Introduction
Researchers and policy makers are faced with a variety of approaches to measuring the extent of poverty in modern welfare states. Ranging from updated absolute poverty benchmarks, to relative income poverty lines, minimum income standards thresholds, material deprivation measures, and various combinations of these approaches, all indicate that there remains a large proportion of society living in poverty (Summers, Reference Summers and Greve2019; Collins, Reference Collins2020; European Commission, 2021; Lister, Reference Lister2021). Policy commitments to reduce the extent of poverty remain commonplace, some targeted at particular groups, like children and older people, while others take a population-wide focus. However, despite some progress, a mismatch between poverty reduction targets and poverty outcomes remains commonplace, and where progress is achieved this is not always sustained as policy priorities change and economic challenges dilute or displace the focus on anti-poverty initiatives (Copeland & Daly, Reference Copeland and Daly2012; Social Justice Ireland, 2020; Pemberton & Arriaga-Garcia, Reference Pemberton, Arriaga-Garcia, Alcock, Haux, McCall and May2022). For example, the Europe 2020 target to reduce by twenty million the headcount of individuals experiencing ‘poverty and social exclusion’ (defined as a combined income, deprivation, and household work intensity indicator), achieved a near eleven million reduction between 2008 and 2019, leaving 105 million people, one in five Europeans, experiencing these conditions (European Commission, 2010, 2021; Eurostat, 2022). Judged over time, the experience in most countries reflects an endless cycle of good intentions, policy initiatives, missed targets, reviews, and new strategies.Footnote 1
Describing the situation in the UK since the ‘Great Recession’ of 2007/08, Pemberton & Arriaga-Garcia (Reference Pemberton, Arriaga-Garcia, Alcock, Haux, McCall and May2022: 220) summarise poverty as being ‘stable yet stubbornly high’, a descriptor that fits the experience of most European welfare states. However, it is of note that despite economic, pandemic, and cost-of-living challenges, poverty has, to date at least, not dramatically increased.
The research literature on poverty highlights the impact on individuals, families, and communities of poverty, costs that are both current to the experience and reflecting its scarring effects. Beyond its direct effect on living standards, poverty has been found to force people to lead a ‘marginal existence’ (Whelan, Reference Whelan2022), stigmatise people and implicitly impacting on their self-esteem, inducing a sense of shame and powerlessness given individual’s circumstances (Walker, Reference Walker2014; Sutton et al., Reference Sutton, Pemberton, Fahmy and Tamiya2014; Greve, Reference Greve and Greve2019a; Boland et al., Reference Boland, Doyle and Griffin2022; Finn & Murphy, Reference Finn and Murphy2022). It also imposes financial stress driven by ‘the relentlessness of the work involved in getting by’ (Gray et al., Reference Gray, Dagg, Rooney and Greve2019: 294). Van Lancker and Vinck (Reference Van Lancker, Vinck and Greve2019: 99) note how these stresses spill over and negatively impact family relationships and the quality of parenting. Poverty also damages people’s health and is associated with a greater prevalence of physical and mental illness, increased risk of chronic illness, health induced limitations to daily activities, and shorter lifetimes (Gordon, Reference Gordon, Hillyard, Pantazis, Tombs and Gordon2004; Marmot, Reference Marmot2004; Layte & Nolan, Reference Layte and Nolan2016; Collins, Reference Collins2020). The association between poverty and lower education outcomes has also been found to impact on life chances including lower earnings and poorer labour market outcomes (Nolan & Marx, Reference Nolan, Marx, Salverda, Nolan and Smeeding2009; Lesner, Reference Lesner2018; Bellani & Bia, Reference Bellani and Bia2019). Research on the intergenerational nature of poverty highlights how poor children are more likely to become parents who live in poverty and how poor children are themselves prone to worse cognitive, social-behavioural development, and health outcomes (OECD, 2018a; Cooper & Stewart, Reference Cooper and Stewart2021; Frazer et al., Reference Frazer, Guio and Marlier2021).
