In this paper I focus on a neglected aspect of Australian political history, the extent to which Australian governments actually redistributed income. The German sociologist Rudolf Goldscheid argued that 'the budget is the skeleton of the state stripped of all misleading ideologies'.^[1] In Australia a party that claimed to represent lower income earners, the Labor Party, was a major political force, but did Labor actually make a difference to the distribution of income across social classes, or did Labor's rhetoric of equity merely serve to incorporate workers into the capitalist system? A quantitative approach to the political history of labour may enable us to escape both nostalgia for old labourism (which the Howard years have encouraged) and a simple and undifferentiated rejection of labourism as a reformist agent of social integration.

     This paper incorporates some material from a 2005 paper that examined overall expenditure patterns and taxation patterns across the states and Commonwealth from 1910 to 1940 but it goes beyond the aggregate approach of this paper to consider the extent which the varying patterns of taxation and public expenditure across Australia impacted on different social classes during the 1930s.^[2] It is very much a preliminary analysis based on existing compilations of taxation statistics. It is a static analysis and does not consider if nominally redistributive taxation and expenditure patterns might be rendered ineffective by consequent interstate migration.^[3]

Sources

This paper is largely based on four sources; a database of commonwealth and state social expenditure prepared by Andrew Podger in 1975^[4]; information on tax rates and levels in the 1930s reports of the Commonwealth Grants Commission (CGC); the 1933 census^[5]; and a series of compilations of state and federal fiscal statistics prepared by Alan Barnard in 1985-86. These publications have been relied on for the comparison of taxation levels.^[6] Barnard and Podger relied largely on the reports of the CGC and the Finance Bulletins of the Commonwealth Bureau of Census and Statistics. I have checked some apparently anomalous figures in Barnard and Podger against these sources and have found them to be correct. One particular problem is the distinction between current and capital expenditure, roughly that financed by taxation and that financed by loans. The analysis in this paper is restricted to current expenditure.

Class and public finance

Since the 1980s little academic attention has been given to the history of Australian fiscal policy in Australia. Until then we could very broadly summarise the historiography of the topic as a shift from a view that regarded party politics as significant in the determination of fiscal policy towards one that emphasised broader economic and ideological determinants. Until the late 1950s commentators identified Labor as the party of initiative in social policy due to its reformist ideology and low-income base.^[7] From the 1950s this came under attack from two flanks. Henry Mayer argued that the thesis of Labor's progressivism was not based on any comparative analysis of the divergent performance of Labor and conservative governments.^[8] From the late 1950s Noel Butlin and his colleagues developed a model of the Australian political economy that challenged the initiative/resistance model. The rise of the welfare state and the decline of public loan-financed infrastructure were seen as an Australian example of global patterns of modernisation. Butlin's later work was more influenced by public choice theory but continued to downplay the role of party politics.^[9]

     Another argument came from those who emphasised the influence of policy paradigms that overlapped conventional political divisions. Francis Castles' influential model of the wage-earners welfare state focused on the distinctiveness of Australia compared to other countries, he described Australia (and New Zealand) as a 'wage-earners' welfare' state that sought to ensure full employment by tariffs and fair income by wage regulation. The result was a laggard development of the Australian welfare state. This model distinguished social democracy from labourism. Social democracy was identified as redistribution via taxation and direct government payments, whereas labourism was defined as redistribution by economic regulation, in particular tariffs, arbitration and loan-financed public infrastructure construction (which boosted employment and wages for unskilled workers). Labourism was seen as focusing on the rights and conditions of male wage earner, social democracy was seen as directed towards a broader constituency of the disadvantaged, including those outside the paid workforce, such as women and children. Feminist critics incorporated the Castles thesis in their critique of labour masculinism. They tended to attribute social policy advances to the cross-party tradition of social liberalism.^[10]

Methods of class analysis

Academic discussion of class in Australia has displayed a tension between a model that opposes a middle class to a working class, defined in terms of occupation, and a Marxist model that identifies the class position of individuals by their relation to the means of production.

