Is the neighbour’s grass greener? Comparing family support in Lithuania and four other New Member States

Abstract

To what extent can a country’s effectiveness in reducing child poverty be attributed to the size of family cash transfers (that is, both benefits and tax advantages) or to their design? In this paper, we disentangle the importance of each of these two factors, focusing on the family support system in Lithuania and comparing it with four other new member states. Both single parent and large families are increasingly susceptible to poverty in Lithuania. This contrasts with other former communist countries, namely Estonia, Hungary, Slovenia and the Czech Republic, which protect these family types much better. This paper examines whether their family transfer systems would achieve similar results in Lithuania. We employ the EUROMOD microsimulation tax-benefit model to swap family policies across countries and test whether size or design has greater effects on child poverty reduction in Lithuania. Hungarian, Slovenian and Czech policies would considerably improve the poverty situation among large families. Single parent families would only gain if Lithuanian spending on family transfers increased radically. Estonian policies would lead to mixed results: small gains for large families and losses for single parent families.

Keywords

child poverty family benefits and tax advantages Lithuania microsimulation new EU member states

Introduction

Child poverty remains a serious problem across the EU, and especially in the new EU member states (NMS), albeit with significant variations in extent and intensity. Children in single parent and large families are the subject of particular policy concern, given that about half of the poor children in the EU live in these two types of households (Commission of the European Communities, 2008). Compared with other EU countries, Lithuania has ‘below-(EU) average performance in all dimensions of child poverty and well-being, and particularly in terms of risk of poverty’ (TÁRKI, 2011). Poverty is especially concentrated among single parent households and households raising three or more children. The design of the Lithuanian family support system has been criticized from the perspective of poverty effectiveness despite numerous past and recent reforms (Cornelius, 1995; Kabašinskaitė and Bak, 2006; Salanauskaite and Verbist, 2009). The implemented policy reforms have not been effective at reducing poverty, especially when compared with the achievements of other new EU member states (NMS), such as Estonia, Hungary, the Czech Republic or Slovenia (TÁRKI, 2011).

Most research on the poverty effectiveness of family support tools has concentrated on Anglo-Saxon countries and ‘old’ EU member states (Kamerman et al., 2003; Levy et al., 2007; Matsaganis et al., 2007). Research within the NMS region is still quite scarce, Förster and Tóth (2001) being one of the few examples. The region is however very interesting not only because of the fast changing socio-economic environment and demographic conditions, but also because of recent reforms in family policy. (Relative) child poverty rates in some of the NMS are actually lower than in a number of richer EU member states.

In this article, we study the poverty effectiveness of family transfers, more specifically child benefits and child-related tax advantages. Existing studies often point to the size of family transfers as the key factor in reducing child poverty. We hypothesize that the interaction with the design of policies is also a crucial factor. Our focus is on the Lithuanian system. We compare its effectiveness in combating child poverty with that of Estonia, Hungary, the Czech Republic and Slovenia. The four countries resemble Lithuanian political and socio-economic circumstances in many ways (for example, Soviet heritage, relatively similar pathways in reforming/introducing tax-benefit policies, and so on), though there are also important differences (for example, in terms of share of single parents and large families, work intensity in households with children, and so on; see also note 8). The four countries are selected as they all have better child poverty outcomes, which are attributed to arguably more effective family policy measures (TÁRKI, 2011). Our interest is to examine to what extent one country’s success story in achieving low(-er) child poverty rates, especially among the most vulnerable household types, can be attributed to the size and the design of the selected transfers. The study is anchored in 2008 – the year when a major family benefit reform was fully implemented in Lithuania (for more details see Salanauskaite and Verbist, 2009).

The paper starts with background information on child poverty in the five NMS. We also review evidence on the poverty effectiveness of family tax-benefit mechanisms. Next, we describe the methodology of policy swapping scenarios within the microsimulation framework of EUROMOD. We then present and analyse the microsimulation results. The final section provides our conclusions.

Child poverty and family support systems: existing evidence

In 2008 the at-risk-of-child-poverty rate (or child poverty) in Lithuania was above the EU and just below the NMS average. However, the at-risk-of-poverty rates (or poverty) of large households and single parent households are, with rates of over 45 percent, extremely high (Figure 1). This is despite the state’s recognition of these household categories as major poverty reduction targets (e.g. National Report of Lithuania on Social Protection and Social Inclusion Strategies 2008–2010, Republic of Lithuania, Ministry of Social Security and Labour, 2008). This poor result contrasts with most other EU countries, where at least one of these population groups has a better income position. Among the five countries we selected for our analysis, Slovenia performs best for these two household types. In Hungary both household types have increased poverty risks, though at much lower absolute levels than in Lithuania. Given their vulnerable income position our analysis focuses on the poverty outcomes of these two groups.

Figure 1.

Poverty among households with children in the selected countries (2008).

Along with socio-demographic characteristics of the child’s family, the parents’ labour market position as well as overall tax-benefit policies are seen as major child poverty determinants (Commission of the European Communities, 2008; TÁRKI, 2011). Even though family cash policies in themselves are often insufficient and are not actually meant to fully eliminate child poverty (Bradbury and Jäntti, 2001; Cantillon and Van den Bosch, 2003), their role is of great importance, with size and design as major factors.

The size of social spending dedicated to families with children is often considered to be the key factor influencing child poverty (e.g. Bradshaw and Finch, 2003; Notten and Gassmann, 2008). Figure 2 confirms that a higher share of GDP spent on tax breaks and transfers to families with children is associated with lower child poverty rates. Among the five selected countries, Hungary spends the largest share of GDP on families, and Lithuania the least. Child poverty levels in both countries, however, are somewhat higher in comparison to other countries with similar spending levels. The best performance is recorded in Slovenia: a relatively low share of GDP spent on transfers corresponding to a very low child poverty risk.

Figure 2.

Generosity of family transfers and (child) poverty in the EU, 2007.

The size and the design of the systems are arguably interlinked, with universal rather than targeted systems having both higher budgets available and larger poverty reduction effectiveness (e.g. Korpi and Palme, 1998; Nelson, 2007). The most recent empirical evidence, though, seems to suggest that this observation might not be valid any more (e.g. Kenworthy, 2011). Furthermore, the final poverty outcomes, are highly country-specific due to other complexities of national policy systems, socio-demographic environments, original income distributions, social insurance arrangements, and so on. The way in which benefits are designed is also very diverse. Countries use universal, categorical or income-selective family benefits. Tax advantages are also increasingly used as an important family policy tool (e.g. Figari et al., 2011). The poverty impact of these diverse designs is often not well assessed, especially for the NMS. Three studies on NMS are particularly interesting in this respect.