These individual, family, and community costs of poverty are stark and understandably receive the majority of attention in the research and policy literature. A further cost, one that is frequently hidden from assessments of poverty and its consequences, is the cumulative and recurring public expenditure associated with current policy responses to poverty (Bramley & Watkins, Reference Bramley and Watkins2008; Griggs & Walker, Reference Griggs and Walker2008; Barayandema & Fréchet, Reference Barayandema and Fréchet2012; Collins, Reference Collins2020). These costs derive from the identification of poverty as a determining factor in the need for, and demand for, a wide range of public services and policies ranging across almost all areas of public policy (Hirsch, Reference Hirsch2008, Pearce, Reference Pearce2011; McLaughlin & Rank, Reference McLaughlin and Rank2018). Poverty triggers needs and responses by the state, not always at the ideal level of adequacy, but nonetheless involving resources that are allocated to addressing current and past experiences of poverty. Overtime, in the absence of poverty, these are costs which can be expected to dissipate (Bramley et al., Reference Bramley, Hirsch, Littlewood and Watkins2016; Collins, Reference Collins2022). The determination of the nominal value of these costs, based on an assessment of public expenditure in one EU-15 state, is the focus of this article.
As a means of identifying those experiencing poverty, this article uses those below the 60 per cent median equivalised income poverty line as a proxy measure for those ‘whose resources (material, cultural, and social) are so limited as to exclude them from the minimum acceptable way of life’ in the society in which they live (Townsend, Reference Townsend1979; EEC, 1985; Piachaud, Reference Piachaud1987). While there are drawbacks to using a measure based on current income (see Predelli et al., Reference Predelli, France and Dearden2008; Hick, Reference Hick2012; ATD Ireland, 2022; Whelan, Reference Whelan2022), and the aforementioned list of alternative measurement approaches, this approach reflects a widely used relative measure which is a useful, if far from perfect, way of achieving Atkinson’s (Reference Atkinson1987) objective of distinguishing the poor from the non-poor.Footnote 2
The overall level of poverty in a society is frequently presented as a policy choice rather than an unavoidable societal outcome (Magadi & Middleton, Reference Magadi and Middleton2005; Esping-Anderson & Myles, Reference Esping-Anderson, Myles, Salverda, Nolan and Smeeding2009; Sinfield, Reference Sinfield and Greve2019; Greve, Reference Greve and Greve2019b), with Lister terming it ‘a preventable social harm’ (2021: 1). If so, the scale of poverty in a society, and the recurring public expenditure costs associated with it, can be considered as either an implicit or explicit policy outcome. Table 1 illustrates how EU-15 states record notably similar pre-distribution poverty rates (40–44 per cent) but the reductions achieved by pensions and social transfers range from 52 to 71 per cent.Footnote 3 While the table presents a point-in-time analysis, and overlooks differences in population structures, labour markets, and longer-term interventions to prevent poverty and its intergenerational transfer, it does underscore how prosperous welfare states starting from a similar position make different societal choices around the scale of intervention to address poverty.
Table 1. Poverty levels and poverty reduction in EU-15 states, 2020
Source: Calculated from Eurostat online database (indicators ilc_li02 and ilc_li09).
Note:
* UK data is for 2018 (the final year of UK data submitted to Eurostat).
This article focuses on one of these EU-15 states, Ireland, and determines an estimate of the annual additional public expenditure occurring as a result of current and past experiences of poverty. Ireland serves as an interesting choice given a near continuous series of national anti-poverty ‘strategies’, ‘action plans’, ‘targets’, and ‘roadmaps’ since 1997.Footnote 4 These plans have been framed with the objective of addressing multiple dimensions of poverty and have achieved some success (Government of Ireland, 2018: 23–24; DSP, 2021). Over that time the headline relative income poverty rate has fallen from approximately 20 per cent (1998–2005) and stabilised around 14–16 per cent (2008–2018) prior to the Covid-19 pandemic (Whelan et al., Reference Whelan, Layte, Maitre, Gannon, Nolan, Watson and Williams2003; CSO, 2022). However, headline targets to reduce those simultaneously experiencing relative income poverty and enforced deprivation (‘consistent poverty’) have continually been missed (Social Justice Ireland, 2020).
This article adds to the existing literature in a number of ways. First, it attempts to widen the lens society brings to considerations of poverty and the costs it imposes on society. The article highlights the nature and scale of the recurring public expenditure costs associated with poverty and in doing so identifies poverty as an issue with implications for almost all aspects of government and public expenditure. The failure of the policy system to understand the recurring fiscal impact of poverty suggests this is a gap in current anti-poverty policy, and it may go some way to explaining the juxtaposition between recurring commitments to tackle or eliminate poverty and policy outcomes which consistently miss these targets (Nolan & Marx, Reference Nolan, Marx, Salverda, Nolan and Smeeding2009; Vandenbroucke & Vleminckx, Reference Vandenbroucke and Vleminckx2011). The findings also strengthen the case for investing in poverty prevention strategies, including impeding intergenerational transfer. Judged from a narrow public expenditure cost-benefit perspective, there are notable long-term fiscal returns to be gained from reducing and preventing poverty, again considerations that tend to be absent from much contemporary anti-poverty policies (Sinfield, Reference Sinfield2012 and Reference Sinfield and Greve2019; Greve, Reference Greve and Greve2019a). Finally, the article adopts a novel approach to estimating a public sector cost of poverty, arguing that it is better to determine a range rather than a point estimate. This reflects the unavoidable uncertainty associated with the determination of some cost estimates and differs from the approaches adopted in previous assessments of this issue.
