Social tax expenditures

We adapt the term ‘STEs’ used by Toder (Reference Toder1999) to refer to tax expenditures that qualify as what Titmuss (Reference Titmuss1958) called fiscal welfare. In his essay ‘The Social Division of Welfare’, Titmuss’ major insight was that the ‘definition, for most purposes, of what is a “social service” should take its stand on its aims; not on the administrative methods and institutional devices employed to achieve them’ (1958: 42). According to Titmuss (Reference Titmuss1958), welfare programmes share the common aim of enhancing the wellbeing of their recipients. He conceived of a social division of welfare (SDW) comprising three welfare systems delivered through different policy instruments: social welfare distributed by the social expenditure of the state, occupational welfare delivered to workers as benefits obtained through employment and fiscal welfare received through tax expenditures and other elements of the tax code (Titmuss, Reference Titmuss1958: 42). We classify tax expenditures as fiscal welfare, and thus STEs, if they allocate resources in the key areas of welfare provision, including: social security, health, education, housing and family and community assistance.

Titmuss (Reference Titmuss1958) developed his SDW thesis in response to right-wing commentators who framed the post-war welfare state as ‘expensive for the middle classes, unnecessary for many of the working classes and potentially debilitating for all social classes’ (Mann, Reference Mann2008: 2). He claimed that these assertions relied on a narrow and misleading conception of the welfare state as only social welfare, which was mainly received by lower-income earners. He was concerned that this partial understanding of welfare reinforced negative stereotypes of the ‘dependent’ recipients of social welfare, while ignoring similar benefits higher-income earners receive through ‘other means’. Addressing these concerns, the SDW thesis drew attention to how individuals receive welfare through their social relationships with the state, the tax system and employment – showing how all members of contemporary capitalist democracies are interdependent (Titmuss, Reference Titmuss1958: 53; Mann, Reference Mann2008: 3). We consider our dual welfare state concept a vindication of Titmuss’ classic thesis.

Accounting for tax expenditures

Broadly understood, tax expenditures provide selective tax breaks to taxpayers who undertake certain actions or belong to particular social groups. Distinct from general tax cuts, tax expenditures provide individuals with incentives to change their behaviour or compensate for social risks. The major types of tax expenditures include: tax exemptions, tax deductions, tax rebates or offsets, tax concessions, allowances, rate relief, provisions for tax deferral and income tax averaging (Smith, Reference Smith2003: 3). The selectivity of tax expenditures means that they are deviations from the tax benchmark, the normative structure of the taxation system that treats similar taxpayers equally (Ingles et al., Reference Ingles, Jackson, Podger and Raymond1982: 6).Footnote ⁹

The resource allocation of tax expenditures is conceptually equivalent to that carried out by social expenditures because of their similar effects on state revenue and resource allocation. Surrey and McDaniel (Reference Surrey and McDaniel1985: 25) conceive of tax expenditures as instantaneous cash transfers involving taxpayers, who pay their full tax liability according to the tax benchmark, and the state which reimburses taxpayers through lower taxes. They contend that the revenue reimbursed by the state through tax expenditures is conceptually equivalent to the revenue expended on social expenditures (Surrey and McDaniel, Reference Surrey and McDaniel1985: 25). Moreover, tax expenditures and social expenditures are both financed by reallocating resources from other programmes, running deficits or increasing taxation (Hacker, Reference Hacker2002: 34), and both increase the resources available to their recipients.

But despite wide agreement over definition, precise estimates of tax expenditures in Australia are difficult to calculate for three main reasons: broader disputes over the composition of the tax benchmark (Smith, Reference Smith2003: 3), the inability of conventional accounting systems to capture them (Auditor-General, 2008: 13) and the range of rebates and concessions not included in the Treasury's Tax Expenditure Statement (TES), particularly hybrid policies that combine elements of tax expenditures and social expenditures. While these measurement issues highlight the importance of identifying data sources (Bradford, Reference Bradford and Bruce1988: 431), they are not dissimilar from those associated with other policy instruments, meaning the tax expenditure concept remains a useful tool to shed light on an effectively invisible form of redistribution. We draw on statistical data from the Treasury, the Economic Planning Advisory Council (EPAC) and the Australian Tax Office (ATO), which measure tax expenditures using the revenue-forgone approach.

