After the tax revolt: why Medicare matters more to middle Australia than lower taxes (1).

Conventional wisdom in western politics has it that the public overwhelmingly supports cutting taxes over increasing public spending. This belief has a conditioning impact on politics and public policy in Australia and elsewhere: tax cuts are the crowning achievement of good government while higher spending is waste--a sign of governments unable to withstand the fiscal claims of special interests. Yet, if that ever was the public view, evidence now points in a different direction. In the February 2003 issue of the Australian Journal of Social Issues, we reported Australian Election Study findings between 1987 and 2001 that indicated Australians have moved away from preferring governments to cut taxes in favour of higher spending on social services. This finding coincides with a lively and ongoing debate in Australian politics over the fiscal priorities of the Federal government. Here, we present new evidence about this preference shift away from tax cuts to social spending available in the Australian Survey of Social Attitudes (AuSSA) conducted among 4,270 Australians in late 2003.

Using data from this survey, we provide a profile of the new electorate for higher social spending, and use multivariate analysis to examine the factors driving this trend. We also present evidence about the public's willingness to actually pay higher taxes for more spending on services, a willingness many politicians and policymakers assume is relatively weak (Saunders 2003; see also Weissberg 2001: 5). In concluding the paper, we draw out the implications of this trend in favour of higher spending, which we argue provides insights into the public mood, revealing in particular strong consensus for maintaining the standard of major public services.

Taxing and spending preferences and the 2003 debate

Since the early 1980s, the Australian National University has collected public opinion data about cutting taxes and spending money in the National Social Science Surveys (NSSS) and, more recently, in the Australian Election Studies (AES). Both surveys have regularly included the following question:

   If the government had a choice between reducing taxes or spending
   more on social services, which do you think it should do?

Although the response categories for this question in the two surveys differ slightly (the NSSS offers six response categories while the AES offers five (2)), we are able to put together a trend-line for the public's most basic orientations about social spending and cutting taxes since the earl), 1980s. These responses are included in Table 1, which reports results collected by the NSSS in 1984, 1987 and 1990 and the AES between 1987 and 2001.

During the 1980s and 1990s, Australians overwhelmingly favoured tax cuts. Quite understandably, Australia's two major political forces concentrated their fiscal and political energies on meeting the tax relief demands of the Australian public. During the 1980s, Labor met its traditional redistributive goals within the constraints of both anti-tax public opinion and the prevailing policy consensus in favour of reducing the size of government (Castles 1993: 6-34). Support for tax cuts remained consistently a majority view until the 1998 election, two years after the election of the Howard Government in 1996, which had cut real expenditures during its first term in office (Department of Treasury 2003: Appendix C). According to the AES series, support for higher social spending started to increase during the early Howard years, reaching about 30 percent in 2001. At the same time, support for tax cuts fell away sharply, from 65 percent at the 1987 Federal election to 42 percent in 2001. Previously, we argued that this shift was a predictable response to a change of government, and one that had committed to real spending cuts (Wilson and Breusch 2003: 49).

Our earlier findings contributed one academic voice to a debate in 2003 about the appropriate level of taxation and expenditures, a debate that started to challenge the political and policy wisdom of further tax cuts. The debate was given further impetus by troubled public services. The Howard Government was forced to play 'catch up' politics to address problems with major services, most critically the declining availability of Medicare bulk billing, which Australians overwhelming endorse (Department of Health and Ageing 2003: Figure 2.3). Our findings were in line with British research that identifies a powerful current of public opinion in favour of higher spending on social services, which first emerged as a reaction to the austerity of the Thatcher years, and then persisted under Blair (Sefton 2003; see also Wilson 2004). We raised the prospect of Australia following a similar trend after a long period of fiscal restraint, albeit less severe than the British experience of the 1980s.

Our findings received a mixed reception. The Federal Labor Party claimed that the new evidence confirmed Labor's emphasis on new spending in public health and education, adding they provided no evidence that the public was ready to accept major tax increases (McMullan 2003). The Howard Government's response was more critical. In an interview, the Treasurer, Peter Costello, responded to this debate by claiming: 'I have never, ever come across anybody in 13 years of public life who believes they should pay more tax.' (Sydney Morning Herald: 31 May 2003). But the strongest criticism came from the Centre for Independent Studies (CIS), which doubted whether our results provided real evidence about voter willingness to pay the taxes required to fund this spending. The Centre then commissioned their own survey during 2003, and CIS Director Peter Saunders argued that the findings showed that few voters would actually pay higher taxes for welfare services. The CIS survey (conducted by AC Nielsen) asked online respondents the following question:

   Some people want the government to increase income tax rates and to
   spend the extra money on welfare and social services. Other people
   say we are over-taxed and that the time has come to reduce taxes
   even if this means reducing total welfare spending. Which do you
   think the government should do? [Raise taxes to increase spending
   on welfare/social services; Reduce taxes and reduce spending on
   welfare/social services; Leave taxes and social welfare spending
   as they are; Don't know/No opinion]