The remainder of the article is organised as follows. The next section provides a background for the study outlining recent trends in relative income poverty in Ireland before setting this analysis in the context of previous attempts to determine these costs. The estimating approach is then outlined and is followed by a presentation of the results. Finally, the article concludes by exploring the broader policy implications of these findings for the anti-poverty agendas of modern welfare states.
Background
Relative income poverty in Ireland
Between 2010 and 2020, on average one in seven of the Irish population lived on an income below the poverty line – approximately 720,000 individuals (see Fig. 2). Much like the experience across the EU-27 and the UK, the risk of poverty has altered little over most of that period with the rate sitting around 15–17 per cent (see Fig. 1). Labour market and welfare changes in Ireland and the UK marginally altered these outcomes in 2018–2019, although in opposite directions (Bourquin et al., Reference Bourquin, Cribb, Waters and Xu2019; Roantree et al., Reference Roantree, Maître, McTague and Privalko2021).
Figure 1. Percentage of population in relative income poverty, 2010–2020.
Notes: Poverty is measured using the 60 per cent median income line and the OECD modified equivalence scale. EU-27 members are as per 2020. UK data stops in 2018.
Source: Eurostat online database (indicator ilc_li02).
While debate persists on the merits and drawbacks of the relative income poverty line measure (Gordon, Reference Gordon, Dermott and Main2018; Kuypers & Marx, Reference Kuypers and Marx2021), the nominal value of these thresholds underscores the limited current resources available to households. In 2020 the weekly poverty line for a single individual in Ireland was €286, a rate that was 41 per cent above the minimum jobseekers payment and equivalent to 35 per cent of gross average weekly earnings. Collins found most of those living in poverty had an income some distance below the threshold, averaging a depth of €55 per week over 2008–2017 (2020: 15–18).
The age distribution of those below the poverty line is also informative – Fig. 2. While most of those living in poverty are of working age (60 per cent) there are a large number of children living with these adults. About 31 per cent of all those in poverty are aged seventeen years and under, averaging 220,000 each year – one in five children. This illustrates the phenomenon of low-income families living below the poverty line and points towards some of the current and intergenerational personal and public policy challenges that poverty represents (Hill et al., Reference Hill, Davis, Hirsch and Marshall2016; Daly & Kelly, Reference Daly and Kelly2015; Daly, Reference Daly and Greve2019). At the other end of the age distribution poverty is less common among those aged over 65 years. About 60,000 pensioners live on an income below the poverty line and represent 9 per cent of the poverty population – one in ten of this group.
Figure 2. Composition of population in relative income poverty, Ireland 2010–2020.
Source: Calculated from Central Statistics Office online database (indicators PEA15, SIA24, 28, 77 and 82).
Despite occasional increases and decreases, overall, the count and age composition of poverty has been reasonably static in Ireland for some time (Summers, Reference Summers and Greve2019; Collins, Reference Collins2020). While longitudinal assessments suggest that within these numbers there are frequent entries and exits from poverty (OECD, 2001; Fouarge & Layte, Reference Fouarge and Layte2005; Biewen, Reference Biewen2014), at any one point in time living life on an income below the poverty line is a reality for a large proportion of the Irish population.
Approaches to establishing a public service cost of poverty
The persistent presence of a large group of people living in poverty impacts on the choices that societies can make regarding the allocation of scarce public resources (Bramley et al., Reference Bramley, Hirsch, Littlewood and Watkins2016; Greve, Reference Greve and Greve2019a). State expenditure to address current and past experiences of poverty, and counter their consequences, is an unavoidable current cost that governments are forced to address (Greve, Reference Greve and Greve2019b); although the adequacy with which they do so is a matter of political choice.
Research focused on establishing the monetary costs of poverty is small relative to that on the experiences and consequences of poverty. Some studies focus on the population as a whole while others concentrate on one group, such as children. Many take a macro/top-down approach where measures of national income, productivity, and earnings are examined to determine the cumulative lost income and economic growth associated with the negative education, employment, and social outcomes of poverty. Others take a more micro/bottom-up approach and build up a picture of costs across a range of key public spending areas before aggregating these to provide an overall estimate. In some cases these estimates are extended to include individual/family opportunity costs and projections of future forgone taxation.
