It is well established in the literature that Australian unions raise their members' wages relative to those of otherwise comparable non‐members by some amount in the range 7–15 per cent. However, it is also known that firm size is positively associated with union density and that firm size is positively associated with relative wages independent of unionism. Using a new, large and rich set of data from the Training and Education Survey 1993 we show that the estimated union relative wage effect is largely comprised of bias due to the omission of firm size as a variable in the wage equations usually estimated.

No abstract

made useful suggestions on an earlier draft. The normal caveat applies. Different levels of measured skills, geographic location and demographicfactors (such as marital status and country of birth) explain almost none of the hourly wage differences of Australian women and men in full-time employment. The major contribution to wage differences is apparently in the different returns paid by employers to men and women for observable characteristics. Usually this is considered as evidence for the existence of direct wage discrimination by employers but—at least for the data of this study—some questions remain as to the extent of this influence. Measurement issues related to both schooling and general labour market experience tend to exaggerate the role of direct employer discrimination, but even extreme assumptions as to the extent of mismeasurement of these variables do not eliminate such discrimination (although it is reduced from 13 to 3.5 per cent). Some part of the overall difference is likely to be a consequertce of (unmeasured) occupation factors and, of much greater contplexity, the possible influence of role-stereotyping and its interaction with family arrangements. This paper investigates the determinants of the male-female wage differential. Until recently little has been known of the determinants of wage differences between women and men in Australia. Broadly based investigations have identified average wage differences by sex (Power 1973), and several analyses have used regression techniques to ascertain the influence of particular variables (Jones 1982;Chapman & Miller 1983;Chapman 1984; Haig 1982). This research, however, typically suffers from problems of specification and inadequate data. Consequently, strong conclusions concerning the nature and extent of so-called 'discrimination' are premature at this stage, as are calls for particular policy initiatives to remedy the situation. This paper reports an empirical examination of an extremely rich and so far unutilized data set that provides new information on the origins of wage differences. A commonly used reduced-form econometric framework is employed to determine the separate contribution of a host of variables to average wage differences between men and women. The novelty and usefulat DEAKIN UNIV LIBRARY on March 15, 2015 jir.sagepub.com Downloaded from

No abstract

Two predictions of the exit/voice model of union activity that have been confirmed by empirical research in the United States are that union workers will have longer tenures and lower quit rates than nonunion workers. This study replicates the methods used in one important U.S. investigation of these “voice” effects to explore whether the same effects are apparent in Australia. The authors' analysis of data on young men from the 1985 Australian Longitudinal Survey yields support for both predictions. Two differences between the Australian and U.S. results are that the union effect on job tenure is considerably larger in Australia than in the United States, and unionism significantly reduces layoff rates in Australia, whereas it has been found to have no significant effect on layoffs in the United States.