The most pressing quandary in using the tax expenditure concept, however, stems from the existence of hybrid policies, such as the private health insurance and childcare rebates. These policies allow recipients to either claim assistance through the tax system or as a subsidy when purchasing services. The structure of assistance also varies. Rebates often refund a proportion of the cost of the service, rather than a proportion of tax paid, and usually have a cap. As there is no definitive method to classify hybrid policies, the distinction between tax expenditures and income transfers remains partially arbitrary. Our approach is to conceive of hybrid policies as fitting along a continuum with policies delivered as tax expenditures at one extreme and income transfers at the other (see Figure 1). We classify public policies as tax expenditures if they meet the criteria of the tax expenditure concept outlined above. We also treat hybrid policies as tax expenditures for analytical purposes if they are framed as such in public discourse, as they are subjected to similar political processes as tax expenditures. Hybrid policies provide further rationale for incorporating STEs within the welfare state by demonstrating their substitutability with social expenditures.

Figure 1.

The continuum of income transfers and tax expenditures

The tax expenditure–social expenditure continuum highlights the importance of examining the detail of policy. Although a detailed account of this diversity is not possible here, caution should be taken when applying this broad account to particular policy domains. We acknowledge that reforms that move a policy along the continuum have important implications for equity and transparency, and can change income distribution (see Spies-Butcher and Stebbing, Reference Spies-Butcher and Stebbing2009). Nonetheless, this conception aids understanding of how the frame of tax expenditures impacts on the structure of the welfare state beyond ‘pure’ tax expenditures.

The rise of tax expenditures

Tax expenditures, along with social welfare programmes, are the ‘two oldest forms of financial assistance provided by the Commonwealth’ (Auditor-General, 2008: 27). In Australia, STEs have a long history in policy domains such as health and retirement incomes. The Fisher Labor government enacted superannuation tax concessions in 1915 (Nielsen and Harris, Reference Nielson and Harris2008: 2). The Menzies Coalition government established tax deductions for private health insurance premiums in 1953 (Kewley, Reference Kewley1973: 361). However, despite the longer history of tax expenditures, we focus on trends from the 1970s because of data limitations. The same measurement issues are even more pronounced at state level, where reliable data are unavailable. Hence, the figures here only refer to federal expenditure.

With the aforementioned measurement issues in mind, analysis undertaken by the Economic Planning and Advisory Council (1986) suggests that tax expenditures declined compared to direct expenditure and as a proportion of Gross Domestic Product (GDP) during the 1970s. As a proportion of public spending, tax expenditures fell from 13.1 per cent of public spending in 1970–71 to 6.4 per cent in 1982–83 (EPAC: 1986: 9). Similarly, tax expenditures dropped from 3.1 per cent of GDP in 1970–71 to 1.9 per cent of GDP in 1982–83 (EPAC, 1986: 9). Because EPAC only collected tax expenditure estimates listed in the Budget, its figures are significantly lower than those appearing in the Treasury's Tax Expenditure Statements (TES) that are compiled annually from the Budget and in-house estimates since 1986.

Subsequently, the TES data indicate that tax expenditures have increased substantially in number and scale since the mid-1980s. The number of tax expenditures reported in the TES has increased by around 75 per cent: from 170 in the 1984–85 financial year to over 320 in 2007–08 (Treasury, 1986, 2009). The budgetary scale of tax expenditures has grown in both monetary and proportional terms. The reported cost of tax expenditures listed in the TES swelled from AUS $7 billion in 1984–85 to more than AUS $73 billion in 2007–08, or from 3.6 per cent to 7.1 per cent as a share of GDP; tax expenditures also rose, as a proportion of public spending, from 11 per cent to 21 per cent (see Figure 2). Although the data are not fully comparable, the apparent decline of tax expenditures in the 1970s and subsequent rise from the 1980s onwards suggest that this represents a novel development.