CIS research found that only 12 percent supported raising taxes to increase spending on welfare, 41 percent wanted to leave taxes and social welfare spending as they are, and 39 percent preferred to reduce taxes and welfare spending (Saunders 2003: 33-34). But their study did not ask respondents about spending on universal services like Medicare and public education where all available evidence indicates solid support. Additionally, the CIS question uses the phrase 'welfare and social services', which we would expect to capture mainly negative public feelings attached to the word, 'welfare'. So we would expect a fairly negative response to the CIS spending proposition, something which our new evidence also confirms, and to which we return later. Even so, the CIS results found that support for reducing taxes and spending was less popular than the middle option--leaving welfare and taxes at their current level. Unintentionally, it seems, the CIS research undermined conventional wisdom too, by providing yet more evidence of the limited appeal of cutting taxes when the consequences for social services are made explicit.

Support for higher social spending in AuSSA 2003

Our present findings contribute additional information about public opinion in this ongoing debate by providing new evidence about the public response to concrete tax and spend alternatives and by investigating public willingness to pay higher taxes for spending on services. We shall deal, first of all, with the public's preference between tax cuts and social spending, and then discuss evidence about voter willingness to pay taxes for improved services. During August-December 2003, the tax and spend preference question was put to 4,270 Australian voters aged 18 years and over who responded to the Australian Survey of Social Attitudes conducted by the Centre for Social Research at the ANU. (3) The stem of the question differs slightly from the one asked in the AES and NSSS, offering respondents a choice between two concrete alternatives for tax cuts and social spending. The question is:

   If the government had a choice between reducing personal income
   taxes or increasing social spending on services like health and
   education, which do you think it should do? [Strongly/mildly
   favour reducing income taxes; Depends; Strongly/mildly favour
   increasing social spending].

So in essence the question asks voters to choose between two tangible alternatives: personal income tax cuts or increases in social spending in areas like health and education.

We find overall support for increasing social spending is a substantial 48 percent--some 18 percent higher than support for spending recorded in AES 2001 (see Table 1). At the same time, support for tax cuts is some 14 percent lower at 28 percent. Table 2 compares support for social spending and tax cuts in the AES 2001 and ASSA 2003 for the five response categories offered to survey respondents, which are consistent in both surveys. The largest shift in preferences is for 'Strongly favour' increased social spending, up 17 percent between the two surveys. This increase comes at the expense of the middle category 'Depends' and the two tax cut categories. The decline in respondents choosing 'Depends' suggests that voters are better able to distinguish between two specific alternatives and to choose one of them. The decline in support for tax cuts--even though the new question specifies personal income tax cuts, which is a popular option--indicates that respondents have switched to social spending either because the new question hypothecates spending or because the secular trend away from tax cuts to spending is continuing.

An analysis of the support for social spending found in AuSSA 2003 enables us to examine the electoral building blocks for this rising support. First, when we examine support among four age groups--18-34, 35-49, 50-64 and 65 and over--(see Table 3), we find support for social spending increases for older Australians and then levels off. Among voters over 50, there are 51 percent in favour of increased spending--around twice the proportion favouring reduced taxes. Older voters are now a solidly 'pro-spending' bloc, and this in spite of their generally conservative voting behaviour. Young respondents are least likely to support social spending (42 percent) and most likely to support cutting taxes (32 percent). This finding is similar to that reported by Tom Sefton (2003: 9) using British Social Attitudes data, who also finds that 18-34 year olds have recorded the smallest increases in favour of increased taxes for more spending on health, education and social benefits between 1987 and 2002. (4)

The fact that young people in both Australia and the United Kingdom are the age group with weakest support for tax and spend policies may indicate that age effects are not merely the product of national politics. There may be structural, rather than ideological, reasons why younger voters are weaker supporters of spending. Lower unemployment rates--the product of expanded tertiary education participation and an improved labour market--have decreased young people's contact with the social welfare state. In Australia, young people receive proportionately less in transfer payments than older cohorts (Whiteford and Angenent 2001: 58-59) and they are understandably less likely to draw on medical services that otherwise prompt support for higher health spending. As less frequent clients of the welfare state, we might expect younger voters to be less self-interested in preferring social spending, preferring instead to reduce taxes. Whatever the reasons for this lower support among young people, these findings must be kept in perspective: the 18-34 cohort still prefers increased spending over tax cuts by a 10 percent margin, and we will see below that full-time students, who are typically younger, are a major constituency, for social spending.

When we consider support for social spending by employment status and household arrangement, we again find predictable patterns. Full-time workers, who pay higher taxes, are slightly less supportive of social spending than the median voter, as are respondents from households with children under the age of 18. Households with children are predominantly families facing high living costs and high full-time labour force participation, both factors that increase support for tax cuts. Still, both groups favour social spending by a margin of 14 percent and, presumably, middle income families are a potential constituency for greater spending in public health and education and other services like publicly subsidised childcare. By contrast, full-time students and retirees--both who directly depend on government services and transfers and who pay less tax--are decisively in favour of higher social spending by margins exceeding 30 percent. Although full-time workers and families with children have different tax and spend preferences than students and retirees, the differences are not large enough for these preferences to be opposed.

