Florida Steel Corp.

National Labor Relations Board - Board Decisions

Dec 14, 1984

273 N.L.R.B. 889 (N.L.R.B. 1984)

FLORIDA STEEL CORP. 889
Florida Steel Corporation and United Steel Workers
of America, AFL-CIO. Cases 11-CA-5324, 11-
CA-5326, 11-CA-5347, 11-CA-5366, 11-CA-
5405, 11-CA-5429, 11-CA-5444, 11-CA-5455,
11-CA-5776, 11-CA-5943, 11-CA-5946, and
12-CA-6593
, 14 December 1984
SUPPLEMENTAL DECISION AND
ORDER
BY CHAIRMAN DOTSON AND MEMBERS
ZIMMERMAN AND HUNTER
On 10 January 1983 Administrative Law Judge
Stephen J. Gross issued the attached supplemental
decision. The Respondent filed exceptions and a
supporting brief, to which the Charging Party and
the General Counsel filed answering briefs. The
Charging Party also filed exceptions and a support-
ing brief and the Respondent filed an answering
brief. The General Counsel filed exceptions and a
supporting brief to which the Respondent filed an
answering brief.
The National Labor Relations Board has delegat-
ed its authority in this proceeding to a three-
member panel.
The Board has considered the decision and the
record in light of the exceptions and briefs and has
decided to affirm the judge's rulings, findings, and
conclusions' as modified and to adopt the recom:
mended Order.
Prior to 1974, it was the Respondent's practice
to grant scheduled general wage increases based in
part on periodic area wage surveys. In the spring
of 1973, the Union began its organizing activity at
the Respondent's Charlotte and Indiantown
plants. 2 In 1974 the Respondent canceled its sched-
uled wage surveys and withheld annual and quar-
terly wage increases or TEAM 3 increases at these
plants. In a series of cases, 4 the Board found that
' Member Zimmerman would find that the 22 November 1976 letters
to John Wells Jr, James W Ashcraft, and Loren D Theodore do not
constitute valid offers of reinstatement because these employees were re-
quired to take a physical examination Absent the Respondent's discnini-
natory discharge, these employees would have continued to be employed
without their employee status being jeopardized by such a requirement
He would also find the reinstatement offer to Bruce Lemmond invalid for
this reason, as well as the reasons set forth by the judge
2 The Union was certified as the bargaining representative in January
1974 at Charlotte and in June 1975 at Indiantown
3 TEAM stands for The Economic Adjustment Match, a policy insti-
tuted by the Respondent in mid-1974 which involved quarterly rather
than annual wage increases
4 220 NLRB 260 (1975), enfd 543 F 2d 1389 (DC Cu. 1976), 220
NLRB 1201 (1975), enfd 538 F 2d 324 (4th Cir 1976), 221 NLRB 554
(1975), enfd 562 F 2d 46 (4th Cir 1977), 226 NLRB 123 (1976) Member
Hunter was not on the Board at the time of the Board's decisions in these
cases. He notes, however, that the Board's decisions were enforced in the
circuit courts, and he therefore accepts these decisions as the law of the
case
the Respondent violated Section 8(a)(3) and (1) of
the Act when it canceled the surveys and withheld
the wage increases. Accordingly, the Board or-
dered the Respondent to cease its unlawful conduct
and to make whole the employees in the appropri-
ate units for losses in pay suffered by reason of the
Respondent's discriminatory practices. After the is-
suance of the Board decisions, the Respondent re-
sumed granting its scheduled increases. 5 However,
the General Counsel and the Union maintain that
the wages at the ,Charlotte and Indiantown plants
still fell short and have continued to fall short of
the wages that would have been paid had the Re-
spondent continued its pre-unfair labor practice
wage policies. A controversy arose over how best
to determine what the appropriate wage rates
would have been at Charlotte and Indiantown but
for the Respondent's unfair labor practices. None
of the 'parties have argued that wages be recon-
structed by replicating the area wage surveys used
by the Respondent before 1974. At the hearing
before the judge, experts presented various tech-
niques for arriving at appropriate wage rates. As
discussed in more detail in his supplemental deci-
sion, the judge found and we agree that the best
approach to reconstructing the wage rates is by a
"proportional" method.6
Another issue raised by the parties concerns the
length of the backpay period. Contending that
wages at Charlotte 'and Indiantowli continue to fall
short, the General Counsel and the Union urge that
the backpay period continue until this proceeding
is brought to an end by the Board. The Respond-
ent maintains that it has ceased its unlawful con-
duct and since it has "commenced to bargain in
good faith, offered and/or reached agreement with
the Union on interim increases, and implemented
those increases, the underlying unfair labor practice
was no longer being committed" and the backpay
period came to an end. 7 In its exceptions, the Re-
spondent maintains that by December 1982-Janu-
ary 1983 it had already complied with the judge's
recommended Order and therefore is not liable for
backpay during the period that has intervened
since the issuance of the judge's supplemental deci-
sion.
We do not agree with the judge's recommenda-
tiOn that the Respondent continues "to incur back-
pay obligations to its hourly employees in the
5 The Respondent began implementing wage Increases in August 1976
at Indiantown and in April 1977 at Charlotte
6 Insofar'as the decision applies the judge's model to wages after the
date of the judge's supplemental decision, we reject the Respondent's
contention that a "combination index" Method is more appropriate
7 We note, as did the judge, that although the Respondent and Union
reached agreement on intenm wages, neither party waived any nghts
the backpay proceeding
273 NLRB No. 118
890 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
Charlotte and Indiantown bargaining units when-
ever the wage levels of those employees fall short
of the levels that -would obtain, in light of the then
current circumstances, pursuant to the wage poli-
cies in force at Charlotte and Indiantown prior to
1974 and that are still in force at [Respondent's]
other facilities." In concluding that the backpay
period be "open-ended," the judge relied on the
underlying Board cases at 220 NLRB 1201 and 226
NLRB 121 In these cases, the Board ordered the
Respondent to pay backpay "until such time as [it]
institutes and eqiially applies the TEAM. program
to the Indiantown etnployees" 8- and "to notify the
bargaining representative of the imminent imple-
mentation of scheduled TEAM increases and cor-
poratewide fringe - benefits and offer to place said
benefits into effect for unit employees at Charlotte
unless the Union as bargaining representative ob-
jects." 9 We note that the Board's broad orders, es-
pecially at 226 NLRB 123, are based in part on
what it describes as Respondent's "history of unfair
labor practices" and "recent history of other types
of flagrant violations," indicating "a course of un-
lawful conduct taken by Respondent in the service
of designs inimical to the collective-bargaining
process."' ° Although the General Counsel argues
that the Respondent's proclivity to violate the Act
has "continued unabated since the underlying
orders in these cases," evidence of the Respond-
ent's alleged continuing unfair labor practices is not
before us. Further, once the Respondent has met its
backpay obligation, it .would be inappropriate to
hold it liable for backpay whenever wages fell
below the judge's model rates. The determination
of future' wage rates may only be established by
good-faith bargaining between the parties and not
mandated by orders of the Board. ," The purpose
of this proceeding is to remedy the Respondent's'
past unfair labor practice, , not to police indefinitely
the parties' collective bargaining. Once 'the Re-
spondent has complied - with the order set forth in
this supplemental' decision the likkpay period will
have ended. Accordingly, we order the Respond-
ent to reimburse employees for wages at the rates
and schedules established by the judge, including
the period from the date of the judge's supplemen-
220 NLRB at 1204
9 226 NLRB at 124
'° 226 NLRB at 124
" In finding that the Respondent "incurs backpay obligations so long
as its periodic general wage increases at Charlotte and Indiantown fail to
at least equal the increases that [Respondent] would have granted had it
not changed its wage policies at those two plants," the judge noted that
the "parties are free to vary Respondent's general wage Increase policies
through good-faith collective bargaining"
tal decision on 10 January 1983 to the date of this
Supplemental Decision and Order."
ORDER
The National Labor Relations Board adopts the
recommended Order of the adminstrative law
judge and orders that the Respondent, Florida
Steel Corporation, Charlotte, North Carolina, and
Indiantown, Florida, its officers, agents, successors,
and assigns, shall take the action set forth in the
recommended Order.13
MEMBER HUNTER, concurring and dissenting, in
part.
concur with my colleagues in adopting the
judge's decision in all respects, including his formu-
la for determining backpay. 1 In adopting the for-
mula, I note that the General Counsel and the
Charging Party took no exception to the formula,
and that the Respondent only excepted to the
extent it applied to any backpay liability post 1982.
In fact, the Respondent states in its brief that it has
implemented the judge's model rate and has made a
final payout of backpay, "which the Regional Di-
rector is now preparing to audit."
While I agree with the adoption of the judge's
formula and other findings, I do not agree with my
colleagues' decision to cut off the Respondent's
backpay liability as of the date of this decision. The
Respondent here contends that the backpay period
terminates when the Respondent attains rates in
conformance with the judge's model since, at that
point, the Respondent will have made its employ-
ees whole for its past violations. I find merit in the
Respondent's contention, and I would cut off the
backpay period as of the date the Respondent
shows it had attained the wage rates established by
the judge's model. At that time, the Respondent
will have restored the status quo ante and this pro-
ceeding should therefore terminate. After the date,
the Respondent should be free to establish rates
and other terms of ,employment after giving the
Union an opportunity to bargain in' goodfaith.
" In finding that the Respondent is liable for backpay for the period
between the date of the judge's decision and this supplemental decision
and order, we reject the Respondent's contention that the backpay period
ended in December 1982—January 1983 when it allegedly complied with
the judge's recommended Order and made a "final payout" The backpay
proceeding cannot be closed until a final determination by the Board
Any reimbursement of backpay by the Respondent will be noted at com-
pliance and applied to its backpay
13 Although we have modified the judge's holding to limit the Re-
spondent's backpay liability to the date of compliance, including the
penod from 10 January 1983 to the date of this Supplemental Decision
and Order, we do not find it necessary to change the judge's order
Thus, I join Chairman Dotson in adopting the judge's finding, for the
reasons stated by the judge, that the Respondent made valid offers of re-
instatement to John Wells Jr, James W Ashcraft, and Loren D Theo-
dore
FLORIDA STEEL CORP 891
As noted before, the Respondent asserts that it
has already made the requisite payout in accord-
ance with the judge's formula and that payout is
now being audited by the Regional Director. How-
ever, the Respondent's assertion is not sufficient to
establish that it has in fact complied with the
judge's 'decision. I therefore leave it to compliance
to determine whether the Respondent has already
satisfied its backpay obligation, or whether it is still
not in compliance in some respects. Accordingly,
for the reasons stated above, I respectfully dissent
from my colleagues' cutting off of backpay as of
the date of this decision.
SUPPLEMENTAL DECISION
STEPHEN J. GROSS, Administrative Law Judge. The
purpose of this proceeding is to determine the amount of
backpay due various persons employed or previously
employed by Respondent Florida Steel Corporation
(FSC or the Company) at its Charlotte, North Carolina
and Indiantown, Florida facilities. The FSC backpay ob-
ligations that are at issue here arise out of a series of
Board orders,' the impact of which will be discussed in
subsequent parts of this decision. Twenty days of hearing
have been held (in Charlotte, Atlanta, and Winston-
Salem). 2 Briefs have been filed by the General Counsel,
by FSC, and by the Charging Party, United Steelwork-
ers of America, AFL-CIO (the Steelworkers . or the
Union).3
While the parties have settled numerous backpay mat-
ters, a considerable number of others remain to be re-
solved here. In terms of dollar amounts at issue, the prin-
cipal remaining matters concern a change in FSC's prac-
tices regarding general wage increases for its hourly paid
Charlotte and Indiantown employees. Those general
wage increases issues will be considered in parts I
through VII of this decision. All other issues will be con-
sidered in part VIII.
I. WAGE RECONSTRUCTION—BACKGROUND AND
BACKPAY PERIOD
As of early 1973, none of FSC's employees were rep-
resented by a union. But in May of that year, the Steel-
workers filed a petition with the Board seeking certifica-
tion as representative of the production and maintenance
employees at the Company's Charlotte facility. (An elec-
tion was held; the Charlotte employees voted to be rep-
resented by the Steelworkers; and in January 1974 the
1 214 NLRB 264 (1974), 220 NLRB 260 (1975), enfd 543 F 2d 1389
(DC Or 1976), 220 NLRB 1201 (1975), enfd 538 F 2d 324 (4th Cir
1976), 221 NLRB 554 (1975), enfd 562 F 2d 46 (4th Cir 1977), 221
NLRB 1008 (1975), enfd in part 551 F 2d 306 (4th Or 1977), 226 NLRB
123 (1976)
2 Hearings were held as follows in Charlotte on February 10 through
12, 1981, in Winston-Salem on March 9 through 12, in Atlanta on May 4
through 7, and again in Winston-Salem on May 27 and 28, June 30, July
1 and 2, and August 5, 6, 18, and 19, 1981 In addition, a preheanng con-
ference was held on January 15, 1981, in Washington, DC before Judge
Joel Harmatz
3 FSC also filed a "Limited Reply Bnef and a "Supplemental Limited
Reply Brief" The Steelworkers filed a reply to FSC's Supplemental Lim-
ited Reply Brief
Board certified the Steelworkers as the Charlotte em-
ployees' exclusive bargaining representative.)
Starting at least as far back as 1966 FSC granted
annual general wage increases at each of its facilities. In
keeping with that practice FSC had granted a general
wage increase to its Charlotte employees in February
1973 and had planned to grant another such increase
about the same time in 1974 But FSC officials changed
their plans and did not grant the Charlotte employees
any general wage increase in 1974. The admitted basis
for that change was the Steelworkers representation peti-
tion FSC reacted similarly to another Steelworkers rep-
resentation petition, filed in September 1973, covering
the production and maintenance employees at FSC's In-
diantown, Flcirida _plant. Again, FSC refrained from
granting a wage increase that it would have granted but
for the Union's presence.
(In mid-1974 FSC began a policy of granting general
wage increases on a quarterly, rather than annual, basis:
FSC called the new policy "TEAM"—an acronym for
The Economic Adjustment Match. The first wage in-
creases at Indiantown that the Company would have im-
plemented but for the Union's activity there would have
been TEAM increases. At Charlotte, as indicated above
the first time FSC refrained from granting a general
wage increase the Company was still following a yearly
increase policy at nonunion plants. Subsequently, howev-
er, FSC withheld quarterly—TEAM—increases from' the
Charlotte employees that it admittedly would have-,im-'
plemented if there had been no union activity there.)
In a series of four cases the Board concluded that
FSC's withholding of annual and TEAM wage increases
from the Charlotte and Indiantown employees violated
the Act. The Board accordingly ordered FSC to cease
its unlawful practices and to make the affected employ-
ees whole for the losses of pay they suffered by reason
of such practices.
A. The Parues'Positions About the Backpay Period
The pages ahead will discuss how best to "recon-
struct" the wage rates that FSC' would have implement-
ed for its hourly employees at Charlotte and Indiantown
had FSC not changed its wage increase policies there.
But a preliminary issue must first be decided. What is the
length of the period for which FSC wages must be re-
constructed.
The problem is this. FSC failed to give any wage in-
creases to employees in its Charlotte and Indiantown
bargaining units until after issuance of the Board deci-
sions finding FSC's withholding of wage increases to be
violations of the Act. While FSC did resume implement-
ing wage increases, the General Counsel and the Steel-
workers contend that FSC's Charlotte and Indiantown,wages still fell short, and have continued to fall short, of
the wages that FSC would have paid had it continued its
pre-unfair labor practice policies. The General - Counsel
and the Steelworkers accordingly urge that the backpay
period continue until this proceeding is finally brought to
an end by the Board.
FSC urges otherwise, on three grounds. First, FSC
long ago resumed granting wage increases to the Char-
892 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
lotte and Indiantown employees (April 1977. at Charlotte;
August 1976 at Indiantown) And "nothing in the
Board's orders suggested that backpay would continue to
accrue beyond the time that Respondent actually ceased
and desisted from unlawfully withholding wage in-
creases." 4 Second, since as far back as the mid-1970s
FSC has been negotiating with the Steelworkers about
numerous matters, including wages. In particular, FSC
implemented "interim" wage increases at Charlotte and
Indiantown on, respectively, January 27, 1980, and De-
cember 23, 1979, after concurrence by the Union. Ac-
cording to FSC: "Once Respondent commenced to bar-
gain in good faith, offered and/or reached agreement
with the Union on interim increases, and implemented
those increases, the underlying unfair labor practice was
no longer being committed" and the backpay period
came to an end 5 Third, an "open-ended" backpay
period "would mean either that the pay rates "established
for the future would be -highly speculative or that, in
order to avoid this, additional back pay proceedings fea-
turing new data, revised models, and more expert testi-
mony would be required."6
B. The Board Orders
The First Charlotte Wage-Increase Case. 7 The first case
concerning FSC's failure to implement general wage in-
creases involved FS's withholding of an annual in-
crease in early 1974 at Charlotte. Administrative Law
Judge Thomas S Wilson concluded that FSC had there-
by violated the Act, After noting that FSC's wage in-
creases at its other -facilities "averaged in 1974 about 40
cents per hour" (and for other reasons spelled out. in his
decision), Judge Wilson ordered FSC to pay to each
Charlotte employee: "a sum of money equal to that
which he would have earned on and after February 1,
1974, at an increase of 40 cents per hour for the hours
worked thereafter . . ."8
The Board affirmed Judge Wilson except for the 40-
cent-per-hour figure he had selected:
•
In view of the record evidence that Respondent
does not give the same general wage increases at all
its 11 facilities, and its concern for maintaining a
competitive position in each labor market, we shall
• not set a specific hourly sum. Accordingly, we shall
Order Respondent to make whole all those employ-
ees in the appropriate unit at Charlotte for any loss
of 'pay they may have suffered by reason of the dis-
crimination practiced against them by Respondent's
refusal to give them a general wage increase for
1974. If. Respondent and the Union cannot agree on
the amount of the general increase and/or the retro-
active _date, it shall be left to the compliance stage
of these proceedings to determine such amounts and
4 FSC Br at 15
5 Id at 18
6 Id at 22-23
220 NLRB 260 (1975), enfd 543 F 2d 1389 (DC Cir 1976)
220 NLRB at 268 and 269
the proper date, or, if agreement cannot there be
reached, to a backpay proceeding.9
The Indian town Wage-Increase Case." The Board
found that FSC had unlawfully failed "to institute the
new quarterly wage review policy [i.e., TEAM] at In-
diantown." It accordingly ordered FSC to "make whole
all those employees in the appropriate unit at Indiantown
for any loss of pay they may have suffered by reason of
the discrimination practiced against them by Respondent
with respect to wages during 1974 and until such time as
Respondent institutes and equally applies the TEAM
program to the Indiantown employees."'
The Second Charlotte Wage-Increase Case. 12 This case
involved, among other things, FSC's refusal to imple-
ment TEAM wage increases at Charlotte. The Board or-
dered FSC to "make whole all those employees in the
appropriate unit at Charlotte for any loss of pay they
may have suffered by reason of the discrimination prac-
ticed against them by refusing to implement the new
quarterly wage review policy "3
The Third Charlotte Wage-Increase Case. 14 By late
1975 it was thus reasonably clear that the Board intended
to require FSC to apply its TEAM policy to its Indian-
town facility and that FSC would incur additional back
pay obligations to its Indiantown employees, without fur-
ther proceedings, until it did so, and did so with wage
increases that were commensurate with the amounts FSC
would have paid had it not stopped its TEAM increases
at Indiantown But the Board's Charlotte orders read dif-
ferently, apparently referring only to particular past
wage increases not implemented by FSC.
That led the General Counsel to seek reconsideration
of the Board's second Charlotte order on the ground that
the order "limited monetary recovery to the pay raises
employees would have received from August 15, 1974 to
April 19, 1975, when the complaint issued, .without or-
dering ongoing implementation of the TEAM policy of
quarterly wage reviews and increases, if applicable "15
By the time all the pleadings were in, the Board had ex-
plicitly before it the General Counsel's request for an
"open-ended remedy" and FSC's contention_ that "it
would be inappropriate to require [FSC] to grant the
TEAM wage increases and benefit improvements to the
Charlotte employees who are represented by a certified
bargaining agent because said increases are mandatory
subjects of bargaining 9,16
The entire Board opinion must be read to. get its full
flavor. But to summarize, the Board, after focusing on
"Respondent's history of unfair labor practices" that in-
dicated "a course of conduct Inimical to the collective-
bargaining process," continued:
9 220 NLRB at 260
10 220 NLRB 1201 (1975), enfd 538 F 2d 324 (4th Cir 1976)
11 220 NLRB at 1204
12 221 NLRB 554 (1975), enfd 562 F 2d 46 (4th Or 1977)
75 Id at 555
14 226 NLRB 123 (1976)
15 226 NLRB at 123 (quoting the Board's descnption of the General
Counsel's motion)
16 226 NLRB at 123
FLORIDA STEEL CORP. 893
• . . we hereby clarify our order by providing that
the Respondent's obligation to . implement the
TEAM wage policy corporatewide, which obliga-
tion was specifically made retroactive, continued
during the period that has intervened from the date
of the original Decision (November 20, 1975) until
the date of this Supplemental Decision. In addition,
Respondent in the future is to notify the bargaining
representative of the imminent implementation of
scheduled TEAM increases and corporatewide
fringe benefits and offer to place said benefits into
effect for unit employees at Charlotte unless the
Union as bargaining representative objects.' 7
C.' Duration of Backpay Period—Conclusion
As I read the Board's orders in 220 NLRB 1201 (In-
diantown) and 226 NLRB 123 (Charlotte), FSC incurs
backpay obligations so long as its•periodic general wage
increases at Charlotte and Indiantown fail to at least
equal the increases that FSC would. have granted had it
not changed its wage policies at those two plants. The
only limitation on that obligation-is one inherent in the
Act: "The parties are free to vary" FSC's general wage
increase policies "though good-faith collective bargain-
ing. "1 8
To that extent FSC's backpay obligation is - indeed
"open-ended," as stated by the General Counsel in the
pleadings leading to the Order in 226 NLRB 123
As for the increasingly "speculative" nature of the re-
constructed wage rates as the years move past us," that
was an argument appropriate for raising in the proceed-
ings leading to the 226 NLRB 123 Order. Given that
Order, it is beside the point here. On the other hand, the
length of the backpay period is a factor warranting con-
sideration when deciding what kind of formulation to use
in reconstructing the wage 'rates that FSC would have
implemented but for its unlawful actions.
