An inquiry on introducing a minimum income scheme in Turkey: Alternating between cost efficiency and poverty reduction

Abstract

Turkey is one of the few countries in Europe without a minimum income scheme. In an effort to simulate the potential economic impact of introducing such a scheme in Turkey, data from the 2007 Statistics on Income and Living Conditions were used to consider various policy scenarios with regard to eligibility conditions and benefits. The impact of a minimum guaranteed income scheme on the extent of poverty, income distribution and the proportion of potential beneficiaries at the individual and household levels were investigated under different policy variations, as well as their corresponding costs. Subsequent analyses provided supporting evidence for the affordability of minimum income schemes in Turkey that would contribute to altering inequalities in income distribution.

Keywords

poverty minimum income income distribution Turkey social assistance

Introduction

Poverty constitutes one of the main challenges faced by welfare regimes. Even before the consolidation of modern welfare states, around the 1950s, local and discretionary poor relief was mostly codified under the poor laws that aimed to reduce poverty. As Ferrera (2005) notes, these ad hoc interventions lost their financial and functional importance by the late 1960s, when the longstanding tradition of discretionary poor relief gave way to the emergence of the welfare state. A new generation of social assistance schemes appeared, mainly in Europe, primarily to close the coverage gaps of social insurance programmes and establish a minimum guarantee for all citizens. These sets of instruments were apparently novel in their rights-based nature, where protection was conditional and offered on the basis of assessed need and people’s inability to cope with such need. In other words, these non-contributory schemes were rights based in principle but grounded in selectivity and residuality (see, for example, Ferrera, 2000, 2005). As Eardley et al. (1996) and Gough (1996) clearly illustrate, most countries established means-tested categorical schemes that targeted the poor, the elderly, the disabled, widows and orphans and the long-term unemployed, and also generalized income support schemes that resembled minimum income guarantees.

Saraceno (2010) highlights the importance of income support for the poor in most EU countries as a part of the social citizenship package where a non-categorical safety net exists for all citizens below a given income threshold. In this context, the broader category of policy measures mainly includes traditional social assistance policies in general and means-tested ‘last resort’ schemes, such as minimum income schemes in particular. These schemes offer individuals and households basic protection by preventing them from falling below a minimum economic threshold for a decent living. Ferrera (2005), more specifically, emphasizes minimum income policies as responses to ‘accidental fall’ situations that are often due to market hazards. An alternative approach, however, refers not to a last resort to safeguard against accidental falls due to the market, but rather to a ‘guaranteed’ means-tested social assistance scheme that aims to ensure a minimum standard of living.

At the European level, poverty and social exclusion are issues that have been increasingly prioritized on the agenda of the European Commission and other Union bodies. Thus far, the presence of a generalized safety net has become a fundamental ingredient of the European social model in all national variants. In their recent work compiling the implementation of minimum income schemes in Europe, Frazer and Marlier (2009) clearly indicate that all member states, with the exception of Greece and Italy, have been trying to implement some form of a national minimum income scheme that aims to offer income support to people who are either ineligible for social insurance payments or no longer entitled to them.

There are almost as many minimum income schemes in Europe as there are member states, and they vary in both coverage and comprehensiveness in terms of selectivity criteria, and effectiveness in terms of their impact on the levels and depth of the income poverty risk (Cantillon et al., 2008; Figari et al., 2010; Rat, 2009).¹ Cantillon et al. (2008) report that minimum income protection is supported by benefits in at least 19 of 25 European Union (EU) member states. Frazer and Marlier (2009) depict four types of minimum income schemes across Europe: simple and comprehensive schemes for those with insufficient means of living (e.g. Portugal); simple and non-categorical schemes with restricted eligibility and coverage (e.g. Hungary); a complex network of different, often categorical and sometimes overlapping, schemes (e.g. Spain); and limited, partial or piecemeal arrangements restricted to very narrow categories of people (e.g. Bulgaria). Saraceno (2010) highlights the importance of income support for the poor in most EU countries as part of the social citizenship package where a non-categorical safety net exists for all citizens below a given income threshold.