These studies vary in the comprehensiveness of their approach. Some consider costs at a point in time, generally one year, such as McLaughlin and Rank (Reference McLaughlin and Rank2018) for children in the US in 2015 (5.4 per cent of GDP), Blanden et al. (Reference Blanden, Hansen and Machin2008 and Reference Blanden, Hansen and Machin2010) for children in the UK (1–1.8 per cent of GDP, 2006 prices), Laurie (Reference Laurie2008) for the social/non-individual costs of poverty in Ontario, Canada (1.8–2.2 per cent of GDP, 2007 prices), Ivanova (Reference Ivanova2011) for British Columbia in Canada (4.1–4.7 per cent of GDP, 2008 prices), Barayandema and Fréchet (Reference Barayandema and Fréchet2012) for the social/non-individual costs of poverty in Québec, Canada (1.9–2 per cent of GDP, 2008 prices), and Ozdemir and Ward (Reference Ozdemir and Ward2014) for working age adults in the EU (0.8–1.2 per cent of GDP, 2010 prices).Footnote 5 Others select key areas where costs arise and sum-up the cumulative poverty induced lifetime amount. For example, Sherman and Edelman (Reference Sherman and Edelman1994) and the Children’s Defense Fund (1996) focus on US children but only estimate lost earnings (1.4–2.8 per cent of 1992 GDP). Holzer et al. (Reference Holzer, Schanzenbach, Duncan and Ludwig2008) take a more comprehensive approach and estimate those costs associated with lost earnings, additional crime, and poorer health among children in the US (3.8 per cent of GDP, 2006 prices). A similar approach, but focused on establishing an annual cost of child poverty to individuals and the state, was concluded by Pearce (Reference Pearce2011) for New Zealand (4.5 per cent of GDP, 2010 terms). Briggs et al. (Reference Briggs, Lee and Stapleton2016) focused on the cost of poverty to the city of Toronto and estimated savings from lower crime and improved health but broadened the analysis to simulate the additional taxes to the state from higher earnings (C$4.4–5.5bn per annum).
In the UK, Hirsch (Reference Hirsch2008) includes forgone future taxes, additional future benefit expenditure, losses to future net earnings, and estimates for poverty induced current spending on a range of public services to assess the costs of child poverty (2 per cent of GDP, 2008 prices).Footnote 6 That study built on work from Griggs and Walker (Reference Griggs and Walker2008) and Bramley and Watkins (Reference Bramley and Watkins2008), with the latter providing a comprehensive estimate of the public service cost of UK child poverty across eleven areas of public spending (0.8–1.4 per cent of GDP, 2006/2007 prices). Bramley et al. (Reference Bramley, Hirsch, Littlewood and Watkins2016) subsequently updated and extended this approach to cover all those in poverty in the UK (3.6 per cent of GDP, 2015 prices) and added estimates of the ‘knock-on’ costs of lost taxation revenue and additional future benefit expenditure (plus 0.5 per cent of GDP, 2015 prices).
This study follows Bramley and Watkins’ (Reference Bramley and Watkins2008) and Bramley et al.’s (Reference Bramley, Hirsch, Littlewood and Watkins2016) approach to estimating the additional state expenditure that partially or totally arise as a result of poverty. Bramley et al. (Reference Bramley, Hirsch, Littlewood and Watkins2016: 5–6) provide a useful categorisation of the types of public service costs of poverty under four headings: Damage caused by current experiences of poverty (e.g. poor health outcomes); Giving help to relieve the difficulties of living on a poverty income (e.g. providing social housing); Preventing the passing on of the effects of poverty (e.g. providing additional resources for disadvantaged schools); and Dealing with the legacy effects of past poverty (e.g. providing adult literacy programmes).
As not all of the expenditure on these areas can be regarded as an exclusive cost of poverty, some would arise in any event, this article is focused on isolating the proportion of expenditure that occurs as a result of poverty and which would, over time, be expected to dissipate in the absence of current experiences of poverty and the reduction and resolution of the scarring effects of past poverty. It aims to determine these costs as experienced by the state (the national government and its agencies) in one year and compares this to annual state expenditure and taxation revenue.Footnote 7 Given the uncertainty associated with projecting future earnings, taxes, and benefits across the lifetime of those currently experiencing poverty, the analysis does not attempt to quantify these future costs.
