The last two decades in Australia have seen a major recasting of tax and social security policies towards families. Primary income inequality over this period has increased (Raskell et al, 1994; Saunders, 1994), although as a result of award and minimum wage arrangements the earnings distribution has remained relatively compressed compared to other OECD countries (Norris and Wooden, 1996). Cass (1996) has described how the battle between principles of horizontal and vertical distribution has been fought out in family policy, with the latter increasingly winning out through substantially increased but more tightly targeted family payments. Castles (1985) has argued that the Australian 'wage earners' welfare state' has been predicated on the maintenance of a degree of income security and equality through central control of the wages system, thus distinguishing it from the other low-tax 'liberal' welfare states of the USA or the United Kingdom. Certainly there is broad agreement that the social wage initiatives which were meant to compensate for wage restraint under the Accord framework did mitigate the effects of growing inequality. The question now is how far such protection for low-income families can be maintained, with the erosion of central wage bargaining, a governmental commitment to less rather than more social welfare expenditure, and a partial shift in direction back towards support for families through tax allowances.

It is therefore an appropriate time to take stock again of where Australia stands in tax and social security policies towards families compared to the other industrialised countries from which many of its inhabitants originally came. This is of interest both intellectually, in terms of our conceptual understanding of welfare states, and practically, as a means of assessing the relative impact and effectiveness of different policy approaches

One way of addressing these questions is to use a 'model family' technique for comparing the structural effects of tax/transfer policies at one point in time. This paper is based on incorporating data for Australia in May 1994 to those collected for the then 12 members of the EU, as part of the work of the European Commission's Observatory on National Family Policies. It should be emphasised that only a preliminary analysis is presented here. More detailed comparisons will be presented in further work incorporating 1995 data and examining changes since 1992, drawing on Bradshaw et al. (1993) and the Australian-focused analysis by Whiteford (1995a). The paper at present is deliberately limited in purpose. It does not attempt at this stage to provide a detailed explanation of differences, although aspects of the wider policy traditions in different countries are touched on. It also does not address broader and important elements of family policy such as provision of parental leave and other arrangements for reconciling work and family life, or support for caring work. Policy developments in these areas for the EU countries are discussed in Ditch et al. (1995a, 1995b). The model families method is described and discussed in the next section. The paper presents a number of analyses which locate Australian policy in relation to the various patterns found in the EU. In the discussion some speculations are offered on the directions in which more recent policy changes on family support are taking Australia.

Simulating policy structures using the model family method

The model family technique allows comparison of the structure and value of 'income packages' which accrue to particular types of family or household as a result of the distribution of market earnings and the fiscal and social security arrangements in existence at a given time in each country. Figures supplied by national experts according to an agreed, common matrix, based on a package of benefits and charges likely to apply to a set of model families in specified, near-identical circumstances, are entered onto a series of spreadsheets. These data enable comparison of disposable income accruing to the model families, and thus of the marginal gains or losses for different types and sizes of family which are implicit in policy structures. They also allow analysis of the proportion of income derived from different sources, relative costs of specified services, replacement ratios and many other elements of income packages.

This approach to policy analysis was pioneered by Kahn and Kamerman (1978, 1983, 1989) in a series of international studies of comparative family policy, and was also used by Bradshaw and Piachaud (1980) in their early study of support for children in Europe. It was further elaborated in a more recent study of child support packages in 15 countries (Bradshaw et al., 1993). Data from that research were also used to compare the structural and incentive effects of social security policies for lone parents (Whiteford and Bradshaw, 1994), and the treatment of married women engaged in housework and homecare (Shaver and Bradshaw, 1995). The same method has since been used in a study of social assistance schemes in the OECD countries (Eardley et al., 1996a, 1996b), in work for the European Observatory (Ditch et al., 1995a) and, most recently, in an examination of the reasons for differing levels of participation in paid work by sole parents in 20 countries (Bradshaw et al., 1996).

For the Observatory study, the matrix included nine family types and six sets of economic circumstances (or Cases). 'Family' is of course an imprecise and contested term. Here it is used to refer mainly to people with dependent children. They are assumed to have no other adults living in the household and are thus close to the 'income unit' concept used for social security. Income units without children are included in order to allow comparison of the value of net tax and transfer packages accruing where dependent children are present in the family.

Family types

The following model family types were chosen for comparison:

1. A single person aged 35
2. A couple both aged 35 (all couples were assumed to be married)
3. A couple aged 35 with one child under three (two years and eleven months ie. pre-school age in all countries)
4. A couple aged 35 with one school aged child (seven years)
5. A couple with two children, aged and eight years
6. A couple with three children, aged seven, eight and 14
7. A sole parent (female, separated or divorced not widowed) with one pre-school child
8. A sole parent (female, separated or divorced not widowed) with one school-age child (aged seven).
9. A sole parent (female, separated or divorced not widowed) with two school-age children (aged seven and eight).

These types were chosen as illustrative of the range of possible families. There are obviously limitations attached to the choices. Some family types are more or less common in different countries; they are all nuclear families and so the impact of tax/transfer policies on more complex households cannot be examined; couples are all married and so different treatment of cohabiting or de facto couples is also outside the scope of analysis (although the European Observatory is currently studying this question). For the purposes of the analysis it is also assumed (contrary to some known evidence) that resources are shared equitably within families. The technique does, however, allow for different countries' policy approaches to be reflected in areas such as individual taxation, and direction of payments for children through tax allowances (often to the man) or through cash payments (usually to the mother).

Income levels

Families were allocated hypothetical market earnings at several levels or alternatively to have no earned income or entitlement to social insurance benefits and thus to be reliant on social assistance payments. The following cases were compared:

1. Couples and sole parents with one earner at half national average male earnings in the manufacturing sector

2. Couples and sole parents with one earner at average male earnings in manufacturing

3. Couples and sole parents with one earner at 1.5 average male earnings in manufacture

4. Couples with one earner receiving average male earnings and the other at 0.66 average female earnings

5. Couples with one earner at 1.5 average male earnings and the other at 1.5 average female earnings

6. Couples and sole parents receiving social assistance (Newstart or Sole Parent Pension in the Australian case)

Clearly it would often be unrealistic for sole parents to be receiving the relevant ratio of average male earnings, but for this analysis it was necessary to avoid confusing the structural effects of tax and social security policies with wage differentials by sex.

