Table 5.6 illustrates the average wage differential between full- and part-time work for both Australian and United States' females. The raw wage differential between full- and part-time workers in Australia is minus 17 percent. For the United States, the raw wage differential between full- and part-time workers is plus 30 percent. That is, in Australia, part-time workers earn 17 percent more per hour than full-time workers. For the United States, part-time workers earn 30 percent less per hour than full time workers. Two questions are apparent from this results. Firstly, why do these differences occur ? Secondly, why are the differences in different directions?
In the previous section, the wage differential is divided into three components. Unexplained differences are estimated to account for a 20 percent increase in full-time wages relative to part-time wages in Australia. For the United States, unexplained differences increased the full-time wage by 32 percent relative to part-time workers. Endowment differences are estimated to have a very small effect on the Australian wage differential between full- and part-time workers. For the United States, endowment
See Lambert (1991) for an analysis of the effect of mismeasurement of experience in a human capital model, and Hirsch and Addison (1986) for a critique on the role of sample selection terms in a wage equations. Kidd (1992) analyses for Canada the effect of incorrectly treating occupational outcomes as exogenous, and the role of Mincer’s experience proxy in the estimation o f the gender wage differential.
TABLE 5.6 : Summary of female wage differential between full-and part-time workers, Australia and the United States (a)________________
Australia United States Observed differential (b)
(In Wft - In Wnt) -0.17 0.30
Decomposition effects :
(1) unexplained differences 0.20 0.32
(2) endowment differences 0.02 0.13
(3) sample selection effects -0.39 -0.15
Notes :
(a) differences in log hourly wages for trichotomous model. The results differ from those reported in Hawke (1992) which utilise the dichotomous model. The implication of this difference in treatment is large differences in the sample selection effects. Since the observed differential is unchanged, the offsetting effect is borne by productivity differences.
(b) full-time work is treated as base for decomposition. See Sloane (1985) for index number problem associated with this decomposition.
differences accounted for a 13 percent wage differential between full- and part-time workers. That is, whilst endowments are relatively unimportant in explaining the wage differences between females working full- and part- time in Australia, they are estimated to be important in the United States. Finally, sample selection effects are estimated to increase the wage of Australian females working part-time relative to females working full-time by 39 percent. For the United States, this effect is in the same direction as Australia, however, the magnitude of the effect is estimated to be 15 percent.
These findings are very interesting. In order to understand the implications of these results, it is useful to recall the discussions of Chapters 2 and 3. Australia differs from the United States in that wages are centrally determined by an industrial tribunal which imposes wage levels and relativities on the Australian labour market. In the preceding chapter
we discussed how these tribunals have determined that part-time workers should obtain wages which are equivalent to the returns to full-time work. Thus, the institutional determination of wage relativities between full- and part-time workers is consistent with the result of an observed raw differential of 17 percent.
Having found that institutions do influence wage outcomes, it would be unreasonable to expect the market not to respond to these effects. In Chapter 4 we analysed one method with which the market may respond - through employment levels. However, as has been found by Gregory et. al.
in the case female wages relative to male wages, an increase in the cost of part-time workers did not lower the demand for these workers, but rather, the number of part-time workers increased. The advantage of this decomposition is that we may gain some insights into why this phenomenon occurred.
As noted earlier, the wage differential between full- and part-time workers does not appear to be explained by differences between the directly observed human capital characteristics of individuals in full- and part-time work23. In terms of the work undertaken in Chapter 2, it would appear that human capital does not have a role in explaining the wage differential between full- and part-time workers. Obviously it is necessary to look at the other components of the decomposition for an explanation of this differential.
Sample selection factors represent the unobserved characteristics of individuals which are thought to influence the wage. One interpretation of these unobserved factors is the quality of an individual. For example, we can measure whether an individual has completed high school, but we are
23 This result is consistent with the findings of Australian studies which analyse the wage differential between males and females working full-time, such as Gregory and Ho (1985) and Chapman and Mulvey (1986)
unable to determine whether this individual is highly motivated or quick to grasp new ideas. In some ways, sample selection may be interpreted as reflecting these unobservable qualities. If this is the case the sample selection result for Australia indicates that the part-time labour market in Australia is adjusting to the institutionally determined wage, not by reducing the number of part-time workers, but by picking the highest quality workers which allow employers to compensate themselves for the imposed wage.
