Manufacturing Earnings and Cycles: New Evidence

While we have found that standard real wages as well as μ and ʌ are
pro-cyclical, we have yet to ascertain whether or not they are in phase with
the various cyclical indicators. In order to establish the degrees to which
phase shifts are important, we estimated the dynamic correlations expressed
in equation 7. Our procedure was as follows. Indicators were selected if they
displayed significant associations - for one or more lengths of cycle - with a wage
component in Table 3. Thus, for example, correlations with respect to all four
indicators were undertaken in relation to μ while only one correlation (with
respect to GFCF) was estimated for W/Pp. The estimated correlations are
presented in Table 4 and, perhaps surprisingly, they indicate a predominance
of signihcant leads between wage components on the various indicators. Taken
alone, however, these results are misleading. We are able to translate phase
shifts into months using the cycle lengths (labelled ‘periods’ in Table 4) at
which the explained variances of the wage components are maximised (see last
column of Table 1). This is achieved simply by multiplying period × relative
phase × 12. In all cases, these calculations reveal phase shifts of less than
one year. In effect, the variables are contemporaneously related at the annual
frequencies.

22

<<   19   20   21   22   23   24   25   >>

More intriguing information