Finally, there is the factual question of whether the
Steelworkers has agreed that periodic general wage in-
creases at Charlotte or Indiantown should be a product
of collective bargaining rather than of FSC's corpor-
atewide policies regarding such increases. The record is
clear on this point FSC and the Union have discussed
every general wage increase that FSC has implemented
at Charlotte and IndiaiitoWn since the Steelworkers' was
certified at those facilities And the Steelworkers- has
agreed to each of the -increases. But the Union has
always done so-
. . . with the , understanding that such wage in-
creases would be on an interim basis. Interim basis
was understood by both. parties [i.e., FSC and the
.Steelworkers] to mean (1) that neither party was
foreclosed from bargaining for additional wage in-
creases nor (2) was either party in any way waiving
any legal rights that they may have or legal argu-
ments they may have concerning the backpay pro-
ceednik which we are currently involved in.2°
" Id at 124
12 226 NLRB at 124
12 See text at fn 6, supra
20 Stipulation at Tr 524
What that adds up to is that the Union has at no time
agreed that FSC may at either Charlotte or Indiantown,
cease following its companywide policy regarding peri-
odic general wage increases
In sum, as of the date of this decision FSC's backpay
period has not ended. Moreover, absent agreement with
the Union to the contrary or further action of the Board
that revises the Board's Orders of October 1975 and Sep-
tember 1976 (220 NLRB 1201 and 226 NLRB 123), FSC
will continue to incur backpay obligations to its hourly
employees in the Charlotte and Indiantown bargaining
untis whenever the wage levels of those employees fall
short of the levels that would obtain, in light - of the then
current circumstances, pursuant to -the wage policies in
force at Charlotte and Indiantown prior to 1974 and that
are still in force at FSC's other facilities.
II. FSC'S WAGE SETTING PRACTICES
FSC has plants at Ft Lauderdale, Ft Myers, Indian-
town, Jacksonville, Miami, Orlando, and Tampa, Florida;
Atlanta, Georgia, Charlotte and Raleigh, North Carolina;
and Aiken, South Carolina. Since at least the mid-1960s
FSC has used the process described below to set wage
levels at all, of its plants except the two at Charlotte and
Indiantown. As to those two. (1) it stopped using that
process at Charlotte in 1973; and (2) it used that process
at Indiantown only from 1970 (when the Indiantown fa-
cility first went into operation) until early 1974.
A. Wage Structure
FSC has analyzed each hourly paid job at each of its
facilities in terms of various "factors," including the
"education," "experience," and "initiative" needed to
handle the job; the responsibilities the job entails, and the
working conditions associated with the job. 21 Each job
is assigned points for each factor. The greater the de-
mands of the job, and the worse the working conditions
associated with it, the greater the number of points
given. Thus, for example, a job that requires great skill
and at the same time confronts the employee with haz-
ardous and disagreeable working conditions is assigned a
high total point count. Determination of what grade to
assign to any given job is based on the job's point count.
The more points, the higher the grade and the better
paid the job
B FSC's Wage Surveys
As touched on earlier, except at those plants where
the employees are represented by a union, FSC has
granted general wage increases at least annually. FSC
starts its wage-setting procedures with surveys of the
wages paid by other' employers. It does so since—"we're
interested in determining the . . . value of our jobs for
recruiting and retentiOn purposes relative to what other
companies are paying, that we would likely attract em-
ployees from, or that we might lose , employees to."22
2i See, eg, GC Exh 31 Hereafter I will use "GCX," "RX," and
"CPX" when citing exhibits Introduced by, respectively, the General
Counsel, FSC,. and the Steelworkers
22 Tr 1805 (FSC witness Hogue)
894 DECISIONS OF NATIONAL LABOR -RELATIONS BOARD
Since FSC sets wages on a plant-by-plant ,basis, each
survey focuses on employers in the locality of the plant
in question. Generally that means companies within
normal commuting distance of the FSC plant.
The process for conducting the initial survey in an
area is different from the process associated with subse-
quent surveys in that same area. If the survey is the ini-
tial one, company officials select a "representative
group" of "significant" companies that are likely to at-
tract the same kinds of persons as employees as does
FSC. 23 Not all of the selected companies will necessarily
agree to participate in the survey. As to those employers
which do participate, one of FSC's officials will visit
each of them to do the following:
(1) Obtain information on such matters as overall pay
practices, fringe benefits, and the like.
(2) Learn what kinds of jobs the employer utilizes and
"select from those jobs representative jobs."
(3) Determine the point count of each type of "repre-
sentative" job in the same way that FSC evaluates jobs
in its own facilities.
(4) For each separate type of job (at each employer in
the "representative" group), determine (a) the number of
employees holding that job and (b) their average wage.
FSC updates its surveys on a more or less annual basis.
If a new employer enters the area, FSC may add it to
the snrvey. If a participant in the previous year's survey
pulls out . of the survey, a replacement company may be
selected. And if an employer in the previous year's
survey creates a new kind of job, that may be added. In
all of those cases one of FSC's officials will visit the
plant to get the necessary information. Otherwise up-
dates are conducted hy telephone, with the surveyor
asking about changes in wages; about changes, if any, in
working conditions associated with previously, surveyed
jobs, and about changes in the number of employees in
each kind of job.
C. Wage Practice Lines
FSC focuses on three kinds of data obtained from the
surveyed employers: the point count of the various jobs;
the wage of each of those jobs (using the average wage
for each kind of job at each participating employer); and
the number of employees in each type of job. But those
data are not useful in their raw form. As Graph I [omit-
ted from publication] indicates, there will inevitably be a
broad range of wage levels for jobs having similar point
counts. 24 Moreover while a higher point count is often
associated with a higher wage, jobs of a particular point
count at one employer may pay more than the jobs of a
higher point count at another employer (as Graph I also
shows).
FSC accordingly processes the data to determine their
"central tendency." To do , that FSC uses a statistical
technique called "linear regression analysis.'.' 23 In graph-
23 Tr. 1806-07 (witness Hogue)
24 Graph I is taken from a 1980 survey of 16 employers located in or
near Indiantown, Florida See part V,B, below
25 "[R]egression analysis is a statistical technique for trying to quantify
the relationship between a variable that you are interested in predicting
and one or more variables that you would hope to use in making the pre-
diction' Tr 971-972 (GC witness Clapper) "Regression analysis can be
ic terms, the output. of linear regression analysis is a
straight line that is placed so that, relative to any other
straight line, it minimizes the total distance between it
and all of the various "observations." 26 Graph II [omit-
ted from publication] shows the regression line superim-
posed on the observations plotted in Graph I. As so used
by FSC, the regression line may be referred to as an
"area , wage curve" (in the Company's terminology), or,
more generally, a "wage practice line "27
Where regression analysis is appropriate for use, the
technique can provide a wealth of precise data That will
be more fully discussed in part IV, below. For purposes
of FSC's wage-setting purposes it need only be noted
that, subject to various exceptions, if the observations
used to plot the wage practice line were drawn from a
representative sample any further observations are as
likely to be above the line as below it. In that respect, a
wage practice line represents the "average" wage at any
given point count.
Once the regression line is drawn FSC can readily de-
termine how the wage rate for each of its grade levels at
its facility compares to the average wage in the commu-
nity for comparable jobs (assuming the validity ' of the
sample and of the assumptions underlying -the regression
formula). All that need be done is. First, determine the
dollar equivalents of the intersections of the regression
line and the job point counts that represent the midpoint
of each grade at the facility (see Graph III [omitted from
publication]); and second, Compare those dollar figures
with existing wages at the facility. -
•
D. How FSC Determines the Amount of Its General
' Wage Increase
Testimony of FSC Witnesses. FSC grants wage in-
creases as soon as possible after completion of ' the
survey, generally within a few weeks. As for the amount
of the increase, FSC witnesses testified that it is compa-
ny policy to set wages at a level equal to 104 percent of
the level implied by the wage practice line But that 104-
percent standard is adopted only where there are no
countervailing considerations. Such considerations ,gener-
ally are related to employee recruiting or retention prob-
lems being experienced by the facility in questiOn. -De-
pending on the, nature of the problem, FSC ,may. set
wages for its lowest grade employees at levels substan-
tially higher than those indicated by the wage practice
line, and the wages for , its highest level employees at
used to -find the equation of a line which , best describes the relationship
betweeti two vanables " GCX-76 at 3 (In this case Job point count
would be the vanable used to "predict" the wage level of a job Thus
point count would be the "independent" or "predictor" variable, wage
rate would be the "dependent" or "criterion" variable See Tr 974) •
26 Graphically an observation here is a point representing the job point
count and wage of one employee Actually the regression equation mini-
mizes the, total of the squares of the distance, measured vertically, be-
tween each • observation and the regression line (A statistician would
speak of minimizing the "errors" or the "residuals" or,of "minimizing the
sum of the squared deviations from the regression- line ") Thus regression
analysis is also termed "least squares" analysis
22 See GCX 39 and Tr 2225 (On that latter page FSC witness
Haworth defines a wage practice line as "the linear regression estimated
formula for wages as a function of the points involved in the jobs that are
included in the sample.") " . ,
FLORIDA STEEL CORP 895
levels below the wage practice line, or FSC may adopt
exactly the reverse course; or wages at both the upper
and lower ends of its structure may be set well above the
wage practice line, with middle level jobs at the wage
practice line levels.
Under any of these approaches FSC may increase
wages on a percentage basis (i.e , each employee's , wage
being increased by, say, 8 percent), or by a fixed dollar
amount (i.e., each employee's wage increased by, for ex-
ample 50 cents an hour).
Finally, wage setting at a newly established plant in-
volves still other considerations. According to FSC's
witnesses, and as will be more fully considered below in
connection with Indiantown wages, the initial set of
wages at a new plant are pegged at 108 percent to 112
percent of wage practice line levels and then gradually
decreased until the 104 percent standard is reached.
As for the organizational steps by which a general
wage increase is determined, the procedure starts with
several alternative recommendations from FSC's com-
pensation manager; thence to discussions between the
compensation manager and the managers of the local fa-
cility; then on to several headquarters offices. "[B]ut the
final approval is an upper management decision."29
In sum, according to the testimony of FSC's witnesses
the wage survey and resulting wage practice line serve
"as a guide" in FSC's wage-setting process. 29 And while
there are company standards, the choice of the wage in-
crease for a facility's various grades is a "judgmental"
decision: 3 ° "in the final analysis, it [comes] down to a
discretionary decision by corporate management."31
Graphic Displays of Actual Relationships Between FSC
Wage Rates and Wage Practice Lines. The record in this
proceeding includes wage data for only 3 of FSC's 11 fa-
cilities: Charlotte, Indiantown, and Tampa. 32 A number
of graphs are in evidence showing the relationship be-
tween the wage rates at those three facilities and the re-
spective wage practice lines developed from FSC's area
surveys. Those graphs are set out in Appendices A, B,
and C to this decision [omitted from publication]. As can
be seen from the graphs, at those three facilities FSC
wages often did maintain a close relationship with the re-
spective wage practice lines. On the other hand FSC
wage rates sometimes departed substantially from the re-
spective wage practice line levels, and wholly parallel re-
lationships were rare.
III. WAGE RECONSTRUCTION—SUMMARY OF
FINDINGS AND CONCLUSIONS
No one suggests that past years' wage levels at Char-
lotte and Indiantown could or should be reconstructed
by attempting to replicate_ the procedures that FSC
would have followed had it continued its pre-1974 ap-
proach to wage setting at those two facilities. Rather, the
28 Tr 1867 (witness Hogue)
29 Tr 1908 (quoting from an FSC brief in an earlier stage of the FSC
litigation)
3 ° Tr 1870 (witness Hogue)
3i Tr 1908, quoting an earlier FSC statement
32 The exception two graphs showing the wage rates and related
wage practice lines for FSC's Fort Lauderdale and Miami facilities for
1974 GCX-40d
parties agree that the best way to reconstruct the wage
levels that FSC would have implemented at Charlotte
and Indiantown but for its unfair labor practices is to
follow a four-step procedure.
1 Compare, on the one hand, wage levels at FSC's
Charlotte and Indiantown facilities during a pre-1974
period with, on the other hand, the wage levels of other
employers during that same period.
2. Determine what relationship existed between the
FSC wage levels during the pre-1974 period and the av-
erage wages paid during that same period by the other
employers. Express that relationship in mathematical
terms
3 Find out what those other employers' wage levels
were in the years since 1973.
4 Apply the mathematical expression determined in
step two to the other companies' wage levels in the post-
1973 years.
Notwithstanding 'the parties' general- agreement on the
process to use in reconstructing wages, the General
Counsel and the Charging Party, on the one hand, and
FSC, on the other, sharply disagree on what those re-
constructed wage levels ought to be. (While the Steel-
workers activity 'participated- in this proceeding, the
Union wholly supports the General Counsel's backpay
formulations. Accordingly further reference in this deci-
sion to the position of the General Counsel should be
read as also including the Steelworkers position, unless
specified otherwise.)
The General Counsel's Position on How to Reconstruct
Wages. The Bureau of Labor. Statistics, United States De-
partment of Labor (BLS) publishes figures on the aver-
age hourly earnings of employees working for manufac-
turers. (BLS's average hourly earnings statistics will
henceforth be referred to as AHE.) The General Counsel
concludes from a comparison of FSC wages and AHE
that during the pre-1974 "base period" FSC wage levels
at Charlotte and Indiantown increased faster than did
AHE levels For example, during the base period on the
average FSC's Indiantown wages increased $1.57 for
each $1 increase in AHE According to the General
Counsel, that base period trend should be reflected in the
Board's reconstruction of FSC's post-1973 wages. Thus
the Board's reconstruction of FSC's Indiantown wages
should be based on the assumption that those wages
would have continued to increase $1.57 for every $1 in-
crease in AHE.
FSC's Position on How to Reconstruct Wage Levels. FSC
focuses on the average difference between, on the one
hand, FSC wages during the pre-1974 base period and,
on the other, the wages paid by other companies during
the base period FSC -does that several different ways,
using both BLS's AHE data and data derived from pri-
vate surveys of employers in the Charlotte and Indian-
town areas. In each of its proposed wage reconstruction
approaches, FSC claims that the proper way to recon-
struct FSC wages is to assume that throughout the post-
1973 period the difference between the AHE or wage
survey data, on the one hand, and FSC wages on the
other, would have been -the same as the average pre-1974
difference between the wage data and FSC wages. For
896 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
example, FSC claims that during a 1971-1973 the base
period, the average wage that FSC paid to its Charlotte
employees was 23 cents higher than the average wage
paid by other companies in the Charlotte area. FSC goes-
on to argue that the proper way to reconstruct Charlotte
wages during the backpay period is to 'determine what
wage rates theise other companies paid, on the average,
during the post-1973 period and to assume that FSC
would have paid its employees 23 cents per hour more
than that
The Wage Reconstruction Method Adopted by this Deci-
sion. For reasons detailed in parts IV and V of this deci:
sion, my conclusion is that none of the wage reconstruc-
tion methods proposed by the parties are satisfactory.
The General Counsel's methodology results in substan-
tially overestimated Wages for employees in the middle
and lower grades. FSC's proposals substantially underes-
timate wages for employees at all levels.
As discussed in part , VI of this decision, I have con-
cluded that trends in .FSC's wages during the pre-1974
years, the circumstances at FSC's Charlotte and Indian-
town plants in the post-1973 period, and the inherent re-,
lationship between AHE and any employer's wages, to-
gether point toward one wage reconstruction method:
Reconstruct FSC wages, based on the relationship that
existed between FSC wage levels and AHE as of the
date of the last wage increase at FSC's Charlotte and In-
diantown plants prior to FSC's unlawful change in wage
policy.
The table [see Appendix G, Table 1] below lists an ex-
ample of the effect of my recommendation for the recon-
structed hourly , wage rate of one middle level grade of
employees at -Charlotte and at Indiantown for one recent
year. The principal wage proposals of the General Coun-
sel and FSC are included in the table for comparison.
IV. THE PARTIES' CHARLOTTE MODELS33
To sum up, at Charlotte and Indiantown prior to 1974,
and at all times at FSC's other locations, FSC set wage
levels for its hourly employees based on a combination
of (a) wage survey data obtained from employers located
near the facility in question and (b) managerial judgment.
As noted earlier, no one suggests that Charlotte and In.;
diantown wage levels be reconstructed for the post-1973
period by attempting to replicate precisely the process
that FSC would have used but for its unlawfully
changed wage policy at Charlotte and Indiantown.
Rather, all parties propose focusing on the relationship
between, on the one hand, Charlotte and Indiantown
wages during the pre-1974 "base period," and, on the
other, some set of statistical time-series data covering
both the base period and the years since then. This part
of the decision will consider the way the General Coun-
sel and FSC use that technique to reconstruct the wages
of the Charlotte bargaining unit hourly paid employees.
33 "Model" is used here in the sense . of "a theoretical projection in
detail of a possible system of human relationships (as in economics)"
Webster's Third New International Dictionary
A The General Counsel's Charlotte Models
The General Counsel retained a statistician, Dr. James
M. Clapper, to devise a means of reconstructing FSC's
wages at Charlotte and Indiantown during the post-1973
period. in , a' report of his wage reconstruction efforts,
Dr Clapper first briefly noted that during the base
period FSC took into consideration area wage levels de-
termined through annual surveys. The report then con-
tinues:
To attempt reconstruction of the wage rates, it
became necessary to identify a measure of wage
levels in each labor market which could serve as a
surrogate for the no longer available Florida Steel
survey of area employers. Once such a measure was
identified, it would be possible to use the statistical
technique of regression analysis to determine the
pre-1974 relationship between Florida Steel wage
rates and the measure of area wage rates. Once es-
tablished, this relationship could be used to recon-
struct what - Florida Steel wage rates would have
been during the [post-1973] period had the historical.
relationship held, that is, if the company had contin-
ued . . . [its] pre-1974 pattern of wage increases vis-
a-vis area wage rates.34
Dr. Clapper and representatives of the General Coun-
sel 'jointly determined to use, as the "surrogate" for
FSC's • surveys, BLS's AHE figures. More specifically,.
Dr. Clapper and the General Counsel determined to use
AHE figures that BLS compiles covering employers that
are in . the "primary metals and fabricated metal prod-
ucts" business and that are located in the Charlotte
standard metropolitan statistical area (SMSA). BLS pub-
lishes such figures 'on a monthly basis, and Dr Clapper,
concluded that FSC's'. base period Charlotte wages
should be compared to 12-month AHE figure for the 12 -
months ended on the date of each FSC general wage in-
crease. (AHE matters will be discussed in greater detail
below.)
The simplest. way of using regression techniques to de-
termine the pre-1974' relationship between FSC's Char-
lotte Wages and AHE figures is to do so one labor grade
at a time. All that takes is a regression correlating (1)
wage level at each point in time during the base period
when FSC granted a wage increase, and (2) the AHE
figures for the months in which FSC granted those in-
creases
. The results of regressions of that type .are set out in
Table 2 [see Appendix G]. "Slope," as used in Table 2
and elsewhere in this decision, refers to the degree of tilt
of the regression line. In the' regressions under consider-
ation here, it indicates how much wages increased for
every unit change in AHE. For example a slope of 1,110
for Labor Grade 3 means that during the base period
Grade 3 . wages increased an average of $1.11 for every
$1 increase in AHE. Note that a slope of 1.110 does not
mean that during the base period Grade 3 wages could
be approximated by multiplying AHE by. 1.110. Rather,
84 GCX-76, p 1
FLORIDA STEEL CORP 897
according to the regression on the average, one could
approximate the amount' of the increase in Grade 3 wages
from any one base period wage increase to the next by
multiplying the increase in AHE over the same time
period by 1.110: 2 5
- The data on which Table 2 is based were provided by
the General Counsel. But they do not represent any part
of the General Counsel's wage reconstruction proposal.
The. problem, said Dr. Clapper, is scarcity of data.