When we turn to support for social spending by political identification and self-identified class placement, we find tentative signs of convergence between the AES 2001 and AuSSA 2003 (see Table 5). Self-identified middle class respondents are now more likely to support higher social spending (52 percent) than working class respondents (44 percent). In the AES 2001, class support was roughly the same. Similarly, support for spending among Liberal-National party identifiers (42 percent) rose faster between 2001 and 2003 than it did among Labor identifiers (53 percent), reducing the margin of difference between the two voting blocs. The relative increase in support for social spending we find among middle class respondents and Liberal-National identifiers between 2001 and 2003 may reflect a number of factors. Increased middle class support is in part the result of rising tertiary education among this group, a factor we have previously identified with increased support for spending (Wilson and Breusch, 2003: 4647). Support for spending among the middle class and conservative electorates may have increased because the spending option in AuSSA is hypothecated, including only two 'safe' and popular spending options and not left as unspecified spending--perhaps unattractive to conservative voters. And middle class support might be rising because they benefited from substantial tax cuts in mid-2000. With demands for lower taxes now satisfied, middle Australia is more attentive to the costs of limiting spending on public services; namely, a growing expectation by government that households will provide for their health and education needs through the market. Finally, Liberal-National identifiers are presently an older electorate, so some of the convergence in the conservative electorate to a centrist view on social spending may mask age effects.

But another factor helps us understand middle Australia's support for higher social spending. Table 6 breaks down support for spending by perceptions of the standards of health and Medicare, and public education respectively. The AuSSA question is:

   Over the last two years, do you think the following have increased
   Or decreased or stayed the same? {The standard of health services
   Including Medicare; The standard of public education} [Increased a
   lot/a little; Stayed the same; Decreased a little/a lot; Can't
   choose].

Here, we uncover an important relationship--those who believe health and education have declined in the last two years strongly support spending. This is an especially significant finding in the case of health and Medicare because some 60 percent of respondents believe healthcare service standards have decreased. Unfavourable publicity about the state of Medicare is probably influencing public opinion. The universal coverage of service like Medicare means that any policy shift designed to promote 'user-pays' healthcare will have its largest (initial) impact on middle-income Australians. So it is possible that middle Australia is starting to register its disquiet about the direction of major public services by indicating support for more spending.

Multivariate analysis

So far we have assembled a profile of preferences for tax cuts and spending, and discussed a few factors influencing Australians in the direction of higher social spending. Multivariate analysis using AuSSA 2003 data allows us to go a step further, and identify, the real 'drivers' in support for spending. We present three regression models that provide a more complete analysis of this support. The response variable is the five point scale recording AuSSA 2003 responses on the tax cut versus increased social spending question, which range from 'Strongly favour reducing taxes' coded as 1 to 'Strongly favour increasing social spending' coded as 5. The models are estimated as ordered probit models, avoiding the assumption of linearity in the response variable that would otherwise be imposed by least squares estimation. However, to simplify the interpretation of the results, the coefficients reported here are not the coefficients of the ordered probit, but rather the marginal effects on the response variable coded 1-5. In other words, we report these coefficients as the number of response categories shifted by a one-unit increase in the explanatory variable, holding the other explanatory variables constant.

These models introduce three blocks of variables that enable us to assess the contribution to the overall explanation made by demographic factors, political attitudes and behaviour, and policy perceptions. (For a complete description of the AuSSA 2003 variables used, refer to Table 7). Model I includes variables for age, gender, respondent income, labour force status, and education status effects. Model II adds political variables: three dummy variables for Labor, Green/Democrat, and No Party identifiers, and a further measure of support for government redistributing income from the well-off to the less well-off. Model III further includes policy perception variables: respondent perceptions that health and Medicare have declined, that education has declined, and that taxes have risen (all over the last two years).

We report results for all three models in Table 8. We find that older respondents are significantly more likely to support social spending. A respondent 50 years of age or over is, on average, around one third of a point (on the five-point response scale) more in favour of spending than one who is under 35, after the other factors in the models I or II are taken into account. Although males seem to be slightly more in favour of reducing taxes, gender is not a significant factor in these models. Higher respondent income is a weak predictor of support for reducing taxes (it is insignificant in all models at 5 percent, when robust standard errors are used, but it is borderline significant at 6 percent in Model 1).

Respondents with higher education are more in favour of increasing social spending. Holding a post-school qualification produces an average shift in responses of one sixth of a point in favour of more spending, while holding a degree produces a shift of half a point or more in the same direction, relative to the base case of a respondent without post-school qualifications. The effect of holding a degree is moderated somewhat when party affiliations and perceptions of health, education and taxes are taken into account in Models II and III, but still tertiary education remains a strong independent factor shifting respondents toward higher spending. This may indicate why middle class-identifying respondents are no less likely to support spending than lower income earners: they tend to be more highly educated, increasing their support for spending, but pay higher taxes, which we would expect to reduce support. Respondents in the labour force are less likely to support social spending by more than a fifth of a response category, presumably reflecting their personal income tax burdens. Home owners are moderately in favour of reduced taxes, but as with gender, this factor is not statistically significant.

