Belgrade EXECUTIVE SUMMARY Analysing Income Inequality in Serbia From Data to Policy MIHAIL ARANDARENKO, GORANA KRSTIĆ AND JELENA ŽARKOVIĆ RAKIĆ November 2017 n  H igh income distribution inequality is bad for economic growth, both directly and as a potential source of significant political and social instability. Serbia has higher income distribution inequality than any country of the European Union, which is the result of low redistributive power of taxes, social transfers and a troubled labour market situation. n  I n Serbia income inequality is greatest in households with a very low labour intensity whose members do not work or work very little(less than 2.5 months a year). n  C oncrete policy reform proposals to decrease income inequality include: increasing employability options for those with a low level of education, increasing the number of good quality full-time jobs in the formal sector, increasing progressivity within personal income tax and improving pension coverage for the female population.

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA High income inequality hurts economic growth and presents a potential source of political and social instability It is now widely accepted that high income inequality is detrimental to economic growth both directly and as a potential source of significant political and social instability(IMF 2014; OECD 2015). For this reason the topic is becoming more present in global public discourse. Judging by the messages coming out of the latest World Economic Forum in Davos, the global elite is also worried about the extent and pace of acceleration of income and wealth inequality. They warn against societal polarization and income inequality spilling over into real-world politics as recent electoral results in the United Kingdom and United States suggest. These trends are likely to have a lasting impact on the way economies act and relate to each other. In most developed economies income inequality has increased during the last three decades. Data for the majority of the OECD countries show that the Gini coefficient has risen by ten percent since the mid-1980s(OECD 2011). The economies of Eastern Europe and Asia have also witnessed an increase in income inequality, while in Latin America inequality has decreased but is still reaching the highest levels in the world(IMF 2015). In rich and emerging economies inequality surge is mostly due to an increased share of the top 10% earners in the total income and a parallel squeeze of the middle class. The improved position of those at the bottom and middle of income distribution was responsible for a reduction in inequality in Latin America. The role of globalization, technological progress and reduced impact of trade unions on the rise in inequality in developed economies The latest research reveals factors on a global level that have, on the one hand, contributed to the real income growth of the Chinese upper middle class and of wealthy individuals originating from the United Kingdom, United States and other developed countries of the West, and on the other hand, that have caused real income stagnation in the lower middle class in developed Western countries, mainly from Great Britain and the United States. These factors are globalization and technological progress which have benefited highly-skilled workers in developed countries by creating new professions while moving jobs, mostly intended for a workforce with lower qualifications, to China or making them completely redundant(Milanovic 2016). Since the 1980s these processes have been combined with policies inspired by Thatcherism and Reaganomics, resulting in privatization, deregulation, reduced tax burdens for the wealthiest and the diminished impact of trade unions. Some of the policy proposals aimed at reducing income inequality in developed countries include the introduction of basic income, universal provision of child allowance and an increase in capital ownership for a larger number of people (deconcentrating capital ownership). Many prominent thinkers also believe that managing the process of technological change should be an explicit concern of policy makers who should encourage those innovations which increase the employability of workers. Additionally, the distributional consequences of international trade should be carefully addressed by policy makers. In other words, programs that directly tackle the needs of those that are the losers from international trade should be considered. Those include active labour market programs such as training schemes, public works programs and employment subsidies (Atkinson 2015; Piketty 2015; Milanovic 2016). The rise in income inequality in Serbia Regarding Serbia, the data suggest that income inequality has increased since 2000 and is currently the highest among European countries. The Gini of 38.6 points in 2016 is significantly higher than the average Gini for the EU-28 countries(31) and is also higher than in neighbouring countries such as Macedonia(35.2) and Croatia(30.6), and in particular greater than the Gini value for Slovenia(24.5). 2

