data gathered in 11 Western countries in 1985Western countries in , 1990Western countries in , 1996Western countries in and 2006. He argues that attitudes towards welfare and government strategies have become more polarized. Deeming suggests the dominant welfare state ideology may account for that (Deeming, 2016, p. 178). There is more support for social welfare programmes in Nordic countries whereas liberal market economies (Australia, the UK) have become less supportive of unemployment protection and government intervention in the labour market (Deeming, 2016, p. 178). A socioeconomic status of individuals also matter. For example, more affluent Australians "appear to be adopting a tougher line on welfare" (Deeming, 2016, p. 179). Lipsmeyer (2003) studied ISSP survey data from Bulgaria, the Czech Republic, Hungary, Latvia, Poland, Russia, and Slovenia obtained in 1996. She notes that most respondents in the above countries support a considerable government role in all welfare policies but the support varies depending on specific policies (Lipsmeyer, 2003, p. 552). Respondents make distinctions and there is hardly any "unanimous consent for government-controlled social policies" (Lipsmeyer, 2003, p. 552). Lipsmeyer points out that helping the unemployed receives little support. In turn, helping the old is considered to be the government's job. She also finds out that there are differences between various age and income groups. Older cohorts favour greater spending on pensions. The poor are more likely to support greater spending on the unemployed (Lipsmeyer, 2003, p. 556). Butkeviciene (2012) analysed the results of two surveys carried out in 2010 and 2011 in Lithuania. She points out that Lithuanians have great expectations for the government in social provision and welfare (Butkeviciene, 2012, p. 7). The government is expected to help the old, the unemployed, the sick and the disabled (Butkeviciene, 2012, p. 7). However, respondents are not consistent in their views. Lithuanians rely on the government in times of distress (unemployment, illness) but they would rather choose to reduce taxes even it would mean cutting social services (Butkeviciene, 2012, p. 15). Morelock's (2016) dissertation has been arguably the most elaborate study of the subject so far. He studied ISSP survey data from 23 nations including a very diverse set of countries like Latvia, the US, Japan, Switzerland etc. The data come from the ISSP surveys obtained in 2006. He points out that respondents from Southern and Eastern Europe are more likely to support social welfare policies whereas those from Western Europe and North America are less supportive of them (Morelock, 2016, p. 180). Morelock argues that the support for social welfare policies is "a result of people's sociodemographic attributes, socio-psychological attitudes, and views toward the public sector" (Morelock, 2016, p. 180). Attitudes are shaped by political party affiliation and economic self-interest, trust in civil servants and perception of governmental performance (Morelock, 2016, p. 180-181). In other words, a person with the right leaning views, driven by economic self-interest may be more sceptical about government interventions in providing social welfare. However, those who trust civil servants and find government to be doing the right thing are more supportive of social welfare policies. To sum up, people in post-socialist countries tend to support government social welfare policies whereas respondents from liberal economies are more reluctant to do so. Self-interest matters, too. Those who benefit or might benefit from those policies tend to support relevant measures. This research study seeks to find out how public attitudes towards social welfare policies have changed over time and what factors account for social welfare policy preferences in Latvia. The author analysed ISSP survey data gathered in Latvia in 1996Latvia in , 2007Latvia in and 2016. He argues that Latvians remain very supportive of government interventions in providing social welfare. However, economic factors like material wellbeing and self-interest have decreased the overall support for social welfare policies over the time span of nearly 20 years. The article provides a long term perspective missing in previous studies on social welfare policy preferences in Eastern Europe.