The Charlote base period ended in February 1973 (the
date of the last general wage increase before FSC's un-
lawful .change in' policy). , And Dr Clapper determined
that the earliest usable wage rates ,were those that FSC
put into effect in May 1969. (Thus references to the Gen-
eral Counsel's Charlotte "base period"- will be to the
period May 1969 through February 1973) 38 During that
1969-1973 period FSC granted five increases, as indicat-
ed in Table 3 [see Appendix G]. 37 That means only five
base period "observations" And five observations, said
Dr Clapper, "will not yield very reliable results." 38 Dr.
Clapper accordingly first determined to include all the
Charlotte labor grades in - one. "multiple regression".
model." That potentially enabled him' to treat each of
the five wage increases for all Charlotte labor grades as
separate observations for one regression equation (for a
total of 90 observations)" But to do that, Dr. Clapper
had to . assume, for . purposes of the reiression, that the
wage rates of all of the various 'Charlotte labor grades
moved - in parallel; and to agsuine that, for wage recon-
struction purposes, the average difference between one
labor grade and the next (hiring the base period would
be maintained during the post-1973 period.
That did not work. The technique does not require
that the labor grade wage rates actually have moved in
parallel But it does require ihat the rrioveinent of the
various labor grades not have varied too much from a
parallel relationship: And the nioyements of the uPper
grades at Charlotte during the bade period were too dis-
parate from the lower grade movements. After some trial
and error, however, Dr. dapper found that the multi-
grade approach would work if the Charlotte grades were
divided into two sets: lower nine and upper nine 41 And
33 Data are , from GCX-121 For reconstructed wage levels Inherent in
those regressions see RX-141
36 Data for a wage increase iranted in 1968 Were also available to the
General Counsel Dr Clapper'could not recall the'reasoning that led him
to reject the 1968 data
37 Table 3 is identical,to the first Rage of RX-72 through 77
38 Tr 1012
33 Previous examples of regression analysis—FSC's technique for de-
ten/lining "wage practice lines" and the grade-by-grade regressions dis-
cussed . above—involve "simple" regression Each 'use only one "predic-
tor" variable (e g, AHE level) and one "cnterion" . variable (e g, wage
level for one grade) "Multiple" .regressiCemeans that the model includes
two or more predictor variables
40 Due to the mathematics of multiSle regressicin, the effective differ-
ence in number of observations, as ' between the grade-by-grade model
and the all-grades model, is something less than 90 mina 5 Nonetheless
the increase is substantial , <4i The upper 9 did not include grade 16 (which was unpopulated
during the base period) or grade 17 (which was populated only sporadi-
cally dunng the base period) See Exh 144 and Tr 710
that has led the General Counsel to' propose a two-model
forecast for Charlotte.
The result of each of the models is a regression-deter-
mined equation that, in effect, shows the trend in the av-
erage, relationship during the 1969-1973 base period be-
tween, on the one hand, wage rates for the labor grades
covered by the model, and, on the other, AHE. The re-
sulting slope for the nine lower grades is 1.03: That is,
according to the regression for every increase of one unit
in AHE during a wage-increase-to-wage-increase time
span in the base period, the wages of the nine lower
Charlotte grades increased by 1.03 units. (Note that that
is a kind of double average: average movement across
the five base period wage increases and average as
among the various labor grades.)
The slope produced by the General Counsel's model
for the upper Charlotte grades was 1.47.
The following table [see Appendix G, Table 4] shows,
for a few illustrative years and grades, the Charlotte
wage rates produced by applying those slopes to AHE
figures for the post-1973 period. (The figures in the
right-hand column are intended to provide some sense of
the relative increase in the reconstructed wage rates gen-
erated by the General Counsel's models. 1973 wages are
actual, not reconstructed)"
Why the General Counsel's Models Are an Inappropriate
Means of Reconstructing Charlotte Wage Rates, The Gen-
eral Counsel's Charlotte models are clearly the product
of considerable effort and statistical sophistication. None-
theless they are not an acceptable means of reconstruct-
ing Charlotte wage rates. While numerous considerations
point to that conclusion, two are particularly notewor-
thy. One has to do with "autocorrelation"—statistical
term that will be defined below. The other concerns the
connection between, on the one hand, base period pat-
terns and, on the other, reconstructed wages: The record
fails to support the General Counsel's assumption that
the trend in the base period relationship between wages
and AHE would have continued into 1974 and beyond
but for FSC's unfair labor practices.
Autocorrelauon. Statisticians call the vertical distance
between an observation and a regression line a "residu-
al." The mathematics underlying regression analysis are
predicated on the assumption that the residuals in a re-
gression model display little if any relationship to one an-
other. To the extent that the residuals in a regression
model form a pattern, that assumption is not met. Statisti-
cians call a departure from the no-relationship-among-re-
siduals requirement "serial correlation of the residuals;"
or "autocorrelation."
42. Table 4 and the other computations in this decision that use Char-
lotte primary metals AHE are based on the following AHE levels:
12 mos ended
Feb
Avg monthly
AHE level
1973 3 20
1974 3 50
1976 4 16
1979 5 23
1980 5 60
1981 626
898 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
A moderate amount of autocorrelation is not a prob-
lem. But a high level of autocorrelation in a regression
model spells trouble: "When you have a serious autocor-
relation problem. . . you want to find out what was left
out of the model or mis-specified. . . . There is probably
something that should have been pin in the model that
isn't there "43 As the quotation indicates, a high level of
autocorrelation suggests a "specification error." "And
specification error . . . in . . a regression analysis
sense, means that the model that you have hypothesized
is not an accurate model."44
What that adds up to is that a high level of autocorre-
lation indicates that the regression model may be improp-
erly constructed And autocorrelation has a further
aspect of importance here Ordinarily regression analysis
provides not only a set of predictions, but also—via
"confidence interval" statements—detailed information
about the probable accuracy of those predictions. 45 But
the information about the predictions' accuracy is not re-
liable where there are high levels of autocorrelation.
As it turns out, the General Counsel's regression
models present serious autocorrelation problems. In fact,
the requirement of lack of pattern among the residuals is
"grossly violated" by those models." As a result of this
autocorrelation Dr Clapper was "not professionally will-
ing to make confidence interval statements." 47 Thus, if
one looks no further than the statistical standards for
evaluating regression models, the General Counsel's
Charlotte models are a failure. Dr. Clapper recognized
that and argued that additional statistical studies (of
FSC's Tampa facility) provide evidence that counterbal-
ance the Charlotte models' autocorrelation problem."
The import of General Counsel's Tampa studies will
be considered below. But before getting there it is neces-
sary to focus on the question of why one should accept
the General Counsel's assumption that the trend in the
base period (1969-1973) relationship between increases in
FSC wages and increases in AHE would have continued
into the post-1973 years had FSC not Violated the Act.
43 Tr 2151 and 2150 (witness Haworth)
44 Tr 3506 (witness Clapper)
44 "If the explanatory variables in the regression equation account for
all major forces acting on the dependent variable, so that the unexplained
variation is the product of many small forces, statistical theory makes it
possible to compute confidence Intervals These permit the assertion,
within a specified level of confidence, that the true value of a de-
pendent variable lies within a certain range of the value estimated by the
regression equation" M Finkelstin, "Regression Models in Administra-
tive Proceedings," 76 Harv L Rev 1442, 1450-51 (1973) (footnotes omit-
ted)
46 Tr 1468 (witness Clapper) See also Tr 1316 (agreement by Dr
Clapper that the models' autocorrelation levels are "very substantially
above the threshold level")
47 Tr 1098 See also Tr _1467-68 and 3651
44 As may be recalled, the record contains evidence of the result of
grade-by-grade regressions that correlated AHE with the base period
wages of the various individual labor grades at Charlotte See Table 2
[Appendix G] The General Counsel suggests that the results of those
grade-by-grade regressions, which show no evidence of autocorrelation,
lend statistical support to the General Counsel's models That does not
appear to me to be the case The grade-by-grade regressions are depend-
ent on but five observations each And that is too few to produce reliable
results See Tr 1012 (witness Clapper) and 2141 (witness Haworth) I
cannot conclude that the results of the statistically unreliable grade-by-
grade regressions validate the General Counsel's statistically unreliable
multigrade regressions
The Lack of Any Economic Rationale Supporting the
General Counsel's Wage Reconstruction As just discussed,
according to the General Counsel's Charlotte models,
during the 1969-1973 base period on the average the
wages of the Charlotte lower grades increased by $1.03
for every $1 increase in AHE, while the upper Charlotte
grades increased by $1.47 for every $1 increase in AHE.
Thus as Table 4 indicates, most of the reconstructed
wages produced by the General Counsel's models keep
increasing, relative to AHE, throughout the backpay
period 49 The question is whether FSC wages would in
fact have followed those same trends in the middle and
late 1970s and into the 1980s if FSC had not unlawfully
stopped granted wage increases in '1974. (Note that the
question would remain even if the General Counsel's
models presented no autocorrelation problems)
One kind of consideration raised by that question re-
lates to the duration of the base period and wage recon-
struction period. Assume, • for the moment, that: (1) the
base period had been a lengthy one—say 15 or 20 years;
(2) the slopes of 1 03 .and 1.47 produced by ,the General
Counsel's models accurately described the wage-to-AHE
relationships throughout that base period; and (3), the
length of the wage rconstruction period was short—just
a few years. Under those circumstances it would be rea-
sonable to assume, without more, that these same wage-
AHE relationships would continue during the wage re-
construction period—unless FSC could prove otherwise
(as by showing that determinative factors present for the
20 years to 1973 had thereafter abruptly changed). But
the Charlotte base period used by the General Counsel is
a short one, as all parties agree . 5 years and five wage
increases. It is so short, in fact, that the , trend . in the
wage-to-AHE relationship developed by the General
Counsel could readily' be a function of a series of isolated
factors. And the wage reconstruction period is a long
one (at least 9 years). In these circumstanCes it seems to
me that it is incumbent on the General -Counsel, to point
to evidence that supports the 'assumption' , that a Wage,
trend that existed during the base period would have
continued in the , post- 1973period but for FSC's unfair
labor praônce.
I have been unable to find any 'such' evidence.
The General Counsel's Assumption of a Nonparallel Re-
lationship Between AHE and FSC Wages. Both AHE and
FSC's Wages tend to reflect the saine kinds of economic
forces. 5 ° As a result, absent information about FSC'S op-
erations suggesting otherwise, one can assume a p.arallel
relationship between changes in AHE and changes in
FSC's wage levels. Over any appreciable period of time
44 Under the General Counsel's principal 'models, the Wakes of the
lower 'grades in each of the Charlotte models increase faster, on a per-
centage basis, than do the ukper grade (That is discussed further in part
VI, below ) The General Counsel provided evidence on how his model
could be used to Increase the reconstructed wages of all Charlotte grades
proportionally GCX-125 Under that approach Charlotte grade I would
increase from 81 percent of AHE in 1973 to 88_ percent in 1981, grade 6
(Mill) would incregelrom 94 percent to 100 percent, grade 7 would in-
crease from 99 percent to 114 percent, and grade 15 from 147 percent of
AHE in 1973 to 157 percent in 1981
See,,e g, Steelworkers witness Roomkm AHE "captures the supply
and demand forces that influence wages' ; (Tr 3468)
FLORIDA STEEL CORP 899
(all other things held equal) the wages of FSC's employ-
ees ought to increase in proportion to increases in AHE.
And while FSC's wages may move in a pattern different
from AHE for a few years at a time, decreases in FSC's
wages relative to AHE ought to be as likely as increases
(again, all other things held equal). These considerations
run counter to the General Counsel's reconstructed
Charlotte wages, of course, since with limited exception
those reconstructed wages increase proportionally faster
than AHE throughout the backpay period.
That, however, does not end the matter since there are
a variety of economic factors that could, in theory, cause
FSC's wages to depart from a parallel relationship with
AHE, even on a relatively long-term basis. For example:
a worsening of working conditions at FSC's plants rela-
tive to those at other facilities, faster expansion by FSC
than by other employers, increasing skill requirements at
FSC's facilities relative to the skills required by the jobs
of other employers, and commumtywide changes in fac-
tors other than wages that would affect AHE levels but
not wage rates 51 If one or more of those factors persist-
ed (and steadily increased in impact) throughout the
1970s at FSC's Charlotte plant, one could reasonably
conclude that the General Counsel's model captured the
impact of these factors during the base period and that it
would make sense to project the model-determined slope
forward. But there is no evidence at all that any of those
factors persisted throughout the 1970s, much less in-
creased steadily in scope.52
In sum, the record provides no explanation of why
FSC's wages would have increased relative to AHE
throughout the 1970s and early , 1980s had FSC not vio-
lated the Act."
Reconstructing wages in analogous circumstances. FSC
operates nine facilities in addition to Iridiantown and
Charlotte. One of those nine facilities is located in
Tampa, Florida. Dr. Clapper accordingly "hypothetical-
ly reconstructed" Tampa Wage rates "using the same
methodology employed in the Indiantown and Charlotte
reconstructions." 54 He then compared the wage rates
Such as amount of overtime worked See the discussion of AHE,
below
32 Another possibility I considered is that FSC, for policy reasons of
its own—such as a possible desire to avoid the unionization of its plants—
deliberately increased its employees' wages faster than other employers
("[A] company under the feeling that it was under the threat of a union
organizing campaign might seek to install a wage premium to move to
the upper area of wages in the 'community as a means of forestalling
unionization" Tr 3277 (witness Roomkin) ) But again, the record is
barren of evidence indicating the adoption by FSC, for that reason, of a
policy whereby FSC wages would , keep increasing relative'to community
wages
33 Two commentators on the use of regression analysis in administra-
tive proceedings suggest that absent such an explanation an agency ought
not accept a regression-determined forecast M Finkelstein, "Regression
Models in Administrative Proceedings," 76 Han L Rev _ 1442, 1446
(1973) ("there must be some underlying theory which supports" the use
of the regression-determined-correlation), F Fisher, "Multiple Regression
in Legal Proceedings," 80 Co/urn L Rev 702, 714, 735 (1980) ("The
measurement provided by least squares regression is a way of making
theoretical assumptions precise or of the testing them, it is not a substitute
for thought one should make sure that the model used is construct-
ed on sound hypotheses based on theoretical considerations generated
from outside the model itself ") Were the Board to adopt that position,
consideration of the General Counsel's models could stop right here
34 Exh GC-79, p 2
produced by his model to Tampa's actual wage rates.55
FSC has never changed its wage-setting procedures at
Tampa—FSC still uses wage surveys, wage practice
lines, and the like. As Dr. Clapper saw it, if he devel-
oped a model for Tampa similar to his Charlotte and In-
diantown models, and if that model then produced "re-
constructed" wages close to the wages FSC actually
paid its Tampa employees, his Charlotte and Indiantown
models would thereby be validated. The General Coun-
sel contends that the results of the Tampa regression
model make up for the Charlotte models' problems of
autocorrelation and lack of an underlying economic ra-
tionale for assuming post-1973 continuation of a base
period trend in wage-AHE relationships.
As indicated by Dr. Clapper, the General Counsel's
Tampa model is methodologically similar to the General
Counsel's Charlotte models. The main differences are: (1)
the Tampa model uses as a base period April 1973-Octo-
ber 1975 (as contrasted to May 1969-February 1973 for
Charlotte); 56 (2) during that base period there were six
wage increases (not five, as in the Charlotte base period);
(3) the Tampa model uses one regression equation for all
Tampa labor grades (unlike Charlotte, which uses two).
The slope produced by the Tampa model is 1.32 (com-
pared to 1.03 and 1.47 for Charlotte).
- A compafison of the "reconstructed" and actual wage
rates for the period April 1976-April 1980 is set out in
Table 5A [see Appendix G]. (The listed months are all
those during the 1976-1980 period in which Tampa mill
employees received general wage increases.) As can be
seen (1) The model starts off (in 1976) with minor over-
predictions -and ends (in 1980) with somewhat larger un-
derpredictions; and (2) in general the model accurately
reconstructs Tampa wages.
Despite' that accuracy, however, I cannot conclude
that the Tampa model provides the necessary proof that
the General Counsel's Charlotte regression models are
designed soundly or that the trend in the base period re-
lationship between wages and AHE shown by the Gen-
eral Counsel's models would have continued but for
FSC's unlawful wage policy changes. The question,
again, is how much the General Counsel has to prove
before FSC comes under the burden of showing why
what may be an analogous situation is not in fact analo-
gous Clearly if the General Counsel had introduced
similar to the Tampa model covering all nine of the fa-
cilities at which FSC continued its wage increase policies
one could thereby accept the General Counsel's Char-
lotte wage reconstructions absent compelling evidence
from' FSC showing why those models did not support
the General Counsel's Charlotte contentions. But in fact
the General Counsel's evidence covers only one of two
sets of wages at 1 of FSC's 11 facilities. That, in my
" The Tampa facility has two sets of wage rates Dr Clapper's Tampa
model uses one set—for the Tampa "mill"
3 ! Dr Clapper selected a Tampa base period beginning in April 1973
because that was the first wage increase granted at Tampa after the facili-
ty's wage structure was revamped (" . in Tampa [pnor to 1968]
they -had a 13 grade system, they compressed It to twelve grades begin-
ning in April of 1968 and in April . of 1973 It was compressed to 10
grades" General Counsel witness Yost, Tr 663)
900 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
judgment, says too little about the probable accuracy or
the General Counsel's models.57
Questions Regarding the Consistency of AHE Data. One
of the assumptions on which the General Counsel's
models rests is that the AHE data the models use are
consistent at all relevant times'—that is, from 1969 to the
present. The fact is that those data are not wholly con-
sistent throughout that period of time. The only real
question is how substantial are the inconsistencies.
Monthly AHE data are derived by dividing the gross
payroll for the month for the various employers included
in a particular AHE series by the total hours worked by
the employees of those employers. Irregular bonuses and
the like are excluded from the gross payroll figures, but
overtime payments are included. All such data are ob-
tained on voluntary basis. BLS presents the data by lo-
cality (such as for Palm Beach County in Florida and for
the Charlotte SMSA in North Carolina), by State and
nationally; and by industry groupings (such as "primary
metals," "nondurable goods," or "all manufactunng").58
The focus in this proceeding is on wage levels in- par-
ticular geographic areas for particular groupings, over
substantial periods of time Under these circumstances all
sorts of inconsistencies are possible. Those inconsisten-
cies include: (1) changes in the definition of the geo-
graphical area covered by the AHE data, (2) changes in
the definition of the industry grouping; (3) changes in the
relative -level of overtime (increasing overtime will result
in higher AHE figures even if wage rates are not in-
creased); (4) changes in the levels of skills demanded by
the employers covered by any AHE series; and (5)
changes in the identity of the companies choosing to file
reports with BLS. -
Overtime Levels, Employee Skill Levels, Identity of Re-
porting Companies. The record provides no evidence one
way or the other on whether there were substantial
changes in any of these factors during periods pertinent
to this proceeding." It is clear that BLS uses statistical
techniques to avoid sudden shifts in AHE data that
might be caused by- substantial changes in any of these
factors. But there remains the possibility that for two pe-
riods several years or more apart such factors could
mean that apparently comparable AHE figures were not
in fact so. And, indeed, one state official involved in the
collection of AHE data advised a representative of the
General Counsel that he (the official) did not consider
AHE data appropriate for the purposes in which . it is
used in the General Counsel's models.
57 Dr Clapper used Tampa wage rates, rather than those of some
other Florida Steel facility, as a check because the General Counsel hap-
pened to have obtained only Tampa—plus Charlotte and Indiantown--.-
wage rates from FSC There is no indication that the General Counsel
could not have obtained access to the wage rates of additional FSC facili-
ties
58 BLS (a Federal agency) does not in fact collect local or State AHE
data Rather, state agencies do so under contract with BLS In Florida,
the Florida Department of Commerce does the AHE surveys In North
Carolina, the Mirth - Carolina Division of Current Employment Statistics,
,Employment Security Commission, does the work In all cases the States
operate pursuant to BLS standards
59 Of the firms that report AHE data 2 or 3 percent drop out annually
Tr 2306 (witneis Hawoi-th) But the significance of that level of dropouts
is unclear
Definitions of Geographical Areas The General Coun-
sel's Charlotte models use AHE for the Charlotte
SMSA. The General Counsel began his wage reconstruc-
tion efforts using -the "all manufacturing" series for the
Charlotte area But in 1973 the Charlotte SMSA was re-
defined to broaden its scope." Upon discovering that
change, the General Counsel discarded the Charlotte all
manufacturing AHE series on the ground that it was too
likely that the SMSA redefinition substantially affected
that series But a BLS official advised a representative of
the General Counsel that the "primary metals" AHE
series for the Charlotte SMSA was only "minimally af-
fected by the change."' It is for that reason that the
General Counsel's Charlotte model uses the primary
metals AHE series But that official's statement about the
effect of the SMSA change on the primary metals AHE
series is untested. No one with any expertise in the
matter testified at the hearing in this proceeding:62
AHE—Conclusion Having said all that, the record in-
dicates that whatever flaws Charlotte SMSA primary
metals AHE data might have for the purposes such data
have been used in this proceeding, they are at least as
good as anything else available. Thus, while my conclu-
sion is that the General Counsel's models do not provide
an appropriate method of reconstructing FSC's Charlotte
wages, that conclusion is not based on those models' use
of Charlotte SMSA primary metals AHE data.