Notwithstanding the debate on how to reconfigure existing minimum income schemes, a related discussion is how to design the eligibility, entitlement and benefit payment modules of such schemes from scratch. Framed in this context, this paper aims to simulate the effects of a potential minimum income scheme (in the specific sense of a ‘guaranteed’ means-tested social assistance scheme that aims to ensure a minimum standard of living) in Turkey – a country that lacks a coherent social safety net. Indeed, the nature of the welfare regime in Turkey assumes an inegalitarian corporatist character, where the claims of citizens are highly dependent on membership in occupationally defined groups built within the social security system, which form the backbone of a regime that fundamentally excludes the poor (Bugra and Keyder, 2006; Yakut-Cakar, 2007). The structure of the welfare regime in Turkey resulted in the formation of a patchy social safety net, which is far from granting universal, adequate and systematic benefit provision to all citizens (Yakut-Cakar, 2010). Here, we emphasize that the welfare regime in Turkey may be considered as sharing similar characteristics with its southern European counterparts (namely Greece, Italy, Spain and Portugal), and contextualize our analysis on the common trait of a lack of a national income safety net and an integrated social assistance programme, along with sets of categorical measures (Andreotti et al., 2001; Arriba and Moreno, 2005; Gough, 1996; Matsaganis, 2005; Matsaganis et al., 2003; Sacchi and Bastagli, 2005; Saraceno, 2002). As Saraceno (2006: 97) notes, the poor who do not fit in the given specific categories (such as those poor who are also, for example, elderly, widows, orphans or living with a disability) cannot count on any general policy measure, but rather have to rely on family solidarity, charity or local provision, and consequently fall through the cracks of the perforated and inadequate safety net in these countries. We argue that these are common traits also observed in Turkey, where the realm of social assistance is dominated by the uncoordinated and patchy coexistence of mainly categorical policy measures that grant entitlement to discretionary, irregular, means-tested cash or in-kind benefits (Yakut-Cakar and Yilmaz, 2009).² In this setting, we find it important to discuss the introduction of minimum income schemes not only to ‘fill in’ the holes of the safety net, but also to put forward the possibility of institutionalizing a coherent set of policy measures by replacing the currently existing patchy schemes.³

In our analysis, we simulated a group of policy measures by using the 2007 wave of the National Statistics on Income and Living Conditions (TR-SILC). The impact of a minimum guaranteed income scheme on the extent of poverty, income distribution and the proportion of potential beneficiaries at the individual and household levels were investigated, as well as their corresponding policy costs. A similar study by Matsaganis et al. (2001) shows the impact of a potential scheme on eliminating extreme poverty in Greece, where the absence of a minimum income scheme translates to considerable pension bias in social expenditures and the allocation of insufficient resources to social assistance. Their analysis presents not only the extent and structure of extreme poverty in Greece, but also the scheme’s impact on reducing poverty and inequality, based on variations of the scheme under several non-take-up and leakage assumptions. In a similar vein, our intention here is to investigate the potentialities of institutionalizing such a scheme in a setting where social expenditures fall much shorter than their southern European counterparts, even when providing basic socio-economic security to those in need (Bugra and Adar, 2008). To this end, our methodology emphasizes the estimated costs of introducing variations of such a scheme, and ascertains the impact of these policy variations on both income distribution and the extent of poverty, through observed changes in the Gini coefficient. The policy variations introduced here will be combinations of several benefit payment thresholds and eligibility criteria.

While we acknowledge that the existence of a minimum income scheme may indeed not translate to decent living conditions with adequate incomes, it may certainly be claimed that income transfers through such schemes will contribute to reducing the resource deficit of the poor in contexts where a fully fledged national social safety net is evidently absent (for discussion on Portugal see, for instance, Baptista and Cabrita, 2009; Pereirinha, 2006). In this respect, we would also like to add to the academic debate on the political grounds of introducing such schemes, by providing empirical justifications for advocating them.⁴

The following sections of the paper describe the data used in the policy simulations, the methodology of the analysis, the results and the policy implications of the simulations. The paper then concludes with a brief discussion of the main remarks in light of the preceding sections.

Data

In line with policy developments to harmonize statistical indicators with the EU as part of the accession process, the Turkish Statistical Institute (Turkstat) introduced the TR-SILC, which was adapted from the EU-SILC, piloted in 2004, and has been fully implemented since 2005. The novelty of this paper lies not only in the implementation of potential policy scenarios, but also in the usage of this new set of data from the survey wave of 2007, which was released relatively recently, in December 2009.⁵

Our analysis utilizes the 2007 TR-SILC household data with all incomes reported for the reference year 2006.⁶ The sample consists of 10,796 households with 42,458 members, and sample weights are attached to each household to align the sample with the characteristics of the national population in the reference year. The household disposable income (HDI) of the household dataset is based on two sources of information:

1. Individual datasets, which provide information on income received by each individual in the form of annual net wages and salaries, annual net income from self-employment, in-kind income from consumption of personal agricultural production, unemployment benefits, pension benefits, survivor benefits, sickness benefits, student aids and all other incomes, including lump-sum tax refunds and earnings received from additional jobs held; and

2. Household datasets, which provide information on income received by the household in the form of formal and private cash and in-kind benefits,⁷ earnings of children under the age of 15 (if any), property and investment incomes and imputed income for the non-responding individuals in the household.