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA Figure 1. Gini Coefficient for European Countries, 2015 Iceland Slovakia Norway Slovenia Czech R. Finland Sweden Belgium Netherlands Austria Denmark Malta Hungary Luxembourg France Germany Croatia Poland EU-28 average Italia UK Cyprus Portugal Greece Spain Estonia FYROM Latvia Bulgaria Romania Lithuania Serbia 0 5 10 15 20 25 30 35 40 45 Note: Gini coefficient for equivalent disposable income. Source: SILC 2015, Eurostat. For Macedonia Gini coefficient is for year 2014 Despite this, policymakers’ interest in the problem of inequality has been quite limited. There is no systematic official procedure for the assessment of the distributional impact of policy reforms measures. The Government’s main concern, recently, has not been to tackle inequality and poverty, but rather to carry out fiscal consolidation, preserve macroeconomic stability and incentivize growth and investment. 1. Arandarenko, M, Krstic, G i ZArkovic Rakic, J. e. 2017. Dohodna nejednakost u Srbiji. Od podataka do politike, Friedrich-Ebert-Stiftung, Beograd. At the same time, academics’ interest in the income inequality topic has been rather meagre. Although limited and fragmented, previous research points to possible determinants of high inequality in Serbia- low redistributive power of taxes and social transfers, on the one hand, and a troubled labour market situation, on the other (Milanović 2003; Krstić& Sanfey 2011; Ranđelović & Žarković-Rakić 2011). What drives high income inequality in Serbia? In the study“Dohodna nejednakost u Srbiji. Od podataka do politike” 1 we analyse the current state of inequality in Serbia and discusses its trends over the last decade using EU member states and neighbouring states as comparator countries. By calculating the disposable(posttax-benefit) income inequality and the market (pre-tax-benefit) income inequality in Serbia and by comparing these results with peer countries, the objective is to establish whether weaker redistribution through taxes and social transfers is the main cause of much higher inequality of disposable income in Serbia compared to that of EU countries or whether it is due to a more unequal distribution of market(pre-tax-benefit) income. The study also looks at different sources of income in order to investigate which of those sources contribute most to high income inequality. Income sources are wages, self-employment income, capital income, pensions and welfare benefits. Special attention is given to exploring the linkage between work intensity of individuals and income inequality, given that the number of people with a very low work intensity, that is, those who work on average less than 2.5 months per year, is continually rising. Income inequality is analysed based on the Survey of Income and Living Conditions(SILC) conducted by the Statistical Office of the Republic of Serbia (SORS) since 2013 according to the standardized Eurostat methodology. This survey has collected information on total household income and of 3

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA household members, and on income components, and therefore presents the best source of data for measuring inequality according to the official methodology of the SORS(and Eurostat). The baseline aggregate for analysing inequality is disposable net household income. This is monetary income received by households, which is available for consumption and saving, after payment of taxes and social contributions. Disposable household income is then divided by an OECD equivalent scale, in order to take into account composition of household and economy of scale, to obtain equivalized disposable income which is the base aggregate for inequality analysis. The study provides several conclusions through monitoring income inequality over the last four years, measured by the Gini coefficient, and by comparing the data for the previous period:  he value of the Gini coefficient spanning 38.0 to 38.6 for the period 2013-2016 indicates a relatively stable but high income inequality in Serbia;  here was an increase in inequality over the period 2006-2016, but growth cannot be quantified because it is partly a result of varying data sources used for the analysis;  of income over the period 20062013 was probably higher than the inequality that could have been monitored on the basis of the Household Budget Survey, but until the emergence of SILC in 2013 there were no adequate survey data designed for the coverage of income that would have shown this(other than LSMS data);  value of the Gini coefficient in Serbia (regardless of survey type) over the period 2006-2016 was above the value of this indicator for the EU average(31%). High percentage of people with very low work intensity To provide an answer to the question of how to reduce income inequality in Serbia, the study considered the potential causes of high inequality. One of the causes of high inequality in Serbia is the high percentage of persons who have a very low work intensity. The share of these persons aged up to 60 years old in Serbia(21.2%) is much higher than the European average(10.5%) and individual EU countries,(with the exception of Ireland at 21.1%). The high rate of very low work intensity of household members in Serbia was due to the high inactivity of the working-age population(15-65 years old), as well as to the fact that a low percentage of such persons live with other adults who work. Serbia has the largest share of people who do not work(50%), especially the unemployed, in the working-age population compared to EU countries. SILC data for Serbia suggest that inequality by work intensity of household members indicates that income inequality is the highest in households whose members do not work or work very little (less than 2.5 months per year) and it decreases with an increase in work intensity of household members. This suggests that a significant reduction in income inequality could be achieved by employing unemployed or inactive working-age members from these households or by increasing the working days/months of employed members. Furthermore, the results indicate that most of the income inequality can be reduced in such a way. An analysis of income inequality(measured by the Gini coefficient) by income sources shows several important findings. First, wages contributed most to total inequality(93%) and increased it. All other income sources, other than capital income which influenced very little, such as self-employment income, pensions, welfare benefits, decreased inequality. The second finding of interest is that taxes have the largest impact on reducing inequality. This is followed by social transfers other than pensions, while income from self-employment and private transfers is less important. 4

