By David Peetz and Alison Preston
First published at Online Opinion , 20 July 2007

In introducing the “WorkChoices” reforms the federal government argued that they would encourage increased wages, particularly through Australian Workplace Agreements (AWAs) which it actively encourages. Indeed, we were told (and still are told) that AWAs pay higher wages than collective agreements and nearly double what awards pay.

We obtained unpublished data from the Australian Bureau of Statistics (ABS) Employee Earnings and Hours (EEH) Survey for May 2006 (released March 2007), to examine whether this was the case: in what circumstances and in what ways do AWAs affect earnings, particularly by comparison with collective agreements, which are actively discouraged by the WorkChoices legislation.

The EEH survey is the most reliable (indeed, the only) source of data on earnings of employees under AWAs. The most representative data in this survey are those concerning average total hourly cash earnings of non-managerial employees.

There are only limited quantitative data on changes in conditions under WorkChoices AWAs published by the Office of the Employment Advocate (OEA). However, data leaked to Mark Davis of the Sydney Morning Herald revealed that in May-September 2006, 68 per cent of AWAs abolished penalty rates (up 26 per cent on 2002-03), 52 per cent abolished overtime pay (up 107 per cent on 2002-03) and 76 per cent excluded shiftwork loading.

They also gave the first indication of the extent to which many conditions have been “modified” through AWAs. For most “protected” award conditions, even among those AWAs that do not abolish that condition, the majority will “modify”, that is reduce it. Hence around nine tenths of AWAs either abolished or reduced penalty rates. Similarly, 88 per cent of AWAs abolished or “modified” overtime rates; 89 per cent either abolished or “modified” shiftwork loading; 82 per cent abolished or “modified” public holiday payments; and 83 per cent abolished or “modified” rest breaks.

In the face of widespread public concern about the loss of conditions under WorkChoices, the federal government on May 4, 2007 announced amendments to WorkChoices, including a new “fairness test” on agreements. However, this test is weaker than the former “no disadvantage” test, covering fewer award conditions. The data in our report probably paint a more positive picture toward AWAs than if the survey had been undertaken in 2007.

One limitation was that most of the wage data are expressed as averages, which can be biased by the inclusion of a small number of observations with very high earnings. Some 69 per cent of AWA employees earn less than average AWA hourly earnings. A more representative indicator of the situation of the “typical” worker is provided by median earnings, and we used these data wherever available.

Some hypotheses

We decided to test some possible hypotheses explaining the use of AWAs.

If AWAs are used predominantly for flexibility to benefit both employees and employers, then they should pretty consistently provide for higher hourly pay than registered collective agreements, across different employer types.

Conversely, if AWAs are predominantly used for cutting labour costs and avoiding unions, we would expect to see wide variations in the relationship between AWA earnings and earnings under registered collective agreements. The highest AWA premiums would be in situations where union avoidance strategies are important, while there would be shortfalls for AWA employees where union avoidance strategies were not important. Where cost-minimisation strategies were preferred, we would expect shortfalls for AWA employees to be most severe among workers with low skills levels or in low demand, highly competitive areas.

At the same time, institutional and market arrangements in each industry and occupation will also influence outcomes. For example, particular occupations within an industry may be traditionally non-union. We had to take this into account in understanding the patterns of earnings by industry and occupation.

Overall patterns

We found the ABS data showed that median AWA earnings were only $20.50 per hour, some $4.00 per hour below median earnings for CA employees. That is, the typical (median) AWA worker earned 16.3 per cent less than the median CA worker.

In 2006 men on median AWA earnings earned 15.4 per cent less than men on median CA earnings. The median earnings for female non-managerial employees on AWAs was 18.7 per cent lower than corresponding median for females on CAs.

Advocates of WorkChoices have repeatedly stated that employees on AWAs earn twice as much as people on awards. Yet median hourly earnings for AWA employees were only 16 per cent above median award-only earnings. For women, median AWA earnings were only 5.6 per cent above median award-only earnings.

Median earnings of non-managerial workers on AWAs grew by 7.9 per cent between 2004 and 2006. This was less than the 9.9 per cent growth in median earnings of workers on CAs over the same period.

Firm size

We found a very stark relationship between firm size and the earnings ratio. In organisations with less than 500 employees, AWAs pay less than CAs. The wage shortfall widens as organisations get smaller. Hence the shortfall is 3.5 per cent among organisations with 100-499 employees, rises to 12.4 per cent in organisations with 50-99 employees, 13.6 per cent in organisations with 20-49 employees and is a very substantial 26.3 per cent in organisations with less than 20 employees.

Among large organisations with more than 1,000 employees (the majority of whom are covered by collective agreements), there is a wage premium for AWAs of 30.8 per cent.

Industry

AWA employees in the majority of industries received a lower average hourly rate than their counterparts did on CAs.

The industry with the highest average wage premium for AWAs (worth 50 per cent) was communication services, the second highest premium was in government administration and defence (33 per cent) and the third highest premium was in finance and insurance (22 per cent). These are all industries with well documented efforts by at least some organisations to use AWAs to reduce union influence, where employees in the more highly remunerated parts of an organisation are hired on individual contracts.