     My approach seeks to combine these two approaches. I identify the following class groups: employers; Own Account or self-employed workers ('own account' was the census terminology); manual workers (designated 'workers); non-manual workers (designated 'employees); and pensioners.

     A theoretical justification for the separation of manual and non-manual workers could be found in an analytical Marxist approach that would point to their different possession of educational and organisational assets or a poststructuralist position that would stress the multiple determinants, cultural and ideological as well as economic of class identity.^[11] A majority of the Australian population were wage and salary earners and their dependants. It is clear, however that most white-collar workers did not identify with the working class. Their levels of union membership and Labor voting were far below that of manual workers. If we treat all wage and salary earners as a group then this means that we are passing up the opportunity to analyse redistribution between employees and workers. In this analysis we are also taking a different approach from a traditionalist Marxist approach in that we are considering capitalists as the recipients of class income rather than as bearers of the logic of capital.^[12]

Quantifying class

In this analysis I have simply accepted the census numbers for employees, own account workers and pensioners. The task of estimating the numbers in the categories of worker and employee is much more difficult. A major problem is that the only breakdown of the census of respondents by occupation included many employers and own account workers as well as recipients of wages and salaries. For this paper I have defined workers as wage and salary recipients (including the unemployed formerly employed in these categories) in the census industry categories of Fishing & Trapping, Agricultural, Pastoral & Dairying, Mining, Industrial, Transport & Communication, Personal & Domestic Service and Entertainment, Sport and Recreation. Employees I define as those wage and salary earners in the remaining industrial groups of Commerce & Finance and Public Administration & Professional. The industrial divisions do not precisely correspond to the manual/non-manual division but the anomalies largely cancel out. Those respondents in the 'No industry' category, principally unemployed school-leavers have been proportionally allocated across these categories. Most of the unemployed were recorded in their original Industrial sectors and I have simply included them in their appropriate group. Those census respondents who were not allocated to an industry were overwhelmingly 'dependants', mostly married women and children not in the workforce. Feminist critics have accurately observed that a substantial number of women classified as 'dependants' were actually unpaid helpers in family businesses or even engaged in significant economic activities of their own, such as taking in lodgers.^[13] A research project I am currently engaged in will develop more accurate estimates of class comparison but in this paper I have simply allocated all dependents proportionately across the categories of employers, own account, workers and employees.

     The variation in the portion of employers from state to state is largely driven by the size of the agriculture sector, as farmers were the largest group of employers. Queensland was the most working-class state, with the big man's frontier of the pastoral industry.^[14] NSW was close behind but its working class contained a larger portion of industrial and mining workers hence the more radical character of NSW Labor. The relatively low representation of workers and the high number of employees in Victoria sheds light on the question of Victoria's alleged political conservatism.

     The extent to which Labor was politically successful in different states did not simply reflect the weight of the working-class in the electorate. The fact that Labor held office less often in NSW than Queensland is evidence of this. These results can be compared to an earlier, more detailed analysis of the class composition of the NSW electorate in 1933. This analysis attempted a more precise matching of occupational groups to the categories of workers and employees, in particular the inclusion of shop assistants as workers, and took account of family structures in the allocation of dependents. This found about 53% of the electorate to be working class. This bolsters the general reliability of the estimates in Table 1, particularly as workers' representation in the electorate was reduced because many more under 21 than for employees.^[15]

Patterns of public expenditure

From 1910 to 1940 per capita levels current of Australian social expenditure in 1911 prices rose 208% substantially more than Butlin's estimate for GDP growth in 1911-39 of 48%.^[16] The introduction of aged pensions, the maternity bonus and repatriation services saw the Commonwealth take a leading position in overall social expenditure, in particular cash payments to individuals, but the states remained dominant in the delivery of in-kind services, particularly education and health and NSW from 1927 introduced widows' pensions and child endowment.

