The procedure calls for the followiñgaefinitions:
JB = Japanese relative export pr ices of bars in the year t
t
JB = the average of Japanese relative export bar prices of the
years in the sample, i.e.,
JB = 1
16
1976
~
t=196l
JB
t
*
and JB JB =
-B - J t t
*
JB =_(JB - yBi
t t
t £ B
t £ R
where B = the set of all years in the sample which are above
average years of steel aemand and R = the set of all years in
the sample which are below average years of steel demand.
The cyclical dual pricing hypothesis predicts that J is *B
t
positive in boom years and contraction years.
-229-
*
Assuine that (JB
1961
an underlying population
, . .
*
. , JB ) is a random sample
1976
from that is normally distributed
with unknown variance. The cyclical dual pricin\j iiypothesis
predicts that the mean of this population is greater than zero.
A 9cneralized likelihood ratio test for this hypothesis
was employed. This test may be reduced to a t- statistic. 77/
Nine separate tests were performed: one for each of the three
."'''. _.
carbon steel products for each of the three re9 ions.
The results are presented in taole 4.20. The numbers in
the table represent the estimated values of the t-statistic
appropriate for the particular test. A high and positive value
of the t-statistic tends to confirm the hypothesis that the
mean is positive; i.e., that there is cyclical dual pricing.
conversely, negative values tend to disconfirm the cyclical
dual pricing hypothesis.
At significance levels of .10 or lower the data call for
rejection of toe hypothesis of cyclical dual pricing--in all
nine cases. 78/ Significance levels higher than .10 are cons ide red poor tests because of an unacceptaoly hi9h probabil i ty
of accepting the al ternative hypothesis when the null hypothes is
is true. Nonetheless, at the significance level of .25, one
77/ For a more detailed discussion of the procedure see: Hogg
and Craig (8, ch. 10, section 11, and especially Mood, Graybill,
and Boes (18, pp. 428-4291.
78/ The significance level of this test is the probability of
ĂŁCcepting the hypothesis that there is cyclical dual pricing
if in fact there is n6ne.
-230-
would accept the cyclical dual pricing hypothesis for EC bars
and plate, and U.S. bars. This study concludes that the data
do not support the hypothesis that Japan, the European
Community, or the United States prices its steel products in
a cycl ical dual pr icing manner. 22/
Explanation of the Results. The hypothesis of greater
relative export price variability for the Japanese and
Europeans emanated from the assumption that they have a larger
share of fixed costs. This in turn assumed. their labor costs
were fixed and that the Japanese debt-equity ratios imposed
significant fixed costs on the Japanese.
TABLE 4.20
Result of Likelihood Ratio Tests
10r Cyclical Dual Pricing Hypothesis
Estimated values of the t-statistic
Product (carbon steel) Country
Japan E.C. u.s.
Bars -1.72 .742 1.206
Cold Rolled Sheet - .174 .254 .339
Plates - .309 .977 .110
Source: See text.
79/ In fact, the negative t-values for Japan (as displayed in
table 4.20) indicate that, if anything, the cyclical effect
goes the other way for Japan--h igher relat i ve expor t pr ices
in contractions and lower relative export pr ices in expansions.
Again, however, exce~r bars, this effect is not statistically significant at- 90nventional levels.
-231-
However, the proportion of Japanese employees who are
"contract" employees has been rising in the past ten years.
There is now more than one contract employee for every two
regular Japanese steelworkers. 80/ These "contract" employees
are not considered regular employees of the steel firm as they
are employed by a subcontractor.
Thus, the steel firms can,
and in fact do, layoff these workers in recessions.
In add i t ion to the var iabil i ty in labor costs prov ided by
,#'¡; .
contract labor, a new element of Japanese labor cost variability
has appea red since July, 1975. A spec i al form of unemployii;ent
compensation is now available to depressed industries on an
economywide basis since the passage of the Employment Adjustment
Assistance Law. These unemployment compensation funds, known
as "KOyo Chose i Kyufuk in," pay one-hal f the salary of the wor ker
for layoff days in the case of large firms and two-thirds the
80/ The most detailed data available on this subject is compiled
by the Tekkororen Steel Workers Federation in their publication
Rodo HandbOOk (annual in Japanese). Data for the proportion of
workers who are subcontracted on a plant by plant basis for the
major Japanese steel firms are published there. On the basis
of these data it appears that contract labor has significantly
increased in recent years reaching 57 percent of regular
employees (or 36 percent of total employees) by December 1973.
The estimates of the Bureau of Labor Statistics, Office of
Productivity and Technology, have also shown a rise in the pro- portion of Japanese contract employees. Their unpublished estimates are that contract workers represented between 43 and 65
percent of regular employees in 1976 compared with 24 to 36
percent in 1964. (Their analogous numbers for contract workers
as a percentage of total employees are 28 to 37 percent in 1976
and 18 to 25 percent in 1964.)
r".~ .,';.
~:j
The authors' interviews with the major Japanese firms have
also suppor ted these e~t iIDates.
-232-
salary in the case of small firms. The firm pays the balance
of the worker's salary. The government limits the number of
days for wh ich these unemployment compensat ion funds are ava ilable to 75 days per six months; moreover, these funds are
available only in those industr ies designated as depressed by
the Labor minister. In fact, the open hearth sector of the
steel industry has been continuously designated as depressed
since the law's inception in July, 1975 '¡~~'other steel sectors
have generally been included on the list of depressed industries.
Thus, as a result of this unemployment compensation, Japanese
steel firms can assume that a portion of their regular employees'
salar ies are not fixed dur ing recessions.