Table 1 shows estimated average gross male and female earnings in local currencies and purchasing power parities in £ sterling at May 1994. Sterling is used here in preference to Australian dollars as the original EU comparisons were produced in sterling. In general the relativities are of more interest than the absolute levels.

Partly this is a function of purchasing power parities, but it is likely that it relates mainly to the level of the 'social wage'. Employers' national insurance contributions in those countries are all particularly high in relation to wages - as much as 55 per cent in France. What seems to be happening is that workers are effectively foregoing higher earnings for the benefits (often in the future) of a generous social wage. High social security contributions can thus be seen as a form of deferred wage. By contrast, earnings are higher in Australia, for example, and to a lesser extent in the UK, partly because employers are not required to contribute so much to a social wage. If this is the explanation for these differences, then as Whiteford (1995b) has argued, when comparing the earnings levels of workers between countries it might be more appropriate to include employers' social security contributions as part of earnings. Higher average earnings in Australia are also a function of the relatively equal distribution of market wages which has characterised the 'wage earners welfare state'. In comparison with countries which have higher primary wage inequality, such as France and Germany, but spend more on social transfers, this tends to exaggerate the equalising effect of welfare spending (Whiteford, 1997 forthcoming).

Table 1 also shows that there are differences in the relationship between male and female average earnings, with the ratio of the latter to the former ranging from 56 per cent in Luxembourg to 85 per cent in Denmark. Australia is among a group of countries with a ratio at around 75 per cent.

The calculation of cash benefits received and tax and social security contributions payable is relatively straightforward, given the model families' income and household circumstances. However, the estimation of other elements of the package requires the establishment of a common framework of analysis, involving detailed assumptions which are not always entirely satisfactory. These involve local taxation, health, education and child care costs, and housing. Some of these can be particularly problematic and require a brief explanation.

Health costs

The base line assumption was that health care at the point of demand was free of charge, available to all regardless of means and of similar quality in every country. Account was then taken of any variations from this assumption, by costing a standard package of health care in each country. This included the cost of prescriptions for three courses of standard antibiotics per person per year; three visits to a general practitioner per person per year, and one visit per person per year to a dentist, for a check-up and one filling. In the Australian case this package was costed on the assumption that all families used GPs who bulk-billed and that health care card prescription subsidies were used where available, but that all families paid in full for reasonably low-cost dentistry. Costings were based on information collected for the Budget Standards project being carried out at the SPRC (Saunders, 1996).

School and child-care costs

It was assumed that school education of an equivalent standard, including basic books, was available free of charge to all children of school age, that parents would have to pay for a midday meal, and that children lived near enough their school not to require school transport. Account was then taken of any essential charges that parents were expected to pay for education and any benefits (including the value of free or subsidised school meals) that they might receive.

In order to take account of the costs of pre-school child care and the value of free or subsidised provision, a standard package was again established for each country. National informants were asked to cost provision according to the most prevalent pattern of formal, full-time, pre-school provision in their country. This resulted in different types of care being costed in each country and no account could be taken of variations in the quality of care. For comparability, pre-school costs were only taken into account in the case of the working lone parent and the dual-earner couple with pre-school age children. This is inevitably an unrepresentative picture in some countries such as Denmark, where most married women would be working and their children would also attend child-care centres. In Australia too, child care has been widely available even if women are not in the labour market full time. For Australia, it was assumed that in the two cases where child care costs were included the parents were using 40 hours per week of long-day care, at $3 per hour.

Housing costs

Housing costs are perhaps the most difficult element to take account of in comparative research of this kind. Costs vary within countries and between countries according to tenure and the size, age and location of the dwellings. In some countries rents may be controlled for people occupying dwellings before a certain date. For owner-occupiers, loan structures and interest rates vary between countries, often according to the stage in the economic cycle, while the level of mortgage interest is also affected by the stage of a purchaser's life cycle. There are also significant differences between countries in tenure distribution at different income levels. Nevertheless, housing costs cannot be ignored. In many countries, help with housing costs is a critical element in the benefit package, and even where such support does not exist, variations in housing costs mean that real income levels differ substantially before and after taking account of housing.

For this exercise, the families were all assumed to be living in rented dwellings - rented from a public authority, housing co-operative or housing association, if they were common forms of tenure in the country, or from a private landlord if that was the most prevalent tenure pattern. In those countries with high levels of owner-occupation, including Australia, this assumption is obviously less representative.

National informants were asked to fix typical or representative rent levels for such dwellings in a given town in their country. Locating the families in a given place helps to structure the comparisons where benefits vary locally, but it can be difficult in some countries to nominate a typical or 'average' location. The size of the dwellings was specified and varied with the size of the model families. Again, this can be an artificial assumption, as families on constrained incomes will in practice make choices according to varying local housing markets. For Australia, families were assumed to be renting units at $110 per week for one bedroom, $140 pw for two bedrooms, and $180 pw for three bedrooms. These figures were based on data for the lower quartile of rents in the Sydney suburb of Hurstville, an area estimated as typical 'middle Australian' for SPRC's Budget Standards project. It should be noted that this assumption differs from that used in Bradshaw et al. (1993) and subsequently discussed further by Whiteford (1995a), where all Australian model families were assumed to be renting from the State housing authority. Although the model rents were deliberately kept low to compensate for the atypicality of the Sydney housing market within Australia, this decision still makes most of the families notionally worse off than in the earlier study, after housing costs are taken into account.