Förster and Tóth (2001) study the evolution of benefit types and their effectiveness in Poland, Hungary and the Czech Republic in the mid-1990s. They find that large and single parent families became particularly income-vulnerable during the economic transition years, with the most dramatic changes for the latter household type. Most of the reforms at that time introduced means-testing, which consequently increased the poverty-reduction effectiveness of the programmes. Nonetheless, the political will to reintroduce universal benefits remained and crystallized in the numerous reforms in times of economic upturn (e.g. as of 2004 in Lithuania). Levy et al. (2009) evaluate the poverty effectiveness of Polish state support to families by comparing it with systems in France, the United Kingdom and Austria using EUROMOD. They find that single parents in Poland would benefit most if the French system (using both universal and means-tested benefits) was adopted, whereas two-parent families would similarly benefit either from the universal Austrian or from the means-tested British systems. TÁRKI (2011) provides the most extensive evaluation of the EU countries’ performances in reducing child poverty. Their study finds low effectiveness of income support to families with children in Lithuania. The means-tested benefits in the Czech Republic and the universal benefits of Hungary produce similar child poverty outcomes. Social transfers in Slovenia are often seen as not specifically targeting children; however, their effectiveness in reducing poverty is high. As such, the latter two studies do not identify any particular benefit type as having higher poverty effectiveness, but rather highlight the great variation in impact under particular national designs and different socio-demographic circumstances.

Methodology

EUROMOD

We use EUROMOD, the European static tax-benefit microsimulation model, to assess the impacts of (quasi-)reforming family policy. Microsimulation models can help to highlight the role of certain family support instruments, be it taxes or benefits, while at the same time allowing for interactions with the remaining tax-benefit structure. This method also enables testing hypothetical public policy designs – a usually complex task due to the effects of various counterfactuals. Static microsimulation means that no behavioural reactions are taken into account. EUROMOD also assumes full take-up of benefits as well as full compliance with taxes and social contributions. Similar approaches to those used here include Matsaganis et al. (2007) for Southern Europe or Immervoll et al. (2001) for a comparison between the United Kingdom and the Netherlands.

Currently (i.e. 2011), EUROMOD embeds the policy designs of 21 EU countries, among them Lithuania, Estonia, Hungary, the Czech Republic and Slovenia.¹ The model was initially designed to cover the 15 ‘old’ EU member states, with the NMS being added progressively. For four countries the policy system of 2008 is included in EUROMOD and is used here.² For Slovenia, 2005 is the only policy year currently available and therefore we use the annual consumer price index to uprate Slovenian benefits to 2008.

Slovenian policies of 2005 are simulated on a sample of 2004 administrative records – SURS (Čok et al., 2008). The Lithuanian micro-database is derived from the EU-SILC (European Union Statistics on Income and Living Conditions) survey with a few imputations on the basis of the national SILC survey (Ivaškaitė-Tamošiūnė et al., 2010). In the Czech Republic, the national SILC additional variables are merged with EU-SILC (Münich and Pavel, 2010). ‘Pure’ EU-SILC is used for Estonia and Hungary (Hegedűs and Szivós, 2010; Võrk et al., 2010). The income reference of Lithuanian, Czech and Estonian datasets is 2005, whereas the one of Hungary is 2006. For simulating a policy year of 2008, the observed income (components) are uprated using country specific information.

Family cash policies for the five NMS in EUROMOD

We identify four major types of non-contributory ‘transfers to children’ in the five selected NMS: birth grants, universal child benefits, large family allowances, which can be labelled as categorically selective, and means-tested family allowances (income selective).³ This covers 17 different national benefits. Among them, only one benefit type is not simulated in EUROMOD: an Estonian child benefit supplement for single parents.⁴ All benefits are tax-exempt. In Hungary, however, the benefit to large families increases the taxable income base. This affects the calculation of personal income taxes (e.g. a higher personal tax rate could be applied) and consequently disposable income. In addition to these four benefit types, all five countries also use either tax credits or tax allowances to support families with children.

The principal design features and state expenses of both benefit and tax support measures are reviewed in Table 1. Lithuania has the most universalistic package, closely followed by Estonia. Hungary uses the most elaborate package of transfers, with a larger expenditure share going to universal/categorical benefits. Slovenia predominantly uses means-tested child benefits, but universal/categorical transfers are also employed, especially when tax advantages for families are taken into account. The Czech Republic exclusively relies on means-tested transfers.

Table 1.

Actual state expenses and beneficiaries of ‘transfers to children’ (2008)^a.

		LT	EE	HU	SI	CZ
Birth grant	Benefit per recipient, EUR	301.1	317.4	270.4	212.0	544.1^b
Child benefit	Age thresholds	18 (24)	17 (20)	~17 (~20)
	Size: Δ^c with child age	↓	No	No
	Size: Δ with no. of children	↑	↑	↓
	Extra1: for single parents	No	Yes	Yes
	Extra2: for young children	Yes	Yes	No
	Extra2: age thresholds	3	3 (8)	No
	Benefit per recipient, EUR^d	388.1	332.1	1117.6
Allowance to large families	Age thresholds		17 (20)	17 (25)	18 (26)
	Eligibility: no. of children		≥7	≥3	≥3
	Benefit per recipient		277.6	1426.0	334.2
Means-tested allowance	Age thresholds			17(25)	18 (26)	18 (26)^e
	Income threshold			1.25*OAP	Average wage	2.4*MLS
	Size 1: Δ no. of children			No	Yes	Yes
	Size 2: Δ other factors			No	↓ Income	↑ Age
	Benefit per recipient, EUR			756.4^f	1031.9	353.6
Tax relief: allowances or credits	Allowance (A)/credit (C)	A	A	C	A	C
	Age thresholds	18	18	17 (25)	18 (26)	18(26)
	Eligibility: no. of children	≥1	≥1	≥3	≥1	≥1
	Size: Δ with no. of children	Yes	Yes	No	Yes	Yes
	Extra: for single parents	Yes	No	No	No	No
	Means-tested	No	No	Yes	No	Yes
	Support per recipient, EUR	48.1	459.0	307.6	738.1	421.1
Country population, millions (P)		3.3	1.4	10.1	2.0	10.2
Expenses on benefits, million EUR (B)		150.0	94.0	1866.9	282.7	369.5
Per capita benefits, EUR (B/P)		45.5	69.6	185.8	124.2	36.2
Expenses on tax relief, million EUR (T)^g		25.2	79.2	52.9	237.8	594.1
Per capita tax relief, EUR (T/P)		7.6	58.7	5.3	118.9	58.2
Total ‘transfers to children’, (B+T)^h		175.2	173.2	1919.8	486.1	963.6
Per capita ‘transfers to children’ EUR		53.1	128.3	191.1	243.1	94.5
Per capita ‘transfers to children’, PPSⁱ		88.5	175.8	285.2	319.8	152.4