In model II, identifying as Labor or Green/Democrat increases support for spending. ALP identifiers shift about one third of a point toward spending relative to the base case, which comprises Coalition and 'Other party' identifiers. Green/Democrat identifiers are a whole point more in favour of spending than the base group. Those who have no stated party identification shift on average about the same amount towards reducing taxes as ALP identifiers shift to more spending (although the effect is not as clear in statistical significance). This finding may indicate that uncommitted or loosely attached voters are still a constituency responsive to the appeal of tax cuts. Supporting government redistribution of income is also a strong predictor, with an effect of similar magnitude to the effect of being in the labour force, though obviously in this instance the shift is toward higher spending not reduced taxes.

With the addition of the attitudinal variables in model III, the separate age effect lessens somewhat for those in the middle age groups but remains as strong for the oldest age group. Respondents who think health and Medicare have declined in the past two years support social spending by almost half a point more on the five-point scale when other variables are taken into account. We find something similar for those perceiving a decline in public education, who move towards spending by a third of a response category. This means that respondents who think both public education and health standards have declined are about three-quarters of a point stronger in their support for more social spending. Conversely, respondents who believe that taxes have risen in the past two years support reducing taxes (but not by as much as perceptions of declining services move respondents in the other direction). The independent contribution of health and education policy perceptions to explaining support for spending was established in our earlier research (Wilson and Breusch, 2003: 52). Their significance in model Ill is further evidence that key public services have a very substantial, independent impact on tax and spend preferences. In other words, the more voters believe services are declining, the more they are inclined to support spending.

Support for increasing taxes to pay for public services in AuSSA 2003

While we have new and compelling evidence about the shift in favour of social spending, these findings only tell us about the public's choice between taxes and spending as a matter of government priority. They do not tell us if the public is willing to pay higher taxes so that the government can fund higher spending on social services. We must resolve a tricky question confronting almost all public opinion research that finds a high level of support for higher public spending. Critics of these findings persuasively argue that respondents are rarely made aware of the consequences higher spending would have on taxes when they respond to survey questions (see, for example, Saunders 2003, 33-34). They believe that when respondents are able to 'optimise' their preferences, in full awareness of the costs of their choices, they reduce their support for spending because the tax consequences are made plain. One question in AuSSA 2003 enables us to find out about voter willingness to increase spending and can meet critics' objections--at least to an extent. The question is:

   Here is a list of areas where the Federal government spends money.
   Please tell us if you would be willing to pay higher taxes so the
   government could spend more in each of these areas ... [Health and
   Medicare, Primary and secondary schools, Environmental protection,
   Defence and national security; Welfare benefits]

The results are presented in Table 9. Our findings do not confirm the critics' doubts: we find voters are willing to pay more tax for health and Medicare, and public education, by very large margins. Almost 70 percent of voters are prepared to pay more tax to increase spending on health and Medicare, reflecting the service's wide use and popularity. This result is identical to the level of support for paying more tax to fund the National Health Service found by British researchers using British Social Attitudes data (BSA 2001). These new findings for Australia contrast remarkably with findings from the AES 1990, which showed that only about one in five voters were prepared to pay higher taxes for health and education services (see Table 9). Although the questions differ between the 1990 and 2003 surveys, ruling out fine comparisons, the increase in support for tax rises is nonetheless quite remarkable. The huge increase in support also communicates something of the reliability of polling evidence. The fact that support for higher taxes for health and education has changed so dramatically over time indicates that respondents have most likely changed their minds about tax increases. If we assumed otherwise--that respondents always express sanguine spending preferences to pollsters--it would be hard to explain why so few voters wanted tax increases in 1990.

A majority of voters are also willing to pay more tax for the environment, and defence and national security. Only welfare benefits attract a minority of support (34 percent). This finding is further evidence that the CIS research, which asked about welfare, could not fully capture the public's willingness to pay higher taxes for public spending. In fact, the present results provide robust evidence about public support for raising taxes when tax increases are modest and when they are clearly hypothecated in areas of public demand.

We are able to further understand support for tax increases by classifying the proportions of the electorate by their basic 'tax and spend' orientations about the broad social welfare functions of the government. (5) To do this, we classify respondents who support higher taxes for three 'welfare' options (health, schools and welfare) in Table 9 as 'Spenders'. We call respondents who support none of these three options 'Anti-spenders'. The rest--those who support one or two areas for tax increases but not all three--are 'Conditional Spenders'. The results of our analysis are presented in Diagram 1. We find that 28 percent of AuSSA respondents are 'Spenders', 19 percent are 'Anti-spenders', and the remaining 52 percent are 'Conditional Spenders'. These results reveal that a surprisingly small number of respondents are consistently opposed to tax increases. Although less than 30 percent turn out to consistently favour tax rises, we find that a majority of voters have an open mind about them. This means a large number of voters would potentially support an increase in taxes to support higher spending, depending on the type of tax increase and spending proposed. In other words, it appears that a large proportion of Australians have an open mind when it comes to considering tax increases.

