be said to have cyclical structure of the length specified.
Our findings are as follows. First, the cycle of the premium mark-up, μ,
is predominantly explained within the shortest (3-5 year) range. Second, the
distribution of explained variance with respect to the proportion of overtime
workers, ʌ, is similar to the C,p-deflated wage cycle, although the 5-7 year
range for ʌ, is relatively more dominant than W/Cp. Third, the production
wage has two significant cycles that fall into the 5-7 and 7-10 year ranges, with
the dominant cycle being in the 5-7 year range. Fifth, the production wage
cycle length is the longest followed by the proportion of overtime workers, the
consumption wage cycle and the premium markup.
One or more of the frequency wave ranges shown in Table 1 displays a signif-
icant share of total variance association with each component of the wage. We
selected four economic indicators to represent these ranges. First, gross Fxed
capital formation (GFCF) was chosen to capture the longer end of the time
spectrum.13 Moreover, as we argue in Section 3.2, it is a potentially appealing
cyclical indicator in relation to the production wage. Second, output (Y) and
employment (TV) represent the middle time band. Output- and employment-
based indicators are typically used to represent the business cycle. Finally, the
change in business inventories, labeled inventory investment (II), is taken to
represent cycles of relatively short duration. We argued in Section 3.1 that
this may well provide an ideal indicator of overtime hours cyclicality, the driv-
ing variable in the overtime premium mark-up. Using the same methodology
as Table 1, the results in Table 2 for the four indicators show the following.
With regard to significant contributions to total variance, the cycle lengths in
ascending order are II, N, Y and GFCF (see last column, Table 2). II has
one signihcant cycle in the 3-5 year range. N. Y and GFCF each have one
signihcant cycle in the 5-7 year range. Additionally, GFCF displays a second
signihcant cycle in the 7-10 year range.
13For the G7 countries, GFCF is dominated by a long cycle in the 7-9 years range, with
the exception of a slightly shorter cycle in the US series (Woitek, 1996), which is in line
with the results in Table 2. This phenomenon can be explained by the existence of built-in-
stabilisers in the US economy which are absent in European countries (Romer, 1999).
17