The earnings components of the HDI are reported as net of personal income tax, property tax and investment income tax. Moreover, any income transferred to other households is deducted from the HDI. Ideally, the impact assessment of a new benefit scheme such as a minimum income scheme would be best examined via the national tax–benefit model. Figari et al. (2010), for example, used the micro-data derived from national income surveys and simulated entitlement to minimum income schemes in the EU by using an integrated tax–benefit micro-simulation model. Meanwhile, for Turkey, Yakut-Cakar (2010) constructed a static national model based on the 2004 Household Budget Survey (HBS) data. In this dataset the earnings and benefits components of the HDI were defined differently to our data from the TR-SILC, which, we argue, is superior to the HBS in relation to its more detailed breakdown of items constituting the HDI. Also, as mentioned above, our intention is to introduce basic computational cost calculation exercises with several policy combinations, rather than to situate this in a fully fledged tax–benefit system. Thus, considering these limitations along with the already existing work in Yakut-Cakar (2010), here, with the methodology outlined below, we deal solely with the benefit aspect of the tax–benefit system without interfering with the income tax element.

Methodology

As mentioned above, the calculations were performed with data from the 2007 wave of the TR-SILC, with reference to incomes from 2006. As our aim was to introduce a means-tested benefit, poverty thresholds had to be determined. We adopted three types of thresholds: 40%, 50% and 60% of median income in 2006, the year for which household-level income data are available from the 2007 wave of the TR-SILC. Thresholds were calculated as 425.74, 532.17 and 638.61 Turkish lira per month for a family of two adults and two children.⁸ The Organisation for Economic Co-operation and Development (OECD) equivalence scale was used to adjust for the adult-equivalent figures for these modal families of two adults and two children. It is useful to note that the official gross monthly minimum wage in Turkey for the year 2006 was 531 Turkish lira. Though the amount is not binding, the level of the official minimum wage may be regarded as a proxy for the poverty threshold, as it refers to the amount necessary to meet the basic needs of the wage earner and dependants. Once the poverty thresholds were set, reference incomes that would act as rules for simulated means testing under the specific schemes were introduced. As mentioned above, the HDI variable contained in the household dataset of the TR-SILC was used to this end. As our ultimate aim was to introduce a scheme to replace certain components of the current social assistance system, we modified the HDI of the dataset by subtracting the formal cash and in-kind benefits from state-provided social assistance component.⁹

Based on the modified and equivalized HDI as the reference income, variations of the means test were applied to our potential schemes vis-à-vis Thresholds 1, 2 and 3.¹⁰ In addition to these three thresholds, we explored two variations introducing households where the household head was not covered by social security, and households where the household head was eligible for a ‘Green Card’, which is a means-tested public health insurance programme for the poor. This resulted in five different groups:

Below 40% of the median income: only those households with incomes below 40% of the median income; 10.4% of the households fell into this category.

Below 50% of the median income: only those households with incomes below 50% of the median income; 16.57% of the households fell into this category.

Below 60% of the median income: only those households with incomes below 60% of the median income; 23.51% of the households fell into this category.

Insecurely poor: the social security system in Turkey is of an inegalitarian corporatist nature that provides contribution-based hyper-protection to insiders and their dependants (Bugra and Keyder, 2006; Yakut-Cakar, 2007). The means-testing criteria of existing social assistance schemes exclude those covered by social security. With this point in mind, we ascertained the social security status of the self-reported head of the household in all households with incomes below 60% of the median income. As any one member anchored in the guarantismo provides social security coverage and access to contribution-based services to other members as dependants (Ferrera, 1996), we then excluded households that were covered through the head of the household. Our calculations revealed that insecurely poor households comprised 15.5% of total households.

Formally poor: as mentioned above, the Green Card scheme provides free healthcare services to poor households lacking social security coverage. The means test for Green Card eligibility uses the gross official minimum wage (531 Turkish lira) as the threshold value and determines benefit entitlement on the basis of per capita income in the household. Green Card take-up was used as a proxy to introduce the possibility of administrative discretion.¹¹ Excluding households under 60% of the median income that were not Green Card holders, 9.45% of the households were calculated to be formally poor.¹²

We note that formally poor households (consisting of Green Card holders without social security) are fully represented in insecurely poor households that are grouped on the basis of the social security coverage for the head of the household.

For each of these five eligible groups of households, we developed two alternative minimum income scheme payments, as follows:

A universal benefit provided to all eligible households with the amount of 40% of the median income. This corresponds to a monthly payment of 425.73 Turkish lira for a family of four (two adults and two children).