Housing costs are not only a problem at the design stage of comparative projects, they are also difficult to handle at the analysis stage, particularly for people with incomes at the social assistance level. The problem arises because, in some countries, housing costs, or a proportion of them, are paid together with social assistance (including Australia, where Rent Assistance is linked to receipt of income-related pensions, allowances or family payments). In other countries, housing costs are subsidised either by a reduction in rent payable (as for public tenants in Australia), or by a housing allowance scheme administered separately from social assistance. If comparisons are made of the level of social assistance before housing costs, then the first group of countries - those that pay the housing subsidy as part of social assistance - will appear to have higher levels than the others. Housing, nevertheless, has to be included, as for many people it has a crucial impact on their disposable income. It would be wrong to compare social assistance only before housing, yet it would also be misleading in certain circumstances to compare social assistance incomes only after housing. One answer is to present the results in most circumstances both before and after housing costs, although for reasons of space and time this is not always done in the present paper. The key point is always to bear in mind that results are affected by the assumptions made.

Limitations of the model families method

The advantage of the policy simulation approach is that it maximises the possibility of comparing like with like - one of the perennial difficulties in comparative research - and allows for an extensive range of analyses. However, the technique clearly also has a number of limitations. These have been discussed in more detail elsewhere (Eardley, 1996). First, this method inevitably produces a description of the way a system should work rather than how it necessarily does. Thus the behavioural effects of policy are largely missing from the analysis. For example, the high cost of formal child care in some countries means that in practice many working lone parents have to find other informal solutions. In modelling their net incomes, it may, therefore, be unrealistic to take full child-care costs into account. The problems mentioned above in relation to housing are particularly intractable. These are limitations which make it necessary always to emphasise that it is the structural features of tax/benefit systems which are being compared, not the real outcomes for actual families. Concentrating on the formal arrangements and the intended impact of the policies one is seeking to evaluate can, nevertheless, be valid.

Secondly, looking at families at one point in time obscures the more complex life-cycle effects of tax/transfer systems. In particular, while employees' social security contributions are taken into account as reducing disposable income, there is no way of modelling the future benefits which accrue from them. In so far as higher contributions may bring better benefits (which is not always the case), this may distort the picture for some countries from a longer-term perspective, especially in relation to retirement pensions.

Thirdly, the more assumptions that are made about the circumstances of the model families in order to achieve comparability, the less representative those families are of actual populations. It has to be accepted that the model families in the study are most unlikely to exist in all their characteristics in any country.

Bearing in mind these limitations, we now move on to discuss a selection of analyses based on the model family matrices.

Results of the model family analysis

First we look at how the tax/transfer systems in 1994 operated to alter the distribution of market incomes for four family types (Figures 1 - 4). The first column for each country represents gross earnings, the second column gives the income after direct taxes (including local taxes or rates where they exist) and any compulsory social security contributions. Australia has no social security contributions as such, but the Medicare Levy fulfils this role so it is included in this column. The third column gives disposable income after any social security benefits have been received, net of any applicable charges for health, education or child care. The final column gives disposable income after net housing costs have been deducted.

Not surprisingly, we find considerable variation between countries at different stages of the redistributive process. The low-earning larger family (Fig. 1) in Australia started off, as we have seen, with earnings roughly on a par with the UK, Belgium, Germany and the Netherlands, faced little direct taxation and then had its income raised to the highest level of all except Luxembourg through net transfers. Other countries offering similar levels of social security support to such families included the UK, Ireland, Belgium and France. After housing, however, because of the limited level of Rent Assistance available to people in privately rented housing, this Australian family's income fell to below that of both Ireland and the UK, among others. The Irish family had the lowest net housing costs at this level and thus ended up with the highest disposable income. It is worth noting that in Portugal and Spain, in spite of a traditional policy emphasis on extra assistance for larger families, the benefit system appears to have had only a marginal effect in raising post-tax income. By contrast, the impact of benefits was substantial in Greece, but was severely reduced by net housing costs.

See Figure 1

Similar re-ordering of countries can be observed in Figure 2, where the sole parent's need to pay for child care in order to work reduced the net benefit package considerably in some countries - notably Ireland and the UK. Consequently, the reality is that sole parents in these countries are often forced to seek cheaper, informal alternative. At average male earnings, the Australian sole parent had lost entitlement to the income-related Additional Family Payment but still benefited from subsidised child care. At this stage she had the highest disposable income (after Luxembourg which is consistently exceptional), but then lost out again as a result of housing costs. It is noticeable that at this earnings level, Denmark, the Netherlands and Germany took a relatively high amount from this sole parent in tax and social security contributions.

See Figure 2

Figure 3 looks at dual earners with a pre-school child. At this moderate level of total market earnings Australia still occupied the same position in the ranking because the relativities between male and female wages are similar to the other higher-waged countries. A combination of separate income taxation and no social security contributions apart from the Medicare Levy left this Australian family relatively lightly taxed compared to a number of the northern European countries, though not the most lightly. Only France and Belgium maintained benefits and child care subsidies at a level where net income after transfers and charges exceeded that post-tax.

See Figure 3

The child payment package

A key element of tax/transfer policies is the combination of tax allowances or credits, cash payments and remission of charges in respect of dependent children. The value of this child-related package is represented in the following charts as the difference between the net disposable resources of a family with children and a childless couple with the same gross earnings. Figure 4 shows how the value of the total package (without considering the different mechanism of delivery for the moment) varied by earnings, for a couple with two school-aged children. Luxembourg is omitted from this chart because its exceptionally generous payments for higher earners distorts the scale. Luxembourg is unusual in a number of ways, being a very small, wealthy country with a high proportion of migrant workers who often do not benefit from its generous social security system.

See Figure 4

It is evident that in Ireland, especially, followed by the UK and then Australia, support for children was highly targeted towards the lower-earning family. In these countries the package did not vary significantly for families with earnings above this level except that it dropped off for the highest earning Australian family when the income test for Basic Family Payment cut in. Belgium and Denmark barely varied their packages by earnings (though such one-earner families are unusual in Denmark), and France and Germany gave more to higher earners, mainly through income-related (and therefore non-progressive) child tax allowances. The figure for higher earners in Italy would be negative because any child-related benefits or allowances did not compensate for educational costs parents are expected to pay. Charges or benefits for after-school care for children of school age are not included here, although such provision is available on an income-related basis in a number of countries - particularly in Australia and Denmark.