Background information on economic and social development in Latvia
Since 1996, Latvia experienced a rapid economic growth. In 1996, GDP per capita was 1904 EUR (in the current prices) while in 2016, the figure reached 12762 EUR (Central Statistical Bureau of Latvia, 2017e). It should be pointed out that there are considerable regional differences in this regard. In Riga (the capital city), GDP per capita stood at 19 912 EUR but, in the poorest region Latgale, the figure was a meagre 5981 EUR (Central Statistical Bureau of Latvia, 2017d). The economic growth created jobs as well. The employment rate in the age group 15-64 grew from 57% in 1996 to 68% in 2016 (Central Statistical Bureau of Latvia, 2017a). The structure of the labour force changed, too. The share of persons with university diploma has increased from 19.6% in 1996 to 37.1% in 2016 (Central Statistical Bureau of Latvia, 2017a). Other figures are less spectacular. The economic development brought wage increases but the average gross monthly wage in 2016 was still only 856 EUR (Central Statistical Bureau of Latvia, 2017b). In this respect, it should be noted that those with bachelor's degrees received significantly more than those with solely primary education. In 2014 (when the most recent detailed information on wages was available), a person with a bachelor's degree earned 1133 EUR (monthly gross earnings) while in turn, those with just primary education earned only 640 EUR (Central Statistical Bureau of Latvia, 2017c). The unemployment rate remained high and volatile throughout the time period from 1996 to 2016. In 1996, the unemployment rate stood at 20.5%. Then it dropped to 6.2% in 2007. At the bottom of the economic crisis in 2009, the unemployment rate reached 19.8%. Since the beginning of the recovery in 2011, the rate decreased to 9.9% in 2016 (Central Statistical Bureau of Latvia, 2017a). Material deprivation has been a serious problem for many people throughout this period. Comparable statistics for material deprivation rates are only available from 2005 when the rate stood at 56.8%. In 2016, the material deprivation rate was still high, i.e. 26.4% (Eurostat, 2017b). A number of factors account for the deprivation. First, low wages as indicated above. Second, compensation mechanisms like social transfers are missing or inadequate. In 2006 (when the last comparable statistics were available), the share of GDP allocated to social protection was 11.6% while in 2015, it was 14.9% (Eurostat, 2017a). Similarly, Latvia allocates very little to health care. Data from the World Bank indicates that the share of GDP spent on health care remained stable around 6% over the period from 2005 and 2014 (when comparable statistics were available) (World Bank, 2016). At last, Latvia was and is still an unequal society. The Gini coefficient slightly decreased from 36.5 in 2004 (when the last comparable statistics were available) to 34.5 in 2016 (Central Statistical Bureau of Latvia, 2017f). As indicated above, differences in wages and regional disparities may account for some of that inequality. To sum up, Latvians are much richer on average in 2016 than in 1996. There are more jobs for more people as well. However, large sways of the population remain poor because they earn very little and/or receive inadequate benefits and services (like health care) from the government or municipalities. There are also considerable disparities between regions, educational and income groups. This could be one of the reasons why many people in Latvia support a greater government role in providing help for those in need.

Methodology
Survey data analysed in this article were gathered in 1996, 2007 and 2016 in Latvia in the framework of the International Social Survey Programme (ISSP). Samples were representative to the 18-74-yearold population in Latvia. The author analyses only those questionnaire items that were included in all three surveys. ISSP studies have a core of items that do not change and also batteries of questions that are included once or twice. The author did not include the latter items because one or two measurements are not enough to establish a trend. All data presented in the following tables are weighted to adjust for the under-coverage of some social groups (young males, persons from nontitular ethnic groups).