When we consider how key constituencies divide between Spenders, Conditional Spenders, and Anti-spenders, we get an idea about which groups have made up their mind about taxes and social spending, and which are more fluid. Green/Democrat voters have the highest number of Spenders (45 percent). More Labor identifiers are Spenders than Coalition identifiers, but like Coalition identifiers, most are Conditional Spenders. As we might expect, more retired voters are Spenders (37 percent) than are full-time workers (3 8 percent). But, overall, the number of voters willing to consider tax increases for some social spending areas is surprisingly large, and has no doubt expanded over the past decade. It would appear that public apprehension about the state of services has been a driving factor in this new willingness to consider paying more tax.

Making sense of the new findings

Our findings challenge conventional wisdom about Australians' attitudes to taxes and social spending; for a long time, it has been to safe to assume Australian voters were pre-occupied by cutting taxes and that it was "natural' for politicians to prioritise them. By way of a conclusion, we reiterate our new findings that contradict received wisdom, and draw out implications for Australian politics and policy.

I. The new electorate for social spending

Our evidence suggests a continuing trend in Australian public opinion away from tax cuts towards higher social spending. Some factors driving this support for social spending are demographic: an ageing, more highly educated population is one more likely to support government spending on social services. Changes to the tax system may be playing some indirect role. A less 'visible' tax mix--lower personal income taxes for middle Australia and a GST in 2000--may have weakened support for further tax cuts (on tax visibility', see Wilensky 2002: 392). But our evidence suggests that the most important policy factor driving support for social spending is the widespread perception that health care and public education are in decline. From these findings we develop a more concrete picture of the constituencies who support and oppose greater social spending. Older voters who are concerned about the state of the health system are one clear constituency for higher social spending. Highly educated voters concerned about the state of education are another. By contrast, workers without higher education who think taxes have risen in the past two years are more in favour of tax cutting. Despite these differences, the shift towards higher social spending is across the board. While middle Australia may still seek tax cuts, these preferences are now balanced by new fiscal considerations, particularly the resources available to public health and education.

As we have suggested, support for higher social spending is not unique to Australia: support for social spending is very strong in the United Kingdom as well. How can we explain support for spending in both countries? Theorists argue that public opinion follows the cycles of politics and policy-voters eventually correct for long-term movements in policy in one direction by moving in the other (Stimson, 1991; Erikson, Mackuen and Stimson 2002: 33-53). When it comes to understanding trends in public opinion about taxes and spending, the British case is particularly illustrative of this fact, and not least because the British Social Attitudes provides researchers with almost 20 years of time series data. By the end of the 1970s, tax cuts were clearly popular. But the Thatcher program of fiscal austerity through the 1980s dented support for tax reductions, in part because tax levels were cut and in part because the public expressed dissatisfaction about the state of British public services, especially the National Health Service. Interestingly, strong support for social spending has persisted under British Labour, and this is in spite of several large increases in public expenditure to improve the NHS (Appleby and Rosete 2003: 38). This tells us that the British public remains disappointed about the state of public services, either because public expectations of service provision have increased (Glennester 1998: 477478) or they are mistrustful of all governments protecting and extending mainstay public services (see Wilson 2004: 18-19). Australia may be following a similar pattern: a period of conservative government, dissatisfaction with standards of health and education, and changing electoral demographics all appear to be contributing to stronger public support for spending.

II. Higher taxes to improve public services?

As we mentioned earlier, one longstanding and legitimate criticism of public opinion research that finds support for higher public spending is that such research rarely asks voters to consider the fiscal costs--usually tax rises--of such spending. Our findings here deal with this criticism directly by asking respondents to indicate their willingness to pay more tax for more spending for a range of public services. Our results show quite overwhelming support for a tax increase if it is modest and spent on Medicare and/or public education. And we find the proportion of Australians who unconditionally oppose a tax increase to improve services in the three social service areas of health, schools and welfare benefits is actually small--19 percent of the electorate. However, our findings do not mean a party offering a pro-public spending program would certainly win an election with these policies. In the United Kingdom, Thatcher and Major both won elections at a time when public opinion was unquestionably hardening against the Conservative fiscal policy. As Robert Weissberg points out, while public opinion data tells us something of the public's general preferences, it is less useful in determining the electorate's preferred set of policies (2001: 3). Nonetheless, our findings do indicate that the public is more likely to consider tax increases to pay for improved services than it has in the recent past. For politics to remain focused only on cutting taxes appears to misread the public mood.

Could a proposition to raise taxes actually assist--rather than hinder--a political party's election prospects? Certainly, any unspecific plan to increase taxes would be a political risk: the constituency for tax cuts has hardly disappeared. However, our results give us more specific insights into the electorate's thinking. Modest tax increases to pay for more spending on Medicare and public schools are actually popular options. This is because the tax increase proposed is both small and hypothecated in favour of public provision that the public wants protected-and even expanded. The nearest thing to such a hypothecated tax of this type is the Medicare levy. Because the purpose of the levy is reasonably well understood by voters, the risk of voter backlash from increasing the Medicare levy to pay for healthcare is probably low. In fact, any party proposing to increase the Medicare levy to pay for health may benefit from being honest and specific with voters about its tax and spend plans.