Notes: ^aAll SI data refers to 2005; local currency amounts converted using the 30 June exchange rate. ^bEstimation: [amount, 2008] × [no. of beneficiaries, 2005]; ^cΔ denotes ‘change(-s)’; ^dEE supplement to the single parents excluded (about 15% of the child benefit expenses). ^ePooled information on two benefits: child and social allowance; benefit per recipient = total expenses/recipients of child allowance. ^fEUROMOD simulated expenses; ^gTax relief, residual of simulated income taxation revenues with or without the measure. ^hAdministrative costs excluded (differences by transfer/country could exist); ⁱPPS, purchasing power standard.

Source: EUROMOD Country reports and MISSOC.

Birth grants are found in all countries, with quite similar benefit rules. The benefit is proportional to the number of newborns in all countries, except in Hungary. The benefit is particularly high in the Czech Republic.

Universal child benefits are provided in Lithuania, Estonia and Hungary, though the rules are quite different both in terms of eligibility and calculation. In Lithuania child benefit is provided to all children up to the age of 18 years, and up to the age of 24 years if a child is still in education and belongs to a large family. The benefit is increased for children up to the age of 3 years if raised in a large family. As such, these two components of the child benefit could be considered as a large family allowance. Lithuania does not provide these benefits separately, indicating that the demarcation between benefit types is not always straightforward. Estonia applies a lower age threshold for children who are still in (higher) education (i.e. under 20 years old). It also provides a benefit supplement for children younger than 8 years, with higher rates applicable to those below the age of 3 years. In Hungary child benefit is not directly linked to a specific age threshold, but depends on the child’s enrolment in education. The benefit size does not depend on the child’s age and has a regressive schedule for numerous children. Overall, Hungary offers the most generous child benefit provision.

Categorical selectivity is most explicit in the form of allowances for large families in three countries: Estonia, Hungary and Slovenia. The Estonian benefit is targeted towards families raising seven or more children. In Hungary families with three children or more are entitled to the allowance, but only if the youngest child is between 3 and 7 years old. In Slovenia all families with three or more children are eligible. In all three countries, the size of the allowance is uniform irrespective of family composition. Hungary offers the most generous support.

Income selectivity is applied in Hungary, Slovenia and the Czech Republic through means-tested child allowances. In the Czech Republic this is the only available benefit type, aside from the birth grant. Here, the means-tested income threshold is family-specific and set in relation to the state determined minimum living standard (MLS, a parameter that depends on the age and the number of family members⁵). The benefit size is set per child and increases with age. Hungary applies the simplest benefit calculation rules: any family with per capita incomes lower than 125 percent of the minimum old-age pension (OAP⁶) is entitled to a uniform benefit amount. The Slovenian means-tested threshold is much higher than in Hungary. The benefit size depends on per capita family income and is gradually reduced to 0, when reaching 99 percent of the average gross wage.⁷ Due to the use of per capita incomes in benefit calculations, larger families receive proportionally larger benefits.

Lithuania, Estonia and Slovenia have personal income tax systems that use tax allowances (i.e. income-independent amounts deductible from taxable income). Tax allowances are increased for families with children. The rules of family tax allowances are relatively similar, though levels differ. Lithuanian tax allowances differ by family type with the most generous support going to large families, followed by single parent families. Families with up to two children receive the smallest allowance. The Estonian family tax allowance assigns an identical amount per child. The Slovenian family tax allowance increases with each subsequent child. Using EUROMOD to calculate the value of these measures, the Slovenian system appears to be the most generous. Here, the tax allowance is the second largest state support to families (after the means-tested allowance). The cost of the Lithuanian tax allowance is relatively modest compared with expenses on benefits. In Estonia, expenses on family tax allowances almost equal the level of expenditures on family benefits.

Hungary and the Czech Republic have tax credits for families with children (i.e. deductions from tax liabilities). All families are entitled to an income-dependent tax credit in the Czech Republic. The credit amount is proportional to the number of children and is subject to a maximum yearly amount. If the tax liability is lower than the tax credit, the difference is paid to the taxpayer, i.e. it is refundable. In Hungary only families with three or more children (about 2 percent of the total population) are entitled to receive an income dependent family tax credit, which is non-refundable, i.e. it cannot be higher than tax liability.

Considering all benefits and tax allowances, Slovenia takes the lead in generosity with €243 per capita. Lithuania has the lowest spending (€53 per capita) on transfers to children. Taking into account differences in purchasing power standards (PPS), disparities in the generosity levels of the identified family benefit systems reduce, but still remain high. Furthermore, the observation presented in the first section that systems with the most universal design of benefits tend to have the largest budgets does not hold across the selected countries.

Policy scenarios

Microsimulation models enable us to test the distributional impact of both existing and ‘what-if’ policies. In this article we exploit both options.

In order to check how effective existing transfers to children are in reducing child poverty, we eliminate them from the country’s tax-benefit system (see Results section). In this setting, the other tax-benefit rules still play a role in further increasing or decreasing household income (e.g. the social assistance safety net may compensate part of the income loss resulting from eliminating family transfers). By comparing poverty outcomes with and without transfers to children we evaluate the first-order poverty effects of existing arrangements.

Swapping policies means that policies of a donor country are integrated into the tax-benefit system of a recipient country instead of the existing family benefit system (see Results section). This enables us to test the effectiveness of diversely composed donor policies given interactions with the remaining tax-benefit structure and socio-demographic features of the recipient country.⁸ We model the impact of transfers, assuming a 100 percent take-up. We think this is a reasonable assumption for at least two reasons. First, this allows comparing the intended systems’ designs. Second, take-up is not always well documented, but for the countries where the information is available, actual take-up of family transfers is very high (e.g. see the corresponding EUROMOD Country reports). Simulation biases also occur, for example, due to imperfections of the underlying survey micro-data resulting in an inability to fully ‘parameterize’ policy rules. Overall EUROMOD simulations provide ‘reasonably consistent’ poverty results in comparison to survey data (e.g. see Figari et al. 2011).