A complementary/top-up benefit payment scheme where the amount of the benefit assigned to a household is the difference between the relevant threshold and household reference income set as modified and equivalized HDI. In other words, the modified and equivalized HDI is topped up to the relevant threshold. The scheme assumes the same means testing procedure as above and implicitly supposes that in implementation, all incomes will be reported as in the TR-SILC.

Obviously, the first alternative implies some universality in the amount of the benefits, while the second inherits the minimum income scheme approaches employed in many EU states. By elaborating the universal scheme, we intend to demonstrate the maximum cost of the benefit schemes under different eligible group alternatives. Though the design of this universal scheme incurs low administrative costs as it offers a flat amount of benefit, it also has the potential to accommodate an inegalitarian perspective given the unequal income distribution across the households below the thresholds. Here, we acknowledge that the generosity of the benefits do not directly translate to lower poverty risk, as the effectiveness of a social assistance scheme also depends on eligibility criteria and the issue of take-up (Cantillon et al., 2008). By assuming full take-up of benefits with stringent eligibility criteria, we intend to demonstrate the cost-effectiveness of both benefit alternatives (that is, topped up or universal), as both could be implemented in the given welfare administration context. Indeed, the most typical minimum income scheme that guarantees a certain level of income is the Portuguese scheme, Rendimento Minimo Guarantido, implemented in 1996, where the benefit amount is attached to the amount of social pension as the guaranteed threshold value (for a detailed discussion, see Baptista and Cabrita, 2009; Bruto de Costa, 2003; Gouveia and Rodrigues, 2002; Rodrigues, 2004). Tables 1a and 1b present the conceptualization of eligible households and policy scenarios, respectively.

Table 1a.

Definitions of poverty

	Below 40% of the median	Below 50% of the median	Below 60% of the median	Insecurely poor	Formally poor
Threshold (for a family of four (two adults and two children per modified OECD equivalent in TL)	425.74	532.17	638.61	638.61	638.61
Additional criteria	None	None	None	Head of household not covered by social security	Holds a Green Card
Share in total HH (%)	10.4	16.57	23.51	15.5	9.45

HH: households OECD: Organisation for Economic Co-operation and Development; TL: Turkish lira.

Source: Authors’ calculations based on the 2007 wave of the TR-SILC.

Table 1b.

Minimum income scheme benefit payments

Policy	Type	Amount of the benefit
Case 1	Universal	40% of the median income
Case 2	Complementary/top-up	Difference between reference household income and relevant threshold

The take-up rate was assumed to be 100% for all policy scenarios where the legal rules were universally known and accepted, all citizens had perfect information about eligibility and entitlement criteria, the costs of compliance were zero and all eligible households received the benefit payment in all combinations.¹³ For each group of eligible households and each minimum income benefit payment developed above, the total cost of the schemes was calculated by adding the benefit transfers provided to all eligible households for every combination of households and benefit payments. These sums were then presented as proportional to GDP and central government expenditures for the same year. The full take-up assumption allowed us to estimate the upper bounds for costs and coverage of the policy combinations. Also, ex-post poverty rates after benefit transfers were calculated to demonstrate the poverty-reducing impact of the policy scenarios. It should be noted that, even after transfers, these households stayed below the median income level; hence, the poverty thresholds remained unchanged. Apparently, as mentioned before, minimum income schemes serve as a last resort in maintaining a basic level of standard of living and, thus, work to eliminate extreme poverty. In this respect, the calculation of poverty rates after transfers deserves attention. In addition, the impact of each scenario on income inequality level was analysed via the Gini coefficient and the S80/S20 ratio calculated at household level.

Here it should be noted that the Gini coefficients presented to demonstrate the impact of minimum income schemes were calculated with reference to the income position of the households, equivalized after receipt of the transfer. Accordingly, the Gini coefficient was found to be 0.4106 prior to any policy simulation.

Although the Gini coefficient is criticized as being too sensitive to the relative changes around the middle/mode of the income distribution, it is relatively simple to interpret and not overly sensitive at the lower ends of the distribution (as mentioned, for instance, by Foerster and d’Ercola, 2005). The primary purpose of the analyses in this paper was to raise the possibility of introducing a minimum income scheme to combat extreme poverty, rather than elaborate on overall distributional impact. In this respect, we used the S80/S20 ratio from the primary Laeken indicators; that is, the ratio of the total income received by the richest quintile to that received by the poorest quintile. The total income of the top quintile in Turkey was calculated as being approximately 8.2 times more than that of the lowest quintile, prior to any policy simulation. This points to a substantial income gap between the richest and the poorest segments of the population, compared with the EU-27 average of 5.0 in 2007 (Eurostat, 2010).

Results

Combinations of the policy simulations with different categories of eligible households are presented in Table 2. The first column represents eligible households, while the second indicates the policy scenarios developed. The following columns provide resulting at-risk-of-poverty rates and costs for each combination of eligible household and type of payment.