Estimating a public service cost of poverty
The cost estimates in this article are categorised into six broad areas of public policy. Within these the analysis examines twenty-five individual areas or expenditure programmes, accounting for €27.9 billion in annual state expenditure. These areas were chosen based on the aforementioned costing literature and a review of the expenditure programmes outlined in the 2019 Irish Budget. They reflect the identification of poverty as a determining factor in the need for, and demand for, a wide range of public services and policies (see Table 2). Some welfare support programmes, those funded from the exchequer rather than social insurance contributions, are also included where these are identified as likely to experience a reduction in demand in the absence of poverty.
Table 2. Estimating the public service cost of poverty
Notes: Figures have been rounded for the purpose of presentation. Annual expenditure data from DPER (2018a and 2018b) except for Health Care (Department of Health, 2018) and those for fuel allowance and household benefits package (Oireachtas 2019b). For some of the smaller welfare support payments it was not possible to establish a range. For some areas of expenditure, the main estimate is established as the same as the conservative low estimate or the high estimate. Supplementary material available at https://doi.org/10.1017/S147474642300043X provides a more comprehensive decomposition of these calculations.
The costings are driven by available statistical and administrative data from the Irish Central Statistics Office (CSO), various government departments and agencies, answers to parliamentary questions and the published results of related national and international studies. Public expenditure data has been principally sourced from the Department of Public Expenditure and Reform’s (DPER) Expenditure Report and Revised Estimates for 2019 (DPER, 2018a, 2018b). Data on population outcomes and incomes is analysed using microdata from the CSO’s Survey on Income and Living Conditions (SILC). This is particularly useful for comparing the situation of those above and below the poverty line.
As a means of dealing with the range of uncertainties and data deficits that arise in determining a public service cost of poverty, the analysis uses this data to identify a range of cost estimates (main, low, and high) for each category of public expenditure. These estimation challenges arise for four interrelated reasons: estimate uncertainty, where it is not possible to conclusively determine how much expenditure would continue in the absence of poverty; incomplete information, where there is an absence of sufficient evidence (research, evaluations, data) to allow the determination of a point estimate; the use of assumptions, where sensitivity controls are needed for cost estimates; and the uncertainty of causality, where the research evidence is unable to determine the proportion of expenditure that is dependent on current and past experiences of poverty (e.g. poverty and poor health). Collectively, these challenges mean that the determination of an accurate point estimate would be difficult, and most likely misleading. This scenario is not dissimilar to that faced by economic evaluators confronting the inclusion of intangibles and uncertainties in a cost-benefit analysis and where using a range of values is common (HM Treasury, 2011: 128; DPER, 2012: 29; OECD, 2018b: 425; HM Treasury, 2022: 59). New Zealand’s Guide to Social Cost Benefit Analysis puts it well: ‘given that valuations are uncertain and approximate, it is more honest to provide ranges rather than point estimates’ (NZ Treasury, 2015: 31). Thus, this analysis sets out to determine an indicative main estimate (ME), built around the available evidence, data, and literature. Often this estimate is established using a proxy indicator that is available within the data and literature. This is determined simultaneous to a low-estimate (LE), which reflects a more conservative approach to determining the cost, and a high-estimate (HE) reflecting the likely upper-limit.Footnote 8 Collectively, the estimates give a range for the overall public service cost of poverty. The low and high estimates also allow readers to ascertain the impact of alternative assumptions of the costing approach.
Across the twenty-five areas of expenditure, four different approaches are taken to derive the cost estimates. These are labelled A to D and are noted next to each item in Table 2. Each method is selected based on the available data and literature for expenditure on that public service.Footnote 9 These methods, each with an example, are as follows:
-
(A) A combination of microdata analysis and estimation assumptions (seven estimates).
Example: Seven of the cost estimates are based on the analysis of microdata and estimating assumptions that derive from the results of that analysis or from previous research. For example, data on the composition of social housing tenants by poverty status, sourced from the SILC, is used to estimate the proportion of social housing capital investment associated with poverty. Based on that data from 2008, 2011, 2013, and 2017, on average 38 per cent of local authority (LA) tenants lived below the poverty line. The analysis assumes that new social housing expenditure associated with poverty arises in line with this composition (38 per cent, ME). A lower estimate, of 34.2 per cent, is based on data for a more broadly defined group of social housing recipients and is the average proportion of individuals living below the poverty line who are either LA tenants or ‘renting below market price or rent free’. The latter category encompasses a range of rental structures that are provide, or supported, by the state including charities and social housing bodies (34.2 per cent, LE). The higher estimate derives from an assumption that new expenditure on social housing is likely to be much more orientated towards those on the lowest incomes. In general, possessing an income below certain income limits is the first criterion listed by LAs when assessing social housing need (NESC, 2014; Citizens Information, 2023). In practice, this is likely to result in a large proportion of new social housing stock being allocated to individuals living below the poverty line, with the balance going to those not in poverty but who still possess limited resources to pay for private rented accommodation and/or have other needs necessitating social housing. In the absence of evidence on the poverty profile of new social housing tenants, the higher estimate is calculated as a proxy measure based on the average proportion of LA tenants in the bottom income distribution quintile over the SILC for 2008, 2011, 2013, and 2017 (48.5 per cent, HE).