In mining, however, AWA employees earned 3.6 per cent less than CA workers.

AWAs paid on average well below CAs in manufacturing, construction, transport and storage, health and community services; property and business services; and “personal and other services”.

Occupation

For the top three occupational groups, AWA employees earn more on average than CA employees. Professionals are clearly a group with high labour market power.

At the other end of the labour market, labourers and related workers experienced a consistent AWA pay shortfall - their wages were 17 per cent lower than wages of workers on CAs. In all, five of the six lowest occupational groups revealed an AWA pay shortfall compared to CAs.

The most disadvantaged group appeared to be female labourers and related workers - those on AWAs were paid 26 per cent less than similar women on CAs. Indeed, female labourers and related workers on AWAs were receiving 20 per cent less even than the award-reliant average for that occupation.

Implications

There are two reasons why we have found the AWA shortfall to be worse than in previous research. First, previous estimates have relied on data concerning registered individual contracts and have included state registered individual agreements, which pay nearly double the average in AWAs. Second, data on hourly earnings of the typical (median) employee have not previously been available. Average earnings data significantly overstate the earnings of the typical AWA employee.

Overall, our findings support the hypothesis that AWAs are frequently used for cost cutting or union avoidance. Very large firms and federal government departments face the real prospect of unionisation and have the resources and sophistication to mount concerted union avoidance strategies. They can use AWAs as part of that, offering wage premiums to induce workers to sign AWAs and/or financially penalising those who seek to remain on collective agreements by refusing to promote them unless they sign an AWA.

Smaller firms are not likely to follow this approach, as unions find them logistically difficult to organise anyway. So smaller firms are more likely to use AWAs as a cost minimisation tool, presumably through cutting penalty rates, overtime pay and other “protected” award conditions.

The industry pattern was also consistent with this proposition, with AWAs generally paying above CAs in industries where union avoidance strategies are important and below CAs in industries where labour cost minimisation is important. Structural factors within industries also played a role in explaining some industry patterns.

Had AWAs been used as a device for promoting flexibility for the mutual benefit of employees and employers, we would have expected that employees in small and medium firms would have gained in roughly similar proportions to those in large firms, and AWA gains would have been seen across all or at least most industries. This clearly is not the case.

The pattern of earnings by occupation was consistent with the hypothesis that workers with low bargaining power in the labour market arising from low skill levels are most adversely affected by individual bargaining through AWAs, while occupations with high skill and short demand appear able to maintain high wages under AWAs and possibly attract a union avoidance premium in some cases.

The effects of AWAs will vary according to the reason for their introduction and the labour markets in which employees are working. The overall AWA (median) shortfall of 16.3 per cent suggests that cost-minimisation is an important element in AWA strategising, and any “flexibility” benefits that exist are not enough to offset the cost-minimisation effects on wages.

Given the known loss of conditions under WorkChoices AWAs, outcomes for WorkChoices AWAs would very likely be worse, even with the operation of the “fairness test”. The data here understate the gap between AWAs and union CAs, as the CA data include non-union CAs (which have, on average, lower wage increases than union CAs) and are also depressed by the impact of free riders on bargaining power of unionised workers negotiating new CAs.

We have mainly spoken about AWAs and collective agreements, and said little about awards. For most industries and occupations, comparisons between award-reliant workers and workers on agreements are of little value in telling us about the effects of agreements on pay. This is because the award-reliant group is not representative of award rates, being in most industries disproportionately concentrated at the lower end of the award pay structure.

For example, we know that award rates of pay for tradespersons are by definition higher than those for unskilled workers, yet the average earnings of award-reliant tradespersons were lower than the average earnings of award-reliant labourers and unskilled workers, because most tradespersons (except some on the lowest classifications) received over-award payments that took them outside the ABS definition of being “award-reliant”. Hence our interest was principally in comparing employees on AWAs with employees on registered collective agreements.

Still, two related groups of workers (in which a large proportion of workers are award-reliant) give us an insight into the effect of AWAs compared to awards. In hospitality, AWA workers receive an average of 2 per cent less than average award wages. Female casual workers on AWAs received average earnings some 7.5 per cent below average award earnings. These figures suggest that AWAs can often lead to earnings falling below the award average. They also reinforce that individual “bargaining”, through AWAs, is especially detrimental for women, particularly when they lack labour market power.

Overall, AWAs are commonly associated with poorer outcomes for typical employees than registered collective agreements. While AWAs sometimes attract wage premiums, associated with union avoidance strategies, these mainly affect a small number of industries and some very large organisations. Where union avoidance is not a common issue, for example in small organisations, the negative impact of AWAs on earnings becomes very stark. The impact of AWAs is worst for those people without unique skills, who do not have strong bargaining power in the labour market.

David Peetz is the author of Brave New Workplace: How Individual Contracts are Changing Our Jobs (Allen and Unwin, 2006), and Professor of Industrial Relations at Griffith University.

Alison Preston is a Professor of Economics at Curtin University, Western Australia.