Figure 2.

Tax expenditures as a proportion of GDP and public spending 1984–2008

Source: Australian Yearbook, ABS (various years) National Accounts and Treasury (various years) Tax Expenditure Statement.

The rise of social tax expenditures

Reflective of the overall trend, the STE subset of total tax expenditures appears to have grown in number and budgetary scale since the mid-1980s. Table 1 displays STEs that have been introduced by Labor and Coalition governments for: low-income earners, families with children, childcare, retirement incomes, health, housing and individuals with disabilities. The number of STEs reported in the TES has risen by more than 85 per cent from 36 in 1985–86 to 67 in 2007–08 (Treasury, 1986, 2009).Footnote ¹⁰ The cost of STEs surged over eight-fold from around $5 billion in 1985–86 to $43 billion in 2007–08, amounting to an increase in GDP from 2.4 per cent to 4.2 per cent.Footnote ¹¹ As a proportion of total public spending, STEs rose from around 7 per cent in 1985–86 to 12 per cent in 2007–08 (Treasury, 1986, 2009). Growth was particularly rapid under the Coalition government during the mining boom years at the beginning of the century. Much of this growth is attributable to the rising costs of superannuation tax concessions from 1.3 per cent of GDP in 1985–86 to 1.9 per cent of GDP in 2007–08 (Treasury, 1986, 2009). Currently the largest category of STEs, superannuation tax concessions amount to $26 billion of revenue forgone (Treasury, 2009: 10). Both the growing number and budgetary impact of STEs highlight their increasing relevance to the Australian welfare state.

TABLE 1.

Direct expenditures and tax expenditures in key areas of Australian welfare

However, as noted before, the TES underestimates the budgetary impact of tax expenditures. In particular, three large housing STEs appear in an appendix of the TES, but not in its total estimates. The Senate estimated the capital gains tax exemption for owner-occupier housing to account for $20 billion, the tax exemption of imputed rent at $15 billion and the land tax exemption for owner-occupier housing at $10 billion in 2007–08 (Senate Select Committee on Housing Affordability in Australia, 2008: 40). If these tax expenditures were accounted for in the TES, STEs would account for $88 billion, which is equivalent to around 21 per cent of direct expenditure or 5.8 per cent of GDP. But this is still likely to understate the value of STEs, as other programmes not costed in the TES have been estimated to exceed several billion dollars (Smith, Reference Smith2003: 9).

The progeny of austerity?

The rapid expansion of STEs is partly a consequence of fiscal austerity, which builds on Pierson's (Reference Pierson2001) concept of permanent austerity without implying inherent stability. By permanent austerity, Pierson (Reference Pierson2001: 411) refers to the political context of capitalist democracies from around the late 1970s in which real or perceived fiscal constraints on the state restricted the growth of social expenditure. Fiscal austerity has increased the appeal of STEs as they allow legislators to extend social provisions while minimising political resistance by reducing the visibility of spending.

Fiscal austerity has presented barriers to expanding or retrenching the welfare state. Economic factors – such as slowing economic growth, the ageing of the population and economic globalisation – limited revenues. The hegemony of neo-liberalism constituted a further ideological constraint (see Harvey, Reference Harvey2005). At the same time, the electoral popularity of existing social programmes among their recipients limited retrenchment (Pierson, Reference Pierson2001: 412).

In Australia, the constraints of fiscal austerity were more pronounced during the 1980s and 1990s when public concern focused on the expanding current account deficit (see Kelly, Reference Kelly1992), but were less evident in the mining boom of the early 2000s. This did not prevent the growth of social expenditure, which increased from 11.3 per cent of GDP in 1980 to 17.8 per cent in 1998 (Castles, Reference Castles2004: 25); rather, it limited this growth to social programmes that could be linked to broader economic reforms (alongside social justice considerations). At the same time, the electoral popularity of welfare programmes prevented widespread welfare state retrenchment (see Mendes, Reference Mendes2009). This reflected the growth in the proportion of the population that relied on welfare for their primary source of income, from less than 5 per cent of the working-age population in the 1960s to more than 16 per cent in 2004 (Saunders, Reference Saunders2007).