See Figure 5

Figure 5 looks at the child payment package in another way. It takes a single-earner couple on average earnings and shows how the package - again before housing costs - varied with the number of children. Luxembourg is again excluded. In Belgium and France the net package was proportionately much more generous for a three-child than a two-child family, reflecting perhaps a history of pro-natalism, whereas in most other countries families received about the same or less for the third child. The level of payments for children at this income level in Australia again illustrates how they were proportionately targeted even more at low earners than in the UK.

Support for sole parents

So far the analysis has concentrated mainly on two-parent families, but sole parents in most countries, including Australia, are among the groups most likely to be poor. Here two aspects of their treatment in tax/transfer systems are considered. First, Table 2 compares their net disposable resources with those of a single person without children at the same gross earnings level. Their treatment varied by earnings and family size in most countries except Denmark, but at half average earnings a sole parent with one school-age child received the most generous treatment in Australia, Ireland and Germany, and the least generous in the Portugal and Spain. This sole parent in Australia would not only have received substantial family payments, but also some Sole Parent Pension, Guardians Allowance, Pharmaceutical and Telephone Allowances. The strength of the targeting principle in Ireland seems to have outweighed Catholic traditions and historical discouragement of sole parenthood, though Germany's position here is perhaps more surprising. At higher earnings levels France also became relatively generous, as were Germany and Luxembourg. Most striking, however, is the way that differential support for sole parents in Australia dropped off at the higher earnings levels.

Table 2: Support for sole parents: difference between net disposable income of a sole parent and a single person, before housing, £ sterling PPPs per month

	Half average earnings 1 ch (7)   2 ch (7, 8)	Average earnings 1 ch (7)   2 ch (7, 8)	Average earnings 1 ch (7)   2 ch (7, 8)
Australia	204       297	59        76	59        76
Belgium	59        161	59        161	59        161
Denmark	122       227	122       227	122       227
France	69        201	124       254	131       282
Germany	195       328	193       347	214       379
Greece	30        39	33        47	31        45
Ireland	209       271	99        109	88        112
Italy	45        63	15        28	-22       -31
Luxembourg	135       250	226       363	192       395
Netherlands	98        150	131       184	138       175
Portugal	4         8	9         24	11        22
Spain	12        25	42        53	58        62
United Kingdom	185       269	99        135	99        135

After housing costs (not shown here), on the assumption that sole parents would have to occupy larger accommodation than single people, they would be worse off in Spain and Italy than single people and little better off in Greece and Portugal. This highlights the difficulties of taking account of housing costs. Precisely because little financial help is available to sole parents in these countries, women on their own with children would have difficulty meeting the rents for the standard size and type hypothetically allocated to them and would in practice be obliged to occupy other, less expensive accommodation.

Since only part of any extra income accruing to sole parents is directly child related, a second way of understanding how they are treated is to look at the difference in net disposable resources of sole parents and couples with the same number of children and the same earnings levels. Table 3 below does this both before and after housing. The tax/transfer systems in most countries treated sole parents slightly more generously than couples with the same earnings. The main consistent exceptions where sole parents were worse off than couples were Belgium and Italy.

Table 3: Net disposable income of sole parents compared with that of single-earner couples

	Half average earnings 1 ch (7)   2 ch (7, 8)	Average earnings 1 ch (7)   2 ch (7, 8)	average earnings 1 ch (7)   2 ch (7, 8)
Australia before housing after housing	%         % 108       110 117       116	%         % 102       103 103       95	%         % 101       103 102       97
Belgium before housing after housing	93        94 89        91	90         91 86        88	92        93 86        91
Denmark before housing after housing	103       115 104       118	98        107 97        108	95        101 94        102
France before housing after housing	108       114 128      119	105       109 116       112	102       107 104       110
Germany before housing after housing	106       117 105       114	104       113 105       116	99        108 99        109
Greece before housing after housing	100       100 101       119	100       100 100       107	100       100 100       104
Ireland before housing after housing	96        105 96        95	102       102 103       110	100       101 99        101
Italy before housing after housing	94        94 92        92	99        98 99        98	100       99 100       98
Luxembourg before housing after housing	100       100 100       100	99        100 99        100	90        91 88        89
Netherlands before housing after housing	102       102 103       103	104        104 104       104	107       107 108       108
Portugal before housing after housing	102       100 104       101	100       101 100       101	96        96 96        96
Spain before housing after housing	102       102 105       107	101       101 101       101	100       100 101       101
United Kingdom before housing after housing	104       104 107       109	103       103 105       105	102       102 103       103

Components of the tax/transfer system

So far the value of the family package has been presented without any analysis of the impact of its different component parts. Breaking down the package into components allows us to see where different countries place their policy emphasis (Tables 4 - 6)

Table 4 shows the structure of the family package for a lower-income, larger family. The figure in each column represents the income difference (either positive or negative) that the presence of children makes to the family compared with a childless couple. Child tax allowances, which would accrue to the main earner (most likely the man), were only significant for families at this earnings level in Belgium and Italy, whereas for most countries except Australia, Ireland and those of Southern Europe horizontally redistributive universal child benefits were a key component.

Table 4: Structure of the family package for a single-earner couple with 3 children (7, 8, 14) on half average male earnings: difference between their income and that of a childless couple, £ sterling PPPs per month

	Tax and social insurance	Non income-tested child benefit	Income-tested child benefit	Health costs	School costs	Other	Total
Australia	9	-	280	-10	-	-	279
Belgium	36	278	-	-7	-18	-	289
Denmark	-	104	-	-1	-	-	103
France	-	162	92	-10	99	-	343
Germany	-	125	55	-	-	24	204
Greece	-	23	-	-1	-	-	22
Ireland	19	57	223	37	9	-	345
Italy	9	-	-	-10	-43	87	43
Luxembourg	-	386	-	-4	-22	-	360
Netherlands	-	165	-	-8	-34	74	197
Portugal	-	22	17	-15	-7	-	17
Spain	-	-	46	-15	-	-	31
United Kingdom	-	116	212	-	-	-	328

Australia and Ireland, in line with the targeted approach, relied primarily on means-tested family payments, while the UK's support was divided between the two instruments. Income-related payments played a significant role in France and Germany too, but were subordinate to the universal payments. Variations in health charges were not large, although Ireland stood out in offering an extra health-related payment for this family type. In addition France and Ireland provided free school meals.