Table 2.

Poverty-reducing impact and costs of policy simulations

Payment to	Amount paid	Relative poverty rate (40% of median)	Relative poverty rate (50% of median)	Relative poverty rate (60% of median)	Share in GDP (%)	Central government expenditures (%)	Cost (million TL)
Initial level		10.4	16.57	23.51
Households below 40% of median poverty threshold	Pay 40% poverty threshold	0	6.47	14.9	1.15	4.90	728
Households below 40% of median poverty threshold	Complete to 40% poverty threshold	0	16.57	23.51	0.33	1.42	210
Households below 50% of median poverty threshold	Pay 40% poverty threshold	0	0.3	8.73	1.87	7.96	1,181
Households below 50% of median poverty threshold	Complete to 50% poverty threshold	0	0	23.51	0.71	3.01	447
Households below 60% of median poverty threshold	Pay 40% poverty threshold	0	0.3	1.79	2.69	11.46	1,701
Households below 60% of median poverty threshold	Complete to 60% poverty threshold	0	0	0	1.28	5.45	809
Households below 60% of median excluding those under SGK	Pay 40% poverty threshold	2.04	4.51	9.6	1.75	7.46	1,107
Households below 60% of median excluding those under SGK	Complete to 60% poverty threshold	2.04	4.24	8.01	0.96	4.10	609
Households below 60% of median and with Green Card	Pay 40% poverty threshold	4.64	8.82	15.3	1.05	4.46	662
Households below 60% of median and with Green Card	Complete to 60% poverty threshold	4.64	8.57	14.07	0.64	2.72	403

SGK: Social Security Institution TL: Turkish lira

As expected, our analyses revealed that policy scenarios with benefits topping up income to the relevant threshold eliminates the risk of poverty according to that threshold, but does not affect the risk of poverty associated with higher thresholds. Universal payment schemes provide more interesting results. Even when the target group is restricted to those earning less than 40% of the median income, relative poverty is reduced from 23.5% to 14.1%. The rate drops further to 8.7% and 1.8% when those below 50% and 60% of the median income are covered, respectively.

The poverty rates in Table 2 reveal that the nature of each policy and eligibility combination was correspondingly reflected in the poverty-reducing impact of the benefit payments. All policy combinations were seen to achieve a substantial reduction in those below 40% of median income. In fact, except for the two restrictive cases (according to social security coverage and Green Card), the risk of poverty for those below 40% of the median income was totally eliminated – as expected. When analysed with respect to other thresholds, the transfers had a poverty-reducing impact in universal payment schemes. When insecurely poor households were targeted, the payment of universal benefits in the amount of 40% of the median income and topping up to the 60% threshold appear to have similar results in both poverty reduction and the costs of the schemes.

As seen in Table 2, the cost of the policy scenarios ranges between 0.33% and 2.69% of gross domestic product (GDP), while their respective shares among central government expenditures were between 1.42% and 11.46%. The total cost of universal payments ranges between 1.05% and 2.69% of GDP, which is significantly higher than the total cost of topped-up payments. Even though benefit schemes that top up benefits to the relevant thresholds seem to cost less than universal payments, they also seem to have less impact on reducing the risk of poverty. Compared with the level of public social assistance expenditures in 2007 (0.58% of GDP), we observe that the benefit scheme that tops up the income of households below 40% of the median income to that threshold is even cheaper in financial terms, whereas the costs associated with the other schemes are higher. Nevertheless, within the contemporary political debates surrounding poverty reduction in Turkey, we find these alternatives politically feasible in light of the recent public statement of Prime Minister Erdogan, who announced a gradual increase in social assistance expenditures to 3% of GDP by 2023 (JDP, 2011: 102).

Regarding their impact on income distribution, all policy simulations with different combinations of eligible households were found to favourably alter income distribution inequalities, as expected, albeit at different levels. That is to say, even in the case of the policy combination that had the smallest poverty-reducing impact – which covered only poor households that were Green Card holders where payments rounded up income to the 40% threshold – the Gini coefficient decreased by approximately 0.0071 percentage points, from 0.4106 to 0.4035. In turn, the most generous alternative, a payment of 40% of the median income to those below the relative poverty threshold, improved the Gini coefficient, decreasing it by 5.794 percentage points to 0.3599. The ratio of the impact in the most influential policy relative to the impact of the least influential policy is 7, in line with the ratio of costs, which is 8.

The ratio of the total income received by the richest quintile to that received by the poorest was found to reveal an improvement in unequal income distribution in all policy combinations (Table 3). Although the decrease rate in the S80/S20 ratio was the least in topping up benefit payments to the HDI for Green Card holders, the inequality-reducing impact of the transfers was clearly apparent in the changes in S80/S20 ratios with universal payments in the amount of 40% of the threshold, for all three groups simulated in the analysis.¹⁴

Table 3.