We analyse three major policy-swapping scenarios (see Figure 3),⁹ distinguishing between a full and a budget-neutral implementation.¹⁰ In all three scenarios we rely on national monetary references, such as the (share of) average gross wage, when converting intermediary monetary parameters (i.e. income brackets, eligibility thresholds, etc.). This allows (partial) policy ‘adaptation’ to national circumstances.

Figure 3.

Policy swap scenarios.

In Scenario I, we implement the benefits of the four other countries (as listed in Table 1) in Lithuania. Swapping of tax support measures is excluded here. In the full swap, the benefit amounts are introduced at the original levels, except for the adjustment for purchasing power standards and currency rates. This leads to an increase in total benefits’ expenses compared with original Lithuanian settings from 1.7 times under Estonian policies to 3.7 times under Hungarian policies. The budget-neutral scenario implies that simulated total expenses are rescaled (at the micro-level) to the budget of the existing Lithuanian system.¹¹

In Scenario II, we replace both Lithuanian benefits and tax allowances to children with the respective policies of the other countries.¹² A comparison between Scenarios I and II highlights the additional influence of the tax support instruments. This swap also measures the effect of benefits and tax measures together. Under budget-neutrality both state expenses on benefits (i.e. using scalars of Scenario I) and income taxation revenue are respectively matched to original Lithuanian levels (i.e. imposing changes on spending proportions between benefits and tax advantages). The scaling factors for tax support instruments are estimated using empirical calibration due to non-linearity in the income tax calculation.¹³

In Scenario III, we shift Lithuanian transfers and tax instruments for children to the other four countries, while keeping the remaining tax-benefit structure of those countries unchanged. We focus on the budget-neutral swapping impacts,¹⁴ using analogous assumptions as already described in the Scenarios I and II. Scenario III shows the effectiveness of Lithuanian policies given different socio-economic and demographic settings, as well as interactions with the remaining tax-benefit system. The composition of spending on benefits and tax advantages is matched to the proportions observed in the recipient country. This scenario allows testing to see if Lithuanian policies are indeed less effective at reducing poverty or whether unfavourable socio-demographic settings drive the observed poverty rates in Lithuania. By ‘adopting’ Lithuanian policies into different and arguably more advantageous (if lower poverty rates are considered) socio-demographic settings we obtain additional information on the policies’ effectiveness.

Policy effectiveness indicators

We evaluate the effectiveness of swapped programmes by their impact on two measures of (relative) poverty, the poverty headcount and gap, before and after implementation of a scenario (see e.g. Foster et al., 1984). The poverty headcount measures the share of individuals with an equivalized household income below the poverty line. The poverty gap indicates the average extent between poor people’s income and the poverty line and is expressed as a percentage of the poverty line. The poverty line (60 percent of the median equivalized income) is recalculated for each scenario. Using a recalculated poverty line means that we maintain the relative character of the poverty line, thus allowing for the potential shift in median income following from changes in the income distribution (see e.g. Marx et al., 2012). In comparison with the poverty line in existing Lithuanian settings (about €216), it decreases by a maximum of 2 percent (Scenario III, Estonia) or increases by a maximum of 5 percent (Scenario II, full implementation of the Czech policies) for different scenarios. Disposable income is the annual sum of total gross household income from labour earnings, plus income from investment and savings, plus all types of simulated or observed contributory and non-contributory benefits, minus simulated social contributions, simulated final personal income taxes and observed other taxes (i.e. property, wealth taxes). It is equivalized with the modified OECD scale (i.e. value 1 is assigned to the first adult; 0.5 to any other person aged 14 years or older; 0.3 to each child younger than 14 years). Standard errors (with a 95 percent confidence level) of poverty indicators are computed using Taylor first order linearization with the STATA DASP programme.¹⁵

A comparison of the poverty outcomes in the baseline and swap scenarios gives the effect of implementing a foreign system. By simulating the budget-neutral implementation, we can distinguish between the design (the baseline in comparison to the budget-neutral implementation) and the size (the budget-neutral in comparison to the full implementation) effects. Statistically significant changes (with a 95 percent confidence level) between the point (poverty) estimates of a baseline and a swap scenario are determined at the micro-level (i.e. comparing poverty status of each person in the two scenarios) by taking co-variation into account.¹⁶

Simulation results

Poverty impacts of existing policies

Without transfers to children (i.e. both benefits and tax concessions) all countries would have higher poverty levels for all groups of interest (see Table 2). The smallest effect of the transfers is observed in Lithuania (an 8 percent reduction in child poverty rate). The Hungarian system has the strongest effect, with a child poverty reduction of more than 60 percent.

Table 2.

Poverty headcount and gap in pre- and post- transfer systems.