The same picture is presented in Table 5, this time for a sole parent on average earnings, compared with a single person. In all countries except Denmark the sole parent paid less tax than a single person, though the difference was marginal in Portugal and Greece. The Netherlands, Denmark and the UK stood out in providing a significant level of universal child benefit, while France's policy of extending income-tested help for sole parents further up the income scale than most other countries is also demonstrated. The considerable variation between countries in the net disposable income available to this sole parent compared with a single person was crucially affected by the cost of child care. In many countries at this earnings level she would have to meet a substantial proportion of the nominal costs, or find cheaper alternatives. Here Belgium and, particularly, France stood out in the provision of subsidised child care, while if the UK sole parent had to pay in full she would be worse off (before housing costs) compared to a single person than in any other country. In Australia, although child care assistance would have met more than four-fifths of the nominal day-care costs, the extra tax allowance and family payment available would still not match what the sole parent had to pay in order to be able to go out to work.


Table 5: Structure of the family package for a sole parent with 1 child (2 years, 11 months), on average male earnings: difference between her income and that of a childless single person, £ sterling PPPs per month

	Tax and social insurance	Non income-tested child benefit	Income-tested child benefit	Health costs	Child care costs	Other	Total
Australia	43	-	21	-4	-92	-	-33
Belgium	14	41	-	-2	-	-	53
Denmark	-	87	-	-	-97	46	-36
France	67	-	91	-4	12	44	210
Germany	76	21	-	-	-42	76	131
Greece	4	4	-	-	-	-	8
Ireland	89	19	-	-9	-184	-	-85
Italy	27	-	-	-3	-73	29	-20
Luxembourg	158	55	-	-1	-102	-	110
Netherlands	93	82	-	4	-164	-	-7
Portugal	5	11	1-	-5	-65	-	-54
Spain	54	-	-	-5	-153	-	-104
United Kingdom	29	71	-	-	-347	-	247

The final table in this series looks at the high income dual-earner family. This family type is of interest in that it is here that we might expect to find clear evidence of horizontal distribution in favour of families with children - which, depending on one's perspective, might alternatively be described as 'middle class welfare'. The results are striking. In two-thirds of the countries, with the exceptions being Australia, Denmark, the Netherlands and the UK, this family benefited from child tax allowances - very substantially in Luxembourg. Universal child benefits were also payable in most countries. In Australia, by contrast, families at this income level were outside the scope of all tax/transfer policies in favour of children and were thus among the ten per cent or so of families with children excluded from the Basic Family Payment. The UK and Ireland, while providing some universal family support, had also withdrawn their income-tested payments by this level of earnings.

Table 6: Structure of the family package for a dual-earner couple with 2 children (7 and 8), at 1.5 average male + 1.5 average female earnings: difference between their income and that of a childless dual-earner couple, £ sterling PPPs per month

	Tax and social insurance	Non income-tested child benefit	Income-tested child benefit	Health costs	School costs	Other	Total
Australia	-	-	-	-18	-	-	-18
Belgium	36	144	-	-5	-	-	160
Denmark	-	70	-	-1	-	-	-69
France	84	63	-	-7	-	-	140
Germany	78	42	-	-	-	-	120
Greece	10	13	-	-1	-	-	22
Ireland	6	38	-	-18	-	-	26
Italy	13	-	-	-7	-66	-	-60
Luxembourg	415	184	-	-2	-	-	597
Netherlands	-	96	-	-40	-	-	56
Portugal	13	22	-	-10	-5	-	20
Spain	19	-	-	-10	-8	-	9
United Kingdom	-	80	-	-	-	-	80

Support for families at the social assistance level

We have looked at the interaction of tax and social security policies for families in work at various income levels. But one of the main causes of poverty among families in all countries is unemployment. The next series of tables (at the back of the paper) provide a summary of the value of means-tested social assistance or minimum income support for families, in purchasing power parities, both before and after housing costs. The assumption here is that families have no other earnings. In countries which have social insurance schemes for unemployment the adults are assumed either not to have acquired entitlement or to have exhausted their entitlement through long-term unemployment. This is not a very representative assumption in some European countries, but it does provide a structural comparison of the 'last resort' safety net. The complexities of social assistance schemes and difficulties in comparisons have been examined elsewhere (Eardley et al., 1996a, 1996b). Assistance rates are not fixed nationally in all European countries and the figures given are estimates of the amounts likely to be awarded to the families in the given locations. Countries where variation and officer discretion are most likely to make a difference are Italy and Spain, while Greece and Portugal do not have general social assistance schemes available to all those in need.

Taking a couple with two school-age children as an example, payments in May 1994 in £ purchasing power parities varied massively, from £35 per month before housing costs in Greece to £1,156 in Luxembourg (Table 7). As before, Luxembourg was an outlier and Australia's payment of £680 per month was otherwise surpassed only by Denmark and, marginally, by Germany. This is perhaps not surprising given the social assistance-based structure of welfare in Australia and the emphasis on targeting. However, unemployment and sole parent payments in Australia also have free earned income areas and tapers which allow for substantially greater additional earnings to be kept than in most European schemes. This means that total disposable income of pensioners and beneficiaries in Australia with access to some part-time work could be relatively even higher than in a number of the EU countries. On the other hand, Australia did much worse after housing (Table 8), because of the limitations of the Rent Assistance scheme compared to the extensive housing benefit schemes in some European countries.

Tables 9 and 10 provide a picture of the implied equivalence scale in social assistance, representing the net disposable income of the different family types as a percentage of that of a couple without children. Pre-school child care costs are not included, but all the other elements of the income package are counted. There was considerable variation in the extra amounts payable for children, but most countries paid sole parents with one child more than a childless couple. The exceptions were Denmark, Ireland, Luxembourg and, particularly, Australia. France, on the other hand, gave sole parents with a child under three substantially more and also stood out as the most generous to couples with two or more children.