-
(B) A combination of the analysis of published/official/administrative data and estimation assumptions (seven estimates).
Example: Published data on the income status of children attending disadvantaged schools, those identified as needing additional targeted state funding and known as DEIS (Delivering Equality of Opportunity in Schools) schools, is used to determine the proportion of expenditure on a school meals programme that is associated with poverty. School meals are targeted, rather than universal, in the Irish education system (Department of Social Protection, 2020). In 2019 1,580 schools and organisations received funding and provided food for 250,000 children costing €57.6m (Oireachtas, 2019a; DPER, 2018b: 175).
Weir and Kavanagh (Reference Weir and Kavanagh2018) provide a basis for an estimating assumption of the proportion of school meal recipients who are in poverty; some may live on an income above the poverty line but receive school meals as they attend a disadvantaged school where all pupils get meals. They found that 61.6 per cent of children in post-primary DEIS schools were in families who possessed a medical card. These cards provide access to free primary care with entitlements predominantly arising as a result of a low-income means test; they also arise because of specified medical need. Although using family medical card possession as a means of identifying children experiencing poverty is likely to overestimate the number of such children, as the 2018 SILC data show that 41.3 per cent of children are in families who possess a medical card and 30.3 per cent of children live below the poverty line, it serves as a useful and available proxy. Consequently, the analysis assumes that 61.6 per cent of school meal expenditure is targeted at children in families living below the poverty line while the remainder benefits others (61.6 per cent, ME). The low estimate assumes that the school meal scheme would remain irrespective of the level of poverty as such needs are likely to persist for many children in disadvantaged families and communities (0 per cent, LE). Conversely, the high estimate assumes that all of the costs of this scheme are associated with the existence of poverty and that in its absence the school meals programme would no longer be needed (100 per cent, HE).
-
(C) Based on existing literature and an assumption of its applicability to Ireland (three estimates).
Example: Identifying the proportion of expenditure on mental health care that is associated with poverty serves as an example where the cost estimates draw on results in the existing literature and assume their applicability to Ireland. In the absence of data and published analysis for Ireland, the study draws on a UK study which used data from the 2012 Poverty and Social Exclusion survey to create an individual index of mental health and used this as the dependent variable in a model that isolated the effects on mental health of current poverty status, current experiences of socio-economic disadvantage, and past poverty experiences (Bramley et al., Reference Bramley, Hirsch, Littlewood and Watkins2016: 20–22). Its results found that 14 per cent of mental ill-health is directly associated with current poverty (14 per cent, LE) and a further 13.5 per cent is attributable to current experiences of socio-economic disadvantage (27.5 per cent, ME). The scarring effects of past experiences of poverty were found to explain a further 7.9 per cent (35.4 per cent, HE). These results are used to determine the poverty related expenditure on mental health care in Ireland on the basis of two assumptions. First, the UK study models mental health experiences rather than service usage and in using its results it is assumed that mental ill-health and mental health care usage are strongly correlated. Second, the study assumes the UK results provide a reasonable proxy for experiences in Ireland; while there are differences in mental health care provision and poverty experiences between both countries, and particular differences in benefit systems (Bolton et al., Reference Bolton, Whelan and Dukelow2022) with the welfare conditionality aspects of the UKs shown to exacerbate mental health challenges (Wright & Patrick, Reference Wright and Patrick2019), we assume that these are unlikely to be so large that they would significantly alter the general pattern.
-
(D) Based on an estimation assumption alone: (1) where evidence is lacking (four estimates); or (2) where expenditure in area X can be assumed to be similar to that estimated for area Y (four estimates).