Although limiting the growth of social expenditure, fiscal austerity provides a reason for the popularity of STEs among legislators since the mid-1980s, as they have provided a policy back door to incrementally increase social provisions. As Toder argues:

STEs also lower political resistance by reducing public accountability. Tax expenditures circumvent many of the oversight arrangements that social expenditure programmes are routinely subjected to in the legislative process, as displayed in Table 2. Unlike social expenditure programmes, tax expenditures are not identified by all Commonwealth agencies, not estimated according to independent standards or subjected to annual audits. Tax expenditures are also infrequently accounted for in the Budget, with many escaping regular review. Moreover, the TES, the principal data source, does not amount to a comprehensive review and requires no vote when tabled in parliament (Wanna, Reference Wanna2003: 3).

TABLE 2.

Oversight arrangements for direct expenditures and tax expenditures

Source: Auditor General (2008: 15).

Tax expenditures also receive little attention in Australian public discourse. The TES has partially enhanced the visibility of tax expenditures, but it receives little media coverage – particularly compared to the Budget. The Treasury maintains that the TES is released near the beginning of the calendar year to inform the Budget (Senate Select Committee on Housing Affordability in Australia, 2008: 32). But its release during this period contributes to its low visibility: in each of the past six years, it was released within a week of Christmas or Australia Day. The lower visibility of tax expenditures in the media means that they often do not receive the public response accompanying new social expenditures, reinforcing STEs as a policy ‘back door’.

Further, STEs have low visibility by design as they do not involve direct transactions and appear to give taxpayers ‘their own money’ back and reduce the size of government. On the one hand, recipients may not understand tax expenditures as state transfers, as they reduce their tax burdens rather than provide cash or a service. On the other hand, tax expenditures allocate resources by reducing revenue rather than increasing outlays, which extend state regulation but reduce the state's Budget. The combined effect of their low visibility in the legislative process, media and their design has reduced the political costs of STEs to legislators.

Neo-liberalism, privatisation and institutional layering

A second explanation for the rapid growth of STEs is their compatibility with neo-liberalism. The widespread influence of neo-liberalism not only contributed to fiscal austerity but also increased the appeal of reforms that privatised public services (Kelly, Reference Kelly1992: 15). STEs appealed to the reformist spirit of neo-liberalism because they provided legislators with instruments to privatise welfare services while leaving popular social welfare programmes intact. We contend that the legislators of the Hawke-Keating and Howard governments used STEs to build additional institutional layers of private social provision on top of existing social welfare programmes.

Many STEs extend consumer choice and shift resources to private service providers. Examples include the Private Health Insurance Tax Rebate (PHITR) and the concessional taxation of superannuation. By requiring consumers to choose between private welfare services to access the benefit, these STEs are framed as encouraging rational economic behaviour while also directly benefiting private enterprises. And, although STEs constitute market intervention, they minimise state bureaucracy.

STEs shift resources to the private sector without dismantling popular public services through a process that Thelen (Reference Thelen, Mahoney and Rueschemeyer2002) calls institutional layering. Institutional layering entails the building of new institutions or policies on top of existing policies and institutions (Thelen, Reference Thelen, Mahoney and Rueschemeyer2002: 225). It is an incremental approach to reform often used by legislators when radical reforms to existing programmes are not politically feasible. Thelen (Reference Thelen, Mahoney and Rueschemeyer2002: 227) contends that layering new policies or institutions on pre-existing institutional arrangements alters the interests of political actors and the trajectory of institutional development. Although Thelen applies this concept to explain the development of individual policies, the same logic is at work here on a larger scale.