Replacement rates of earnings

In the context of high unemployment, both in Australia and in many countries of the EU, it is difficult to look at assistance benefits for people of working age without considering the incentive structures implicit within them. It has often rightly been pointed out that the factors influencing labour market decision making are more complex than simple rational calculations of financial advantage, even if assistance recipients are in a position to make accurate assessments of this kind (see, for example, McLaughlin et al., 1989). There are also difficulties involved in comparing benefit replacement rates cross-nationally, since neither gross nor disposable incomes are independent of the processes of redistribution in different countries (Whiteford 1995b). Nevertheless, such a comparison does provide useful indicators of the incentive or disincentive effects built in to the structure of assistance payments.

Replacement rates are usually calculated by comparing the levels of benefit to some measure of income in work, thus showing what percentage of earnings is 'replaced' by benefits. They can be thus be altered either by changing the rates of benefit or by altering the level of disposable income in work. Several countries have important schemes to provide incentives through in-work benefits, including Family Credit in the United Kingdom, Additional Family Payment (as it was known in 1994) in Australia and Family Income Supplement in Ireland. These in-work benefits are most common in the English-speaking countries with extensive assistance schemes, and are mainly directed to families with children, although in a significant move the UK has begun the piloting of an income supplementation scheme for people without children (Earnings Top-up).

Table 11 compares the level of social assistance payable as a proportion of the net disposable resources of the same families at half average male earnings, before housing. This is an indicator of the replacement rate for social assistance on the assumption that long-term unemployed people receiving assistance are likely to be in the market mainly for lower-paid employment. The extent to which this represents a common experience depends to some extent on how important social assistance is for unemployed adults. It also depends on how common it is for people to be in full-time employment at this level. In Australia, for example, the half-average male earnings point would not be common in practice for full-time workers in manufacturing because most Award-based wages would be above this level. Also, in countries like Denmark, where dual-earner couples are the norm, it might be more realistic to base the replacement rate on earnings by both partners.

Replacement rates varied by family type, but for couples with children, before housing costs they were particularly high in Denmark, Luxembourg, Germany and the Netherlands, and in all except the latter some families would be better off on social assistance than in work at this earnings level. Australia has managed to maintain relatively low replacement rates (before housing costs at least) both by paying income-tested in-work family payments which extend some way up the income scale and by having relatively high average earnings.

The extra income effect of second earners

Given that dual earning among married couples with children is becoming increasingly common, the net income effect of having a second earner is of considerable interest in providing an indication of the employment incentive for partners of the main earner. Table 12 shows the proportionate increase in net disposable income resulting from second earnings for three family types at two earnings levels, before housing costs.

Table 12: The income effect of second earners: extra disposable income of dual-earner couples as a percentage of that of one-earner couples, before housing

	Average male + 0.66 female earnings			1.5 average male + 1.5 average female earnings
	Couple no child	Couple + 1 child (under 3)	Couple = 1 child (7)	Couple no child	Couple + 1 child (under 3)	Couple + 1 child (7)
Australia	49	43	50	80	68	72
Belgium	46	43	43	63	51	61
Denmark	45	30	43	63	51	61
France	43	39	43	67	68	68
Germany	39	32	37	55	50	54
Greece	45	44	44	60	59	59
Ireland	35	11	35	54	35	53
Italy	51	28	48	78	65	81
Luxembourg	56	39	52	52	33	50
Netherlands	46	15	44	72	45	70
Portugal	28	32	47	69	50	67
Spain	38	18	38	52	38	52
United Kingdom	50	12	47	69	42	66

The first part of the table shows that in Luxembourg, as a result of the interaction of taxes and transfers, extra earnings received by a female partner at two-thirds of the average female wage increased net family income by 56 per cent, compared with only 35 per cent in Ireland. This is in spite of the male/female wage relativity being much more favourable in Ireland than in Luxembourg. At 49 per cent, the extra earner incentive in Australia compared reasonably well against most EU countries. The child care effect is evident again in the second column, which shows that in countries with high net pre-school day care costs the second earner incentive was much reduced.

It is particularly interesting to observe the effects of marginal taxation at the higher earnings level in the right-hand part of the table. Here the net benefit of second earners was greatest in Italy, the Netherlands and Australia - somewhat surprising effect for the first two countries given that they have amongst the lowest levels of female labour force participation in the industrial world, though one that supports a similar analysis for 1992 by Shaver and Bradshaw (1995). Belgium stood out as having an apparently rather low incentive for extra earnings at this level, whether families had children or not. The Australian result stems from the couple being able to take advantage of separate taxation, reducing the combined marginal rate, as the man's income was just below the top tax threshold. At this income level no spouse tax rebate was allowable.

Discussion

As was stated earlier, there are limitations to the model family technique. While it provides a basis for understanding comparatively the complex interactions of policies affecting families in different countries, it is nevertheless describing implicit policy structures rather than actual outcomes for families. Model family results have to be qualified or set in context by looking at the actual distribution of family types at different earnings levels and in different circumstances. For example, most sole parents in Australia are not in the full-time work force. The fact that their incomes on social assistance are higher in purchasing power terms than in a number of European countries does not prevent them from being poor, if poverty is measured by incomes as a proportion of the national median or against a community benchmark like the Henderson poverty line.

Nevertheless, the overall picture suggests that structurally at least Australia is continuing to maintain one of the most progressive tax/transfer systems among the countries observed. Horizontal redistribution between family types and towards those with children is by no means insignificant but is weaker than the vertically redistributive impact of assistance targeted to those on low incomes.

On the other hand, comparative evidence also suggests that Australia has one of the highest levels of child poverty in the OECD (Rainwater and Smeeding, 1995). There are a number of technical problems with the approach taken in such studies which may tend to exaggerate the poor performance of Australia. Nevertheless, it is hard to argue that poverty among families with children is not a serious problem in this country. This suggests that it may the 'quantum' of redistribution available under a low-tax, low expenditure welfare system which is insufficient.

