With the introduction of enterprise bargaining in 1991, decentralised wage
determination in Australia was generally expected to widen the gender wage gap (see
Chapter 3). However, as discussed in Chapter 4, the research that underlies this
expectation is typically based upon aggregated data and suffers from a number of
deficiencies. In contrast, this dissertation utilises unit record data from the extensive
1995 Australian Workplace Industrial Relations Survey (AWIRS95) commissioned by
the former Department of Workplace Relations and Small Business to test the
hypothesis that enterprise bargaining has affected the gender wage gap in Australia.
Whilst the passing of four years between 1991 and the time of data collection may not
have allowed for the landmark industrial relations change to have worked itself through
the labour market, a noteworthy and major feature of AWIRS95 is that it identifies
workers and workplaces operating under enterprise bargaining agreements, as well as
containing a female/male split of the enterprise bargaining status (see Chapter 5).
Along with hourly earnings data derived from AWIRS95, a clear assessment can be
made of gender wage gaps for employees under enterprise bargaining and those not
employed under enterprise bargaining. The thesis uses OLS earnings regressions to
identify the part of any gender wag gap that can be justified by the difference in
measured characteristics between males and females, as well as identifying the part that
remains unexplained (see Chapter 7). Given the potential that workplace characteristics
can affect the integrity of OLS results, a random effects model is also used (see Chapter
8). Interestingly, the OLS and random effects results are virtually identical (see
Chapter 9).
It needs to be noted that the component of the gender wage gap that is unable to be
justified by direct statistical reference to the regression model has been attributed to
discrimination in the labour market. However, this is something of a misnomer as the
unjustified component also captures the impact of:
� model misspecification, including excluded variables;
� mismeasurement; and
� errors of calculation.
111
Every effort has been made to reduce these effects. Nevertheless, there may be an
element of discrimination in the regression model utilised in this thesis that is not
discernible through the observable and measurable variables (see Chapter 3).
Results of analysis undertaken in this thesis indicate that the gender gap, as well as the
unexplained component thereof, are larger for employee data associated with enterprise
bargaining than is the case for workers not so employed. Even so, the result is not
deemed to be statistically significant, as is further supported by extensive sensitivity
testing (see Chapters 7 and 8).
Further research is needed to support the posed hypothesis. Nevertheless, the thesis still
provides a wide range of interesting outcomes in providing a greater understanding of
an observable gender wage gap in Australia, as well as the associated and contributing
characteristics of employees and employers. It is in this capacity that the research work
recorded in this thesis provides a new level of knowledge and understanding,
particularly given the thorough use made of recent microdata and the observed earnings
effects of selected variables. As a consequence, the results of this thesis will form a
solid foundation upon which further gender wage gap debate, policy formulation and
labour economics research can stand.
| Identifer | oai:union.ndltd.org:ADTP/219284 |
| Date | January 2000 |
| Creators | Reiman, Cornelis Adriaan, n/a |
| Publisher | University of Canberra. Management & Technology |
| Source Sets | Australiasian Digital Theses Program |
| Language | English |
| Detected Language | English |
| Rights | ), Copyright Cornelis Adriaan Reiman |
Page generated in 0.0026 seconds