Impact of potential minimum income schemes on income distribution

Payment to	Amount paid	Gini Coefficient	S80/S20 Ratio	Change in mean HDI for Q1 (%)
Initial level		0.4106	8.2
Households below 40% of median poverty threshold	Pay 40% poverty threshold	0.3882	6.36	28.91
Households below 40% of median poverty threshold	Complete to 40% poverty threshold	0.4035	7.52	9.04
Households below 50% of median poverty threshold	Pay 40% poverty threshold	0.3747	5.69	44.09
Households below 50% of median poverty threshold	Complete to 50% poverty threshold	0.3955	6.85	19.70
Households below 60% of median poverty threshold	Pay 40% poverty threshold	0.3599	5.15	59.31
Households below 60% of median poverty threshold	Complete to 60% poverty threshold	0.3836	6.00	36.57
Households below 60% of median excluding those under SGK	Pay 40% poverty threshold	0.3777	5.94	38.13
Households below 60% of median excluding those under SGK	Complete to 60% poverty threshold	0.3907	6.44	27.25
Households below 60% of median and with Green Card	Pay 40% poverty threshold	0.3915	6.72	21.97
Households below 60% of median and with Green Card	Complete to 60% poverty threshold	0.3980	6.98	17.52

HDI: Household Disposable Income; SGK: Social Security Institution

The minimum income scheme had a considerable impact on average income level in the lowest income quintile. The policy that provided a universal level of benefits to all below 60% of the median income – the most generous policy – increased the mean income level in this quintile by 59%. In contrast, the smallest improvement, a change of only 9%, was observed when the policy addressed only those below 40% of the median income, with a top-up benefit payment completing the income to that threshold. Though not presented here, an improvement was also observed in the average income level of those in the second-lowest income quintile when policies with wider coverage and more generous benefits were considered.

Concluding remarks

The analysis presented in this paper should be regarded as a ceteris paribus exercise in introducing variations of a minimum income scheme in Turkey that will alter existing social assistance mechanisms, rather than a test of the effectiveness of the social protection system as a whole.

Our calculations demonstrate that the costs of introducing a minimum income scheme vary between 0.33% and 2.69% of GDP in its extreme forms, with the lowest benefit amount paid solely to those earning below 40% of the median income, and the highest provided to all households with incomes below 60% of median-equivalized HDI. The effects of these policy combinations on income inequality (in the form of Gini coefficients, S80/S20 ratios and mean HDI figures) were then computed.

Similarly, using different thresholds, we have shown that the poverty-reducing impact of the policy combinations is significant even in the least generous benefit payment combination. As expected, top-up benefits that increased household income to the relevant threshold eliminated the risk of poverty defined relative to that threshold, and at risk of poverty rates reduced significantly with universal payments. We should note here that even the most generous payment, that is topping up benefits to 60% of the median threshold, is approximately 300 euros per month and relatively lower than currently existing schemes in the EU. Thus, the high cost associated with this scheme is solely due to the high prevalence of poverty in Turkey, as the number of households with incomes below the threshold is considerably high. In this respect, we believe our analysis could be carried further to examine poverty profiles and income distribution under the most cost-efficient and poverty-reducing policy combinations. As noted above, the reference income calculations in our analysis exclude formal cash and in-kind benefits received by the household. Our policy variations implicitly illustrated the total costs of social assistance expenditures composed solely of the hypothetical schemes introduced.

As argued above, contemporary actors involved in social assistance in Turkey constitute not only several state institutions but also local municipalities and charitable organizations, where the role of non-state actors may be considered marginal, as state institutions assumed 85% of total social assistance expenditures in 2007. We noted that state-provided social assistance comprises mainly categorical schemes that target the elderly and disabled poor, conditional cash transfers that target children in poor families and discretionary ad hoc cash and in-kind transfers (Yakut-Cakar and Yilmaz, 2009). As mentioned, our policy simulations inherently assume that currently existing social assistance mechanisms would be replaced by the minimum income scheme introduced. Thus, considering that the proportion of actual social assistance expenditures by state institutions constituted approximately 0.58% of GDP in 2007 (Yenturk, 2009), we argue that for several combinations of benefit levels and eligible households, minimum income schemes are both affordable and politically feasible in Turkey.

Our analysis has certain limitations: we assumed that all components of the HDI would be fully and accurately reported. We also chose not to consider the administrative costs of introducing a minimum income scheme. Instead of a complex and interrelated network of schemes, we introduced simplistic but comprehensive policy variations that primarily dealt with the cash benefit component. In this respect, we did not take into account the activation/labour market integration component, which constitutes one of the main pillars of contemporary schemes throughout Europe. Finally, we refrained from fully discussing the funding for the scheme, opting to consider it as a rough replacement for current social assistance schemes offered by various state institutions.