Example D1: In the case of four relatively small welfare payments the absence of detailed data on the household income status of recipients impedes a detailed empirical examination of their poverty status. Rather than exclude this expenditure, the cost of poverty calculations include it based on an estimation assumption alone. For example, the Supplementary Welfare Allowance payment is an emergency means tested payment to people in need of immediate assistance. In 2017 there were 17,993 recipients of this payment averaging €5,421 (DEASP, 2019: 33). The cost of poverty estimates associated with this payment derive from assumptions relating to the expectation of needs under this scheme continuing to arise in a context where income poverty is absent. The low estimate assumes that as this is exceptional expenditure all of it would remain (0 per cent, LE). The main estimate assumes that a sustained period where all have an income at or above the poverty line would see some of these emergency demands dissipate. However, as emergencies are still likely to arise, the payment would still be needed. It assumes that half of the cost of this scheme is associated with current and past experiences of poverty while the other half relates to short-term and unavoidable emergencies that would remain (50 per cent, ME). The high estimate assumes that even more of the scheme is associated with poverty with the proportion set at 80 per cent.
Example D2: Four cost estimates, for areas with data limitations, are based on the assumption that the cost of poverty estimates already established for a closely related area of expenditure can be used as the basis for estimates for these areas. For example, the proportions of police (An Garda Síochána) expenditure on crime and disorder, derived as per approach C, are used to provide the estimates for expenditure on crime and disorder by the courts and prison system.Footnote 10
Results and comparisons
Table 2 presents the results of the cost of poverty analysis. For each area of expenditure (column 1) the estimated proportions associated with poverty (columns 4–6) are each multiplied by the total annual expenditure in that area (column 2) to determine a main, low, and high cost of poverty estimates for that category of state spending. The final row of figures in the table sums these calculated estimates to provide a main, low, and high cost. The main estimate approach finds that the annual public service cost of poverty to Ireland is almost €4.5 billion per annum. While this estimate is intended to provide a reasonable indication of the poverty related costs currently experienced by the state, it sits between a conservative low estimate (€3 billion) and an upper-limit high estimate (€7.2 billion).
Importantly the main, low, and high figures produced by these approaches are estimates. They are not intended to be measures of precision and are somewhat tied to the adequacy of public expenditure commitments to address poverty and disadvantage. Where these are deficient, current expenditure by the state is lower than ideal, giving a mirage of poverty costs that are lower than what should be their true level. However, as the objective of this study is to determine an indicative cost of poverty based on current state expenditure commitments, the general scale of the overall poverty cost estimate is unlikely to be very different to the range determined.
One implication of this analysis is to highlight the hidden nature of these costs, which while often visible individually are infrequently considered cumulatively and compared to other demands and priorities for the state. Therefore, Table 3 considers the estimates more broadly. Expressed in per capita terms the main estimate implies that poverty imposes a public service cost equivalent to a sum of €913 per person each year. Relative to the number of households, the annual cost sits between €1,800 and €4,200 per household with the main estimate being just over €2,600. We can also compare the public service cost of poverty to the projected 2020 income and expenditure of government (pre Covid-19). Table 3 finds that the main estimate is equivalent to 5.1 per cent of total general government revenue and 5 per cent of total general government expenditure. Put another way, €1 in every €20 collected by the state from taxes, social insurance and charges end up being allocated by the state to make up for the way that poverty damages people’s lives. Comparing the main estimate to the allocated 2020 budget for each government department, this hidden annual public service cost of poverty comes in at less than the individual budgets of the three largest spending departments (Welfare, Health, and Education) but larger than that of each other government department including Housing, Justice, and Transport (DPER, 2018a and 2018b; Collins, Reference Collins2020).
Table 3. The public service cost of poverty in context
Notes: Calculated using CSO population projections for April 2019, CSO household count from Census 2016, CSO national income measures for 2020 and budget 2020 projections for (pre Covid-19) general government revenue and expenditure (Department of Finance, 2019: 58). GNI is used for national income comparisons as Ireland’s GDP is artificially inflated by MNC activity.
The costings have some drawbacks worth noting. First, while they include most major areas of expenditure, some smaller areas are excluded. For example, there are other areas of health care expenditure (e.g. elderly care, maternity care), which are likely to include some additional poverty related expenditure. Second, the costs are essentially first round effects, reflecting current annual spending associated with poverty, and do not capture the subsequent knock-on public sector costs of current poverty on future fiscal outcomes. Third, there is no guarantee that the estimated costs will directly transfer into expenditure savings for the state should poverty be reduced or eliminated; in the context of under-funded public services with waiting lists or other forms of excess demand, it is likely that in many areas the expenditure reductions triggered by lower poverty may simply free-up resources to address other needs rather than provide exchequer savings. Fourth, the costs are estimated based on household surveys. As these surveys miss those not living in households, often those who are among the most disadvantaged in society, they understate the true poverty population and skew the cost estimates downwards.