III. Private or public services? The battle over 'middle class welfare'

In A Self-Reliant Australia, Peter Saunders claims that:

   Welfare services, such as public education and healthcare, benefit
   high and middle income earners as much as (and sometimes more than)
   low income earners. Many recipients do not need this help from
   government. Even under current arrangements, most "middle income
   households" could afford to purchase their healthcare, education and
   retirement annuities from outside the state system ...' (2003: 46)

Saunders sees a solution in 'phasing out direct provisioning in favour of subsidized purchases (for example, through tax allowances or vouchers to offset schooling costs)' (2003: 46).

Higher levels of private health insurance and a trend to private education might be cited as supporting evidence for Saunders' model, an approach reflected in Howard Government policy. One might say that the market is speaking for itself. But it must be remembered that governments have encouraged private provision through tax incentives for private health and redirected spending priorities to private schools (Smith 2003: 16; ABS 2002: 9-10). One could thus argue that the choice to 'exit' the public system is being encouraged by the powerful incentives offered by current government policy.

Our findings in fact provide every indication of widespread support and confidence in government provision: the public appears willing to accept, and perhaps expects, tax rises to preserve public health and education systems. Some of this support may indicate a strong normative commitment to universal public health and education. But, equally, voters may be willing to pay higher taxes for public services because they calculate that paying higher taxes might still be cheaper than paying market prices for the private alternative. The relationship we find between perceptions of service decline and support for higher spending indicates strong voter resistance to poorly funding health and education services, a result consistent with findings elsewhere (see Pierson 1994: 166-175; Wilson 2004).

We conclude by rejecting the belief that Australians always support tax cuts over social spending. The political cycle has clearly turned. And the public's preparedness to consider specific tax increases might also be a timely resource for governments currently forced to address long-term policy problems with limited government revenues; especially as these revenues still lag behind most other OECD nations (see Keating 2004: 6-7; see also Department of Treasury 2004: Appendix I). Middle Australia is now seeking a new fiscal agenda in Federal politics that reflects its ongoing reliance on, and support for, major public services.

Table 1: The trend towards social spending in Australia

                           1984 *   1987   1987 *   1990 *   1993

Increase social spending       24     15        9        8     17
About the same/Depends         23     20       32       29     27
Reduce taxes                   53     65       59       63     56
n =                          2909   1740     1479     6084   2942

                           1996   1998   2001   2003

Increase social spending     17     26     30     48
About the same/Depends       26     27     28     24
Reduce taxes                 57     47     42     28
n =                        1727   1804   1951   4080

Source: * National Social Science Surveys 1984, 1987, 1990; Australian
Election Studies 1987, 1993, 1996, 1998, 2001; Australian Survey of
Social Attitudes 2003.

Table 2: Swing to spending between 2001 and 2003

                                             2001   2003   Change

Strongly favour increasing social spending     15     32      +17
Mildly favour increasing social spending       15     16       +1
Depends                                        28     24       -4
Mildly favour reducing taxes                   15      9       -6
Strongly favour reducing taxes                 27     19       -8
n =                                          1951   4080

Source: Australian Election Study, 2001, Australian Survey of Social
Attitudes, 2003

Table 3: Attitudes to taxes and spending by age

                          Social             Reduce
                        spending   Depends    Taxes

18-34 (n = 714)               42        26       32
35-49 (n = 1238)              45        25       30
50-64 (n = 1233)              51        22       27
65 and over (n = 895)         51        25       24

Source: Australian Survey of Social Attitudes, 2003

Table 4: Attitudes to taxes and spending by work and household status

                                      Social             Reduce
                                    Spending   Depends    Taxes

Working full-time (n = 1608)              46        22       32
Working part-time (n = 539)               49        24       27
Retired from paid work (n = 811)          56        22       22
Household duties (n = 384)                41        30       29
Full-time student (n = 115)               60        21       19
Households with children under 18
  (n = 1475)                              45        24       31

Source: Australian Survey of Social Attitudes, 2003

Table 5: Attitudes to taxes and spending by class and party ID (change
from 2001)

                                           Social              Reduce
                                         Spending   Depends     Taxes

Middle class (n = 1958)                   52(+22)    22(-4)    26(-17)
Working class (n = 1626)                  44(+16)    25(-4)    31(-12)
Liberal-National identifiers (n = 1416)   42(+21)    23(-5)    35(-16)
Labor identifiers (n = 1379)              53(+15)    23(-6)    24(-9)
Green identifiers (n = 190)               77(+14)    13(+3)    10(-17)
No party identification (n = 801)         43(+17)    30(-2)    27(-15)

Source: Australian Election Study, 2001, Australian Survey of Social
Attitudes, 2003

Table 6: Attitudes to taxes and spending by perceptions of health and
education services

                                            Social             Reduce
                                          Spending   Depends    Taxes