The significance for within-family distribution of resources of the direction of most family payments to women as the main carers of children should not be underestimated. Since May 1994, a number of further changes have been introduced, including the 'cashing out' of the dependent spouse rebate into first Home Child Care Allowance and then Basic Parenting Allowance, paid to the main carer with little or no income in their own right but not tested against their partner's income. The couple rate of unemployment allowances has also been split and the couple income test for unemployment and sickness allowances partly individualised. These changes reflect changes in assumptions about the dependence on women on men, while modifying the progressivity of social security payments towards the family as a unit. The addition of the child care cash rebate gave a further boost to support for working parents. On the other side of the equation, the introduction of compulsory superannuation has introduced a new element into the calculation, which increases the level of redistribution across the life cycle, bringing Australia closer to some of the European countries.

Under the 1996 Budget, progressivity has been further modified by the introduction of the Family Tax Initiative, which gives a per child addition either through the tax system or as a cash payment to all families with incomes below around $70,000 per year, plus a flat rate extra payment to families with only one earner - including sole parents but clearly also aimed at moderate income 'male breadwinner' families. This complicates the structure of family support and represents a policy shift away from full-scale targeting, while still providing a significant increase in assistance to low income families. On the other hand, support for child care is being reduced, through an end to direct subsidies for Centres and tighter limits for Child Care Assistance.

The impact of all these changes can be modelled in comparison with changes in other countries using the model family method and further work is planned along these lines. An important question to examine for Australia is whether it can maintain the relative effectiveness of its tax/transfer system for families without a commitment to the necessary revenue and expenditure.

References

Australian Bureau of Statistics (1994) Average Weekly Earnings: States and Australia, May 1994, Catalogue No. 6302.0, AGPS, Canberra.

Bradshaw, J., J. Ditch, H. Holmes, and P. Whiteford (1993) Support for Children: a Comparison of Arrangements in 15 Countries, DSS Research Report No. 21, HMSO, London.

Bradshaw, J., S. Kennedy, M. Kilkey, S. Hutton, A. Corden, T. Eardley, H. Holmes and J. Neale (1996) Policy and the Employment of Lone Parents in 20 Countries, Family Policy Studies Centre, London.

Bradshaw, J. and D. Piachaud (1980) Child Support in the European Community, Bedford Square Press, London.

Cass, B. (1996) 'A family policy 1983-1995', Just Policy, No. 6, May, pp.16-25.

Castles, F. (1985) The Working Class and Welfare: Reflections on the Political Development of the Welfare State in Australia and New Zealand, 1890-1980, Allen and Unwin, Wellington.

Castles, F. (1996) 'The institutional design of the Australian welfare state', paper presented to the Reshaping Australian Institutions Workshop held as part of the Research Committee 19 (Poverty, Social Welfare and Social Policy) of the International Sociological Association, Annual Meeting, 19-23 August 1996, Australian National University, Canberra.

Castles F. and D. Mitchell (1991) Three Worlds of Welfare Capitalism or Four?, Australian National University Discussion Paper 21, ANU, Canberra.

Ditch, J., H. Barnes, J. Bradshaw, J. Commaille and T. Eardley (1995a) A Synthesis of National Family Policies in 1994, European Observatory on National Family Policies, Social Policy Research Unit, University of York, York.

Ditch, J., J. Bradshaw and T. Eardley (1995b) Family Policies in the European Community: Trends and Developments in 1994, European Observatory on National Family Policies, Social Policy Research Unit, University of York, York.

Eardley, T. (1996) 'Lessons from a study of social assistance schemes in the OECD countries', pp. 51-62 in L. Hantrais and S. Mangen (eds.) Cross-national Research Methods in the Social Sciences, Pinter, London.

Eardley, T., J. Bradshaw, J. Ditch, I. Gough and P. Whiteford (1996a) Social Assistance in OECD Countries: Synthesis Report, DSS Research Report No. 46, HMSO, London.

Eardley, T., J. Bradshaw, J. Ditch, I. Gough and P. Whiteford (1996b) Social Assistance in OECD Countries: Country Reports, DSS Research Report No. 47, HMSO, London.

Esping-Andersen, G. (1990) Three Worlds of Welfare Capitalism, Polity Press, Cambridge.
Kahn, A. and S. Kamerman (1978) Family Policy: Government and Families in 14 Countries, Columbia University Press, New York.

Kahn, A. and S. Kamerman (1983) Income Transfers for Families with Children, Temple University Press, Philadelphia.

Kamerman, S. and A. Kahn (1989) 'Single parent, female-headed families in Western Europe: social change and response', International Social Security Review, No. 1/89, 3-34.

Lewis, J. (1992) 'Gender and the development of welfare regimes', Journal of European Social Policy, 2, 3, pp.159-173.

McLaughlin. E., J. Millar and K. Cooke (1989) Work and Welfare Benefits, Avebury, Aldershot.

Mitchell, D., Harding, A. and Gruen, F. (1994) 'Targeting welfare - a survey', The Economic Record, 210, September, pp.315-340.

Norris, K. and M. Wooden (1996) The Changing Australian Labour Market, Commission Paper No. 11, Economic Planning Advisory Commission, AGPS, Canberra.

Rainwater, L. and T. Smeeding (1995) Doing Poorly: The Real Income of American Children in a Comparative Perspective, Luxembourg Income Study Working Paper No. 127.

Raskell, P., J. McHutchinson and R. Urquhart (1994) Decomposing Income Inequality Change in Australia, Centre for Applied Economic Research, University of New South Wales, Sydney.

Saunders, P. (1994) Welfare and Inequality: National and International Perspectives on the Australian Welfare State, Cambridge University Press

Saunders, P. (1996) The Development of Indicative Budget Standards for Australia: project Outline and Research Methods, Budget Standards Working Paper No.1, Social Policy Research Centre, University of New South Wales, Sydney.

Shaver, S. and J. Bradshaw (1995) 'The recognition of wifely labour by welfare states', Social Policy and Administration, 29, 1, 10-25.

Whiteford, P. (1995b) The Use of Replacement Ratios in International Comparisons of Benefit Systems, Social Policy Research Centre Discussion Paper No.54, SPRC, University of New South Wales, Sydney.