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA As compared to EU countries, all income sources 2 decrease income inequality as in the EU, the only difference appears in income from selfemployment. In most EU countries income from self-employment increases inequality(ILO 2015), while in Serbia it reduces it. This is because most of the self-employed in Serbia earn a low income, which is not the case in more developed EU countries. If income from self-employment is unequally distributed towards better off households in Serbia, as are wages, it would increase income inequality like in EU countries. 3 Although the Gini coefficient of disposable income in Serbia is significantly higher than the average for the EU-28, but also higher than in individual EU countries, it is interesting to note that the Gini coefficient for market income is within the EU average(55.1 vs 55.2). This indicates that the main reason for such a high inequality of disposable income in Serbia is the low redistributive role of social transfers and taxes. Low coverage of pension for female population Where social transfers are concerned, the largest difference in the impact on income inequality in relation to the EU is recorded for pensions. The effect of pensions on inequality is significantly lower in Serbia than in the EU(10.9 vs 17.2). One possible explanation is the smaller coverage of the population by pensions in Serbia than in the EU, since 93% of men and only 79% of women in Serbia receive a pension according to 2012 Census data. 1. Except wages whose effect on inequality is not considered, as wage distribution is an initial, reference distribution. 2. Income from self-employment can decrease or increase inequality like any other income source. Income from self-employment in Serbia is predominantly distributed towards poorer households, which decreases inequality. Let’s assume that initially 30% of households with the lowest income receive 15% of total income. After receiving income from selfemployment, 30% of households will receive, for example, 18% of total income. Lorenz curve moves closer to the line of 45 degrees(where the same% of households receive the same% of total income G=0), which means that inequality decreases. 4. Average amount of monetary social assistance per household was 33% of the minimum wage and average amount of child benefit per household was 20% of the minimum wage. Low progressivity of the income tax system and low spending on social transfers Other social transfers are slightly less efficient in reducing inequality in Serbia than in the EU(3.5 vs. 3.7 percentage points, respectively) which is the result of the low coverage of monetary social assistance and child benefits and the low amounts of these benefits in Serbia. 4 This difference is even higher for taxes(2.7 vs. 4), which is explained by the very low level of progressivity of the Serbian income tax system. Larger number of good quality jobs could reduce income inequality in Serbia… First of all, the troubled labour market situation is mirrored by the increasing number of individuals aged under 60 with very low work intensity(Krstic 2016). The majority of them are unemployed or pensioners, and most have low levels of education. The results show that the biggest impact on reducing income inequality is achieved by a reduction of income inequality within households with a different work intensity. Boosting employment might be expected to reduce income inequality as the number of people with no salary or relying on unemployment benefits falls. However, it is important to increase the number of good quality, full time jobs in the formal sector as only these can lead to decent wages, provide social security coverage and offer good working conditions. Policies that contribute to greater employability of individuals with a very low work intensity are particularly important. These include active labour market policies conducted through the National Employment Service such as employment counselling, trainings, public works and wage subsidies for new employment. Spending on active labour market policy measures should be increased, as it currently stands at a mere 0.1% of GDP. 5