Conclusions for the anti-poverty policy agenda
Recurring policy commitments to tackle poverty underscore how anti-poverty strategies remain an important issue for modern welfare states (Greve, Reference Greve and Greve2019b). Yet despite almost universal support for poverty reduction/eradication, few countries have made great progress in achieving these aims and, as Table 1 shows, in the EU-15 living on an income below the poverty line is the reality for between one in four and one in eight people (Nolan & Marx, Reference Nolan, Marx, Salverda, Nolan and Smeeding2009; Summers, Reference Summers and Greve2019).
Within the research literature much attention has been given to the individual and community implications of experiencing poverty with less focus on the recurring state expenditure that arises as a result of current and past experiences. This analysis highlights the scale of these costs for one country, providing a heretofore absent benchmark for Ireland. Although differences in the nature and scale of the welfare state, and the underlying demographic context, make the precise transferability of costings between countries challenging, the findings highlight how a large amount of money raised and spent by the state each year is associated with the ways that poverty damages people’s lives. This public service cost of poverty is unlikely to be dramatically different in other states with similar levels of poverty.
The analysis carries a number of implications for those arguing for anti-poverty policies and for their broader design and public policy justification. First, the lack of focus on the scale of these recurring public expenditure costs, and the limited appreciation of their implications for fiscal policy choices, is a gap in current policy making. Responding to poverty is too frequently classified as a welfare policy issue while its broader public service costs are overlooked; in fact, as this study shows, these costs fall much harder on the health, education, housing, and justice budgets. Given the need to pay for these public service needs, their presence limits public expenditure options and these costs are experienced by all members of society, whether they are above or below the poverty line. Second, there is a greater need to frame anti-poverty strategies intended to reduce, prevent, or disrupt poverty as being able to create a virtuous circle for society, where the lives of the least well-off are improved while freeing up government resources to focus on other areas of unmet need or on other policy priorities. Some of these expenditure reductions have the potential to arise in the short-term while others, in particular those associated with the legacy impacts of poverty and disadvantage, will take time to appear. In some areas the state will need to ‘invest to save’, pursuing what Greve terms ‘social investment in people’ (Reference Greve and Greve2019a: 400), by increasing commitments and spending so that substantial medium to longer-term expenditure reductions can be realised. Such an approach would be analogous to recent initiatives around investments in early childhood education and care (Sylva et al., Reference Sylva, Melhuish, Sammons, Siraj and Taggart2010; OECD, 2011; Bouget et al., Reference Bouget, Frazer, Marlier, Sabato and Vanhercke2015). Both of these implications underscore the need for a broader narrative around the adoption and implementation of anti-poverty initiatives, one which breaks away from an exclusive focus on helping the poor to one that articulates that objective alongside the advantages associated with saving public resources or using them in a different way to pursue other socio-economic policy objectives. Indeed, if states judge anti-poverty policies in a similar way to infrastructural investment decisions there is the potential for greater policy buy-in as it would become clearer that ‘it simply makes more sense to pay to address poverty directly than to wait for its longer-term consequences to arise and to pay to deal with them then’ (Ivanova, Reference Ivanova2011: 39).
Third, the large and multidimensional nature of the public service cost of poverty is relevant to assessments of policy initiatives to restrict or reduce welfare supports, policies that invariably reduce the income of those already disadvantaged and increase poverty (Collins & Murphy, Reference Collins, Murphy, Murphy and Dukelow2016; Millar & Bennett, Reference Millar and Bennett2017; McGann et al. Reference McGann, Nguyen and Considine2020; Graven, Reference Graven2021). While such measures may generate short-term expenditure reductions, they are likely to trigger additional expenditure in the medium-term dealing with poverty’s consequences and legacy effects. Judged over time, the latter may significantly overshadow the former.
Finally, the methodological approach taken in this article, and the cost of poverty literature in general, has implications for how researchers might consider other hidden in plain sight social costs, such as those associated with, for example, inequality. The determination of a range of recurring public expenditure costs imposed on all of society as a result of these social experiences has the potential to generate the evidence needed to further nurture a political willingness to respond. For poverty, this analysis shows that there are benefits for all in society associated with successful strategies to address and successfully reduce it and its consequences.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S147474642300043X.
Acknowledgements
The author wishes to thank the Society of St Vincent de Paul that funded this research, and participants at the 2021 SPA annual conference for their valuable comments and suggestions.
Open access