Whiteford, P. (1995a) 'Families, benefits and taxes: support for children in a comparative perspective', Social Security Journal, June, pp.49-86.

Whiteford (1997 forthcoming) 'Targeting Welfare: a Comment', The Economic Record.

Whiteford, P. and J. Bradshaw (1994) 'Benefits and incentives for lone parents: a comparative analysis', International Social Security Review, 3-4/94, pp.69-89.

Figure 7: Net disposable income for a sole parent with 1 child (aged 7), after housing.

Table 7: Net Disposable Income for All Family Types on Social Assistance Before Housing, £ppps Per Month

	Single	Couple	Couple + 1 (2 years 11 months)	Couple + 1 (7)	Couple + 2 (7, 8)	Couple + 3 (7, 8, 14)	LP+1 (2 years 11 months)	LP+1 (7)	LP+2 (7, 8)
Australia	287	518	599	599	680	787	426	426	507
Belgium	312	414	473	480	572	703	486	493	587
Denmark	317	566	812	801	836	886	516	506	610
France	191	271	354	357	456	555	420	307	393
Germany	323	486	585	585	690	790	509	509	632
Greece	0	0	17	17	35	53	17	17	35
Ireland	242	387	492	492	565	641	355	355	437
Italy	120	239	341	335	440	546	239	233	330
Luxembourg	515	782	909	946	1156	1429	641	678	887
Netherlands	372	531	510	568	620	654	563	522	574
Portugal	158	199	205	204	231	239	210	209	213
Spain	184	209	245	247	286	319	224	227	265
UK	198	311	422	438	522	641	332	348	432

Table 8: Net Disposable Income for All Family Types on Social Assistance After Housing, £ppps Per Month

	Single	Couple	Couple + 1 (2 years 11 months)	Couple + 1 (7)	Couple + 2 (7, 8)	Couple + 3 (7, 8, 14)	LP + 1 (2 years 11 months)	LP+1 (7)	LP+2 (7, 8)
Australia	137	295	386	380	398	516	213	213	225
Belgium	152	224	283	290	359	489	296	303	374
Denmark	216	426	647	636	658	735	421	411	515
France	99	141	217	220	350	480	321	208	288
Germany	184	307	407	407	489	589	331	331	431
Greece	-102	-102	-113	-113	-125	-107	-113	-113	-125
Ireland	226	362	464	464	535	607	336	336	415
Italy	67	133	235	229	282	387	133	127	171
Luxembourg	313	528	654	691	848	1120	387	423	578
Netherlands	281	438	516	475	521	555	469	428	476
Portugal	100	142	148	146	174	181	153	151	155
UK	198	311	422	438	522	641	332	348	432

Table 9: Implied Equivalence Scale of Social Assistance, Net Disposable Income as Percentage of that of a Childless Couple, Before Housing Costs

	Single	Couple	Couple + 1 (2 years 11 months)	Couple + 1 (7)	Couple + 2 (7, 8)	Couple + 3 (7, 8, 14)	LP+1 (2 years 11 months)	LP+1 (7)	LP+2 (7, 8)
Australia	55	100	115	115	131	152	82	82	98
Belgium	75	100	114	116	138	170	117	119	142
Denmark	56	100	143	142	148	157	91	89	108
France	70	100	131	132	168	205	155	113	145
Germany	66	100	120	120	142	163	105	105	130
Greece*	-	-	117	117	135	153	117	117	135
Ireland	63	100	127	127	146	166	92	92	113
Italy	50	100	143	140	184	228	100	97	138
Luxembourg	66	100	116	121	148	183	82	87	113
Netherlands	70	100	115	107	117	123	106	98	108
Portugal	79	100	103	103	116	120	106	105	107
UK	64	100	136	141	168	206	107	1123	139

* People without children receive no social assistance, but zero payment for couple is treated as index of 100 for comparison with other family types.

Table 10: Implied Equivalence Scale of Social Assistance, Net Disposable Income as Percentage of that of a Childless Couple, After Housing Costs

	Single	Couple	Couple + 1 (2 years 11 months)	Couple + 1 (7)	Couple + 2 (7, 8)	Couple + 3 (7, 8 14)	LP+1 (2 years 11 months)	LP+1 (7)	LP+2 (7, 8)
Australia	46	100	131	129	135	175	72	72	76
Belgium	68	100	126	129	160	218	132	135	167
Denmark	51	100	152	149	154	173	99	96	121
France	70	100	154	156	248	340	228	148	204
Germany	60	100	133	133	159	192	108	108	140
Greece*	-	-	-	-	-	-	-	-	-
Ireland	62	100	128	127	146	164	93	93	115
Italy	50	100	177	172	212	291	100	95	129
Luxembourg	59	100	124	131	161	212	73	80	109
Netherlands	64	100	118	108	119	127	107	98	109
Portugal	70	100	104	103	123	127	108	106	109
UK	64	100	136	141	168	206	107	112	139

* People without children receive no social assistance.

Table 11: Replacement Rates: Net Disposable Income on Social Assistance as a Percentage of Net Disposable Income on Half Average Earnings, Before Housing

	Single	Couple	Couple + 1 (2 years 11 months)	Couple + 1 (7)	Couple + 2 (7, 8)	Couple + 3 (7, 8 14)	LP+1 (2 years 11 months)	LP+1 (7)	LP+2 (7, 8)
Australia	51	85	85	85	87	88	58	46	59
Belgium	65	36	83	83	84	87	92	92	91
Denmark	74	52	149	149	147	146	101	92	93
France	41	25	63	73	78	69	68	58	59
Germany	70	39	95	95	103	104	82	78	80
Greece*	-	-	6	6	11	9	6	6	11
Ireland	57	36	74	74	77	79	78	56	63
Italy	29	21	67	68	87	105	54	51	69
Luxembourg	74	43	111	113	122	127	84	81	93
Netherlands	75	40	99	98	99	98	98	88	89
Portugal	71	32	93	93	101	103	102	93	93
Spain	44	19	58	58	66	72	81	53	60
UK	39	22	64	66	70	74	97	50	56

* No social assistance for people under retirement age without children.