     The story of social expenditure at the state level is largely one of NSW innovation associated with Jack Lang's first government and Victorian parsimony. A major difficulty in the comparison of the states is that of unemployment relief. Once the immediate fiscal crisis of the early 1930s was past governments reduced direct unemployment relief (paid from recurrent expenditure) in favour of relief works financed by loan expenditure which largely substituted for the public works programs of the 1920s, which had paid award wages.^[17] An accurate determination of how generous governments were towards the unemployed would require a careful comparison of public works expenditure from the 1920s to the 1930s. Unemployment relief expenditure was sometimes arbitrarily shifted between loan and capital accounts. A comparison of state expenditure, both loan and revenue, per unemployed person, reveals more about the greater success of some states in returning to the loan market, but in the early 1930s when states were largely shut out of the loan market NSW was the most generous in unemployment relief.

     There are three notable periods of upheaval, first the post-war inflationary boom that increased wage costs (then maintained at this higher level by industrial arbitration) combined with the influenza epidemic, and then the emergence of NSW as a social policy leader in the late 1920s followed by the upheavals of the Depression.^[18]

Allocating social expenditure by classes

Redistribution by the tax/transfer system depends on three factors; the progressivity of the taxation system; the distribution of government cash payments; and the absolute level of government in-kind and cash payments. When pre-tax and transfer incomes are unevenly distributed then equal per capital social expenditure benefits will have redistributive impact.

     A functionalist Marxist might argue that state human services expenditure does not benefit its recipients as it merely serves to sustain class relations. The extent to which public education operated as a form of class streaming might support this, on the other side of the ideological fence a public choice theorist might argue that public expenditure was the tool of special interests and acted as form of redistribution from workers to employees. Public sector wages were higher in NSW due to a shorter working week and union-friendly reforms to the state arbitration system. Conservative governments did not entirely reverse workers' gains from Labor governments. In 1933-34 the CGC found that average male public service clerical wages in NSW were about 20% higher than the all states average.^[19] However in this paper I have assumed that public expenditure on health and education is a real benefit to citizens. I have allocated expenditure on the following basis. First, pensions and reparation. With the exception of NSW widows pensions these were overwhelmingly a Commonwealth responsibility. This expenditure is simply allocated per capita among the census category of Pensioners as identified in the census. Unemployment relief is excluded. Second, in-kind government services, which are nearly all health and education. Most of this expenditure is by state governments but there were small federal programs. The latter I have simply divided between the states in the ratio of their population. In this period government cash payments were largely restricted to pensioners, compared to the present this made the task of redistribution among social classes more difficult, but with an uneven distribution of income even equal in-kind receipts from government services made an impact. To estimate the impact of in-kind services we must first estimate the total income received by each social class.

     The 1933 census asked respondents to place their annual income within one of seven ranges. For most of their ranges I have assumed that all respondents received the midpoint of the income range, but have followed Ian MacLean and Sue Richardson's estimates of average income levels in the lowest bracket and their estimate, produced from national income statistics, that the average income in the open-ended bracket above £260 above a year was £656.^[20] This enables us to estimate income levels for employers, own account and wage and salary earners for each state. Unfortunately the income levels for wage and salary earners by industry groups are only available at the national level. The national totals show clear income discrepancy between workers and employees. The high-income levels for females in the fishing and trapping, mining and forestry groups are for a tiny handful of white-collar administrative staff and government regulatory staff in these areas.

     I have assumed that across each state the incomes of wage and salary earners by industry group are the same ratio as nationally and then adjusted their absolute levels so as to match the overall levels for wage and salary earners recorded in each state. From this it is possible to estimate the total levels of income received by each class in each state and the proportion of total income received by each class. The resultant estimates highlight the fact that workers received the largest share of overall income and that employees received more than employers. The varying levels of income received by employers and own account workers are predominantly driven by the size of the agricultural sector, but there are signs of Victorian exceptionalism apparent in the relative income shares of workers and employees. From our knowledge of the total incomes received by each class it is then possible to estimate the extent to which social expenditure increased class incomes. Because we have assumed that social expenditure has an equal per capita benefit than it has a similar impact on different social classes but a greater proportional benefit to workers because their income is lower.