The Howard government's introduction of the PHITR in 1999 provides an example of institutional layering. Despite opposing Medicare for over ten years in opposition, the Coalition promised to maintain Medicare in 1996 in recognition of its electoral support (Elliot, Reference Elliot2005). Rather than reforming Medicare, the government enacted the PHITR, a tax rebate for private health insurance premiums, as part of a package of reforms to boost private health insurance coverage. In 2006–07, the PHITR shifted $3.4 billion to the private health sector (AIHW, 2008a: 34) and (alongside other reforms) has increased private health insurance coverage. This example shows how STEs can subsidise private services and facilitate private provision by building an additional layer on existing institutions without alienating constituencies that support public services.

However, most STEs enacted by Australian governments typically place few restrictions on resource allocation. For example, the superannuation tax concessions form part of the second layer that superannuation adds to Australia's retirement income system. These STEs provided $26 billion of subsidies in 2007–08 to the superannuation industry, but did not require individuals to invest in funds that met certain policy objectives or provided specified benefits. The lack of restrictions placed on how the subsidies allocated through STEs are spent effectively reduces both the transparency and accountability of public expenditure.

The new politics of class coalition: appealing to the aspirationals

A further influence on the rapid growth of STEs has been influenced by the ability of political actors to develop alliances in support of expansion. Social policy scholarship has focused mostly on the influence of the working class, the group perceived to benefit most from welfare programmes. However, the middle class, broadly conceived of as non-poor social groups, has also played important roles in social policy development as welfare recipients and self-interested agents (Goodin and Le Grand, Reference Goodin, Le Grand, Goodin and Grand1987: 203). Baldwin's (Reference Baldwin1990) comparative analysis of welfare state development shows that the establishment of cross-class coalitions helps explain the structure of social policy. In countries where the working class developed coalitions with farmers or small business owners, this allowed progressive dominance of the state. These coalitions facilitated solidaristic welfare policies, but also often required alliances to be built, with farmers and business people demanding access to benefits offered to workers, and these demands being met through universal provision.

In Australia, STEs represent an alternative strategy for building political support that decouples the interests of middle- and low-income earners by linking middle-class interests with higher-income earners, who gain most from STEs. Higher-income earners pay higher income tax rates and tend to spend more on private welfare. As STEs often return money to taxpayers in proportion to their marginal tax rate and subsidise private welfare spending, higher-income earners tend to benefit most from these policies. Unsurprisingly, STEs are predominately a strategy of the conservative parties, but the changing political context discussed earlier may help explain the more modest, but still important, support for many of these policies by Labor governments.

The largest STEs in Australia involve housing and retirement costs. Here the benefits are largely uncapped and require private spending to gain access to benefits. For example, superannuation contributions and investment earnings are taxed at the concessional rate of 15 per cent. Most employers are obliged to contribute 9 per cent of wages towards these funds on a worker's behalf. For those earning over $180,000 per annum this concession is equivalent to a 30 per cent rebate for investments and earnings, as the top marginal tax rate for these income earners is 45 cents in the dollar. But for low-income earners on under $30,000 per annum, the scheme provides no benefit. Richard Denniss (Reference Denniss2007) estimates that over their life a person consistently earning over $72,000 per annum who relies on compulsory employer contributions would accrue more from STEs than they would have received from a full public pension.Footnote ¹² This should cause us to reassess the size of Australia's welfare state. Superannuation tax benefits are comparable to pensions in their budgetary impact, but are not included in OECD comparisons that show Australia to be a low spender on the aged. Including superannuation also changes our assessment of the distributional effects of Australia's welfare spending.

Even for STEs with hybrid schemes that provide benefits proportional to private spending – rather than tax rates – similar distributional effects emerge. For instance, Denniss (Reference Denniss2005) shows that the PHITR is highly regressive: while 69 per cent of households earning over $100,000 per annum held private insurance and received the rebate, only 24 per cent of households earning less than $25,000 per annum held such insurance. STEs for childcare, health, pensions and housing are structured to provide rebates or concessions to those purchasing private welfare services. As a result, the dual effects – of reversing the tax scale and restricting benefits to those who purchase private services – mean that STEs have radically different distributional effects than the core of Australia's targeted welfare state. A more complete picture of Australia's welfare state must therefore take account of social expenditure and STEs.

