RDP 2000-08: Nominal Wage Rigidity in Australia 8. Assessment

The labour market in Australia displays clear features of downward nominal rigidity, for both wages and, to a lesser extent, broader measures of earnings. The importance of this depends, in the first instance, on whether some of the observed rigidity is artificial.

The incidence of zero wage changes may be inflated by the prevalence of long-term contracts. We have focused on the distribution of annual wage changes. Although these changes are calculated over a window that allows for different salary review dates by firms, and typical lags in contract renegotiation, many contracts are for periods of two years or more. These longer-term contracts do not always provide for a wage rise in each year of the contract. Some lead to a wage increase only when they are renegotiated, resulting in relatively infrequent step-wise increases in the level of wages that add to the concentration of observations at zero and positive skewness of the distribution. This pattern of wage setting is a feature of many individual contracts and wage earners with these contracts are likely to be over-represented in our sample. This might lead to estimates of skewness that are biased upwards and overstatement about the extent of downward nominal rigidity.

We may also observe nominal wage rigidity because of the self-selection evident in reported wage changes. We observe only the distribution of accepted wage offers to those remaining in the same job. Some wage offers are not accepted and an employee quits. Since such offers are more likely to be below the median (that is, below the ‘wage norm’), the resulting distribution becomes positively skewed.^[32] There is an additional reason why self-selection might be a source of skewness in our data. Some participants in the survey are seeking information about prevailing rates of pay for a given job, with an intention of offering a wage above the median to retain valued staff. Consequently, if wage changes below the median were truncated by staff turnover, analysing a sample of stayers would lead to estimates of skewness that are biased upwards. Again, this would lead to overstatement about the extent of downward nominal wage rigidity.

So, part of the rigidity we observe is artificial. But when we assess the economic consequences of rigidity, other factors warrant consideration. Foremost, there are ways in which employers may prevent downward nominal wage rigidity from affecting their compensation costs. For example, they may vary working arrangements, in particular the span of working hours that are considered to be standard. Alternatively, they may promote or terminate staff to achieve desired adjustments in their wage bills, especially if wages tend to rise with tenure more so than productivity (Wilson 1999). We have also observed downward nominal rigidity in an environment of positive inflation and sustained growth in productivity. Perhaps it is unreasonable to expect that nominal wage cuts would occur in this environment, other than for firms in distress (Gordon 1996; Poole 1999).

So how can we tell if the rigidity we observe matters? While a growing body of empirical evidence suggests that downward nominal wage rigidity is a pervasive feature of economies, there is much less evidence about the effects of this rigidity. At the micro level, analysis of its effects on layoffs, promotions and relative wage growth is inconclusive (Altonji and Devereux 1999). So too is analysis of its effects at the macro level. There are those, such as Gordon (1996), who find the Phillips curve to be unaffected by rigidity and ‘resolutely linear’, and those who find it to be non-linear, even in the long run (Akerlof et al 1996).

Identifying the effects of downward nominal wage rigidity in Australia warrants a separate inquiry that is recommended for further research. However, there is evidence to suggest that the short-run Phillips curve for Australia might be a curve rather than a line (Debelle and Vickery 1997; Gruen, Pagan and Thompson 1999). This implies that the economy might function less efficiently at zero inflation than it does at the small positive rates of inflation that are consistent with the current inflation target.

Footnote

The effects of self-selection on skewness have been identified by Weiss and Landau (1984). [32]