When the target group of beneficiaries were households with Green Card holders or those without social security, our simplistic analysis illustrated that the benefit schemes had significantly less impact on reducing the risk of poverty for all households. Especially for social security, this restriction does not qualify households as being eligible for the benefit scheme where the household head is formally employed. This observation points to the prevalence of in-work poverty in Turkey, where the income of approximately 18% of the working population is lower than 60% of the median. Thus far, policy combinations with restrictions over the eligible households presented here could not fully address the problem of in-work poverty.

Against the backdrop of the Southern European experience in relation to the introduction of anti-poverty policies in general, and minimum income schemes in particular, that have so far produced different outcomes in different countries, we have simulated the introduction of a minimum income scheme for Turkey and analysed its potential impacts for society at large and especially the poor. Our exercise will be seen, we hope, as a further contribution to the ongoing debate on social protection systems that are designed to address the risk of poverty and to reduce income inequality.

Footnotes

Acknowledgements

The authors would like to thank Manos Matsaganis as well as two anonymous referees for their comments, criticisms and suggestions. The usual caveat applies.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Andreotti

Marisol Garcia

Gomez

Hespanha

Kazepov

Mingione

(2001) ‘Does a Southern European Model Exist?’, Journal of European Area Studies 9(1): 43–62.

Arriba

Moreno

(2005) ‘Spain: Poverty, Social Exclusion and Safety Nets’, in Ferrera

(ed.) Welfare State Reform in Southern Europe: Fighting Poverty and Social Exclusion in Italy, Spain, Portugal and Greece. London: Routledge.

Baptista

Cabrita

(2009) ‘Portugal: A Study on the Minimum Income Schemes’, Expert Report for Peer Review in Social Protection and Social Inclusion. Brussels: European Commission.

Bruto da Costa

(2003) ‘Minimum Guaranteed Income and Basic Income in Portugal’, in Standing

(ed.), Minimum Income Schemes in Europe. Geneva: International Labour Office.

Bugra

Adar

(2008) ‘Social Policy Change in Countries without Mature Welfare States: The Case of Turkey’, New Perspectives on Turkey 38: 33–52.

Bugra

Keyder

(2006) ‘The Turkish Welfare Regime in Transformation’, Journal of European Social Policy 16(3): 211–228.

Bugra

Sinmazdemir

N.T.

(2004) ‘Turkiye’de Uygulanabilecek Bir Nakdi Yardim Politikasinin Maliyeti Uzerine Farkli Senaryolara Dayanan bir Calisma’ [A Study on the Costs of a Cash Support Policy in Turkey based on Different Scenarios], Bogazici University Social Policy Forum Discussion Paper. Istanbul: Bogazici University.

Cantillon

van Mechelen

Schulte

(2008) ‘Minimum Income Schemes in Old and New Member States’, in Alber

Faney

Saraceno

(eds) Handbook of Quality of Life in the Enlarged EU. London: Routledge.

Eardley

Bradshaw

Ditch

Gough

Whiteford

(1996) ‘Social Assistance Schemes in OECD Countries: Volume I Synthesis Report’, Department of Social Security, Research Report 46. London: HMSO.

10.

Eurostat (2010) Online Database on Population and Social Conditions – Living Conditions and Social Protection Available from: http://epp.eurostat.ec.europa.eu/portal/page/portal/living_conditions_and_social_protection/data/database (accessed 10.05.2010).

11.

Ferrera

(1996) ‘The “Southern Model” of Welfare in Social Europe’, Journal of European Social Policy 6(1): 17–37.

12.

Ferrera

(2000) ‘Reconstructing the Welfare State’, in Kuhnle

(ed.), Survival of the European Welfare State. London: Routledge.

13.

Ferrera

(2005) ‘Welfare States and Social Safety Nets in Southern Europe: An Introduction’, in Ferrera

(ed.) Welfare State Reform in Southern Europe. London: Routledge.

14.

Figari

Haux

Matsaganis

Sutherland

(2010) ‘Coverage and Adequacy of Minimum Income Schemes in the European Union’, EUROMOD Working Paper 2010–37. Colchester: ISER, University of Essex.

15.

Foerster

d’Ercola

M.M.

(2005) ‘Income Distribution and Poverty in OECD Countries in the Second Half of the 1990s’, OECD Social, Employment and Migration Working Papers 22. Paris: OECD.

16.