B. FSC's Charlotte Regression
Central Piedmont Industries (CPI) is an employer as-
sociation in the Charlotte, North Carolina area 63 CPI
conducts annual wage surveys. Employers in the Char-
lotte area which are willing to do so provide CPI with
wage information, by labor grade. (CPI asks employers
to categorize the jobs at their facilities on the basis of
National Metal Trades Association grades—a 10-grade
scale.) FSC retained Dr Joan Haworth to develop wage
reconstruction methodologies, and she concluded that
the CPI surveys provided the best times series for pur-
poses of reconstructing FSC's Charlotte wages." As
will be discussed below, FSC's model does so by deter-
mining: (1) the average wage level in the Charlotte area;
(2) the average gap between one labor grade and the
next; and (3) . the average dollar difference between
FSC's Charlotte wages and average area wages during a
1910-1973 base period.,
Average Area Wage Level., FSC's regression model de-
termined: (1) the average Charlotte area wage paid in
1970, and (2) the difference between that average 1970
65 Gaston County was added to the SMSA at that time
61 Tr 694 (witness Yost)
62 In FSC's annual wage surveys leading to wage increases, FSC
never limited the survey participants to metal products companies Thus
the use of the pnmary metals AHE series runs contrary to the General
Counsel's goal of using a time series as close as possible in nature to the
wage surveys FSC previously conducted
63 Since 1977 CPI has been known as "Central Piedmont Employers
Association, Inc" But the parties all refered to "CPI" throughout the
hearing
64 The model uses most, but not all, of the wage data produced by the
CPI surveys The selection of CPI data for use in the model was made
by Dr Haworth
FLORIDA STEEL CORP

901
wage and the average area wage in subsequent years.66
Thus to derive the average Charlotte area wage for
1980; one would take the 1980 figure generated by the
model and add it to the model's 1970 figure.
Labor Grade Variables Using CPI survey data, FSC's
model determines the average wage of each labor grade
for each year subsequent to 1970. Then it determines the
gap between each adjacent labor grade for each such
year. Finally the model determines the average of the
gaps between all adjacent labor grades. (Assume that
there are only three labor grades; that the year is, say,
1979; and that the gap between the average wage for
Grade 1 and the average wage for Grade 2 was 50 cents,
while the gap between the average wage for Grade 2
and the average wage for Grade 3 was $1. Roughly
speaking, the model would use 75 cents as the labor
grade variable for 1979.)66
FSC Add-On. Doctor Haworth concluded that the re-
lationship between FSC's Charlotte wages and the CPI
wages for the same year correlated less well than FSC
wages and the previous year's CPI wages. Accordingly
the FSC model compares the average FSC Charlotte
wage for 1971, 1972, and 1973 with, respectively, the av-
erage CPI wage for 1970, 1971, and 1972. (One base
period " "year" thus includes FSC data from 1971 and
CPI data from 1970; or 1972 and 1971; or 1973 and 1972.
For ease of reference this decision will accordingly refer
to base period years in the form, e.g., 1972-1971.) The
average difference over those three years-1971-1970,
1972-1971—between the average FSC wage and the av-
erage CPI wage was 23.1 cents. 67 (It may be recalled
that the General Counsel's model used the 5 years 1969
through 1973 as a base period. But CPI data for 1968
and 1969 were not available. So the FSC model does not
use FSC's Charlotte wages for 1969 or 1970.)
Reconstructing Charlotte Wages with FSC's Regression
Model. The following is an example of how one would
use the FSC model to arrive at a reconstructed wage—in
this case, solely as an example, FSC Grade 5 for 1979.68
1" Base year figure (referred to as the "intercept"):
$1.965.
2. Year add-on for 1979: $1.232.
3. Average gap between labor grades for 1979: $0 390.
4. FSC add-on: $0 231.
5 $1.965 plus $1 232 plus five times $0.390 plus $0.231
equals $5.378.
Thus the FSC model's reconstructed wage for (trans-
formed) labor grade five for 1979 is $5.38 per hour.
As Table 5B [see Appendix G] indicates, the recon-
structed wages produced by FSC's regression models are
65 FSC uses a regression model to make these and the other determina-
tions discussed in this part of the Decision But unlike the General Coun-
sel's model, FSC's model does not correlate FSC wages with any predic-
tor series Rather FSC uses a regression model as a handy way of deter-
mining averages As that suggests, one could obtain results close to those
of the FSC regression model by applying routine anthmetic manipula-
tions to the data used by the FSC model
66 See part VI, below, for further consideration of the wage structure
aspects of FSC's model
67 Dr Haworth calls It the "weighted average difference between
Flonda Steel and the CPI for all the labor grades" for the 3-year base
period Tr 2530
69 The figures are from RX-61
generally lower than those of the General Counsel's
mode1. 6 9
The Major Faults of FSC's Regression Models. FSC's re-
gression model presents three serious problems. The, first
is the use of a highly abbreviated base period. The
second is the use of a constant dollar figure ($ 231) as the
difference between FSC wages and area wages. And the
third is the manner in which the model determines the
base period relationship between FSC wages and area
wage rates.
Abbreviated Base Period. FSC has operated the Char-
lotte facility since the mid-1960s. The General Counsel's
model uses as a base period the years 1969-1973. As
noted earlier, the record is silent about why the General
Counsel's model discards data for years prior to 1969.
Nonetheless the base period used by the General Coun-
sel's model gives the General Counsel 5 years—and five
annual wage increases—of FSC wages to compare with
data from the predictor series (the AHE primary metals
series for the Charlotte SMSA) covering that same
period of time.
Even that 5-year base period is exceedingly brief for
purposes for reconstructing backpay period wages, as
touched on above. Yet FSC's Charlotte regression model
uses a base period of only 3 years. There is nothing sus-
pect about FSC's reasons for using a 3-year base period.
FSC did so because adequate CPI data were not avail-
able for the years prior to 1970. 70 Nonetheless, the FSC
regression model's base period is 40 percent shorter than
the base period both of the General Counsel's Charlotte
model and of FSC's "index" models. (FSC's index
models will be discussed below ) That difference in base
periods alone casts considerable doubt on the utility of
FSC's regression models.
The Impact of a Constant Dollar FSC Add-On. FSC's
Charlotte regression model uses a constant dollar figure
($.231) to determine the difference between the recon-
structed FSC wage for any given labor grade and in any
given year and the respective CPI wage. Because of the
considerable growth in wage levels during the backpay
period, the result of that constant dollar approach is that
the FSC model forecasts FSC wages as declining sub-
stantially, in percentage terms, relative to area wages.
For example, according to the FSC regression model,
the average wage of a CPI Grade 4 employee in 1974
was $3.48. The model reconstructs the FSC Grade 4
wage for 1974 as $3.71, or 6.6 percent higher. In 1981
the model shows an average CPI Grade 4 wage of $5.94.
The reconstructed FSC Grade 4 wage in 1981 is $6.17,
or only 3.9 percent higher. Nothing in the record justi-
fies the implicit assumption inherent in FSC's regression
69 FSC figures are for January of the indicated year The General
Counsel figures are for February
If FSC's model matched FSC data for a given year to CPI data for
that same year, the model would have had a 4-year base period 1970,
.1971, 1972, and 1973 But as noted above, Dr Haworth found that FSC
and CPI data correlated better if FSC data were matched against the
prior year's CPI data Thus a base period of 3 years 1970-1971, 1971-
1972, and 1972-1973
902 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
model that FSC's wages would have declined in percent-
age terms relative to area wages in that fashion."
Manner in Which FSC's Regression Model Determines
the Relationship Between FSC Wages and Area Wage
Rates. As discussed earlier, the AHE primary metals
series is a reasonably good indicator of trends in wages
in the Charlotte area. And throughout the period 1969-
1973, .FSC's Charlotte wages increased faster than did
AHE.- One way of seeing that is simply comparing the
average FSC wage with AHE figures Table 6 [see Ap-
pendix G] does that.72
Another approach IS to consider the results of the indi-
vidual wage rate regression shown in Table 2. The re-
gressioti-developed slopes indicate that .nearly all of
FSC's Charlotte wage rates increased proportionally
faster than did AHE during the 1969-1973 base period.73
The spread between FSC's wage rates and the CPI
wage rates used in FSC's regression models is an average
of the difference'between FSC wage rates and CPI wage
rates for the 3 years 1970-1971 through 1972-1973. .That
approach wholly washes out any trends in the relation-
ship between FSC wage rates and CPI wage rates. (As
the .General Counsel puts it: "The averaging process ob-
scures any fundamental changes in wage ratios during
the base period": Br. 161.)
FSC's only justifications for the use of a 3-year aver-
age figure are that no meaningful trends are discernable
in the base period in the relationship between FSC
wages and CPI wages, 74 and that "the arithmetic aver-
age is the one that requires the least justification." 76 But
the fact is that FSC's base period Charlotte wages do
display a pattern. They increased faster than both AHE
and average CPI wages. And the dimensions of that pat-
71 If the FSC regression model were otherwise attractive, this particu-
lar fault could be easily remedied by switching from a dollar to percent-
age relationship
72 Average wage an AHE data are from RX-73
73 That clearly delineated trend does not show up well in the figures
used by FSC's Charlotte regression model For example, based on CPI
figures collected in GCX-105 the wages of only one of the nine FSC
grades (as transformed for use in FSC's model) steadily increased faster
than did the respective CPI wage rates for the full 1970/1971-
1972/1973 base period FSC grade 6 And three wage rates were lower,
relative to their . CPI counterparts, in 1972-1973 than in 1970-1971 FSC
grades 4, 7, and 8 (The remaining FSC wage rates—grades I, 2, 3, 5, and
9—dipped relative to CPI wages but then increased more than they had
dropped ) In part that may be a function of the way the wages of the
companies in the CPI survey moved during the base period But almost
surely it is also a function of the very short base period used by FSC (as
discussed earlier) Further, apparently by reason of the way that Dr
Haworth transformed FSC's actual Charlotte wages for the base -period
Into a nine-grade system, the transformed base period wages used by the
FSC model Increase less rapidly than do the actual FSC Charlotte wages
For example, the average FSC wage at Charlotte (based on actual wages)
increased by 92 cents, or 35 percent, between 1969 and 1973 (See, e g,
RX-73 ) But the average Charlotte wage, as transformed for use in FSC's
model, increased only 81 cents, or 30 percent (Average wage figures de-
rived from the base penod FSC wages as transformed for use in Fk's
regression model can be found in RX-79, p 1) For all that, even the
.figure used by FSC's model show that FSC's wages were more often
than not higher at the end of the base period, relative to CPI wages, than
they were at the start of the period
74 "[There's no point in just looking at historical patterns because we
don't have very many The sequential pattern does not tell me any-
thing". Tr 2198, 2535 (witness Haworth)
75 Tr 3775 (witness Haworth)
tern are sufficiently substantial that they must be taken
into account in reconstructing post-base period wages.
In discussing the General Counsel's models, I conclude
that there is no evidence that would indicate that the
ever-widening gap between FSC wages and AHE pre-
dicted by the models would continue in the years after
1973. But, similarly, there is no evidence that that trend
would reverse itself so that the difference between FSC
wages and area wages that existed at the end of the base
period would diminish. Yet FSC's regression model im-
plicitly assumes that such a reversal would have oc-
curred. Since that assumption is not supported by the
record, FSC's regression model would have to be reject-
ed even if it were otherwise faultless.
CPI Problems
Lack of Representativeness. CPI does not try to use sta-
tistically valid sampling methods in obtaining wage rate
information. - CPI simply sends out questionnaires to em-
ployers in the Charlotte area76 and then compiles with-
out modification the data that are returned by whichever
employers care to respond to the questionnaire. 77 More-
over, the CPI data themselves suggest that the CPI
survey results do not perfectly. reflect the Charlotte area
labor market in that the average wage rate of sOme labor
grades for some years, as shown in the CPI surveys, is
higher than the average wage rate for more senior labor
grades. (tor example, the 1970 CPI survey shows the av-
erage Grade 3 . employee earning more than the average
Grade 4 employee. And the 1979 survey shows that, on
the average, Grade 2 employees earned more than Grade
3 ,and Grade 4 employees.) 78 It is unlikely that such in-
versions would have occurred were the CPI data truly
representative of the Charlotte area labor market 7 9
Consistency. The identity of the employers responding
to the CPI surveys changes substantially from year to
year." While that does not necessarily mean that the
CPI data are not reasonably consistent from one year to
the next, the data again contain a suggestion of a lack of
consistency: The average wage rate for some grades in
some years was lower than for that labor grade the pre-
vious year. ,(For example, the survey shows the average
Grade 4 wage to be $3.90 in 1973 and $3.37 in 1974.)
Given the substantial inflation of the 1970s, that would
76 CPI tries to send the questionnaire to all area companies with 50 or
more employees
77 The 1972 survey covered about 7500 employees RX-45 The 1980
survey covered about 17,000 employees Tr 1509 (witness Dowd) (In
1980 there were about 80,000 employees doing "manufacturing and pro-
duction" work in the Charlotte area Tr 2322 (witness Haworth) )
78 There are over a dozen such inversions in the CPI survey data for
the years 1970 through 1980 See GCX-105
79 All parties agree that the validity of FSC's model hinges on the
extent to which the CPI data are representative of the labor market from
which FSC drew its employees See witness Haworth, Tr 2253-54 (re-
sults of model "dependent upon the survey representing the community
that Florida Steel was allegedly trying to meet or match" when setting
wages)
80 For example, 86 companies responded to CPI's 1971 survey Of
those 86, the names of only 48 showed up in the 1972 survey See CPX-3
and Tr 3214-15 Some of the remaining 38 companies might have been
in the survey, but under changed names (due to merger, etc ) But it is
unlikely that name changes represent more than a small proportion of the
missing companies
FLORIDA . STEEL CORP 903
not have occurred were the CPI surveys wholly consist-
ent from one year to the next 81
Problems Regarding Grade Structure Information. The
wages for each labor grade as reported by the companies
responding to the CPI survey vaned enormously
Almost surely that was in part a function of ambiguous
grade slotting instructions on the CPI questionnaires and
of the fact that CPI left to each responding company's
judgment decisions about how to report the labor grade
category of each job. As a result, the CPI data do not
permit a determination of the spread between one labor
grade and the next. (That is why FSC based its wage re-
construction on the average spread between labor
grades
Infrequency of CPI Surveys CPI conducts its wage sur-
veys once a year That is a problem in 'that, while the
wage increases at FSC's Charlotte facility also occurred
only annually during the pre-1974 period, they occurred
at varying times of the year, as follows.
Wage Increases at FSC's
Charlotte Facility
Yr. of increase Mo. of increase
1971

May
1972

January
1973

February
Yet FSC's regression model makes no adjustment for
those variations in its determination of the relationship
between FSC wages and CPI survey wages.82
CPI Data—Conclusion Obviously there are a lot of
problems with the wage data derived from "the CPI
series. And all things considered, the AHE primary
metals series for the Charlotte SMSA probably more
closely reflects area wage level changes than do the CPI
data. But I cannot conclude that for purposes of this pro-
ceeding the quality of the CPI data is substantially inferi-
or to AHE data.
For one thing, the problems that showed up in the
CPI data are in part a result of close examination of the
raw CPI data. No comparable scrutiny of the AHE data
was possible, partially because of the absence of any wit-
ness fully knowledgeable about the raw data that em-
ployers submit to BLS. Second, the CPI approach avoids
AHE's intermixing of overtime wages With • regular
wages, and unlike AHE, probably at least partially ad-
justs for skill mix changes (since some of such changes
ought to show, up as changes in job grade). Third, the
rate of increase of average • CPI wages tended to be fairly
close to the rate of increase in AHE during the period of
8 FSC determined to use less than all available CPI data (see fn 64,
above) in order to increase the consistency of the data used by the model
"In order to have consistent data bases I did select from the CPI sur-
N;eys] those occupations which were present in all of the years" Tr 2289
(witness Haworth )
82 A related problems is that the CPI questionnaires are not necessanly
sent out the same time each year
The CPI questionnaire was sent out in 1973 in the month of February,
in 1974, in July, in 1975, October, in 1976, November; in 1977, Septem-
ber, in 1978, August, in 1979, again in August -
The record does not indicate the date of the 1971 survey Based on the
dates of the completion of the surveys, in 1970 and 1972 CPI sent out the
questionnaires early in the year, and in 1980, in late summer
1970-1980. (That is depicted in both graphic and tabular
form in RX-109.) And where the rate of growth of CPI
survey wages did depart from the AHE rate of growth,
more often than not it was because the CPI survey
wages grew faster than AHE (which, in FSC's model,
would be reflected in higher reconstructed FSC wages).
Fourth, it is clear that the CPI staff did not seek to bias
the results of the surveys. CPI aimed for wage data that
reflected the Charlotte labor market."
Finally, the CPI surveys do provide us with wage
structure information. And while that information may
not perfectly reflect the actual labor market, it certainly
appears to provide at least an approximation of changes
in the average gap between labor grades in the Charlotte
area AHE, on the other hand, provides no wage struc-
ture information at all.
, FSC's Charlotte Model—Other Considerations From
time to time during this proceeding FSC suggested that
its regression model should be adopted because it resem-
bled the wage-setting methods that FSC used prior to
the Union's arrival at Charlotte. But on several counts
that suggestion is incorrect. First, as discussed in part II
and as portrayed in Appendices A, B, and C [omitted
from publication], it is apparent that FSC wages were
much more a function of management discretion than of
the results of ' FSC surveys and wage practice lines.
Second, while the model's focus on average CPI wage
data to some extent resembles parts of .FSC's wage-set-
ting processes, the model does not•use , (or create) wage
practice lines. And third, the FSC models do not repre-
sent an attempt to emulate FSC's wage-setting practices.
In fact, Dr. Haworth, who was responsible for the
design of FSC's models, never even saw any of FSC's
wage surveys or wage practice lines until Jong after she
had completed the design of the models. - '-
The General Counsel's Dr. Clapper 'complained that
the wage rates reconstructed by FSC's regression model
for any given year did not take account of the previous
year's wage structure To Dr. Clapper that means that
"you can have a situation where in one year the labor
grades are very close together and the next year the
labor grades are spread apart and the next they come
back together, and they can oscilliate all over creation":
Tr. 2729:- But for three reasons that is no real concern:
(1) to tome extent the base period FSC labor grade
structure did oscillate; (2) big swings could occur only if
such- swings occurred in the labor market (as shown in
the CPI data) and such swings accordingly are unlikely;
and (3) iii fact, the average gap between CPI labor
grades behaved in a reasonably orderly fashion through-
out the 1970s.
C. FSC's "Index" Models
FSC's presentation does not- rely solely on its regres-
sion model Rather, FSC also proposed several ways of
reconstructing Charlotte wages based on the percentage
8 3 There is no doubt that CPI's president held an antiunion viewpoint
But there is no indication that that viewpoint affected the CPI surveys
904 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
relationship that existed between FSC wages and AHE
during the pre-1974 base period."
FSC's "Proportional" Index Approach. Under the most
obvious of the index, approaches, FSC determined the
rtverage. percentage relationship during the base period
between, on the one hand, the wages of each FSC labor
grade and, on the other, AHE. For example, during the
period August 1968 through September 1974, the wages
of FSC's Grade 3 Charlotte employees averaged 96.3
percent of AHE. 85 Grade 3 wages for the post-1973
period may then be reconstructed by multiplying .96.3
percent times the AHE figure for the desired date. Thus,
since AHE was $5.60 in February 1980, the reconstruct-
ed wage for the Charlotte Grade 3 employee would be
$5.39 ($5.60 times .963).
That approach maintains the same percentage spread
between labor grades that existed, on the average, during
the base period. It thus assumes that the net impact of all
wage increases that FSC would have given in the post-
1973 period (but for FSC's unfair labor practice) would
have been to maintain a constant percentage relationship
between FSC's various labor grades'at Charlotte.
The "Across-The-Board" Index Apiroach. FSC recog-
nized that, alternatively, FSC could haV'e given general
wage increases during the post-1973 years by increasing
every employee's wage by the same dollar amount. Ac-
cordingly, FSC provided an alternative to the first, ap-
proach. Under-the alternative FSC determined the aver-
age percentage - difference during the base period be-
tween 'the average FSC wage and AHE The CharlOtte
percentage worked out to be 102.6 percent' of AHE.88
FSC then went through the following steps to recon-
struct wages for each relevant postbase period point in
time:
1. Determine the dollar increase in AHE that oc-
curred since the previous point in time. For exam-
ple the Charlotte SMSA primary metals AHE
creased by 37 cents between February 1979 and
February 1980.
2. Multiply the base-period differential figure
times the AHE increase. (E.g., 102.6% times 37
cents.)
