Appendix 2: Estimation with an Alternative Real Wage Measure | RDP 8806: Employment, Output and Real Wages

RDP 8806: Employment, Output and Real Wages Appendix 2: Estimation with an Alternative Real Wage Measure

Bill Russell and Warren J. Tease

September 1988

Download the Paper 974KB

This appendix reports the results of estimating the preferred equation using real labour costs instead of real unit labour costs as the explanatory variable.

Table 1
Variable	Coefficient	Standard Error
a₀	0.55	0.31
lnY_t	0.14	0.03
lnW_t	−0.07	0.01
InE_t−1	0.81	0.04
T	−0.0004	0.0002
RBAR² = 0.99	SEE = 0.005	SSR = 0.0015
h = 1.58	Q(24) = 27.15

The variables of interest are significant and of the right sign. The main difference between the results is that the coefficients on the real labour cost term and the time trend are lower. (The significance level of the trend has also fallen.)

A potential problem with this result is possible collinearity between the trend term, real labour costs and GDP. Both real labour costs and GDP have a strong upward trend for most of the period. To reduce this problem the trend term was dropped from the equation, yielding the results in table 2.

Table 2
Variable	Coefficient	Standard Error
a_o	1.17	0.36
lnY_t	0.10	0.02
lnW_t	−0.08	0.02
lnE_t−1	0.79	0.063
RHO	0.24	0.14
RBAR² = 0.99	SEE = 0.005	SSR = 0.0015
h = −0.006	Q(24) = 22.65

The parameters on output and the real wage are now the same order of magnitude. The implied elasticities are both low relative to results reported earlier and to recent Australian studies. This is probably due to some remaining collinearity between GDP and real labour costs. (Examination of the correlation matrix suggests that this is the case.) This problem does not arise when real unit labour costs are used because they do not exhibit the strong upward trend of real labour costs (and output).

The important point to note from these estimations is that the qualitative relationship between employment and real wages is not dependent on the choice of a real wage variable. When the trend term is eliminated, real labour costs and output have a similar (though opposite) effect on employment. This was the case when real unit labour costs were used as the explanatory variable.