However, these effects have been offset to some extent by the differences in the coefficients of the full- and part-time wage equation. As mentioned earlier, these differences are often thought to result from either discrimination or productivity differences. In the male/female case, many analysts have interpreted this effect as discrimination. If this analysis of full- and part-time workers is identical to the male/female analysis, this result would be interpreted as representing discrimination against part-time workers, in favour of full-time workers, to the extent of a 20 percent wage difference. However, in this case we have the added insights provided by the sample selection term and the analysis from Chapter 4. Thus, given that the sample selection term indicates that the better quality workers are being employed on a part-time basis, it could be expected that better quality workers would be more productive. Therefore, the 20 percent wage difference which results from unexplained factors, suggests that perhaps discrimination is very important, given the high quality of part-time workers.
For United States, where institutions are not as important in the determination of wages as Australia, the wage of full-time workers is estimated to exceed that of part-time workers by 30 percent. This result is consistent with the theories of wage determination discussed in Chapter 3. Unlike Australia, the United States results indicate that the characteristics
of full- and part-time workers do contribute to the wage differential by around 13 percent. That is, the measurable characteristics of full-time workers are, to some extent, better than that observed for part-time workers. This difference results in the wages of part-time workers in the United States being 13 percent lower than their full-time counterparts.
Sample selection differences in the United States do lead to a reduction in the wage differential between full- and part-time workers by 15 percent. As for the Australian case, we may interpret this effect as representing to some degree, difference in the quality of the individuals in full- and part-time work. In the United States, part-time work is sometimes thought of as being 'bad' in terms of the jobs that are undertaken and the workers who are employed. This result suggests that whilst part-time jobs may be considered bad as a result of their hourly wage being lower than full-time jobs, it is not possible to categorise the females working part-time as poor quality workers.
Finally, the unexplained differences between full- and part-time workers contribute to a 32 percent wage difference between these groups. This contribution is larger than the raw differential. As for Australia, it is not possible to ascribe the total of this effect to discrimination, however, since the sample selection contribution is lower than that estimated for Australia, it is plausible that the discriminatory component in the United States case is higher than for Australia.
5.7 Conclusions
Previous chapters have analysed the role of part-time work over time in Australia and the United States. This chapter focussed upon 1986 for Australia and 1987 for the United States in an attempt to examine possible explanations for the wages of part-time workers being around 20 per cent
more per hour than full-time workers in Australia, and approximately 30 per cent less per hour in the United States.
This chapter utilised a human capital model to explain the wages of individuals in full- and part-time work. Factors such as education, experience, occupations, martial status and geographic location are used to explain the wages individuals receive. Additionally, the results from Chapter 4 are included in the specification of the models to account for sample selection effects. The results from the full- and part-time wage equations are then compared. For Australia, the returns to education and experience in part-time work are estimated to exceed the returns to full time work. For the United States, the reverse effect is estimated to occur.
Utilising the estimates from the wage equations for both countries, the wage differential between full- and part-time workers is decomposed into three factors. These factors represented endowment differences, sample selection differences, and finally differences which are unexplained by the model but which may be thought to be influenced by factors such as discrimination, productivity differences or mismeasurement.
The main conclusions from the wage decompositions involve three points. Firstly, human capital does not explain the differences between the wages of full- and part-time workers in Australia to any significant degree, but is an important explanator in the United States.
Secondly, institutions have an important role in the determination of wages in Australia, and the effect of their intervention has been to increase the wage of part-time workers relative to full-time workers. Since previous chapters have illustrated that employers did not respond to this intervention by reducing the role of part-time work in the workforce, further explanation is required. Sample selection is interpreted as reflecting the unobserved characteristics of individuals, and using the results from the trichotomous choice model of the previous chapter, it is estimated that employers adjust
to the higher part-time wage by employing better quality workers. For the United States, whilst the endowment effect indicated that part-time workers did have lower levels of human capital than full-time workers, they could not be categories as poor quality workers.
Thirdly, in both countries a relatively large proportion of the wage difference between full- and part-time workers remained unexplained by the model. To some extent this effect may represent differences in productivity or discrimination between full- and part-time workers. However, other factors may also be important in explaining this wage differential. One of the main influences on wages is occupations, and in the following chapter, the effect of different occupational distributions on the wages of females working full- and part-time is investigated.