84 FSC's evidence includes index models that vary by AHE series
used, by whether or:not AHE monthly data is smoothed by using a
moying average, by the weighting , given to the various base period wage
increases, and by the number of times during the base period that FSC
wages are compared to AHE in order to get an average See RX-72
through 79A-2, revised I am convinced that the most appropriate index
models of all those introduced by FSC are those that use smoothed data
from the Charlotte primary metal AHh sene's, that weight each FSC
wage Increase during the base period based on the number of months the
increase was in effect, and that compare FSC wages to AHE solely at
the time of each Increase in FSC wages Accordingly, the following dis-
cussion is predicated on RX-79A-2, revised
85 See RX-79A-2 revised, Table 2A The exhibit is based on a 9-grade
"transformed" wage structure Thus grade 3 as used in the exhibit repre-
sents a much higher pay scale than grade 3"in the 19-grade structure ac-
tually in place during the base period
86 That is the base period relationship.between AHE and the average
wage in a transformed rate structure The actual average Charlotte FSC
base period wage differs from the average wage under FSC's transforma-
tion, and thus the actual base period relationship between AHE and the
average Charlotte FSC wage would also differ
3. The resulting figure (38 cents in our example)
constitutes the across-the-board increase.
Uniform dollar across-the-board wage increases com-
press the percentage difference'between one labor grade
and- another. ,For example, under this alternative the re-
constructed wage rate for 1980 for Charlotte Grade ,9
employee (the highest grade in the 'transformed wage
structure) is 40 percent higher than the Grade 1 wage
rate. Under the "proportional" increase approach, the
1980 Grade 9 wage rate is 65 percent higher than Grade
1.
• The "Combination" Index Approach. The last variation
proposed by FSC is a combination of the first two. FSC
explains this approach as follows:
During the 1975-1980 period, there were 50 wage
increases at the various plants of Florida Steel. Of
these increases, 33 were across the board increases
which granted the same dollar increase to employ-
ees regardless of grade. The remaining 17 wage in-
creases provided for larger increases in dollar terms
(though perhaps not in percent terms) to employees
in the higher salary grades in order to maintain pro-
portional wage differentials Our estimates [from the
proportional index model] provides for the mainte-
nance of proportional wage differentials. [The
across-the-board index model] provides for only
aeross the board wage increases. The above esti-
mates allow for both, with a weight of two attached
to the- across the board estimates and a weight of
one granted to the proportional wage increase esti-
mates. These weights reflect the 33 across the board
increases compared to the 17 proportional wage in-
credses _which were granted at Florida Steel plants
during the 1975-1980 period 87
Thus the "combination index" reconstructed wage for,
say, Charlotte Grade 1 for February 1980 is determined
as follows:
1. The reconstructed wage increase between Feb-
ruary 1973 (the last base period wage increase) and
February 1980 for Charlotte Grade 1 'employees
under the proportional index approach is $1.98;
under the across-the-board a[3proach it is $2.47:
2. $1.98 plus two times $2.47, all divided by 3,
equals $2.31. The combination index-Model's recon-
structed 1980 wage for 'Charlotte Grade 1 accord-
ingly is $1.31 plus the last base period wage ($2.81)
or $5.12
As an example of the impact of the combination ap-
proach, it produces a 1980 Charlotte Grade .9 to Grade 1
ratio of 1.47:1. In contrast, the comparable proportional
index ratio is 1.65 . 1. The comparable across-the-board
index ratio is 1.40:1.
Table 7 [see Appendix GI lists a sampling of the 1980
reconstructed wages produced by FSC's various index
87 E g, revised RX-79A-2 (emphasis in original)
FLORIDA STEEL CORP. 905
models and the analogous wages from FSC's regression
mode1. 8 8
Advantages of FSC's Index Models
As discussed earlier, changes in AHE levels reflect the
net impact of the various forces impinging on the labor
market. Over any relatively brief time period changes in
any one company's wage levels may be—in fact are
likely to be—different from changes at that same p- omt'm
AHE. But over a period of years those differences, on
the average, will diminish absent some long-term set of
circumstances unique to that company. Moreover,_ for
comparable reasons changes in any one company's wage
levels are no more or less likelY to be proportionally
smaller than the respective • clianges in AHE levels than
they are to be larger than such AHE changes—again,
absent circumstances at the company that push one way
or the other. (It bears repeating that the reference is to
changes in wages versus changes in AHE- No one con-
tends that any single employer's wage levels will tend to
be the same as AHE.)
Those considerations make FSC's index approach an
attractive possible answer to the wage reconstruction
problem facing the Board in this proceeding. And there
is an additional attraction: From the vantage point of
those affected by the reconstruction approach adopted
by the Board, the index method is relatively easy to un-
derstand. That certainly is not the case with either of the
regression approaches offered by the parties.
Problems With FSC's Index Models
--
Use of Multi-year Average. There is one overriding
problem with all of FSC's index models: They all are
based on the average difference during the base period
years between FSC wages, on the one hand, and AHE
levels, on the other. And as discussed in connection with
FSC's regression model, that use of multi-year averages
to establish the difference between FSC wages and AHE
levels is unacceptable. It fails to reflect the growth in the
FSC's wages relative to AHE 'during the course of the
base period.
Lack of Statistical Measurements of the Probable Accura-
cy of the Forecasts. Regression models can provide statis-
tical measures of the probable accuracy of the models'
forecasts. (See fn. 45, above, and the accompanying text.)
FSC's index models cannot do that, and the General
Counsel attacks FSC's index models because of that.
In some settings regression analysis clearly provides
the best way to make sense out of numerical data. 89 In
this proceeding, for example, regression analysis would
have been the ideal method for determining the extent of
the accuracy of FSC's claim that in general it seeks to
set wages at 1 -04 percent of wage practice line levels. (As
it happens, no one used regression analysis for that pur-
pose.)
88 The figures in Table 7 [see Appendix G] are taken froin RX-79A-2,
revised, and RX-62 All figures are as of February 1980 except for the
FSC regression figures, which are as of January 1980
89 See Coble v Hot Springs School District No 6, 682 F 2d 721, 731-732
(8th Or 1982)
But for a variety of reasons regression analysis did not
turn out to be a useful means for reconstructing FSC
wage rates. The subject has already been discussed (con-
nection with the General Counsel's model). But to reca-
Iiitulate briefly: (1) no party could provide an appropri-
ate basis for assuming the continuation in the backpay
Period of -the nonparallel relationship between FSC
wages and AHE levels that the General Counsel's re-
gression found to exist during the base period; and (2)
the base period provides insufficient data for regression
purposes. That second factor—insufficient data—led the
General Counsel to use a model design that presented se-
rious autocorrelation problems. And that, in turn, means
that the statistical indicators of accuracy produced by
the General Counsel's model are meaningless. As Dr.
Clapper, the author of the model, testified, it would be
inappropriate to rely on any of such indicators.
• As for FSC's regression models, they do not use re-
gression techniques to determine the nature of the rela-
tionship between FSC wage rates and area wage levels.
Rather; FSC used regression analysis as a way of arriv-
ing at averages in sets of data. Thus the statistical meas-
ures produced by FSC's regression models are beside the
point.
The contention that regression analysis is a superior
technique because of the "statistical tests grounded in
mathematical proofs of the model's overall correct-
ness'-' 9 ° accordingly fails to come to grips with the impli-
cations of the autocorrelation inherent in the General
Counsel's models and with the topsy turvy way that
FSC chose to use regression analysis.
What all that adds up to is that none of the Charlotte
wage reconstruction models submitted into evidence by
the parties offers a satisfactory way of computing back-
pay But the record does suggest a workable alternative.
That alternative will be considered in part VI, after dis-
cussion of the parties' Indiantown wage reconstruction
models.
V. INDIANTOWN
A. The General Counsel's Indiantown Model
The General Counsel's Indiantown model, like his
Charlotte model, uses a multiple regression approach to
find the relationship between FSC wages and AHE
during a base period The Indiantown model replicates
the methodology of the Charlotte model except in five
respects. First, the AHE figures that Dr. Clapper corre-
lated with FSC's Indiantown wages were taken from the
"all manufacturing" AHE series for the State of Flori-
da. 9 ' Second, the wage 'rates of all of the Indiantown
grades moved in a sufficiently parallel fashion during the
base period to enable Dr. Clapper to use one regression
equation for all grades (instead of two equations, as in
the case of his Charlotte models). Third, the Indiantown
model uses a base period of September 1970 through De-
cember 1973 (The latter date was the effective date of
the last wage increase at Indiantown prior to FSC's un-
99 Steelworkers Br 5
91 Dr Clapper originally used the all manufacturing AHE series for
Palm Beach County That will be discussed below
906 DECISIONS OF NATIONAL LABOR -RELATIONS BOARD
lawful change in wage policy. As for the September
1970 date, FSC opened the plant in March 1970. But the
pre-September wage rates were not included in the
model since they were "not developed in the same fash-
ion" as the later rates.) 92 Fourth, instead- of reconstruct-
ing Indiantown's wage structure based on the average
dollar difference between one grade and the next
throughout the base period, as in the Charlotte models,
Dr Clapper used the average difference during the last 2
years of the Indiantown base period. Fifth, the Indian-
town model includes a "jump shift" adjustment to one of
the wage increases granted by FSC . during the base
period.
In that last respect, Dr. Clapper discovered that the
model (without the jump shift) resulted in a disappoint-
ing "fit" (relationship between regression equation and
data). An investigation into the causes of the problem led
Dr. Clapper to ask whether the February 1973 increase
at Indiantown was an unusual one. Told that it was, Dr.
Clapper made an adjustment to the model based on the
assumption that part of the February 1973 increase was
an extroardinary, one-time event (The February 1973 in-
crease will be further considered below )
The model, as adjusted, produced a slope of 1.57. That
is, during the base period, averaging among the various
labor grades, for every one'- dollar increase in AHE,
FSC's Indiantown wages increased by $1.57. 93 -Table 9
[see Appendix G], shows examples of the resulting re-
constructed wages. 9 4
The General Counsel's Indiantown model presents the
same problems as his Charlotte models. And thus, like
his Charlotte models, the Indiantown model is not air ac-
ceptable approach to reconstructing the wages of FSC
92 Tr 1067 (witness Clapper) All parties concur in the use'of Septem-
ber 1970 as the start of the Indiantown base period Actual base period
wage rates for FSC's Indiantown plant are set forth in Table 8 [see Ap-
pendix G]
93 In contrast, grade-by-grade regressions produced the following
slopes (Source GX-122)
Grade Slope
1 1 52
2 1 44
3 1 50
4 i53
5 i57
6 i59
7 i68
8 0
94 The table uses wages as of December since December 1973 was the
effective date of the last general wage increase prior to F.SC's unlawful
change in wage policy (Thus, the December 1973 wage shown in the
table is the actual wage payed by FSC, not a reconstructed wage ) Table
9, and the other computations in this decision that use Honda statewide
all manufactunng AHE, are based on the following AHE levels
12 mos ended
December
Avg monthly
AHE level
1973 3 45
1974 3 76
1976 436
1979 5 48
1980 5 §8
employees for the backpay period. The model is beset by
high levels of autocorrelation. And again the record fails
to indicate why one should assume, just because Indian-
town wages increased faster than AHE between 1970
and 1973, that that trend would have continued thereaf-
ter but for FSC's wrongfiil change in wage-policy.95
AHE Considerations—Change in the Scope of the Indus-
try Grouping. The General Counsel originally used the
"all manufacturing" AHE series for Palm Beach County,
Florida, in his Indiantown model.-. Subsequently, howev-
er, the General Counsel learned that in 1972 BLS had
changed the scope of all the manufacturing series. The
change, which was made on a nationwide basis, had
varying impacts on AHE data, .depending upon geo-
graphical area. For reasons not clear on this record the
impact was particularly severe in the case of Palm Beach
County data, and BLS accordingly "declared a disconti-
nuity . . . or interruption . . . in the West Palm Beach
series,"99 That led the General Counsel to redo the
model using all manufacturing, data for the "Florida
statewide" area. 9 7
But again, while the AHE all manufacturing series for
the State of Florida has its imperfections, it appears to.be
at least as good a time series, for purposes of reconstruct-
ing Indiantown wages, as any other series suggested by
the parties. 9 8
95 As Table 9 [see Appendix G] indicates, under the General Counsel's
principal Indiantown model the wages of the reconstructed lower Indian-
town grades Increase relatively faster than do the upper grade wages
(That is comparable to the General Counsel's reconstructed Charlotte
wages see fn 49, above ) Under a revised General Counsel model in
which the wages of all grades increase proportionally, the growth of the
reconstructed FSC wages relative to AHE would be as follows grade I
would increase from 10 percent of AHE in 1973 to 112 percent in 1980,
grade 5 would increase from 136 percent of AHE in 1973 to 151 percent
in 1980, and grade 9 would increase from 178 percent of AHE in 1973 to
197 percent in 1980 See GCX-126
96 Tr 3728 (witness Haworth) Additional discontinuities were de-
clared in the Palm Beach all manufacturing series again in 1977 and 1978
93 The model that used the Palm Beach County data produced much
lower projected wage rates than did the model that used Florida state-
wide data E g, the Palm Beach model produced a Grade I wage of
$4 67 and a Grade 9 wage of $7 32 for December 1978 The comparable
figures from the Florida statewide model are $6 01 and $8 66 Thus the
General Counsel's switch in the AHE series was understandably upset-
ting to FSC (The Company had quickly agreed to pay the backpay
amounts resulting from the Palm Beach model's wage reconstructions
and in fact put the model's then current wage reconstructions Into effect
as the wage rates for its Indiantown employees ) But the General Counsel
made the change in timely fashion and for the reasons discussed above,
not for arbitrary or capnclous ones Accordingly it is the Inchantown
model based on Florida statewide data, rather than the model based on
Palm Beach County data, that is appropriate for consideration here Trin-
ity Valley Iron & Steel Co v NLRB, 410 F 2d 1161 (5th Or 1969)
98 FSC's Tampa facility is in the city limits of Tampa, unquestionably
an urban setting The Indiantown facility is in a rural area Generally
speaking, wage rates in urban areas are Wier than wage rates for com-
parable jobs in rural areas Nonetheless the Inchantown wage forecast by
the General Counsel's model are considerably higher than the wages at
the Tampa facility FSC claims that these circumstances argue against the
reasonableness of the General Counsel's forecasts As indicated above, I
agree with FSC that the General Counsel's Indiantown model should not
be adopted But that is not because I think that FSC's Tampa argument
has any validity The location of the Indiantown facility is such that the
majority of its employees have to commute relatively long distances (as
will be discussed below) No one claims that that is the case at FSC's
Tampa facility And the record is clear that employers must pay a premi-
um to attract workers to jobs necessitating long-distance commuting
FLORIDA STEEL CORP 907
, B. FSC's Indiantown Regression Model
From 1970 through 1973 FSC conducted wage sur-
veys in the Indiantown area FSC conducted no such In-
diantown wage surveys from 1974 through 1979. But
then, in late 1980, TSC conducted another wage survey
in the Indiantown area (solely for the purpose of prepar-
ing eVidence . for submission in this proceeding).
FSC's Indiantown regression model is much the same
as its Charlotte regression model, but with Indiantown
survey data playing the role in FSC's Indiantown model
that ,CPI data played in FSC's Charlotte model. The
major difference is that while CPI data are available for
all years after 1969, postbase period survey data are
available at Indiantown only for 1980.
As in its Charlotte model, FSC Indiantown model gen-
erates reconstructed wages by determining: (1) average
annual area wage level; (2) base period between the aver-
age the FSC wage rate and the average Indiantown
survey wage rate.
Thus to use FSC's Indiantown regression model to re-
construct the wages for, say, Indiantown Grade 5 for
1980, one proceeds as follows (using the figures in RX-
65):
(1) Intercept (base year figure): $1.659.
,(2) Year add-on for 1980: $2.125.
(3) Average gap between labor grades for 1980:
$.633 Five times $.633 (since we are here recon-
structing grade 5) equals $3.165.
(4) FSC add-on: $.447
(5) The reconstructed 1980 wage rate for FSC's
Indiantown Grade 5 is $7.40. $1.659 plus
$ 447)
Wage Rates for 1974 Through 1979. As touched on
above, FSC's surveys of Indiantown area wages are
available for 1970-1973 and 1980. No surveys were con-
ducted in the years 1974 through 1979. One way of han-
dling the absence of survey data for the years 1974
through 1979 would be a simple, straight-line, interpola-
tion. And FSC does provide reconstructed wage 'rates
determined in that manner. 99 But FSC points out that
while average area wages • increased each year in the
period 1974-1979, the rates of growth varied from year
to year. To take account of that variance, FSC used
AHE data to determine year-to-year rates of growth and
then applied those growth rates to the reconstructed
wage rates produced by the model. (AHE data show
that growth rates were much higher in 1978 and 1979
than in the 1974-1977 period. The impact of that pattern
on reconstructed wage rates is to diminish backpay
amounts relative to the amounts produced by a straight-
line interpolation.)100 The resulting revised version of
99 RX-66
1 °° See Tr 1970-72 and 2597-98 for Dr Haworth's explanation of her
interpolation procedure Dr Clapper took issue with that procedure and
proposed an alternative See Tr 2784-87 and GCX-112 If I had to
choose, I would pick Dr Clapper's interpolation method as being a more
logical and straightforward approach
the reconstructed wage rates is FSC's proposal for the
wage rates the Board should adopt in establishing back-
pay amounts for Indiantown employees. Those revised
rates are set out in RX-68.
Table 10 [see Appendix G] compares FSC's and the
General Counsel's proposed reconstructed Indiantown
wage rates for illustrative years and labor grades.
The Major Faults of FSC's Indiantown Model
Use of a Differential Between FSC Wages and Survey
-Wages Based on a Base-Period Average. FSC's Indiantown
model computes one average FSC Indiantown wage for
the 3 years 1970, 1971, and 1973, and compares that
figure to the average wage paid by the surveyed compa-
nies in those 3 years. (Actually the mechanics of that
computation are different, and more complicated, than
that description.) It turns out that FSC's wages averaged
44.7 cents higher than the surveyed companies' wages.
Under FSC's approach, wages for the post-1973 period
are reconstructed by first reconstructing survey wages
for the various labor grades and the various post-1973
years and then adding the 44.7 cents to each such recon-
structed wage rate. , .
The use of a constant dollar figure (44.7 cents) rather
than a percentage figure as the FSC difference presents
obvious problems. (See the discussion of this point in
connection with FSC's Charlotte model.) But that can be
put aside as easily curable. The real problem with FSC's
approach relates to the changes during the course of the
base period in the relationship between, on the one hand,
FSC wages and, on the other, the average level of wages
paid by other employers. The nature of those changes is
most readily seen in a straightforward comparison of av-
erage FSC Indiantown wages during the 1970-.1973
period with AHE data for that same period. As can be
seen [see Appendix G, Table 11], average FSC wages in-
creased faster than AHE both in dollar and percentage
terms.1°1
The implicit assumption in FSC's use of multi-year
averages to arrive at the FSC differential (44.7 cents) is
that the differences between FSC Indiantown wages and
the average wages of other employers that existed at the
end of the period are likely to have returned to earlier
.levels (had FSC not changed its wage-setting pattern).
But that assumption is contradicted by the substantial in-
crease in FSC wages relative to those paid by other em-
ployers
FSC made no attempt to explain why the differential
between its Charlotte wages and average area wages
generally was larger at the end of the base period than at
the beginning. But as to the Indiantown situation, FSC
did go to considerable lengths in an effort to show that,
for two reasons, the relatively rapid increase in Indian-
101 Average FSC Indiantown wages and AHE levels are taken from
RX-71A-1 The General Counsel's grade-by-grade regressions show that
practically all of the various FSC Indiantown wage rates grew propor-
tionally much faster over the base period than did AHE (see fn 93,
above) Looking at grade-by-grade comparisons of FSC wages and the
wages paid by the companies in the Inchantown-area surveys (see GGX-
106), against most of the FSC wage rates increased dunng the base
penod relative to their survey counterparts
908 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
town wages during the base period was without long-
term significance. The two reasons put forth by FSC: (1)
the fact that the Indiantown facility first started oper-
ations in 1970; and (2) FSC's response to wage controls.
"Start-up" Wages at Indiantown. FSC claims that its In-
diantown facility was in a "start-up" phase throughout
the base period According to the testimony of James
Hogue, FSC's vice-president for industrial relations, it is
FSC's policy to set initial wages at a new facility at be-
tween 108 percent and 112 percent of the relevant wage
practice line. The Company then aims to get back to its
usual 104-percent standard within 4 or 5 years.
That policy, Hogue said, developed out of its experi-
ence at Indiantown. Accordingly, while a start-up wage
premium was not implemented immediately at the In-
diantown facility, one was put into effect "sometime
after the plant started.' ,102 What that means, argues
FSC, is that the large differential that existed at the end
of the base period between FSC's Indiantown wages, on
the one hand, and, on the other, average area wages, was
a function of FSC's start-up needs.
FSC points out that a start-up premium was essential
at Indiantown for several reasons. First, as is the case
with any new facility, it must be staffed at least partially
with employees drawn away from other employers, and
some premium is necessary just to overcome inertia.
Second, Indiantown is too small a community to have
had within it any large proportion of the people that
FSC needed to staff the new facility. Thus FSC had to
convince prospective employees either to move to the
Indiantown area or to commute relatively long distances.
And third, the kind of work offered at the FSC plant
was different from the types of work potential employees
in the area were accustomed to.
But the record as a whole paints a different picture
from the one drawn by the Company.