Health and Medicare same or better
  (n = 1531)                                    35        28       37
Health and Medicare declined (n = 2375)         57        20       22
Public education same or better
  (n = 2002)                                    41        25       34
Public education declined (n = 1696)            59        19       22

Source: Australian Survey of Social Attitudes, 2003

Table 7: Variable descriptions

Variable              Description

Tax and spend         If the government had a choice between reducing
(response variable)   personal income taxes or increasing social
                      spending on services like health and education,
                      which do you think it should do? (5 point scale)

Age                   Age groups: 35-49, 50-64, and 65 and over (three
                      binary dummies; omitted category: age 18-34)

Income                Gross annual income, in four intervals,
                      represented by a single variable containing the
                      group index, 1-4

Male                  Dummy variable for male respondent

Post-school           Holds post-school qualifications, but not a
                      degree (dummy)

Degree                Holds bachelor degree or higher (dummy)

Labour force          In paid work or unemployed, looking for work
                      (dummy)

Own home              Owns their own home or has mortgage (dummy)

ALP                   ALP party ID (dummy)

Green/Democrat        Green or Democrat party ID (dummy)

No Party              No party ID (dummy)

Redistribution        Agrees or strongly agrees with the proposition
                      'Government should redistribute income from the
                      better-off to those who are less well-off' (dummy
                      recoded from 5 point scale)

Health and Medicare   Thinks the standard of health care including
declined              Medicare has decreased a little or a lot over the
                      last two years (dummy recoded from 5 point
                      scale)

Education declined    Thinks the standard of public education has
                      decreased a little or a lot over the last two
                      years (dummy recoded from 5 point scale)

Taxes up              Thinks taxes have increased a little or a lot
                      over the last two years (dummy recoded from 5
                      point scale)

Table 8: Ordered probit models for dependent variable 'tax and spend'

                       Model I    Model II    Model 1111

Age 35-49              0.206 **    0.163 **    0.103
Age 50-64              0.351 **    0.345 **    0.286 **
Age 65 and over        0.344 **    0.359 **    0.344 **
Male                  -0.038      -0.028      -0.029
Respondent income     -0.060      -0.032      -0.043
Post-school            0.163 **    0.161 **    0.124 *
Degree                 0.599 **    0.541 **    0.446 **
Labour force          -0.222 **    0.232 **    0.212 **
Own home              -0.117      -0.058      -0.055

ALP                                0.338 **    0.317 **
Greens/Democrats                   1.001 **    0.866 **
No Party                          -0.346 *    -0.454 *
Government should
  redistribute                     0.213 **    0.190 **

Medicare and Health
  declined                                     0.436 **
Education declined                             0.347 **
Taxes up                                      -0.382 **

Number of cases        3771        3771        3771
Log-likelihood        -5696.014   -5612.623   -5504.850

Notes: * 0.01 <p<0.05, ** p<0.01 using robust standard errors. The
coefficients are the marginal effects on the dependent variable of
the respective explanatory variables. The dependent variable has a
5 point ordered scale where 1 = Strongly favour reducing taxes;
2 = Mildly favour reducing taxes; 3 = Depends: 4 = Mildly favour
increasing social spending; 5 = Strongly favour increasing social
spending.

Table 9: Paying taxes for services

                              Quite                           Can't
Are you willing to pay        a bit   A little    No more   choose/
more tax for ...           more tax   more tax        tax   Neither

Health and Medicare              17         52         28         4
Primary and secondary
  schools                        14         49         31         6
Environmental protection         11         43         40         6
Defence and national
  security                       10         44         40         5
Welfare benefits                  6         28         60         6

The government should
increase taxes and spend
more on ...                              Agree   Disagree   Neither

Health care (1990)                          18         51        31
Education (1990)                            21         40        39

Source: Australian Election Study, 1990, Australian Survey of Social
Attitudes, 2003

Table 10: Spenders by selected categories

                                   Spenders   Conditional   Anti-
                                              Spenders      Spenders

Liberal-National
  identifiers (n=1484)             22         57            21
Labor identifiers (n=1435)         35         49            17
Green/Democrat identifiers
  (n=303)                          45         44            11
Full-time workers (n=1654)         21         57            22
Retired (n=865)                    38         45            16
Respondents from households with
  children under 18 (n=1517)       25         55            20

Source: Australian Survey of Social Attitudes, 2003

[GRAPHIC OMITTED]

Footnotes

(1) We thank Mr Peter McCarthy (Research School of Social Sciences) for editorial comment. We also thank Dr Pippa Norris (Harvard University) and Dr Gabrielle Meagher (Sydney University) for their helpful suggestions and references.

(2) The five AES response options are 'Strongly favour' or 'Mildly favour' reducing taxes, with similar options for spending more on social services, and a middle category of 'Depends'. The outer two options on each side of the question are grouped to produce the three categories in Table 1. The six NSSS options are 'Very strongly for', 'Fairly strongly for' or 'Mildly for' reducing taxes, with parallel options for increasing spending, The outer two options on each side are again grouped, and the middle two options are grouped to become the middle category of three. This probably overstates the frequencies in the middle category in the grouped NSSS results compared with AES, but the taxing and spending options are treated symmetrically for both surveys.