	LT-pre	LT-post	(Pre–Post)/Pre	EE-pre	EE-post	(Pre–Post)/Pre	HU-pre	HU-post	(Pre–Post)/Pre	SI-pre	SI-post	(Pre–Post)/Pre	CZ-pre	CZ-post	(Pre–Post)/Pre
Poverty headcount (%)
Total	20.8	20.3	2.5%	18.5	17.5	5.7%	17.8	13.3	33.8%	18.3	15.7	16.6%	9.8	8.5	15.3%
	(0.78)	(0.78)		(0.59)	(0.59)		(0.52)	(0.47)		(0.57)	(0.54)		(0.50)	(0.47)
Children	28.2	26.2	7.6%	26.5	20.1	31.8%	32.6	19.6	66.3%	23.6	15.0	57.3%	15.3	10.8	41.7%
	(1.63)	(1.61)		(1.22)	(1.13)		(1.18)	(1.02)		(1.18)	(0.98)		(1.07)	(0.97)
In large (≥3) families	48.0	44.3	8.4%	40.1	20.2	98.5%	60.2	30.6	96.7%	44.9	16.2	177.2%	31.8	20.5	55.1%
	(6.22)	(6.27)		(4.59)	(3.41)		(3.48)	(3.39)		(5.57)	(4.17)		(4.78)	(4.55)
In single parent families	49.3	45.1	9.3%	55.7	45.6	22.1%	44.5	30.2	47.4%	39.8	25.6	55.5%	32.9	27.9	17.9%
	(6.29)	(6.40)		(4.48)	(4.62)		(3.93)	(3.54)		(5.01)	(4.79)		(3.50)	(3.41)
Poverty gap (%)
Total	6.3	5.9	6.8%	5.3	4.9	8.2%	6.0	3.2	87.5%	4.7	3.4	38.2%	1.7	1.5	13.3%
	(0.41)	(0.39)		(0.28)	(0.26)		(0.29)	(0.16)		(0.22)	(0.15)		(0.14)	(0.13)
Children	8.7	7.5	16.0%	8.3	6.2	33.9%	12.0	4.5	166.7%	6.5	2.5	160.0%	2.7	1.9	42.1%
	(0.76)	(0.68)		(0.57)	(0.50)		(0.67)	(0.30)		(0.45)	(0.19)		(0.28)	(0.26)
In large (≥3) families	14.8	12.0	23.3%	10.9	5.2	109.6%	26.1	6.3	314.3%	12.3	1.9	547.4%	6.0	4.3	39.5%
	(2.49)	(2.19)		(1.54)	(0.99)		(2.17)	(0.83)		(2.03)	(0.55)		(1.26)	(1.24)
In single parent families	13.9	13.1	6.1%	18.0	14.1	27.7%	18.2	6.7	171.6%	11.3	3.1	264.5%	5.3	3.7	43.2%
	(2.07)	(2.01)		(1.89)	(1.85)		(2.33)	(1.14)		(2.18)	(0.61)		(0.73)	(0.62)

Notes: Here and further on the poverty headcount measures the share of individuals with an equivalized household income below the poverty line; the poverty gap indicates the average extent between poor people’s income and the poverty line and is expressed as a percentage of the poverty line; standard errors are in parentheses; children are defined as persons under the age of 18 years. Shaded cells show significant pre- and post-transfer poverty (gap) changes. The ratio ‘(Pre–Post)/Pre’ indicates a percentage reduction in poverty rate due to the selected transfers.

Source: own calculations.

The analysed systems have varied poverty gap and headcount effects for vulnerable household types. The Slovenian system is particularly effective for large families as it reduces pre-transfer poverty by almost three times. Overall, all countries but Lithuania seem to manage the poverty risk of this family type by means of transfers to children. The poverty reduction rate of large families in Lithuania is only 9 percent. Single parent families have lower income protection in comparison with large families in all countries; the largest poverty reduction is achieved by the Slovenian (55 percent) and Hungarian (47 percent) systems.

Slovenian policies drastically reduce poverty depth for large families (by more than six times) and for single parent families (by more than three times) – the largest reduction among the five selected countries. In the Czech Republic the pre-transfer poverty gap is already small. Its means-tested system, though, achieves less for large and single parent families compared with the Slovenian system. The Estonian system is able to halve the poverty gap among children in large families. Results for Lithuania reveal the lowest capacity to reduce the poverty gap.

Overall, simulation on the poverty effectiveness of these countries’ existing transfers does not support the above-mentioned literature (e.g. Korpi and Palme, 1998; Nelson, 2007) that observes that greater targeting achieves less poverty alleviation.

Are ‘borrowed’ policies more effective in poverty reduction?

The poverty outcomes of swapping foreign policies are displayed in Table 3 – both benefits and tax advantages – into Lithuania. Our findings show that the relative importance of the size and design effects of these hypothetical reforms depends on the system, the household type and the poverty index.

Table 3.

Poverty headcount and gap under the ‘borrowed’ policies.

	LT baseline	Scenario I: swap of benefits								Scenario II: swap of benefits and tax advantages
		Budget-neutral				Full				Budget-neutral				Full
		EE	HU	SI	CZ	EE	HU	SI	CZ	EE	HU	SI	CZ	EE	HU	SI	CZ
Poverty headcount (%)
Total	20.3	20.3	19.9	18.9	19.4	20.1	18.5	17.6	17.1	20.3	19.6	18.9	19.6	20.0	18.4	18.5	16.9
	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)	(0.79)	(0.79)	(0.78)	(0.78)	(0.78)	(0.78)	(0.78)
Children	26.2	26.8	25.6	23.1	24.7	25.9	20.4	18.6	18.9	26.8	24.4	23.0	25.4	25.5	19.7	18.6	16.6
	(1.61)	(1.62)	(1.60)	(1.57)	(1.62)	(1.61)	(1.52)	(1.49)	(1.53)	(1.62)	(1.57)	(1.56)	(1.62)	(1.60)	(1.46)	(1.47)	(1.50)
In large (≥3) families	44.3	42.6	40.3	32.6	42.9	41.2	24.9	20.4	27.2	42.6	34.4	31.1	45.0	41.2	24.1	19.8	25.3
	(6.27)	(6.29)	(6.27)	(6.08)	(6.29)	(6.29)	(5.82)	(5.60)	(5.88)	(6.29)	(6.11)	(6.07)	(6.27)	(6.29)	(5.80)	(5.55)	(5.85)
In single parent families	45.1	49.3	45.9	45.0	43.6	48.1	39.5	38.1	40.2	49.8	45.3	45.2	44.7	47.8	33.2	39.8	29.2
	(6.40)	(6.29)	(6.38)	(6.40)	(6.44)	(6.33)	(6.29)	(6.50)	(6.53)	(6.28)	(6.39)	(6.40)	(6.41)	(6.34)	(5.78)	(6.46)	(6.66)
Poverty gap, %
Total	5.9	6.0	5.7	5.4	5.4	5.9	4.9	4.8	4.4	6.0	5.6	5.4	5.4	5.9	4.8	5.0	3.9
	(0.39)	(0.39)	(0.37)	(0.35)	(0.35)	(0.38)	(0.32)	(0.31)	(0.29)	(0.39)	(0.36)	(0.35)	(0.35)	(0.38)	(0.31)	(0.32)	(0.26)
Children	7.5	7.7	7.1	6.2	6.7	7.5	4.9	4.3	4.3	7.9	6.6	6.1	6.6	7.4	4.3	4.4	2.9
	(0.68)	(0.69)	(0.65)	(0.56)	(0.62)	(0.67)	(0.48)	(0.43)	(0.46)	(0.69)	(0.59)	(0.56)	(0.61)	(0.67)	(0.42)	(0.45)	(0.32)
In large (≥3) families	12.1	12.1	10.7	7.5	10.6	11.1	5.9	3.1	5.7	12.4	8.5	7.2	10.5	10.9	3.8	3.3	3.4
	(2.19)	(2.20)	(2.16)	(1.63)	(2.02)	(2.10)	(1.45)	(0.99)	(1.38)	(2.23)	(1.82)	(1.61)	(1.98)	(2.08)	(0.97)	(1.09)	(0.84)
In single parent families	13.1	14.3	12.1	12.0	11.9	14.7	8.8	9.5	9.0	14.6	11.0	12.1	11.5	14.8	7.8	10.1	4.8
	(2.01)	(2.10)	(1.82)	(1.83)	(1.92)	(2.16)	(1.50)	(1.58)	(1.77)	(2.11)	(1.73)	(1.85)	(1.82)	(2.18)	(1.56)	(1.68)	(1.02)