To begin with, based on FSC's description of its wage-
setting procedures, it appears that the Company did use a
start-up premium in establishing its Indiantown wage
rates in September 1970. The Indiantown wage structure
in September 1970 was generally parallel to, and about
12 percent higher than, the wage practice line 103 That
closely fits Hogue's description of start-up wages.
Had FSC then followed the start-up procedures por-
trayed" in Hogue's testimony, FSC wage levels would
have gradually decreased, relative to the wage practice
lines, to a level just slightly higher than area standards.
But what in fact happened is that in December 1971 FSC
precipitously dropped the middle and upper grade wages
to well below the level of the wage practice line: At the
same time FSC increased, on a relative basis, entry level
wages. Moreover, there was not hint of an overall de-
cline in FSC wages relative to area wages at any time
during the rest of the base period (No one with any
knowledge of the employment situation at the Indian-
town plant explained that series of relationships between
FSC's wage structure and the relevant wage practice
lines.)
"2 Tr 1894
103 See Appendix B, for a graphic indication of that relationship [omit-
ted from publication]
One could speculate, I suppose, that as late as the end
of 1973 FSC was still trying to find a wage level high
enough to fill substantial numbers of jobs that had re-
mained vacant from the time the Indiantown plant got
underway. That would mean that FSC's management ini-
tially badly miscalculated the wage levels necessary to
fill the newly created jobs and then continued to miscal-
culate for the next 3 years, to the point of seeing needed
work go undone because of 3-year-long job vacancies
If that were in fact the case, it would be inappropriate
to assign much relevance, as with regard to predicting
future wage Patterns, to change during the base period in
FSC wages relative to area wages. But no FSC witness
testified that it took FSC 3 years to find the initial em-
ployees for many Indiantown job slots And the record
contains no employment data showing that jobs at the
Indiantown plant went unfilled from their creation in
1970 through 1973.
I am willing to assume that FSC found it harder than
it expected to retain its Indiantown employees, particu-
larly in the lower grades, and to fill openings when em-
ployees left the Indiantown plant.'" But retaining and
replacing employees (as opposed to the extensive recruit-
ing needed to fill new jobs in a new plant) are routine
functions in an ongoing plant. Thus, the fact that it was
soon after the plant went into operation that these prob-
lems may have cropped up has little if any significance
when evaluating the meaning of changes in base period
FSC wage rates relative to area wage levels
Wage Controls. Federally imposed limits on wage in-
creases were in effect for much of the base period. In
1972 wage increases of up to 5.5 percent for most work-
ers, and larger increases for the lowest paid workers
were permitted without specific governmental authoriza-
tion. FSC concluded that a 5.5-percent increase would
not be sufficient at the Indiantown plant and sought per-
mission to implement a 7-percent increase. Permission for
the increase was granted and FSC implemented it in Oc-
tober 1972. Then, in 1973, FSC sought permission for an
even larger exception to wage control limitations Per-
mission again was granted, , and FSC implemented wage
increases as high as 17 percent.105
104 The difficulties in retaining and replacing employees could be the
very same that FSC witness Hogue said made it hard to fill the newly
created Jobs when the plant opened in 1970 Thus, according to the testi-
mony of FSC's compensation manager in an earlier stage of this case, a
sizable wage Increase was needed in December 1973—
. because [Indiantown] is a very small community, also, it's
way out from any population center, and you need to have some-
thing to attract people to drive that far and work out in -the boon-
docks We have to have some sort of an incentive for people to
drive that far, to work out a great distance from the population cen-
ters, and also to work in an industry which is foreign to the area,
with quite different working conditions, hours, type of work, and ev-
erything else concerned with a steel mill, in relation to industry
common to the area We had to have some sort of a premium, in
other words GCX-39, p 268 See also GCX-38, p 77
Indiantown wages increased as follows in February 1973 (com-
pared with October 1972)
Grade, % increase Increase in cents
1,13% 35
2,13% 40
Continued
FLORIDA STEEL CORP. 909
The wage increases received by FSC's Indiantown em-
ployees in October 1972 and February 1973 were sub-
stantially larger than the increases received during that
period by the average worker in the Indiantown area.
But that, argues FSC, was merely a function of FSC's
ability to obtain special relief from Federal wage con-
trols. Most firms, and in particular small firms, did not
have that same capability. For that reason, so the argu-
ment goes, the faster-than-average growth in FSC wages
in 1972 and 1973 says nothing about the appropriate posi-
tion of FSC wages relative to average area wages in the
period.
But it does not seem to me that that argument holds
up under inspection For one thing, the record suggests
that small firms may have been at least as able as large
firms to put big pay increases into effect during the wage
control era. 1 " For another, many employees whose
wages were included in surveys against which FSC
wages may be compared were employed by companies
whose resources do not compare unfavorably to
FSC's. 107 But beyond all that, no one forced FSC to
obtain special wage increase exceptions. Rather, in late
1972 and early 1973 FSC obviously thought it necessary
to increase its pay scale at Indiantown relative to area
wages—which, it knew, were being capped by Federal
controls. And that is precisely the kind of circumstance
that ought to be considered when attempting to deter-
mine the appropriate level of FSC wages relative to av-
erage wages in the area.
What all that adds up to is that nothing in the record
suggests that the relationship between FSC's Indiantown
wages and area wages", as that relationship existed at the
end of the base period, was a fluke. Similarly, nothing in
the record suggests that that end-of-the-base-period rela-
tionship was more likely to return to earlier levels than it
was to increase still further. Yet FSC's regression model
is predicated on the assumption that FSC wages would
decline relative to area wages in the years after 1973.
Since that assumption is unfounded, FSC's Indiantown
regression model is unacceptable.
The 1980 Indiantown Survey. In December 1980 FSC
obtained wage rate data from employers in and around
Inchantown." 8 Data were sought only from firms that
3, 8% 28
4, 7% 27
5, 9% 35
6, 12% 50
7, 16% 75
8, 0
9, 17% 82 -
average wage,
all grades, 11% 43
'" See Tr 1991-92, 2277-78 (witness Haworth)
1 ° 7 Included in FSC's area surveys W R Grace. Coca-Cola, South-
ern Bell, and Florida Power & Light The largest employer in the Indian-
town area is another major corporation, Pratt & Whitney And while that
company was not included in FSC's surveys, its wage levels presumably
affected area wages
1 " FSC contracted with Hotton & Associates, Inc, to conduct the
1980 survey Robert L Brintle did the actual surveying Brintle and
George Hotton "calculated the data" Hotton "was responsible for the
preparation of the final report" on the survey Tr 1634 (witness Hotton)
Bnntle did not testify Hotton did
had been included in FSC's 1973 survey.'" And only
data from job categories included in the 1973 survey
were used for the 1980 survey.
The 1980 survey is obviously less likely to be repre-
sentative of area wages than even the base period FSC
surveys, given the constraints placed on the firms and
jobs to be included in the survey. 110 And no one con-
tends that the base period surveys were in fact represent-
ative samples. But for all that, I am not convinced that
1980 AHE figures are necessarily more reflective Of the
wages that FSC would have paid (absent FSC's unfair
labor practices) than the figures derived from the 1980
survey. The point is that during the base period FSC did
pay heed—to some undeterminable extent—to its area
surveys And the 1980 survey much more closely ap-
proximates the survey that FSC would have conducted
in 1980 had it continued its pre-1974 practices than does
any AHE series.
The main problem with FSC's use of Indiantown
survey data accordingly is not wholly a, function of the
quality of those data. Rather, the difficulty with using
survey data for wage reconstruction purposes is that no
surveys were conducted from 1974 through 1979. Thus
all of the wage reconstructions for . the entire post-1974
period produced by FSC's regression model are depend-
ent on one survey, the one conducted in 1980. Given the
substantial possibility that the survey's figures failed to
capture the labor market situation in and . around Indian-
town, that is simply putting too many eggs in one basket.
C. FSC's Inchantown "Index "Models
As discussed earlier, FSC proposed several ways , of re-
constructing its Charlotte wages utilizing the percentage
relationships between AHE data and FSC wages that ex-
isted during the 1969-1973 base period FSC submitted
evidence reconstructing the wages of its Indiantown fa-
cility using precisely the same approaches. Again, FSC
provided wage reconstructions based on the assumptions
of (1) continuous percentage increase in wages; (2) con-
tinuous across-the-board wage increases (i.e. same dollar
amount of increase for each wage grade); and (3) a com-
bination of percentage and across-the-board (in a ratio of
two across-the-board increases for every -percentage in-
crease) Table 12 [see Appendix G] shows the recon-
structed wages produced by FSC's index models for the
highest and lowest Indiantown grades.111
But as is the case with all of FSC's other wage recon-
struction efforts, its Indiantown index models are unac-
ceptable in that they reconstruct post-1973 wages on the
basis of the average base period relationship between
FSC wages and those of other employers.
159 The 1980 survey included data from 16 companies There were 18
companies in the 1973 survey But 1 of the 18 had gone out of business
And with another employer, based on the information obtained by Brin-
tle, "we just could not match jobs as they existed in 1973 with what they
purported to have in 1980" Tr 1641 (witness Hotton)
110 That is not an attack on FSC's motivation in putting those limits
on the survey Had the survey included companies or jobs not on the
1973 list, the 1980 survey surely would have been criticized for its lack of
consistency
" 1 Source RX-71A-1, RX-68, and GCX-100
910 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
VI. WAGE RECONSTRUCTION—CONCLUSION
The previous two parts of this decision considered the
General Counsel's Charlotte and Indiantown regression
models, FSC's regression models, and FSC's various
index models. Those were the only proposed approaches
to wage reconstruction examined in any detail at the
hearing. Yet as discussed above at some considerable
length, none of them is a satisfactory way of computing
backpay. The massive amounts of factual data submitted
by the parties and the parties' veryi thorough analysis of
that data do, however, point to a way of reconstructing
FSC wages that adopts the most cogent contentions of
each of the parties.
^
A. Use of End-of-Base-Period Relationships
Any method of reconstructing -FSC's Charlotte and In-
diantown wages for the post-1973 years must, it seems to
me, take into account these considerations:
(1) A comparison must be made with some set of
earnings data available both in the pre-1974 and
post-1973 periods; and BLS's AHE is the best such
set.
(2) Absent specific circumstances to the contrary,
at any given-time an employer's wages are no more
likely to increase faster than AHE than to decrease
faster than AHE (and vice versa); and over a period
of, years the employer's wage levels will tend to
remain in the same proportional relationship with
AHE.
(3) During the post-1973 period, there were no
circumstances at play at FSC's Charlotte or Indian-
town plants that would indicate that FSC wages
would have proportionally changed either faster or
slower than AHE.
(4) Miring, the pre-1974 period, wages at FSC's
Charlotte and Indiantown plants increased propor-
tionally faster than did AHE, so that the difference
between the average FSC wage and AHE was sub-
stantially greater at the end of that period than at
any other time in that period.
(5) There is no evidence that the level of FSC's
wage rates, relative to AHE, at the end of the base
period was a temporary phenomenon and that
FSC's wages were thus likely to retreat relative to
AHE. Rather, the.generally steady upward trend of
FSC wages relative to AHE throughout_ the ,base
period is a positive indication that a further advance
in FSC wages relative to AHE was just as probable
as a decline ,(had FSC not changed its wage poli-
cies).
While the General Counsel's and FSC's models are
each consistent with some of those considerations, none
is consistent with all of them. There is, however, one ap-
proach—basically a modification of FSC's "index"
. models—that would be consistent with each of those
considerations. That is, reconstruct post-1973 FSC wages
based on the relationship between FSC wages and AHE
as that relationship existed at the end of the base period.
That approach: (1) uses AHE data; (2) takes into account
that a company's wage' levels are no more likely to in-
crease proportionally faster than AHE than to decrease
faster than AHE absent specific circumstances to the
contrary, (3) takes into account that no such circum-
stances were shown to have been present at FSC's Char-
lotte or Indiantown plants during the backpay period;
and (4) reflects the rise in FSC wages relative to AHE
during the base penod.
The Steelworkers referred on brief to an end-of-the-
base-period index model. FSC responded as follows.
Reliance on the last pre-violation period is erro-
neous and unreasonable . . . because such reliance
assumes that only one point in time accurately re-
flects long-term Florida Steel policy and that earlier
market adjustments are not relevant. Clearly, an ac-
curate reflection of Florida Steel policy would re-
quire more than one point in time. . . It is par-
ticularly anomalous that the [end of the base period]
. . . was during the lifting of wage and price con-
trols, when Florida , Steel was able to change' its
wage rates significantly. Other firms, reporting
earnings to the BLS for the AHE series appear not
to have acted as quickly. 112' -
FSC's contentions are not persuasive. For one thing, it
simply is not "unreasonable" to base a prediction of a
company's future wage levels on the relationship be-
tween AHE and the company's wage rates at one point
in time (absent information showing that that particular
relationship was idiosyncratic).
[U]nless there is reason to believe that you are .-
out of line with the market . . . I would expect 'that
[the percentage relationship between AHE and the
Company's wages] would be about the same as -it
was before. I would assume that we were in equilib-
rium at that moment, unless I had other inforina-
tion 113 -
An index model based on a number of sets of FSC-AHE
relationships would ordinarily be preferable to a model
based on only one. But that is only if each such set of
relationships was likely to be as good a predictor of
future FSC behavior as each of the others. And it is' that
qualification that points to the use of an end-of-the-base
period model. As reiterated throughout parts IV and V
of this decision, base period patterns show that the rela-
tionships between AHE and FSC wage rates prior to the
time of FSC's last base period wage increases ,are inap-
propriate for predicting FSC's probable postbase period
behavior.
As for FSC's wage control contentions, as noted in
part V they are premised on the assumption that the in-
crease, during the wage control period, in FSC's wage
rates relative to the wage levels at other firms was the
'fortuitous result of FSC's efficiency in 'dealing with gov-
ernment regulation, rather than the product of managen-
-al choice. That would be a dubious -prOpositiOn under
any circumstance. It is particularly so in the case of
FSC, whose management has been shown to be thor-
' 2 FSC's Limited Reply Br 4
" 3 Tr 3882 (FSC witness Haworth)•
FLORIDA STEEL CORP 911
ough, knowledgeable, and deliberate in all matters relat-
ing to employee compensation.
As one witness said, and as the record as a whole
proves, finding a way to reconstruct FSC's Charlotte
and Indiantown wages is "a very horrible task." " 4 It is
clear that a wage reconstruction model that reconstructs
FSC wages based on end-of-base-period relationships to
AHE will not precisely duplicate the wages that FSC
would have implemented had it not violated the Act. But
no method of reconstructing FSC wages could ever do
that. And the end-of-base-period approach is a more rea-
sonable one than anything - else anyone has proposed.
That is enough for me to conclude that it should be used
to reconstruct, for backpay purposes, FSC's Charlotte
and Indiantown wage rates for the years since 1973
B. Wage Structure
As may be recalled, FSC provided three different
ways of using an "index" approach to wage reconstruc-
tion . (1) a straight percentage approach; (2) an "across-
the-board" model, and (3) a combination of (1) and (2).
The main difference between them is that under method
(1) the percentage difference between the various grades
stays the same (as wage levels increase over time) while
the dollar difference between grades increases. Under
method (2) the dollar difference between grades stays the
same while the percentage difference decreases. (Method
(3), of course, has some of the elements of both (1) and
(2).) Those same three approaches can be used with an
end-of-base-period model. And that means that it is nec-
essary to focus on wage structure: Had FSC not unlaw-
fully changed its wage policies, would the grade-versus-
grade wage relationships that existed during the pre-1974
base period have changed; and if so, how.
General Counsel's Wage Structure Contentions—Char-
lotte. As discussed earlier, the General Counsel's Char-
lotte wage reconstruction is based on two regression
models. Both of those models assume that the various
grades would have maintained a constant dollar differ-
ence from one another throughout the post-1973 years
but for FSC's unlawful behavior. For example, under the
General Counsel's proposed wage reconstruction, al-
though the Charlotte Grade 7 wage increases from $3.69
in February 1974 to $7.47 in February 1981, it remains
exactly $1.27 less than the General Counsel's proposed
Grade 15 wage throughout that time. Thus, the percent-
age difference between those two projected wage rates
decreases. from 34 percent in 1974 to 16 percent in 1981.
The General Counsel based the projected grade-to-
grade dollar differences on the average grade-to-grade
dollar differences between the various Charlotte labor
grades during the 1969-1973 base period. (I e., the pro-
jected difference between, say, Grade 4 RF and Grade 5
RF was the average dollar difference between those two
grades during the base period.)' 15
The use of a, constant dollar difference between grades
is contrary to the grade-versus-grade behavior at Char-
lotte during the base period. Throughout the base period
114 Tr 3147 (Witness Yearn)
115 Actually, the model determined the average dollar difference be-
tween each grade and Grade 1 But the effect is the same
the dollar difference between grades steadily increased.
And the percentage spread between grades, which
would have decreased if a constant-dollar-difference had
been maintained, instead remained relatively steady. 116
Table 13 [see Appendix G] shows the percentage and
dollar difference between the highest and the lowest
grades covered by the General Counsel's Charlotte
models.
The General Counsel once again points to Tampa as
support for his projections. And the fact is that at the
Tampa mill the General Counsel's assumption of fixed
dollar differences between grades is borne out through-
out the 1976-1980 period "reconstructed" by the General
Counsel's Tampa model: The dollar difference between
any two labor grades remained constant (which resulted
in a substantial decrease in the percentage spread). But I
cannot conclude that the Tampa data permit the conclu-
sion that but for FSC's unlawful wage policy change
FSC would have maintained a fixed dollar difference be-
tween the nine lower Charlotte labor grades and be-
tween the various upper labor grades. First, there re-
mains the fact that Tampa represents but 1 of FSC's 11
facilities. And second, nothing during Charlotte's base
period hints at such a trend. The Tampa projection, on
the other hand, precisely reflects the fact that during the
Tampa "base period" FSC did maintain a constant dollar
difference between grade levels.
Wage Structure—FSC's Regression Model As discussed
earlier, the CPI wage surveys include wage data on a
labor-grade-by-labor-grade basis. AHE surveys, of
course, do not. FSC's regression model uses the CPI
data to obtain, on a yearly basis, the average gap be-
tween labor grades. (For example, FSC's models show
the average gap between adjacent CPI labor grades in
1974 to have been about 28 cents. The FSC model then
uses the CPI average-gap-between-labor-grades figure
without modification to reconstruct FSC wages. Thus
FSC's 1974 forecast for the wages of Grades 5 and 6 are,
respectively, $3.99 per hour and $4.27 per hour. i e., 28
cents apart. 1 17)
There are several problems with FSC's approach to
reconstructing Charlotte's wage structure. But of most
interest to us here is that FSC's regression model's proc-
esses inextricably ,mix together the predicted average
yearly increase in wages (for all labor grades) and the
predicted average gap between labor grades (Statisti-
cians call the problem one of "multicollinearity. ")118 As
116 The grade-by-grade regression models for Charlotte (see Table 2
[Appendix G]) show with relatively minor exception a steady increase in
slope from low labor grade to high That further reflects the invalidity of
the constant-dollar-spread assumption
" See RX-62
118 Dr Haworth defined_ multicollineanty this way (referring to multi-
ple regression models) "[W]hen two vanables are related to each other,
the coefficient of one of them in most cases [is] likely to take up some of
the effect of the other one The coefficient [of] that particular variable is
really supposed to be measuring the impact of that variable But perhaps
it's measuring a Joint Impact of both variables That's called confounding,
because both vanables' effects may be incorporated or Imbedded in a
single coefficient " (Tr 3799-3800)
912 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
a result there is a possibility that, for any given year,
part of the grade-to-grade gap figure shown in FSC's
model in fact should be ascribed to the year-to-year
change. The reverse is true too. It could be that for any
given year the grade-to-grade gap should be increased to
include part of what the model shows to be the year-to-
year difference.113
The premise behind FSC's use of CPI data to recon-
struct the relationship of the various FSC grades to one
another is that FSC's wage structure will inevitably
refect market forces. The premises is an accurate one.
And it points to examining CPI data directly, so as to
avoid the multicollinearity problems generated by FSC's
regression model.
Wage Structure Information Produced by the CPI Sur-
veys. The General Counsel has compiled grade-by-grade
CPI data for grades 1 through 9, for the years 1970
through 1980. 120 The data show that dunng the 4 years
1970 through 1973 the average CPI grade 9 wage rose
from 66 percent higher than the average CPI grade 1
wage to 81 percent higher. [See Appendix G, Table 14.]
Generally speaking, the average CPI grade 9 wage in-
creased relative to the CPI grade 1 wage during the
years following 1973. [See Appendix G, Table 15.]
Charlotte Wage Structure—Conclusion. As noted earli-
er, using an end-of-the-base period index model, and the
three most thoroughly considered index methods, the
choice is among models producing the following rela-
tionships between FSC grades 1 and 15.
Method

Grade 15 as Percent of Grade
1 As of Feb. 1980
"Proportional"

181%
"Across-the-Board"

140%
"Combination"

153%
FSC's witness Haworth suggested that, if an index
method is to be used, she would recommend the "combi-
nation" approach 'since it reflects the way FSC granted
wage increases at its other plants And a combination ap-
proach would be closer to the preference of the General
Counsel and the Steelworkers than would the propor-
tional approach.