Empirical results
In 1996, 2007 and 2016, a battery of statements about social welfare policy preferences was included in the questionnaires (see Table 1). Respondents were asked to evaluate whether the government should be responsible for providing jobs, health care, and housing for those in need as well as reducing income differences and keeping prices under control. A four point Likert scale was used to identify the support for the government's intervention in the above policy areas where 1 represents "definitely should be" and 4 -"definitely should not be". It should be stressed that smaller values indicate a greater support for government intervention. Positive differences in means should be interpreted as less support for government intervention when comparing the data from 1996 and 2016. Negative differences in means indicate more support for a greater government role in a particular policy area. Table 1 provides the descriptive statistics of those responses. Except one policy area, the overall support for government interventions in various social welfare policy areas has slightly declined over the last 20 years. However, it should be stressed that the mean values below 2 identify strong preferences for a significant government role, for example, in providing health care, helping the old and students from low income families. The exception to the trend is an item related to the government's role in reducing income differences. In 1996, there was no demand for a more even income distribution. In 2016, most Latvian respondents believed that the government should assume a greater role in tackling the inequality problem. No systematic study has been carried out in Latvia to find out why the income inequality has become a more relevant topic for many Latvians. As pointed out above, Latvia has long been an unequal society. One of the explanations for this change might be a public debate on taxation and a proposed tax reform (adopted by Latvian parliament in 2017), which subsequently lead to the introduction of a more progressive personal income tax (Swedbank, 9.08.2017.).
To identify what accounts for social policy preferences, the author created a dependent variable from the above items by using a principal component analysis. Missing values were replaced with mean values because, if they were not included, the effective sample size would have been reduced by one third. Only factors with eigenvalues greater than 1 were extracted (rotation method -varimax with Kaiser normalization). The analysis identified a dimension with initial eigenvalues 3.038 (1996), 3.545 (2006) and 3.571 (2016). Five of the eight items strongly load on the dimension, i.e. a preference for a greater government role in particular social welfare policy areas. Factor scores were saved using the regression method. They are used in multiple linear regression analysis presented in Table 3. Factor loadings, the values of Cronbach's alpha and relevant descriptive statistics for the dependent variable are included in Table 2. Source: Author In the subsequent multiple regression analysis, the author uses only those questionnaire items that 1) were included in all three ISSP surveys and 2) correlated with the dependent variable measuring social welfare policy preferences. Items have different scales (a number of points, scale direction), therefore, a number of items were dichotomized for the simplicity and robustness of interpretation (see Table 3). A positive unstandardized beta value indicates that a particular independent variable has a positive effect on social welfare policy preferences. Negative beta values indicate an opposite effect, i.e. a decreased support for government intervention in particular social welfare policy areas. Factors like education and belonging to a non-titular ethnic group have a strong effect on the social policy preferences over the last 20 years. Those with higher education are less supportive of government intervention. In the Latvian context, higher education is also a proxy variable for one's income. A large proportion of respondents refuse to or provide answers like "hard to say" when they are asked to indicate their incomes (also in the ISSP surveys) but official statistics suggest higher levels of education often correlate with higher incomes. Those respondents belonging to non-titular ethnic groups (Russians, Ukrainians, Poles etc.) strongly advocate for a greater government role in providing help for those in need. There are no statistical data or a study that would compare incomes of ethnic Latvians and non-titular ethnic groups. However, some empirical evidence suggests that nontitular ethnic groups are more likely to point out to the prevailing social injustice in Latvian society (Bela, 2014, p. 30). The significance of other factors has varied. Residence in the capital city (more affluent with less unemployment) was a statistically significant factor in 1996 and 2006 but not in 2016. Similarly, trust in civil servants had an effect on one's social policy preferences in 1996 and 2006 but not anymore. A good understanding of political issues previously had a negative effect one's preferences, i.e. those with a supposedly good understanding of politics were less supportive of government intervention. In 2016, it had no impact on one's social policy preferences. In 2016, older respondents, the employed and students were less likely to support the above social policies. In turn, females and those trusting MPs were more supportive of a greater government role in providing social welfare. These changes are hard to explain without necessary empirical data and in-depth analysis. However, it should be emphasized that none of the above regression models do a particularly good job of explaining social policy preferences. The values of determination coefficients range from 0.091 in 1996 to 0.16 in 2006. The author concludes that sociodemographic variables are not sufficient to explain one's social policy preferences. Unfortunately, the core of questionnaire items included in all three ISSP studies is too small to find out which non-demographic factors (attitudes, values) may account for the rest of the variance.

Discussion
The analysis presented above, at least in part, supports the thesis that material well-being and selfinterest are contributing factors that accounts for the decline of the overall support for social welfare policies in Latvia. Latvians are now better off and there are fewer would-be welfare recipients. This argument is in line with the self-interest thesis advocated by many economists (see Blekesaune and Quadagno, 2003). However, most Latvians still favour government intervention in providing social welfare. The author's empirical analysis supports the arguments put forward by Lipsmeyer (2003), Butkeviciene (2012) and Morelock (2016). They point out that there is greater support for social welfare policies in post-socialist countries than in Western countries. Unemployment, poverty, inadequate public services are still problems often encountered in most post-socialist countries. Two socioeconomic variables have a consistent impact on people's social welfare preferences in Latvia, i.e. education and belonging to a particular ethnic group. In the Latvian context, attained education levels correlate with income levels. The highly educated often enjoys higher incomes. They are less inclined to support social welfare measures. There is no evidence that non-titular ethnic groups are discriminated in terms of payment or access to social welfare programmes and public services. They are no more or no less likely to be welfare recipients. However, it seems that those groups are more supportive of social welfare policies. Further studies would be needed to establish why it is so.

Conclusion
This study sought to find out how social welfare policy preferences have changed over time and what factors influence them in Latvia. The author analyses ISSP survey data gathered in 1996, 2007 and 2016. Thus, the article also provides a long term perspective missing in previous studies on social welfare policy preferences in Eastern Europe. The data analysis shows that the majority of Latvians support a greater government role in providing social welfare. However, the overall support for social welfare policies have declined over the time period of nearly 20 years. The empirical analysis shows that material wellbeing and self-interest are contributing factors to such a decline. The author also identified a factor previously overlooked in other similar studies, i.e. belonging to a non-titular ethnic group. Respondents in non-titular ethnic groups are more likely to support government interventions in providing social welfare.