Note: shaded cells show significant changes between baseline and swap scenarios.

Source: own calculations.

The full implementation indicates that the Hungarian, Slovenian and Czech systems lead to significantly better poverty outcomes than the existing Lithuanian system. This is the case for the swap of benefits, as well as for the combined swap of benefits and tax advantages. These three countries include means-tested transfers in the child transfer package. An introduction of the Slovenian system (both for the benefits-only and for the combined benefit-tax advantage scenarios) leads to the best results for large families: poverty is halved. Much smaller poverty changes are noted for single parent families and across all three swapped systems. The poverty effects of the Estonian system, that resembles the Lithuanian design most closely, are highly heterogeneous: no significant changes are noted for population and child poverty with a swap of benefits; a small but significant poverty reduction is noted for children in large families when tax advantages are added; and a small but significant increase in poverty is recorded for children in single parents families. The latter result is the only negative effect across all swaps. The overall results indicate that the effect of the benefit swapping tends to be stronger than the effect of swapping tax advantages. Exceptions are the Hungarian and the Czech systems, where adding tax advantages leads to a significant reduction in poverty risk for single parents. It is however important to recognize that Hungary and the Czech Republic have a tax credit rather than a tax allowance.

A quick glance at Table 3 would tempt the reader to think that the transfer size is the major determinant for the reduction of poverty risk, as budget-neutrality leads to fewer significant changes. We want to make some qualifications, however, regarding the perceived dominance of size. First, the results are dependent on the system’s design. For example, overall child poverty reduction is comparable across all borrowed systems (except Estonia); however, the Slovenian benefits’ (only) system leads to the largest reduction in poverty risk for large families. The latter system does not offer the most generous benefits’ package, though. Second, when tax advantages are included, all systems achieve significant changes in child poverty. The direction of the change is varied. Adding Slovenian tax advantages, for example, implies a small worsening in poverty estimates for single parent families, while the Czech tax measures achieve a coherent and large poverty drop for all concerned groups. Third, the choice of indicator matters too: with the poverty gap we measure significant reductions for all three ‘successful’ systems (HU, SI, CZ), both under budget-neutral and full implementation scenarios.

While the overall design effect tends to be smaller than the size effect, the final comparison is highly dependent on the population group and the system. Under the Slovenian regime, for example, both the design and the size effects are of equal importance for large families: when moving from the baseline to the budget-neutral scenario (i.e. the design effect), poverty is reduced by about 12 percentage points, and moving from the budget-neutral to the full scenario (i.e. the size effect) yields a poverty reduction of similar magnitude. The design effect is even stronger for the poverty gap. As previously discussed, Slovenia pays considerable attention to large families. It offers an allowance to large families as well as a means-tested child benefit. This last benefit is particularly advantageous to large families because the benefit size is linked to per capita income. Hungarian tax and benefit measures reveal equally important size and design effects (about 10 percentage points each) for large families too. Overall, the design effect seems to be considerably interlinked with the size effect of policies: the systems with the strongest design effects (i.e. Slovenia and Hungary) for large families are also able to achieve the strongest size effects.

The fact that the budget-neutral swap of the Czech system does not introduce significant poverty changes for the two vulnerable groups may come as a surprise, given the fact that this system has only means-tested transfers. The difference from the better scoring Slovenian system relates to how the benefit size is determined, as under the Czech system it is not differentiated according to income. In addition, the income threshold for benefit entitlement in the Czech system is family type specific, while uniform thresholds are applied in Slovenia and Hungary. Czech tax measures under the full implementation (hence the size effect is dominant), though, achieve the best poverty headcount score for single parent households. We find that this better performance is partly due to the fact that the Czech tax credit – contrary to its Hungarian counterpart – is refundable.

In general, the situation of children living in single parent households is least or even negatively (i.e. under the Estonian system) affected by the policy swaps. This is in line with the designs of the systems, which rarely have advantageous provisions for single parent families (especially in comparison with large families). Furthermore, original Lithuanian measures include preferential tax rather than benefit measures for single parents. The Hungarian and Czech tax measures provide the largest relative income improvement for single parents, but only at the cost of a considerable increase in the tax relief compared to the Lithuanian baseline.

Poverty reduction effectiveness of Lithuanian policies in other countries

The budget-neutral implementation of Lithuanian policies in the other four countries worsens child poverty, though to different degrees (see Table 4).

Table 4.

Poverty headcount and gap, swapping Lithuanian policies into four NMS.

	EE		HU		SI		CZ
	Post	LT	Post	LT	Post	LT	Post	LT
Poverty headcount (%)
Total	17.5	17.9	13.3	13.8	15.7	17.5	8.5	10.6
	(0.59)	(0.58)	(0.47)	(0.50)	(0.54)	(0.55)	(0.47)	(0.50)
Children	20.1	21.0	19.6	21.3	15.0	17.8	10.8	15.9
	(1.13)	(1.15)	(1.02)	(1.06)	(0.98)	(1.05)	(0.97)	(1.07)
In large (≥3) families	20.2	21.4	30.6	36.8	16.2	23.8	20.5	31.9
	(3.41)	(3.47)	(3.39)	(3.51)	(4.17)	(4.94)	(4.55)	(4.74)
In single parent families	45.6	47.1	30.2	32.2	25.6	26.7	27.9	35.8
	(4.62)	(4.60)	(3.54)	(3.68)	(4.79)	(4.80)	(3.41)	(3.55)
Poverty gap (%)
Total	4.9	5.1	3.2	3.7	3.4	4.1	1.5	1.9
	(0.26)	(0.26)	(0.16)	(0.18)	(0.15)	(0.18)	(0.13)	(0.14)
Children	6.2	6.7	4.5	5.6	2.5	4.0	1.9	2.8
	(0.50)	(0.52)	(0.30)	(0.38)	(0.19)	(0.29)	(0.26)	(0.29)
In large (≥3) families	5.2	6.6	6.3	10.0	1.9	5.2	4.3	5.7
	(0.99)	(1.21)	(0.83)	(1.24)	(0.55)	(1.15)	(1.24)	(1.29)
In single parent families	14.1	14.8	6.7	9.4	3.1	6.1	3.7	5.8
	(1.85)	(1.86)	(1.14)	(1.51)	(0.61)	(1.29)	(0.62)	(0.78)

Note: shaded cells show significant changes between baseline and swap scenarios.