The greatest benefits of these STEs flow to higher-income earners, but moderate-income earners benefit too: over 40 per cent of Australians hold private health insurance (PHIAC, 2008), roughly 70 per cent benefit from the concessional taxation of owner-occupied housing (ABS, 2006) and over 54 per cent of wage earners receive concessional treatment on their private retirement savings (ATO, 2009). STEs can thus potentially unite the interests of higher- and middle-income earners.

By uniting these interests, STEs have a potential to gain political support and limit redistribution. Korpi and Palme (Reference Korpi and Palme1998) identified a paradox between targeting and redistribution: targeted welfare states redistribute more proportionately but tend to be smaller and redistribute less overall. Korpi and Palme (Reference Korpi and Palme1998) contend that the apparent restriction on the growth of targeted welfare states may reflect the lack of middle-class support for targeted schemes, as they receive few benefits. However, rather than simply restricting welfare state growth, an analysis of STEs allows us to conceptualise welfare state development as a battle between competing welfare models that offer alternative strategies for incorporating middle-class voters.

Iversen and Soskice (Reference Iversen and Soskice2006) contend that majoritarian systems consolidate around two large parties that usually represent worker and business interests, which must then secure middle-class support to gain an electoral majority. On this reading, Australia's generously targeted system stems from attempts by the Labor Party to incorporate middle-class voters into a system of social support that delivers most to low-income earners but still benefits many middle-income earners.Footnote ¹³ Alternatively, STEs largely represent a Coalition alternative to tax cuts that deliver social protection in a manner that maximises support for high-income groups while extending considerable benefits to some middle-income earners. Middle-class voters are unlikely to accept the absence of social support, especially considering growing new social risks, so business parties are required to offer some form of social policy. Also, the low transparency of STEs helps subvert the legitimacy constraint that prevents redistribution from poor to rich (Iversen and Soskice, Reference Iversen and Soskice2006). As a result, STEs are likely to be advocated by business parties as a mechanism to build class alliances while advancing the interests of high-income earners and, we would expect, are more likely to appear in majoritarian electoral systems and liberal welfare regimes.

This largely reflects the Australian experience, as most STEs were initiated and/or extended during Coalition governments, with two significant exceptions: the introduction of compulsory superannuation and the child care rebate. Labor with the unions extended superannuation to most workers; however, they used private schemes that largely retained existing tax concessions, significantly increasing STEs. The choice of a private scheme over a state or union-run scheme is probably explained by the political constraints of fiscal austerity and neo-liberalism discussed earlier. In the second, Labor's position partly reflects the preferences of professional women who are a growing Labor constituency.

There is evidence that support for STE-type policies among middle-income earners helps explain their growth. The wage-earners’ model provided middle-income earners with stable and well-paid jobs. However, middle-income earners have faced increasing uncertainty over recent decades because of the winding back of the wage-earners’ model, the collapse of full employment, and the emergence of new social risks. Many of the new social risks connected with issues such as maternity leave, housing affordability and income stability (see Taylor-Gooby, Reference Taylor-Gooby and Taylor-Gooby2004) are greater issues for middle-income Australians than in previous decades. At the same time, young middle-income families living in outer-suburban electorates play an increasingly significant role in Australian elections. Australian commentators have dubbed this group of voters ‘aspirationals’ (Manning, Reference Manning2005), arguing they support policies that encourage private provision and the neo-liberal message of social mobility. There is some empirical support for these claims, with social attitudes research suggesting this group is more likely to support social policies linked to STEs (Wilson et al., Reference Wilson, Meagher and Hermes2008).

STEs, then, provide a political framework for uniting the interests of middle- and higher-income earners. Their low visibility reduces political resistance to these policies, making them less subject to the legitimacy constraints that typically prevent regressive social spending. Their low visibility may also reduce awareness by middle-income earners of the larger assistance directed to higher-income groups. In a political context where middle-income voters are key to electoral success, face growing social risks and receive relatively little from a targeted welfare state, STEs are likely to enjoy greater political support.