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA …as well as equal access to good quality education for all It is, however, important to act before individuals enter the world of work and that means providing access to high quality education for the largest possible number of people. When discussing inequality, education is important not only because a higher level of education increases the probability of having a better paid job, but also because education should prevent the reproduction of inequality. It should level the academic achievement of pupils coming from different socio-economic backgrounds so that children of economically disadvantaged parents could have equal chances of finding a good job and earning a decent living when they grow up. The current education system in Serbia has failed to offset the effects of socio-economic inequalities. Indeed, it has even deepened the inequality. The results for Serbia in the PISA survey suggest that families in the bottom 20% of the population, with the lowest socio-economic status, have twice as many functionally illiterate children as the wealthiest 20% of families. 5 Children in the former group lag two whole school years behind their peers from well-off families. The support mechanisms for pupils and students usually focus on top performers, rather than those who need assistance because of their socio-economic status. A major overhaul of the mechanisms of scholarships, student loans and cost of living assistance to students is needed(Čekić, Marković & Jokić, 2015). Higher progressivity of the personal income tax system could reduce inequality… In order to reduce inequality, Serbia’s income tax could be made more progressive by introducing tax deductions for supported family members (at present this option is available only to those who pay annual income tax, i.e. those with a high income) and by increasing the tax-exempt 5. http://www1.worldbank.org/poverty/visualizeinequality/PISA/ cov_gaps.html threshold from the current 25% of the average wage to 50%(Arandarenko& Vukojevic, 2008). Crucially, however, the existing so-called schedular tax system, which is becoming increasingly rare in modern tax systems, should be replaced with a system that integrates income from labour and capital and applies progressive tax rates, which should range from 15% to 30%. …as well as the reform of the system of social transfers Low amounts and limited coverage of the two main social benefits in Serbia, monetary social assistance and child allowance, hamper their effectiveness in reducing inequality. While it would be unrealistic to expect an increase in social spending in the coming years, not least because of austerity measures, it is nevertheless possible to promote better targeting of benefits by ensuring less leakage of resources to not-sopoor individuals(e.g. those who earn income in the shadow economy) and diverting the funds to those who need them most. Although the reduction of wages in the public sector and the reduction of pensions in 2014 protected employees and pensioners with the lowest incomes… Recently Serbia has seen important fiscal consolidation measures and far reaching changes in labour legislation, with most important among them having a positive impact on income inequality. On the other hand, the foundations of the tax system, and the labour taxation system in the first place, which we have singled out as one of the main culprits for very high income inequality, have remained almost unchanged. Among the measures of fiscal consolidation, the reduction of wages in the public sector and pension cuts are certainly the two most important measures, with an anticipated significant impact in the direction of reduction of income inequality, especially in the case of wages cuts. 6

Belgrade ANALYSING INCOME INEQUALITY IN SERBIA …other recent reform measures have mostly had a negative distributive effect On the other hand, most other fiscal consolidation and structural reform measures have dominantly worked in the direction of increased inequality. As shown for the measures we have analyzed, although their individual reach was limited, their cumulative effect was strong enough to prevent the Gini coefficient and quintile ratio from falling in 2015. Actually, it could be said that in the longer term perspective the manner in which the two key fiscal consolidation measures, public sector wage reduction and reduction of pensions. have been designed represents a relatively rare exception from the general rule that reforms being carried out either increase or, at best, have an ambivalent or neutral impact on income inequality. One of the similar exceptions was the labour taxation reform in 2007, when the relatively modest zero tax bracket was introduced, slightly improving the hitherto regressive system of labour taxation characterized by a single proportional tax rate and no personal deductions at all. After the introduction of the zero tax bracket, to the amount of 18% of the average gross wage, labour taxation became slightly progressive. to develop and adopt systematic formal procedures for the evaluation of the distributive impact of any future reform measure. …as well as to develop public discourse for identifying and understanding social and economic inequalities Public discourse related to the perception and understanding of social and economic inequality is underdeveloped in Serbia. Although preferences toward greater equality are dominant among the population at the level of individuals, as elsewhere, they are not articulated enough into a set of social preferences moving toward the reduction of income inequality and the promotion of equality of possibility. We hope that our analysis will go some way to providing a contribution in that direction. It is necessary to systematically ex-ante assess, and ex-post evaluate the impact of reforms on inequality and poverty... The fact that two key and socially most painful fiscal consolidation measures at the end of 2014 were designed in such a way as to preserve the living standards of the most vulnerable parts of affected populations(public sector employees and pensioners with low pension incomes), can thus be ascribed to certain random circumstances, such as enlightened decision makers(in the first place within the Finance Ministry) or specific political economy considerations(for example anticipation of parliamentary elections), rather than to systematic assessment of their impact on inequality. This certainly should not be the standard practice. Instead, it would be necessary 7

About authors Mihail Arandarenko is president of the Board of directors of FREN and full professor at the Faculty of Economics, University of Belgrade. Gorana Krstić is member of the Board of directors of FREN and full professor at the Faculty of Economics, University of Belgrade. Jelena Žarković Rakić is director of FREN and associate professor at the Faculty of Economics, University of Belgrade. Imprint Friedrich-Ebert-Stiftung| Belgrade Office Dositejeva /51/1 11000 Belgrade Responsible: Ursula Koch-Laugwitz| Director, Regional office for Serbia and Montenegro Tel.:++381(11) 3283 271| Fax:++381(11) 3283 285 www.fes-serbia.org To order publications: info@fes-serbia.org Commercial use of all media published by the Friedrich-EbertStiftung(FES) is not permitted without the written consent of the FES. The views expressed in this paper do not necessarily reflect the views of the Friedrich Ebert Stiftung or the organization which the authors represent.