Frazer

Marlier

(2009) ‘Minimum Income Schemes across EU Member States’, Report prepared by EU Network of National Independent Experts on Social Inclusion. Available at: www.peer-review-social-inclusion.eu/network-of-independentexperts/2009/minimum-income-schemes (accessed 01.02.2010)

17.

Gough

(1996) ‘Social Assistance in Southern Europe’, South European Society and Politics 1(1): 1–23.

18.

Gouveia

Rodrigues

C.F.

(2002) ‘The Impact of a “Minimum Guaranteed Income Program” in Portugal’, Public Finance and Management 2(2): 250–271.

19.

Immervoll

(2009) ‘Minimum Income Benefits in OECD Countries: Policy Design, Effectiveness and Changes’, OECD Social, Employment and Migration Working Papers 100. Paris: OECD.

20.

Justice and Development Party (JDP) (2011) Resource is Turkey: Target 2023 – 2011 Elections Manifest. Ankara. Available from: www.akparti.org.tr/beyanname2011.pdf (accessed 19.05.2011).

21.

Matsaganis

(2005) ‘Greece. Fighting with Hands Tied behind the Back: Anti-poverty Policy without a Minimum Income’, in Ferrera

(ed.) Welfare State Reform in Southern Europe. London: Routledge.

22.

Matsaganis

Papadopoulos

Tsakloglou

(2001) ‘Eliminating Extreme Poverty in Greece’, Journal of Income Distribution 10(1–2): 40–57.

23.

Matsaganis

Ferrera

Capucha

Moreno

(2003) ‘Mending Nets in the South: Anti-poverty Policies in Greece, Italy, Portugal and Spain’, Social Policy and Administration 37(6): 639–655.

24.

Matsaganis

Paulus

Sutherland

(2008) ‘The Take-up of Social Benefits’, Research Note 5, European Observatory on the Social Situation and Demography. Brussels: European Commission.

25.

Pereirinha

(2006) ‘Poverty and Anti-poverty Policies in Portugal: The Experience of the Guaranteed Minimum Income (GMI)’, in Petmesidou

Papatheodorou

(eds) Poverty and Social Deprivation in the Mediterranean Area: Trends, Policies and Welfare Prospects in the New Millennium. London: Zed Books.

26.

Rat

(2009) ‘The Impact of Minimum Income Guarantee Schemes in Central and Eastern Europe’, in Cerami

Vanhuysee

(eds) Post-Communist Welfare Pathways: Theorizing Social Policy Transformations in CEE. Basingstoke: Palgrave Macmillan.

27.

Rodrigues

C.F.

(2004) ‘The Redistributive Impact of the Guaranteed Minimum Income Programme in Portugal’, Working Paper. Lisbon: Research Centre on the Portuguese Economy (CISEP), Technical University of Lisbon.

28.

Sacchi

Bastagli

(2005) ‘Italy. Striving Uphill but Stopping Halfway: The Troubled Journey of the Experimental Minimum Insertion Income’, in Ferrera

(ed.) Welfare State Reform in Southern Europe. London: Routledge.

29.

Saraceno

(ed.) (2002) Social Assistance Dynamics in Europe: National and Local Poverty Regimes. Bristol: Policy Press.

30.

Saraceno

(2006) ‘Social Assistance Policies and Decentralization in the Countries of Southern Europe’, Revue Française des Affaires Sociales 5: 97–117.

31.

Saraceno

(2010) ‘Concepts and Practices of Social Citizenship in Europe: The case of poverty and income support for the poor’, in Alber

Gilbert

(eds) United in Diversity? Comparing Social Models in Europe and America. Oxford: Oxford University Press.

32.

Yakut-Cakar

(2007) ‘Turkey’, in Deacon

Stubbs

(eds) Social Policy and International Interventions in South East Europe. Cheltenham: Edward Elgar.

33.

Yakut-Cakar

(2010) ‘Comparative Analysis of Poverty and Inequality in Turkey’, Unpublished PhD Thesis. Department of Economics, Institute of Social Sciences, Marmara University.

34.

Yakut-Cakar

Yilmaz

(2009) ‘Social Assistance in Turkey: On the Margins of Redistributive Justice?’, paper presented at Beyond Social Inclusion: Towards a More Equal Society? Annual International Conference, Loughborough University, 22–23 January.

35.

Yakut-Cakar

Yilmaz

(2010) ‘Gelir Hakki: Neden Sol Bir Talep?’ [Right to Income: Why is it a Demand from the Left?], Turk-Is Dergi 389: 60–66.

36.

Yenturk

(2009) ‘Sosyal Koruma Harcamalarini İzleme Kilavuzu’ [Guidance Note for Monitoring Social Protection Expenditures], NGO Training Notes on Monitoring Public Budget 3. Istanbul: Istanbul Bilgi University.