Source: own calculations.

We observe the largest deterioration in poverty rates for all concerned groups in the Czech Republic. Poverty rates in Hungary and Slovenia would also increase under Lithuanian policies. In both countries, however, the poverty outcomes of single parent families does not change, indicating that neither of these countries has a more effective state support package for this group. Estonian budget-neutral policies had mixed results in Lithuania. The reverse swap worsens the poverty situation slightly, except for the insignificant change for single parent families. It is noteworthy that Estonian policies also implied a worsening poverty situation for Lithuanian single parent families.

If Lithuanian policies were implemented, poverty gaps would worsen across all countries and would imply particularly negative changes for large families. The worst performance would be seen in Slovenia. The poverty gap would also widen for single parent households (though to a lesser degree than for large families), especially in Hungary and Slovenia. This is a somewhat surprising finding, given that a much smaller effect was detected when swapping foreign policies into Lithuania.

Conclusions and policy suggestions

This study is the first application of a full tax-benefit microsimulation model for testing the poverty effectiveness of family transfers within a comparative setting of five NMS. EUROMOD allows swapping policies from one country to another. Although such a model has a number of limitations, the advantage is its comprehensive structure in handling the cross-national analysis of distributional policy impacts. The policy systems differ across the five countries in terms of the size and design of their non-contributory transfers to children: birth grants, universal child benefits, large family allowances (categorical benefit), means-tested child allowances and tax advantages to families. An advantage of using EUROMOD is that the distribution of tax measures can also be captured; this is an often neglected factor that can have a significant impact on poverty. This is illustrated by the results from the Czech and Hungarian systems.

Current literature usually points to the transfer size as the primary determinant of child poverty. Our results confirm that it is of high importance. Nevertheless, we find the design effect can be of equal significance and can reinforce the size effect. The strength of the size and the design effects are also highly dependent on the composition of the selected policy measures (universal, categorical, income selective) and the parametric choices of the policies’ inner design (i.e. thresholds, benefit size determination, etc.).

Our study confirms that the best poverty score is not necessarily achieved by the most extensive means-tested systems, in line with the observations of Korpi and Palme (1998). On the other hand, ‘pure’ universal systems are found to be the least poverty effective. A mix of means-tested and categorical benefits that are sensitive to the characteristics of poor families can act as a highly effective tool for poverty reduction. Examples of such ‘good practices’ are the large family allowance in Slovenia and the tax credit to large families in Hungary. The common features of these two instruments are (1) a high coverage of large families and (2) that the level of the benefit is not dependent on the child’s age. Another favourable feature is a higher income threshold for means-tested benefits, which ensures that only the most vulnerable families are eligible. The key to the success of the Slovenian child benefit’s design is found in the combination of a generous means-testing threshold with a benefit level that depends on per capita family income, thus providing higher benefits for large families. This is a major difference when compared with the means-tested benefits in other countries, where the benefit level depends on the child’s age (such as in the Czech Republic) or is uniform for all eligible families (in Hungary).

Our simulations do not reveal any significant design features that would reduce child poverty among single parent families in Lithuania. We expected more positive outcomes given the results of baseline policies (e.g. Slovenia’s system reduces child poverty among single parent families by 55 percent). Apparently, only an increase in size is able to reduce the prevalence of poverty among single parent families in Lithuania. Poverty gap analysis reveals small positive changes, except under the Estonian system. The latter design worsens both poverty score and depth among single parent families both under budget-neutral and full implementation settings.

Finally, we want to stress that it is not only benefit design and size criteria that affect poverty outcomes, but that also the interaction with the specific national context is of great importance. This is illustrated by a comparison of the Lithuanian and Estonian contexts. Although these two countries have similar non-contributory family benefit and tax measures, Estonia achieves a much better poverty reduction for both large and single parent families. This means that the policies in Estonia are better aligned to its own socio-demographic settings than is the case in Lithuania. As such, the observed size and design effects interact with a number of country-specific characteristics, such as socio-demographic settings and other tax-benefit policies.

Footnotes

Acknowledgements

The paper is part of the GINI project, which studies the economic and educational drivers and the social, cultural and political impacts of increasing inequality with novel contributions on the measurement of income, wealth and education inequality (EU FP7 Research Support; more information can be found on ). The results presented here are primarily based on EUROMOD versions F2.38 and F3.0. EUROMOD is maintained, developed and managed by the Institute for Social and Economic Research (ISER) at the University of Essex in collaboration with national teams from the EU member states. We are indebted to the many people who have contributed to the development of EUROMOD and to the European Commission for providing financial support for it. This paper also benefited from useful comments on earlier drafts by the participants of IMA conference (Stockholm, May 2011), and of the EUROMOD research workshop (Riga, October 2011). We are particularly grateful to Chris de Neubourg, András Gábos, Tim Goedemé, Romas Lazutka, Ninke Mussche and Raymond Wagener for constructive suggestions. The results and their interpretation are of course the authors’ responsibility.

Notes

References

Araar

Duclos

J.Y.

(2007) DASP: Distributive Analysis Stata Package. Sainte-Foy, Canada: Université Laval, PEP, CIRPEE and World Bank.

Bradbury

Jäntti

(2001) ‘Child Poverty across the Industrialized World: Evidence from the Luxembourg Income Study’, in Vleminckx

Smeeding

T.M.

(eds) Child Poverty, Child Well-being and Child Policy in Modern Nations: What Do We Know? Bristol: Policy Press.