(3) The Australian Survey of Social Attitudes (AuSSA) was conducted by the Centre for Social Research at the ANU and directed by Rachel Gibson, Shaun Wilson, Gabrielle Meagher, David Denemark and Mark Western. The survey took the form of a mail questionnaire, posted to a random sample of Australian voters registered on the Commonwealth Electoral Roll, which was partially updated in 2002. All responses were collected between September and December 2003. Basic results of the survey and demographics are available online at <assda.anu.edu.au>.

(4) Sefton (2003:11) speculates whether weaker support for taxing and spending policies among young people captures a one-off generational impact of the Thatcher years or represents 'the beginning of a long-term decline in support for the welfare state'. Our own comparisons between NSSS 1984 and AuSSA 2003 (not reported in detail here) reveal a similar trend of declining support for spending policies among the 18-34 year cohort. However, understanding these changes over time would require a separate paper.

(5) We exclude the two other items, defence and national security and the environment, because they both represent different spending types from the social spending items that are directly relevant to this paper.

References

ABS (2002), Centenary Article--Australian schools: participation and funding 1901-2000, (authors: G. Burke and A. Spaull), Commonwealth of Australia: Canberra.

Appleby, J. and Rosete, A. (2003) 'The NHS: keeping up with public expectations?', in Park, A, Curtice, J., Jarvis, L and Bromley, C. (eds.), British Social Attitudes: the 20th report--Continuity and Change over Two Decades, London: Sage, 29-44.

British Social Attitudes 2001 (2001), Online available: <www.data- archive.ac.uk>.

Castles, F. (1993) 'Changing Course in Economic Policy: The English-Speaking Nations in the 1980s', in Castles, F. (ed) Families of Nations: Patterns of Public Policy in Western Democracies, Dartmouth: Aldershot.

Department of Health and Ageing (2003) Annual Report, 2002-2003. Online available: <www.health.gov.au>.

Department of Treasury (2003) Final Budget Outcome, 2002-2003. Online available at: <www.treasury.gov.au>.

Department of Treasury (2004) Australia's Demographic Challenges. Online available at: <www.treasury.gov.au>.

Erikson, R.S., Mackuen, M.B., and Stimson, J.A. (2002) 'Public Opinion and Policy: Causal Flow in a Macro System Model', in Manza, J., Cook, F.L. and Page, B.I. (eds), Navigating Public Opinion: Polls, Policy, and the Future of American Democracy, Oxford University Press: Oxford, 33-53.

Glennester, H. (1998) 'A Reply to "Things Can Only Get Better"', Policy and Politics, vol. 26, no. 4, pp. 477-478.

Keating, M. (2004), 'The Case for Increased Taxation', Australian Academy of Social Science, 2004. Available online: <www.assa.org.au>.

McMullan, B. (2003), 'Paying for Fairness and Equity: Encouraging Enterprise', Address to the Australian Fabian Society, Melbourne, 31 May.

Pierson, P. (1994), Dismantling the Welfare State?: Reagan, Thatcher and the Politics of Retrenchment, Cambridge University Press: Cambridge.

Saunders, P. (2003) A Self-Reliant Australia: Welfare Policy for the 21st Century, CIS Occasional Paper 86, Centre for Independent Studies: Sydney.

Sefton, T. (2003) 'What do we want from the welfare state', in Thomson, K., Park, A., Jarvis, L., Bromley, C., and Curtice, J. (eds), British Social Attitudes: the 20th Report, Sage, London, Chapter 1.

Smith, J. (2003), 'Tax Expenditures: the $30 Billion Twilight Zone of Government Spending', Research Paper No. 8 2002-03, Department of the Parliamentary Library, Commonwealth of Australia: Canberra, May 26.

Stimson, J. (1991) Public Opinion in America: Moods, Cycles and Swings, Westview Press: Boulder, CO.

Weissberg, R. (2001) 'Why Policymakers Should Ignore Public Opinion Polls', Policy Analysis, No. 402, May 29. Available online: <www.cato.org>.

Wilensky, H. (2002) Rich Democracies: Political Economy, Public Policy and Performance, University of California Press: Berkeley.

Whiteford, P. and Angenent, G. (2001) 'The Australian system of social protection-an overview', 2nd edition, Occasional Paper no. 6, Department of Family and Community Services: Canberra.

Wilson, S. (2004) 'Explaining Rising Support for Social Spending in Australia and the United Kingdom: Do Universal Healthcare Systems Make a Difference?', paper for the Welfare State section of the American Sociological Association Annual Meeting, San Francisco, August.

Wilson, S. and Breusch, T. (2003) 'Taxes and Social Spending: The Shifting Demands of the Australian Public', Australian Journal of Social Issues, 38, 1, February, 39-56.

Dr Trevor Breusch

Trevor Breusch is a Senior Research Fellow in the Research School of Social Sciences at the Australian National University. [e-mail]

Dr Shaun Wilson

Shaun Wilson is a Research Fellow in the Research School of Social Sciences at the Australian National University. [e-mail]