Class and taxation

The low level of government cash payments meant that most of the redistributive burden was borne by the tax system. The most substantive exposition of Labor taxation policy came from Queensland railways minister James Larcombe in a 1927 pamphlet. He defended the principle of 'ability to pay' and argued that in practice this meant steeply progressive tax rates and exclusion of as many wage earners as possible from income tax. The policies of the first generation of Labor governments were consistent with this approach: they increased thresholds, especially for heads of families and introduced steeply gradated tax schedules.^[21] These Labor governments faced conservative Legislative Councils, but the Councils were generally reluctant to challenge Labor governments on taxation policy, governments could always spend and force conservative acquiescence in tax increases to balance the budget. Fear of Council obstruction may have made Labor governments reluctant to put more radical options on the table, but overall public finance was probably an area of more legislative autonomy for Labor governments. The low levels of taxation and expenditure in Victoria where Labor only rarely formed government supports this hypothesis. It is difficult to establish any clear connection between Labor governments and levels of taxation, although Victoria and Queensland tended to be at opposite ends of the spectrum.

     However Commonwealth grants meant that the smaller states did not have to rely on taxation alone. In 1940 7.7% of South Australia total revenue, 17.5% of Western Australian total revenue and 17.4% of Tasmanian total revenue was made up of Commonwealth grants.^[22] Commonwealth grants generally contributed towards an equalisation of the resources available to state governments. The rationalisation of Commonwealth grants that followed on the creation of the Grants Commission encouraged convergence between the states. From the early 1930s state taxation levels and the total of taxation and Commonwealth grants (which represented the total of state fiscal resources) converged notably.

     The absence of significant cash payments to distinct groups apart from pensioners limited the redistributive potential of public expenditure. Dramatic increases in social expenditure, as in NSW, had to be financed. NSW Labor and conservative governments increased taxation despite complaints from unions and business groups.^[23] Rob Watts in his analysis of federal Labor's fiscal record in the 1940s suggests that the practical requirements of fiscal policy trumped Labor's rhetorical commitment to taxation equity.^[24]

     In the early 1930s the collapse in economic activity and the costs of unemployment relief forced governments to extend the tax net. This fiscal crisis ended Labor's effort to exempt workers from taxation. A federal Labor government that introduced a national sales tax in 1931 despite the party's theoretical hostility to indirect taxation. In the early 1930s all states increased income taxes or established special unemployment relief taxes that drew most wage earners into the income tax net for the first time and which employed methods of deduction at source.^[25] Divergences still persisted in the treatment of higher-income earners with Queensland imposing the highest rates at the top of the spectrum. However the ultra-rich are a small minority and very high tax rates on them will not raise much revenue. Such taxes are a form of populist political symbolism.^[26] One contemporary commentator suggested that the even distribution of wealth in Australia meant that Labor governments had to be cautious on taxation.^[27] The discrepancies between income reported in the census and that declared to taxation commissioners, varying tax rates across the states, the wide variety of income tax deductions and the fact that most state level income tax was actually paid by companies make any comprehensive estimate of the taxation liabilities on different social classes an arduous task. In this paper I can only report on a first instalment of this project. Table 5 is based on an application of the tax rates described in the 1934 CGC report to the average class incomes per head estimated from the 1933 census. There are discrepancies between income reported in taxation statistics and the census. As a result, at this stage, I would place more value in a comparison of the values in the table rather than their absolute values, but it does suggest that NSW's high expenditure meant higher taxation on workers than Victoria's parsimony. ^[28] A next stage of the analysis will be to estimate the incidence of company taxation and to combine the estimates of the class impact of taxation derived from the census and CGC reports with the state taxation statistics.^[29]

Conclusion

By the late 1930s there were clear divergences in the levels of social expenditure between the states but the low levels of cash payments meant that the redistributive impact of social expenditure was limited. Higher expenditure meant a broader casting of the tax net rather than a concerted attack on capital, but it could be argued that within a capitalist economy Labor governments showed a realistic appreciation of the limits on taxation policy.