Bradshaw

Finch

(2003) ‘Child Benefit Packages in 22 Countries’, paper prepared for the 4th International Research Conference on Social Security, Antwerp, Belgium.

Cantillon

Van den Bosch

(2003) ‘Social Policy Strategies to Combat Income Poverty of Children and Families in Europe’, in Krause

Bäcker

Hanesch

(eds) Combating Poverty in Europe: The German Welfare Regime in Practice. Aldershot: Ashgate.

Čok

Kump

Majcen

(2008) EUROMOD Country Report: Slovenia (2005). Colchester: University of Essex, EUROMOD.

Commission of the European Communities (2008) The EU Social Protection and Social Inclusion Process: Thematic study on policy measures concerning child poverty, Policy Studies Findings (10), available at http://ec.europa.eu/employment_social/spsi/child_poverty_en.htm#childpoverty.

Cornelius

(1995) ‘Cash Benefits and Poverty Alleviation in an Economy in Transition: the Case of Lithuania’, Comparative Economic Studies 37(2): 49–70.

Figari

Iacovou

Skew

Sutherland

(2012) ‘Approximations to the Truth: Comparing Survey and Microsimulation Approaches to Measuring Income for Social Indicators’, Social Indicators Research 105(3): 387–407.

Figari

Paulus

Sutherland

(2011) ‘Measuring the Size and Impact of Public Cash Support for Children in Cross-National Perspective’, Social Science Computer Review 29(1): 85–102.

10.

Förster

Tóth

(2001) ‘Child Poverty and Family Transfers in the Czech Republic, Hungary and Poland’, Journal of European Social Policy 11(4): 324–41.

11.

Foster

Greer

Thorbecke

(1984) ‘A Class of Decomposable Poverty Measures’, Econometrica 52(3): 761–6.

12.

Hegedűs

Szivós

(2010) EUROMOD Country Report: Hungary (2005–2008). Colchester: University of Essex, EUROMOD.

13.

Immervoll

Sutherland

de Vos

(2001) ‘Reducing Child Poverty in the European Union: the Role of Child Benefits’, in Vleminckx

Smeeding

T.M.

(eds) Child Poverty, Child Well-being and Child Policy in Modern Nations: What Do We Know? Bristol: Policy Press.

14.

Ivaškaitė-Tamošiūnė

Lazutka

Salanauskaitė

(2010) EUROMOD Country Report: Lithuania (2005–2008). Colchester: University of Essex, EUROMOD.

15.

Kabašinskaitė

Bak

(2006) ‘Lithuania’s Children’s Policy in the Period of Transition’, International Journal of Social Welfare 15(3): 247–56.

16.

Kamerman

Neuman

Waldfogel

Brooks-Gunn

(2003) ‘Social Policies, Family Types and Child Outcomes in Selected OECD countries’, OECD Social, Employment and Migration Working Papers 6. Paris: OECD, Employment, Labour and Social Affairs Committee.

17.

Kenworthy

(2011) Progress for the Poor. Oxford: Oxford University Press.

18.

Korpi

Palme

(1998) ‘The Paradox of Redistribution and Strategies of Equality: Welfare State Institutions, Inequality, and Poverty in the Western Countries’, American Sociological Review 63(5): 661–87.

19.

Levy

Lietz

Sutherland

(2007) ‘Swapping Policies: Alternative Tax-Benefit Strategies to Support Children in Austria, Spain and the UK’, Journal of Social Policy 36(4): 625–47.

20.

Levy

Morawski

Myck

(2009) ‘Alternative Tax-Benefit Strategies to Support Children in Poland’, in Lelkes

Sutherland

(eds) Tax and Benefit Policies in the Enlarged Europe: Assessing the Impact with Microsimulation Models. Ashgate: European Centre Vienna.

21.

Lietz

Mantovani

(2007) ‘A Short Introduction to EUROMOD: an Integrated European Tax-BenefitModel’, in Bargain

(ed.) Microsimulation in Action: Policy Analysis in Europe Using EUROMOD. Maryland Heights, MO: Elsevier.

22.

Marx

Vandenbroucke

Verbist

(2012) ‘Can Higher Employment Levels Bring Down Poverty in the EU? Regression-based Simulations of the Europe 2020 Target’, Journal of European Social Policy 22(5): 472–486.

23.

Matsaganis

O’Donoghue

Levy

Coromaldi

Mercader-Prats

Rodrigues

C.F.

et al. (2007) ‘Child Poverty and Family Transfers in Southern Europe’, in Spadaro

(ed.) Microsimulation as a Tool for the Evaluation of Public Policies: Methods and Applications. Bilbao: Fundación BBVA.

24.

Münich

Pavel

(2010) EUROMOD Country Report: Czech Republic (2005–2008). Colchester: University of Essex, EUROMOD.

25.

Nelson

(2007) ‘Universalism Versus Targeting: The Vulnerability of Social Insurance and Means-Tested Minimum Income Protection in 18 Countries, 1990–2002’, International Social Security Review 60(1): 33–58.

26.

Notten

Gassmann

(2008) ‘Size Matters: Targeting Efficiency and Poverty Reduction Effects of Means-tested and Universal Child Benefits in Russia’, Journal of European Social Policy 18(3): 260–74.

27.

Republic of Lithuania, Ministry of Social Security and Labour (2008) National Report of Lithuania on Social protection and Social Inclusion Strategies 2008–2010, available at www.socmin.lt/index.php?-311688494.

28.

Salanauskaite

Verbist

(2009) ‘Reforming Child Allowances in Lithuania: What Does Microsimulation Tell Us?’, in Lelkes

Sutherland

(eds) Tax and Benefit Policies in the Enlarged Europe: Assessing the Impact with Microsimulation Models. European Centre Vienna: Ashgate.

29.

Sutherland

(2007) ‘EUROMOD: the Tax-Benefit Microsimulation Model for the European Union’, in Gupta

Harding

(eds) ‘Modelling Our Future: Population Ageing, Health and Aged Care’, International Symposia in Economic Theory and Econometrics 16: 483–88; Maryland Heights, MO: Elsevier.

30.

TÁRKI (2011) Child Well-Being in the European Union: Better Monitoring Instruments for Better Policies. Report prepared the State Secretariat for Social Inclusion of the Ministry of Public Administration and Justice, Budapest.

31.

Võrk

Paulus

Lüpsik

(2010) EUROMOD Country Report: Estonia (2005–2008). Colchester: University of Essex, EUROMOD.