This paper estimates the range of equilibrium rates of unemployment for Australia. The estimation technique nests a unique equilibrium rate of unemployment as a special case. It is found for the period 1965^97 that a range of equilibria of at least 6.6 percentage points of unemployment exists in Australia. The lower limit of this range, which is the minimum rate of unemployment consistent with non-increasing in£ation, was 2^3 per cent in the 1960s, jumped in the early 1970s and was about 5.6 per cent during the 1990s. I IntroductionA number of papers have argued that a range of equilibrium rates of unemployment is more realistic than a unique equilibrium (or natural) rate of unemployment. For example, McDonald (1995) reviews evidence covering a number of countries and time periods, including, most dramatically, the interwar years. More recently Crosby and Olekalns (1998, p. 124) conclude that their estimates of the in£ation process in Australia are consistent with the idea of a range of equilibria.With a range of equilibrium rates of unemployment, a region of low rates of unemployment in which in£ation tends to be increasing is separated from a region of high rates of unemployment in which in£ation tends to be decreasing by a region of intermediate rates of unemployment in which in£ation tends to stay constant. 1 In this paper, the region of low rates of unemployment below the range is called the peak, referring to the peak of the business cycle, and the region above the range is called the trough. However, even although empirical evidence for the existence of the range has been put forward, no formal estimation of the size of the range has yet appeared in the literature. This paper attempts to provide such an estimate for Australia. The estimation technique nests the unique equilibrium as a special case. It is found that, for Australia for the period 1965^97, a range of at least 6.6 percentage points of unemployment exists. The lower limit of this range, which is the minimum rate of unemployment consistent with nonincreasing in£ation, was 2^3 per cent in the 1960s, jumped in the early 1970s and was about 5.6 per cent during the last three years of the period of estimation, i.e. during 1995^97. During the entire estimation period, 1965^97, the actual rate of unemployment was never above the range, i.e. the Australian economy was never in the trough. Wells and two referees for comments on earlier drafts. 1 The constancy of the rate of in£ation in the range is at a given rate of unemployment. Other in£uences, such as changes in the rate of unemployment, can have one-o¡ e¡ects on the rate of in£ation.

This paper investigates the potential of models of the range of equilibria to yield a pattern of equilibrium rates of unemployment that is consistent with the observed mix of cycle with no trend in the rate of unemployment. To do this the paper considers a model in which wages are determined by a group of unionised insiders bargaining with an employer. In the model the supply of labour is endogenous. Speci®cations of the utility function are found which are consistent with the mix of cycle and no trend in the rate of unemployment. Of these it is concluded that loss aversion in the evaluation of wage relativities is a crucial ingredient for models of the range of equilibria. I. I n t ro d u c t i o nThis paper investigates the potential of models of the range of macroeconomic equilibria 1 to yield a pattern of equilibrium rates of unemployment that is consistent with the observed mix of cycle with no trend in the rate of unemployment. The presence of a cycle and the absence of a trend are evident in the pattern of rates of unemployment across industrialised economies. 2 In models of the range of equilibria, a cycle in the rate of unemployment can be explained by cyclical movements in the demand for labour. 3 These cyclical movements can be caused by movements in aggregate demand, that is by the Keynesian mechanism. 4 But movements in the demand for labour will also be caused by the trend factors of capital à Financial support was provided by the Australian Research Council. 1 Models of the range of equilibria include models based on customer markets in the selling of goods and services, see Negishi (1979), McDonald (1987, on sectoral spending asymmetries, see Dixon (1988), on loss aversion in wage bargaining, see Bhasker (1990) and on holdout in wage bargaining, see Holden (1997). It has been argued that models of the range of equilibria offer an improved understanding of the relation between in¯ation and unemployment, see McDonald (1995). Lye, McDonald and Sibly (2001) estimate a range of equilibria model for Australia for the period 1967 to 1997. They ®nd that the range model dominates the (nested) unique equilibrium model and that the size of the range is at least 6.8 percentage points of unemployment. 2 Layard, Nickell and Jackman (1991, pp. 1±5) and Lindbeck (1994, especially chapter 2, pp. 8±9) present evidence for cycles but no long run trend. 3 In the Dixon (1988) model movements in the rate of unemployment are caused by changes in the composition of aggregate demand. 4 In the real business cycle literature cyclical movements in labour productivity cause cyclical shifts in the demand for labour.