Nevertheless it seems to me that the proportional ap-
proach is the more appropriate one. FSC's Charlotte
grade 15 wage rates ranged from 174 percent to 181 per-
cent of the grade 1 wage rates during the base period.
Thus the proportional model more closely squares with
Charlotte's wage structure during the base period than
does any other index approach. And it is the closest to
market realities, as shown by CPI data.
I accordingly conclude that for backpay purposes the
wages of FSC's hourly paid bargaining unit Charlotte
employees should be reconstructed by using an index
" 9 "Because of the way Respondent's model is built and the difinition-
ally Induced multicollinearity, it's virturally impossible to say with any
reliability what the wage structure would be at any point in time, because
you just can't disentangle the wage structure issues from the tyear-to-
year] trend issues" Tr 2767 (witness Clapper)
120 GCX-105 Under the CPI system the higher the grade number the
lower the skills and wage level But throughout this proceeding the par-
ties revised the CPI numbering system to track FSC's (under which the
higher the grade number the greater the skills and wage level) This deci-
sion follows the parties' practice
model based on the percentage relationships that existed
in February 1973 between, On the one hand, each of the
various FSC•wage levels and, on the other, the Charlotte
SMSA primary metals AHE fOr February 1973 (12-
month moving average). -
Table , 16 [see Appendix G] gives some examples of the
reconstructed wage _rates produced by the model adopt-
ed by this decision, together with analogous wage rates
produced by the principal models . of the General Coun-
sel and FSC.'21
Appendix D gives further examples of the reconstruct-
ed wage rates produced by the model adopted by this
decision:
Wage „Structure—Indiantown. During the base period
the percentage _difference between Indiantown grades 9
and 1 (the highest and lowest grades) ranged between 76
and 91, as Table 17 [see Appendix G] shows.'22
As with his Charlotte models, the General Counsel's
principal Indiantown wage reconstructions maintain a
constant dollar difference between grades, resulting in an
enormous decrease in the percentage difference. (The
General Counsel's reconstructed grade 9 for 1980 is only
36 percent greater than his reconstructed grade 1.) _Since
the evidence supporting that approach at Indiantown
no more convincing than it was in the case of the Gener-
al Counsel's Charlotte wage reconstructions, it must be
rejected.
FSC's model, utilizing Indiantown wage surveys for
the base period and for 1980, produces reconstructed
wages that result in grade 9 starting 78 percent higher
than grade 1 and increased gradually so that in 1980 it is
91 percent higher than grade 1. But as with FSC's Char-
lotte regression, -its Indiantown model presents multicol-
linearity problems. -
As for the Indiantown survey itself, it is based on so
small a sample that the average -grade-by-grade wage
rates 'that it produces bounce around far more dramati-
cally than they ought to. But for whatever it is worth,
the 1980 survey suggests that in the Indiantown area the
grade 9 relationship to grade 1 for that year was compa-
rable to that of the pre-1974 period 123
IndiantoWn wage structure—Conclusion. The "pro'por-
tionl" indei approach would produce the same relation-
ship between grade 9 and grade 1 at each point in the
post-1973 period as existed at the , end of the base period;
thus grade 9 would remain 76 percent greater than grade
1. • The "across-the-board" index approach would
produce a 'grade 9 wage that steadily declined relative to
grade 1, so that' by 1980 grade 9 would be only 39 per-
' 2 ' The labor grades referred to in the table are the greles based on
the old, I9-grade, structure I have used a four-place index figure (e g,
8125 for Charlotte grade 1 and 1 472 for Charlotte grade 15) The figures
given for FSC's regression model are as, of January; of the indicated
years, not February
122 The various individual Inchantown grade-versus-grade relationships
varied considerably For example. grade 3 started out (in 1970) 11 2 per-
cent higher than grade 2 and ended the base period only 5 3 percent
higher Grade 7, on the other hand, started 80 percent. higher than grade
6 and climbed to 11 7 percent higher at the end of the base period But
the grade 9 relationship to grade 1 is the best one to focus on for our
purpose in that It summarizes all the others
' 23 Data from GCX-106
FLORIDA STEEL CORP 913
cent higher. (Under the "combination" approach, grade
9 would be 50 percent higher than grade 1 as of 1980.)
Given the , base period wage relationships at FSC's In-
diantown plant plus the scanty data from the 1980
survey, my conclusion is that, • as with the Charlotte
wage reconstructions, the most appropriate index model
- one that maintains the end 7of-the-base-period grade-
versus-grade relationships—that is, the proportional ap-
proach. Thus my conclusion regarding the reconstruc-
tion of the wages of FSC's Indiantown hourly paid bar-
gaining unit employees is that such reconstruction be
based on an .index mOdel utilizing Florida statewide all
manufacturing AHE (12-month moving average); and
that the reconstructed wages reflect the relationship be-
tween the average level of such AHE for the 12 months
ended December 1973 and each of the ESC wage rates
that were retroactively made effective in that month.
Examples of the impact of that approach are set out in
Appendix G, Table 19, and in Appendix E., _
VII. POST-i9;3 DEVELOPMENTS AFFECTING WAGE
RECONSTRUCTION
"TEAM"
Up through early 1974 FSC followed -a policy of
granting general wage increases annually. But as inflation'
rates climbed in 1974 FSC concluded that it should in-
stead implement general wage increases quarterly. -FSC
called the new policy "TEAM. "124. FSC continued to
conduct wage surveys on an annual' basis (at plants other
than Charlotte and Indiantown), and thus each year one
Of the four TEAM' increases was influenced by the
survey. The size of the other three increases each year
reflected whatever tentative information could be col-
lected plus, of course, corporate decisKins about corpo-
rate needs.
FS,C resumed its annual wage Increase policy in late
1976.
The parties agree - that; but_ for FSC's unlawful change
in wage policy, FSC would have granted TEAM in-
creases at Charlotte during the period November 30,
1974, through November 27, 1976. The comparable In-
diantown dates are September 28, 1974, and September
25, 1976 125
As for reconstructing wage levels on TEAM dates,
the most reasonable approach is simply to reconstruct
each TEAM increase on the basis of the 12-month
moving average AHE figure for the month of the
TEAM increase—that is, use exactly the Same method
for reconstructing quarterly increases as is used for re-'
constructing annual increased (For' example,' the Chai-
lotte SMSA primary metals AHE -figure for the 12
months ended May 1975 was $3.91. Since FSC's Char-
lotte' grade 12 (mill) wage rate was 126.6 percent of
AHE at the end of the base period, the reconstructed
grade 12 wage 'rate for May 1975 is $4.95.) -
Transformation of the Charlotte Grade Structure. As dis-
cussed earlier, the Charlotte grade structure during the
pre-1974 base period specified 19 separate grades.. (See
Table 3, Appendix G.) ,FSC continued. to use that struc-
,"4 See part I, above
125 See Appendix F for the precise dates,of the reconstructed wage
increases
ture until April 1977. At that point FSC, with the Steel-
workers' concurrence, adopted a 9- or 10-grade structure
for the Charlotte plant. (FSC says it is a 9-grade struc-
ture; the -General Counsel says it is a 10-grade structure.)
That raises the question of how to transform the old 19-
grade structure' inte the new 9- or 10-grade structure for
wage reconstruction purposes. In this proceeding FSC
first transformed the Charlotte ,base period grades into a
9-grade structure, built its principal wage reconstruction
models around the transformed grades; and then re-trans-
formed the 9-grade structure produced by those models
back into the old 19-grade structure for purposes of de-
termining reconstructed Charlotte grades for the period
1974 to April 1977.
The General Connsel took a very different approach.
For one thing, the General Counsel built its Charlotte
wage reconstruction models around the old structure,
and then determined post-April 1977 wages by trans-
forming the , old structure into the new one solely from
that point on Second, the General Counsel made the
transformation in a noncomplex fashion. To determine
the reconstructed wage- levels for the transformed
grades, the General Coungel assumed that old grade 1
became, new grade 1, and that old grade 17 became new
grade 10. As for - new grades 2 through 9, the General
Counsel divided the reconstructed wage . difference be-
tween grades 1 and 17 by 9 and used the resulting figure
as the dollar interval between each grade. The following
is an example of the General Counsel's approach (using,
for illustration purposes, the General Counsel's recon-
structed wage rates for February 1978 and transformed
grades 1, 2, 3, and 1.0).
1. General Counsel's
reconstructed
Grade 1 wage
(untransformed) $4.31
2. General Counsel's
reconstructed
Grade 17 wage
(untransformed) $7.30
3 Grade 1 recon-
structed wage
(transformed
structure) $4.31,
4. Grade 10 recon-
structed wage
(transformed
structure) $7.30
5. Grade 10 wage
minus Grade 1
wage . $2.99
6. Difference from
step 5 divided into $2.99 divided by 9
nine -intervals - equals $.332
7. Grade 2
(transformed) - $4.31 plus $.33
reconstructed wage equals $4 64
8. Grade 3 (trans-
formed) recon- $4.31 plus two times
914 DECISIONS OF NATIONAL EABOR RELATIONS BOARD
structed wage • $ 33 equals $4.97
. The 'General Counsel's approach to.the'transformation
of Charlotte's wage structure is relatively simple to do
and easy to understand; and it is in line with FSC's inten-
tions in shifting to i new grade structure as those inten-
tions were described to representatives of General Co' tin-
sel. I accordingly conclude that the General Counsel's
approach ' to the transformation of Charlotte's ' wage
structure should be adopted by the Board for purposes of.,
reconstructing Charlotte wage rates.
Merit Pay Adjustment. Prior to the arrival of the Steel-
workers at FSC's Charlotte plant, FSC ,used a merit
system there. Under that system employees who were
evaluated as doing better than satisfactory work received
extra pay. FSC put that system on hold in 1974 (see sec.
VIII, G, below). Then, at the time of the grade structure
transformation in 1977, FSC eliminated the merit system
(with union concurrence) FSC thereupon increased the
base rate of pay (i.e., non-merit rate) for its Charlotte
employees by 4 percent.
The General Counsel proposes that that circumstance
be handled by adding 4 percent to all post-April 1977
(reconstructed) wage levels produéed by whatever Char-
lotte rriodel the Board choses..(For example, the general
Counsel's regression model produces a reconstructed'
Grade 1 wage rate for February 1978 of $4.31. Adding
the 4 percent merit adjustment brings the rate to $4.48,
which is the rate on Which the General Counsel urges
that backpay be based.)
FSC does not take issue with the 4-percent adjustment.
Accordingly the reconstructed Charlotte wages pro-
duced by the end-of-the-base-period index model de-
scribed earlier should be increased by 4 percent to com-
pute the backpay of the Charlotte hourly employees for
the period beginning April 24, 1977.
VIII. OTHER ISSUES
A. Dates on Which FSC Offered Reinstatement to
Unlawfully Discharged Employees
On various dates in 1973 Florida Steel fired a dozen
employees because of the employees' -union activities
The Board subsequently required FSC to offer the dis-
charged employees "reinstatement to their former or sub-
stantially equivalent positions without prejudice to their
seniority and other rights and privileies" and to "make
the . . . employees whole for any loss of earnings suf-
fered by reason of their discharge. "126 -
Issues remain regarding the dates on which FSC made
valid reinstatement offers to six of the discriminatorily
discharged employees. This section will deal with those
issues.
Bruce N Letifinond. Lemmond had worked as a truck-
driver at FSC prior to his discharge in July 1973. On
November 22, 1976, FSC Regional Industrial Relations
Manager James S. Rogers wrote a letter to L emmond
that included the following language:
126 214 NLRB 264, 281 (1974) The Board's order covered 13 employ-
ees That order of the Board was enforced in respect to 12 of the 13 dis-
charged employees in 551 F 2d 306 (4th Cir 1975)
You are hereby offered full; immediate 'and un-
conditional reinstatement to your former, or a sub-
stantially equivalent, position with this Company,
without prejudice to your seniority or other rights.
If you desire reinstatement, contact the under-
signed no later than Friday, December 3, 1976, for
instructions as to when and where to report for
work. . .
Lemmond called Rogers on December 3 and asked to
come in and talk to Rogers about the offer. Lemmond
wanted further information about the job that FSC had
planned, for him. Rogers refused, saying that the Novem-
ber `22 letter was "self-explanatory" and that there was
"no need" to meet on the 'natter.'" In the course of the
telephone conversation Rogers did add 'that: (1) Lem-
mond ,would receive the same rate of pay that he had
been getting in 1973; (2) Lemmond's reporting date
would be December 20, 1976; and (3)' there were no
tiuckdriver job vacancies—which left Lemmond with
the impression that he was being offered some sort of
"equivalent" job, not reinstatement into his old job.
Rogers did not tell Lemmond exactly what kind of job
Lemmond would get or on what shift Lemmond would
work.
On December 7 Rogers called Lemmond to advise
Lemmond of a change in the reporting date. Rogers told
Lemmond to report on December 13 rather than Decem-
ber 20. Lemmond responded equivocally, saying that he
was going to talk to his then current employer about the
situation and that he would be back in touch with
Rogers.128
On the date of that second telephone call, Rogers sent
another letter to Lemmond:
This letter will confirm our conversation of today
. . at which time we established your return to
work date as Monday, December ,13, 1976. You
- should report to my office at 8 a.m at which time
you will receive an appointment for your return to
work physical as well as your job assignment.129
Lemmo'nd did not communicate further with %piers
about the job offer and did not show up for work on De-
cember 13.
Representatives of the General Counsel subsequently
advised FSC that they did not consider FSC's communi-
cations with Lemmond to be a valid offer of reinstate-
ment. In response, in late 1977 FSC' , again offered em-
ployment to Lemmond, and all parties agree that that
1977_ offer was a valid offer of reinstatement effective
December 19, .1977. The issue is whether the letters and
telephone conversations of November and December
1976 constituted a valid offer of reinstatement to Lem-
mond. If they did, FSC's backpay obligations toward
Lemmond stopped running at that time. If, on the other
127 Tr 256
122 Lemmonil testified that he had only one telephone conversation
with Rogers, not 'two, and that it was Rogers who said that he would be
back in touch with Lemmond But I credit Rogers' account over Lem-
mond's in this respect, as the text above indicates
1 ? 9 GCX-11
FLORIDA STEEL CORP 915
hand, those communications did not amount to a valid
offer of reinstatement, FSC's backpay obligations contin-
ued to run until the time of FSC ?s . 1977 job offer, .not-
withstanding Lemmond's unexplained failure to show up
for work on the reporting . date: See, e.g., Craw & Son,
244 NLRB 241, 242 (1979), W. C. McQuaide, Inc., 239
NLRB 671 (1978)
In dealing with that issue one question that needs , to be
resolved is whether the reference in Rogers' December 7
letter to a physical examination rendered the job offer in-
valid for purposes of meeting the Board's reinstatement
requirements.. For in most circumstances a job offer that
is conditioned on the discriminatee passing a physical ex-
amination does not comply with a Board reinstatement
order. e.g., Craw & Son, supra. The question is a close
one since FSC does not routinely , require that its em-
ployees take periodic physical . examinations. Thus if
Lemmond had not been unlawfully discharged and had
'remained at FSC without interruption, he would have
come under no duty to take a physical examination. -
But FSC did not tell Lemmond that his reemployment
'was conditioned on his passing a , physical examination.
In fact the job offer letter (of November 22) said nothing
about a physical examination at all. Rather, the 'reference
to a physical examination (in the December 7' letter) indi-
cated that it would be given after' Lemmond reported to
work. And FSC nowhere stated that its employment of
Lemmond was Conditioned on his passing the -physical.
FSC requires phygicals of all 'emPloyees who
return to 'work after any substantial' period aivay" from
the job. Thus there was nothing discriminatory about the
requirement of a physical examination nor did it suggest
that FSC was treating Lemmond 'as a new employee.
Compare Woodlawn Hospital, 233 NLRB 782 (1977).
Under all these circumstances FSC's reference to a phys-
ical examination in its December 7 letter to Lemmond
did not invalidate the job offer, assuming that FSC's job
offer otherwise met the requirements imposed by the
Board's reinstatement order.
FSC's employment offer, however, did not otherwise
comply with the Board's- order of reinstatement.
FSC's November 22 letter ("You -are hereby offered
full, immediate and unconditional reinstatement to your
former, or a substantially equivalent, Position ") was
enough to require some sort of response from Lemmond.
If Lernmond had said nothing, or had declined the offer,
FSC's backpay obligation towards Lemmond would
have thereby , ended . e.g., Knickerbocker Plastic -Co., 132
NLRB 1209 (1961). But Leminofid did - respond, indicat-
ing that he wanted to find out more about what" FSC•had
-in mind FSC 'refused to provide the information that
Lemmond sought. None of FSC's communications to
Lemmond in November or December 1976 specified the
job or even the shift that FSC -planned for Lemmond.
That put -Lemmond in the position of- having to decide
whether to revamp the circumstances . of his life (by
13 ° References in this section to backpay will relate exclusively to,the
period between unlawful discharge and the effective date of a valid rein-
statement offer It should be understood, however, that the amount of
backpay owed by FSC to the six employees discussed in this .section
affected by FSC's wrongful withholding of general wage increases, as
discussed in parts I through VII of this decision
giving up his existing employment and returning to work
at FSC) without, any information about crucial aspects of
-his prospective work at FSC. I appreciate that FSC felt
that it would be less disruptive to its employees as a
whole if FSC could delay selecting jobs for returning
discriminatees until after FSC determined just who
would be returning. But that does not excuse FSC's fail-
ure to meet its obligation to be specific in offering rein-
statement: see Powell Valley Electric Cooperative, 236
•NLRB 1040, 1043-44 (1978); Pace Motor Lines, 260
NLRB 1395 (1982).
I accordingly conclude that the backpay owed by FSC
to Lemmond should be _ computed On the basis that
FSC's first and only valid reinstatement offer was effec-
tive on December 19, 1977.131
Alfred Haynes. F_ SC fired Haynes in May 1973. At the
time, Haynes was a Grade 9 forklift operator. He was
being paid $3.35 per hour base pay plus a shift differen-
tial (Haynes . was on a rotating shift) of 16 cents per hour.
On February 4, 1977, FSC sent a letter to Haynes that
reads, in substantial part: -
You are hereby offered full, immediate and un-
conditional reinstatement to your former, or a sub-
stantially equivalent position with this Company,
without prejudice to, your ,seniority or other rights.
If you, desire reinstatement, contact the under-
signed no later than Friday, February 18, 1977, for
..instructions as to when and to where to report . for
work.
Haynes responded with , a visit to Rogers' ,office.
Haynes credibly testified that:
-I told him' [Rogers] that I came to listen to see
what I was offered. . . . I told him I wanted to see
. what was offered and what I was going to be doing
. and what job I would have and what not and what
- I would be paid.'" -
Rogers responded .by saying that the job that Haynes
held at the time of his discharge had been discontinued
and that FSC was .not then in a position to tell Haynes
what job he would be assigned to, what department he
would be working in, or what shift he would be on.
Rogers did tell Haynes that his rate of pay would be
$3 35 per hour, the same base rate that Haynes had been
receiving when he was fired in 1973.133
Haynes then mentioned that, due to family needs, he
could work only during the day and argued that in view
of his seniority with the Company he ought to be enti-
tled to his choice of shifts. Rogers 'responded that the
Company did not take seniority into account in assigning
131 FSC initially placed returning discriminatees in entry level Jobs
(while paying them their old wage rates) Had Lemmond been aware of
that practice that alone would have been enough to invalidate FSC's Job
offer But there is no evidence that Lemmond (or any of the other em-
ployees referred to in this section) knew of that practice
13 ? Tr 266 ,
FSC had granted no general wage increases at its Charlotte facility
between the time Haynes had been fired and February 1977
916 DECISIONS OF NATIONAL ' LABOR RELATIONS BOARD
employees to shifts and reiterated that he could not tell
Haynes what shift Haynes would . be assigned.
That ended Haynes' conversation with Roger g . On
February 18, 1977, Haynes sent the following letter to
Rogers:
This is to inform you and the. Florida Steel Cor-
poration that I am not interested in returning to the
employ of the Company due to the pay scale you
offered me in our conference on February 15, 1977.
The amount per hour is not what I was making
when I left the Company and is not what I now
make at my present job.
Your offer at the conference does not meet with
what I was making when I left the Company's
employ in 1973 and therefore, this is not offering
me full and unconditional reinstatement with the
Company.134
As in Lemmond's- case, the Cbmpany's first letter to
Haynes (on February 4, 1977) was, on its face, a valid
reinstatement offer. But also as in Lemmond's case,
Haynes' questioning put the Company under an obliga-
tion to be more specific: Haynes was entitled to be told,
at the very least, what job he was to be assigned to if he
returned to work with the Company. Rogers' failure to
provide that information brought to an end any require-
ment on Haynes' part that he further respond to the
Company.
-• Haynes did respond, however, saying that he was not
going to return to FSC because of his dissatisfaction with
the wage FSC offered. And the letter says nothing about
declining to work for FSC because of uncertainty about
the nature of the prospective job. But the reasons
Haynes gave for refusing to return to FSC are irrelevant.
Due to Rogers' lack of specificity, at the end of the
meeting between Haynes and Rogers no valid reinstate-
ment offer was outstanding: 138 And a valid reinstate-
ment offer must be outstanding in order for an employ-
ee's refusal of employment to toll the backpay- period:
e.g., Craw & Son, supra; W. C. McQuaide, Inc., supra..
In sum, neither FSC's commtinications to Haynes in
February 1977 nor Haynes' February 18 letter to FSC in
which he rejected FSC's offer brought FSC's backpay
obligations toward Haynes to an end.136
Charles D. Frazier. FSC fired Frazier on May 4, 1973.
Using its usual language the Board ordered FSC to rein-
state Frazier and to pay him backpay. All parties agree
that in December 1977 FSC made a valid reinstatement
offer to Frazier and that, accordingly, Frazier's backpay
period ended no later than December 19, 1977. But as
with Lemmond and Haynes, FSC contends that it made
an earlier valid reinstatement offer to Frazier and that
Frazier's backpay period thereby ended in December
1976.
134 Rogers responded to Haynes' February 18 letter with a letter dated
February 22 In that letter Rogers reconfirmed that Haynes' "base rate of
pay" at the time of Haynes' discharge was $3 35
133 See Powell Valley Electric Cooperative, supra
136 All parties agree that FSC made a valid offer of reinstatement to
Haynes in December 1978, which Haynes declined Thus FSC's backpay
obligations to Haynes ended on Decembbr 21, 1978
On November 22, 1476, Rogers wrote a letter to Fra-
zier that was identiCal to the letter he -had sent to Lem-
mOnd on that same day. Rogeis sent the letter by certi-
fied mail, return receipt requested, to Frazier's address at
the time he was -fired, as listed in the Company's person-
nel records. The Postal Service ultimately returned the
letter to Rogers on December 8, with the indication that
the Postal Service had tried to deliver it but that the
'letter had gone "unclaimed."137
FSC made-ho further attempt to contact Frazier.
As discussed earlier, the language used in the letter
sent to Frazier represents a valid reinstatement offer.
That being the case, if FSC's transmittal of the letter rep-
resented a reasonable, good-faith effort to. contact Fra-
zier, FSC's backpay obligations to Frazier ended on the
last day he could have responded to the offer, December
3. 138 The question is whether FSC did make a reasona-
ble, good-faith effort to contact Frazier. And that ques-
tion is a close one.
On the one hand transmittal of a reinstatement offer by
certified mail to an employee's list known address is an
appropriate means of attempting to communicate without
,more satisfies the , obligations placed on an employer.by a
Board order requiring reinstatement of an employee: see,
e.g., Knickerbocker Plastic Co., supra; General Iron Corp.,
218 NLRB 770 (1975).
But in view of the circumstances present here, my
conclusion is that FSC did not satisfy its notification ob-
ligation by its one letter to Frazier. For one thing, the
letter was returned to FSC by the Postal Service, so that
FSC knew that Frazier had not received the notification.
For another, FSC had discharged Frazier 3-1/2 years
prior to its transmittal of the November 1976 letter to
him, and the address used in that letter was taken from a
file that had not been updated since Frazier had been dis-
charged. Third, the plant's employees had a bargaining
unit representative, the Steelworkers, and it would have
been a simple and obvious thing for FSC to have enlisted
the Steelworkers in attempting to locate Frazier. But
FSC did not do that Finally, the Steelworkers affirma-
tively sought, information from FSC concerning the iden-
tity of the discharged employees who were receiving re-
instatement offers. FSC refused to provide the informa-
tion.139
That, it seems to me, adds- up to something less than a
good-faith effort by FSC to reach Frazier , with a rein-
statement offer. I accordingly conclude, that FSC's back-
pay obligation. towards Frazier was not tolled until De-
cember-19, 1977. , .
'John Wells, Jr. On November 22, 1976, FSC sent a
letter to Wells.worded identically to the letters it sent on
that date to Lemmond . and Frazier. -,
. According to Rogers, Wells telephoned Rogers on
December 1, 1976, -in response to the November 22
letter. The focus of the conversation was on Wells' -pro-
spective reporting date at FSC. Rogers testified that
"7 RX-7
138 Southern Household Products Co, 203 NLRB 881 (1973) No one
argues-that FSC's demand in its November 22 letters that the addressees
respond by December 3 was unreasonable
139 See Florida Steel Corp. 242 NLRB 1333 (1979).
- FLORIDA STEEL CORP :917
Wells did not ask for more details about the job to which
FSC planned to assign him. But according to Rogers he
did indicate to Wells -that Wells "would not necessarily
•be going back to the same identical position which he
was in'.' at the time he was fired, but that "he [Wells]
would be making the same rate of pay as that which he
was making at the time he left."
Wells did not testify
Following Rogers' December 7 conversation with
Wells, Rogers sent the following letter to Wells:
This letter will confirm our telephone conversa-
tion of today, December 7, 1976, at which time we
established your return to work date as Monday,
December 13, 1976. You should report to my office
at 8 a.m. at which time you will receive an appoint-
ment for - your - return to work physical as well as
• your job assignment. -
As we discussed previously, your rate of pay will
be $3.17 per hour.14°
;Wells did not report to work on December 13, 1976,
and there is nothing in the record about why he failed to
report
My recommendation is that the Board conclude that
FSC's communications to Wells in 1976 constituted a
valid reinstatement offer and that FSC's backpay obliga-
tions towards Wells were tolled on December 13, 1976,‘
the date Wells failed to report for work.
As indicated earlier, the language that FSC used in its
initial reinstatement letters—"You are hereby offered
. reinstatement to your former, or a substantially
equivalent. position"—while ambiguous, is sufficient to
require some sort of response from the employee. But if
the employee does respond, asking for further details
, abotit the job to which he will be assigned, a failure of
the employer to provide that detail precludes the em-
ployer from claiming that the letter was a valid reinstate-
ment offer. Here Wells did respond, and in the course of
his conversation' with Rogers, Rogers said that Wells
would not "necessarily" be getting his old job back and
said nothing about what job Wells would be getting. But
Wells did not ask for any information about the kind of
job that FSC had planned for him (at least insofar as the
record indicates). And since Wells did not ask for more
information about the nature of the Work he would be
given, my conclusion is that the initial validity of FSC's
letter offering reinstatement stands ummpaired."1
James W. Ashcraft and Loren D. Theodore. On Novem-
ber 22,- 1976, Rogers sent letters to Ashcroft and Theo-
dore that were identical in wording to those he sent on
s that day to Lemmond, Wells, and Frazier. 142 The letters
were sent by certified mail and, according to the Postal
Service, both were received by the addressees a few
days after they were sent- As.hcraft did not respond to
the letter 'and, as far as the record indicates, never com-
140 GC-12.
. 141 As for the reference to a physical examination in Rogers' Decem-
ber 7 letter to Wells, see the above discussion about Rogers' letter to
Lemmond
1 42- Copies of both letters are included in the record as RX 7 1 (Ash-
craft) and RX-2 (Theodore)
municated thereafter with FSC. Theodore responded to
Rogers' letter with a letter of his own declining the job
offer 143
As discussed earlier, if Ashcraft and Theodore had re-
sponded to Rogers' letters with expressions of interest
and if they had inquired further about the nature of jobs
they were to be offered, Rogers almost surely would
have failed to provide them with a reasonable amount of
detail about their forthcoming job. Moreover, if they had
nonetheless accepted the job offers, the record is clear
that FSC would initially have placed them in entry level
positions—in violation of the terms - of the Board's order.
But there is no indication whatever in the record that
either Ashcraft or Theodore was aware that those cir-
cumstances would occur if he responded favoi-ably tb the
November 22 letters. And since the November 22 letters
contained a reinstatement offer that was "valid on its
face," those letters ended the backpay obligation of FSC
toward Ashcraft and Theodore: Eastern Die Co., 142
NLRB 601, 603 (1963).
As for the precise date that FSC's'backpay obligations
toward Ashcraft and Theodore ended, Rogers' letter re-
quired the two employees to contact him nO later than
December 3, 1976. Since Ashcroft did not reply it is
clear that FSC's backpay -obligation tolled on' December
3: Eastern Die Co., supra; Southern Household Products
Co., supra. As for Theodore, if we knew the date on
which he rejected FSC's November 22 order, that date
would be the date of the tolling of FSC's backpay obli-
gation. Southern Household Products Co., supra. Since the
record does not provide that information, FSC's backpay
obligation toward Theodore will be deemed to have
tolled on December 3, as in Ashcraft's case.
B. Robert Early's Missed Promotions.
FSC fired Early in August 1973. The Board -ordered
FSC to reinstate Early, and FSC did return him to work
at his previous wage rate in December 1976.
The General Counsel contends that, had Early not
been fired, he would have been promoted in 1975 and
then again in 1976. If the General Counsel's contentions
are correct, the computation of the amount of backpay
that FSC owes to Early should take into account those
missed promotions: Pepsi Cola Bottling Co., 187 NLRB
1017. (1971), enfd 467 F.2d 164 (9th affd. 414
U.S. 168 (1973). FSC denies that Early would have been
promoted.
The Period to November 1979. Early was a Grade 8
crane operator when FSC fired him." 4 At that time (in
1973), FSC operated three different kinds of cranes in
the area of the Charlotte facility in which Early worked.
Each of the cranes had different functions, controls, and
capacity. Then in May 1975 FSC changed the manner in
which it produced steel. That, in turn, required FSC to
eliminate one kind of crane, add another, and make
143 The only evidence on this point is Rogers' testimony Neither
Theodore nor Ashcraft testified Theodore's letter was not placed Into
evidence
144 Early was employed at FSC's Charlotte facility The grade levels
referred to in this section of the decision refer to the 19-grade system
used at Charlotte until April 1977 (see part VII, above)
918 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
changes in the operation, of the remaining two. But, for
present purPoses, it is enough to know that, at the time
FSC fired Early, there were two different kinds of
Grade 8 crane operator jobs and one kind of Grade 11
crane operator job. That was altered in May 1975 so that
from then until November 1979 there was a Grade 9
crane operator position -as well as two different kinds of
Grade 8 crane operator jobs and One kind of Grade fl
Crane operator job: ''
There is no dipute that in the normal course of things
(after May ,1975) FSC'filled Grade . 11 crane operator
openings from among the Grade 9 crane operators, and
filled .Grade . 4 crane oPerator openings from among the
Grade 8 crane operators. As for how FSC selected a
Grade 8 or Grade . 9, crane operator for promotion, it is
clear that the -Company. considered both ability and se-
niority But the record .says nothing about what factors
FSC look :into account in evaluating the relative ability
of competing crane operators. .
Early's Performance. As of August 1973 (when Early
was fired), Early was the most senior of all the Grade .8
crane operators. From time to time he had performed
Grade 11 crane operator duties and had done so satisfac-
torily Since 1970 Early had been graded "better than
, satisfactory", on all of his periodic performance reports.
(The record contains no information on his previous per-
formance evaluations.)
The. Other Crane Operators. FSC employed six Grade 8
crane operators (in -addition to Early) at the time the
Company Tired Early. 145 ' Another employee -became a
Grade 8 crane operator about a month after Early's-dis-
charge. 146 And still another employee became a Grade 8
crane operator about ,a year after Early was fired. 147 Of
, those eight, 'six were Promoted to the Grade 9 crane op-
erator job prior to the time that Early was reinstated.148
And of .those six, three.,..became Grade 11 crane opera-
, tors." 9 . -
The record contains. information on the performance
evaluations of four of those eight other crane opera-
tors—three .who became Grade 11 operators and one
who never made it beyond Grade 9. While the record is
less than complete on point, the available information
shows that Early's performance evaluations were at least
'-as . good as those given to the other crane operators, and
'probably better.
In terms of relative experience, as indicated above
Early had more time on the job as a Grade 8 crane °per-
, ator at the time he was fired than any of the other FSC
- Grade 8 operators. On the other hand two of the other
'operators came to FSC as Grade 11 operators. While
both lost their Grade 11 status for considerable periods
of time (for reasons not stated in the record) it was they
(plus one other employee) who became Grade 11 Crane
145 The six Brewer, Hauser, Johnson, Jones, Marshall, and McIlwain
146 Employee Nilsson
147 Employee Anderson
148 The promoted employees Anderson, Brewer, Hauser, Johnson,
McIlwain, and Nilsson
149 Johnson (in April 1976), McIlwain (in August 1976), and Nilsson
(in June 1976) ,
operators during the period that Early . was away from
-FSC.15°
The Period to November 1979—Conclusion. The Gener-
al Counsel has the burden of showing that Early would
have been promoted but for his unlawful- diicharge by
FSC. 151 Since FSC presented virtually no evidence on
point, the question is whether the General Counsel' has
made out a prima facie case My conclusion' is that for
the peiiod prior to November 1979 the Geheial'Counsel
did so for the Grade 9 crane operator job, but did not do
so for the Grade 11 position , . t.
. Turning to the Grade . 9 . job first, as indicated earlier,
of the six emloyees who were, with Early, Grade 8
crane operators at the time of Early's discharge, four
. were Promoted to the Grade 9 position. And of the two
employees who became .Grade 8 operators after _Early's
discharge, both became Grade 9, crane, operators Given
those statistics, the reasonable inference. is ,that any em-
ployee who was a Grade 8 crane, operator at the time of
Early's discharge would have become a Grade 9 opera-
tor unless there was some specific reason precluding the
promotion. Nothing about Early's 'career at the time he
was discharged suggests any such factor that would'have
prevented his promotion. That, in My view, adds up to a
prima facie case. see S.. E. Nichols of Ohio, Inc., 258
NLRB 1, 9-10 (1981); Golay -& Co. v. NLRB, 447 F.2d
290 (7th Cir. 1971). - -
The situation is quite different for the General Coun-
sel's contention that Early would have been 'pr'omoted to
Grade 11 crane operator during the peiiOd he' was away
from FSCI Of the six employees who held Grade 8 crane
operator positions at the time Early was fired, three
eventually became Grade 11 - operators. BM; as 'noted
above, of those three, two had originally been hired as
Grade 11 crane operators. Their promotion to Grade 11
crane operator indicates nothing about Whether Early
would have been prompted to that position. 'A fairer way
of looking at the question is that at the time Early was
fired, there were four Grade 8 crane operators who had
not previously held the position of Grade 11 operators.
Of those four, only one was ever promoted to the Grade
11 position.1 52
Given the fact that .seniority was only one consider-
ation used by ;FSC in selecting employees for promotion,
given the absence of information in the record about the
factors FSC used in selecting Grade 11 crane operators,
and in view-of the foregoing stankics, my, conclusion is
that the General Counsel failed to present, a prima facie
case that, but for Early's unlawful discharge,,FSC would
• have promoted Early to . Grade 11 crane operator at any
time prior to November 1979. ;
150 The two emplo'kees Johnion (hired in' August 1968 as a Grade:11
crane operator and became Grade 8 crane operator in 1970), and Mai-
wain (hired as a Grade 11 crane operator in October 1969 and became a
Grade 8 crane operator in 1972) -
151 parties have agreed that the General Counsel has the burden
of proof on this issue See also Decker-Truck Lines, 139 NLRB 65 (1962),
modified on other grounds 322 F 2d 238 (8th Or 1963)
152 Employee Nilsson He became a Grade 11 crane operator in June
1976 He was terminated as an FSC employee (for reasons not set forth
in the record) in August 1976 • . .• •
FLORIDA STEEL CORP 919
The Period November 1979 to January 1980. On No-
vember 4, 1979, FSC reclassified the Grade 9 crane oper-
ator job so that employees holding the position were
thereafter paid at the Grade 11 rate.' 53 Absent evidence
to the contrary, I can only assume that, had Early not
been unlawfully discharged, and had he been a Grade 9
crane operator at the time of the 1979 reclassification (as
discussed above), he would have been promoted to
Grade 11 in keeping with the reclassification.
FSC did in fact promote Early to Grade 11 crane op-
erator on January 27, 1980 (directly from Grade 8). That
ended FSC's backpay obligation toward Early.'"
The Dates on Which Early Would Have Been Promoted.
As touched on earlier, the record indicates that six FSC
employees who were Grade 8 operators at the time
Early was fired,'_ and two FSC employees who became
Grade 8 operators after Early was fired, became Grade 9
crane operators prior to Early's reinstatement. The
record contains information about the dates of the pro-
motions of those employees only for four of the employ-
ees Of those four, three became Grade 9 crane operators
in early May 1975, when the Grade 9 position was first
created 1 5 5 And the fourth was promoted to the Grade 9
position in June 1976 In the absence of further informa-
tion, the most reasonable inference to draw—having con-
cluded that Early would have been promoted to the
Grade 9 position at some point in time, is that Early
would have been promoted at the beginning of- May
1975, when the Grade- 9 position came into being In
view of the lack of information about the precise date
when such promotion would likely have occurred, and
in view of the specification's reference to a May 17, 1975
date, I shall recommend that the Board conclude that the
backpay amounts owing . Early be based on a May 17,
1975 hypothetical promotion date; as stated in the Gener-
al Counsel's specifications.
As for the date on which Early would have been pro-
moted to Grade 11, as stated above that would have oc-
curred on November 4, 1979.
C. Are "Bending Coordinators" Members of the
Bargaining Unit
" FSC's Charlotte facility is made up of two divisions
The "Mill" (where steel is produced) and the "rebar
foundry." (Rebars are steel bars used for reinforcing con-
crete.) The pertinent part of the Board's description of
the Charlotte bargaining unit reads as follows: "All pro-
duction and maintenance employees . . . employed at
the Employer's [Charlotte] N.C. steel mill and fabricat-
153 The November 4 date is based on a posthearing stipulation that I
am including in the record as Exh ALJ-4 The upgrading of the Grade 9
crane operator job is also referred to at Tr 334 and 341 (witness Moore)
The Charlotte wage structure had shifted in 1977 to a 10-grade system
from a 19-grade structure, so that that ,actual reclassification of the crane
operator job was from Grade 4 to Grade 5 For the sake of clarity, the
discussion in this section refers to jobs in their pre-1977 designations,
even for the 1979-1980 period
"4 That reference to the' end of FSC's backpay obligation toward
Early does not take into account the backpay that FSC owes to Early
based on FSC's unlawful change in wage policies
"5 Johnson and Nilsson became Grade 9 crane operators on May 4,
1975, Anderson became a Grade 9 crane operator on May 1 1, 1975
ing division, excluding . . . fabricating shop leadmen
. . . and supervisors as defined in the Act."
At issue here is whether persons occupying the posi-
tion of "bending, coordinator" in the rebar foundry divi-
sion are members of that unit. If they are, persons who
occupied the bending coordinator position at any time
since February 1974 are entitled to backpay (because of
FSC's unlawful change in wage policies) Bending coor-
dinators are not members of the unit (and not entitled to
backpay) if they are supervisors within 'the meaning of
Section 2(11) of the Act if they are "fabricating shop
leadmen," within the meaning of the Board's unit de-
scription. They are members of the bargaining unit if
they do not fit within either of those categories.'56
Administrative Law Judge Harmatz, who conducted
the prehearing conference in this proceeding, ruled,
without objection, that the General Counsel had the
burden of proving that persons occupying bending coor-
dinator positions are members of the bargaining unit 157
The question to be resolved in this section of this deci-
sion is whether the General Counsel has met that
burden.
Bending Coordinators as "Fabricating Shop Leadmen."
Insofar as the record here indicates, there is no entity at
the Charlotte facility known as the "fabricating shop."
There is, however, a "rebar shop" that handles the "fab-
rication" of the rebars. 158 That fabrication, in turn,
"consists of shearing [the rebars] to certain lengths and
then some of it has to be bent in different shapes and de-
signs."'" As that quotation suggests, there is a "bending
department" in the rebar shop. Until about 1973 or early
1974 the senior worker on each of the shifts in the bend-
ing department (under the rebar shop foreman) was
known as the "bending crew chief." (The bending crew
chief was in fact senior to everyone in the entire rebar
shop, other than the foreman.) He assigned work to the
10 or so other bending department workers on his shift;
guided and directed them (including instructing inexperi-
enced employees); checked on their output; coordinated
"work flow through [the] bending operations"; 61 0 made
hiring and firing recommendations to his superiors; kept
production records, and did some setup work
Under these circumstances there is no doubt that the
bending crew chiefs were the "leadmen" in the rebar
shop's bending division. And since the reference in the
unit designation to "fabricating shop" must be construed
to mean, or at least include, the rebar shop, the bending
crew chiefs were excluded from the unit by virtue of the
unit designation.
The "bending crew chief' title was subsequently
changed to "bending coordinator." (Actually the title
was changed on some categories of company documents
156 Three persons occupied the position of bending coordinator during
periods in which bargaining unit employees at the Charlotte plant became
entitled to backpay by reason of FSC's wrongful withholding of pay in-
creases J A Tabor (from October 26, 1973, to February 9. 1975), W T
Newell (from April 7, .1974, to February 9, 1975), and C R Chastain
(from January 1, 1975, to February 9, 1975, and from May 11, 1975. to
November 1979)
157 Transcript of January 15, 1981 preheating conference at 18-19
"8 Tr 608 (witness Hunt)
"9 Id
160 GCX-31
920 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
by early 1973, which was prior to the issuance of the
Board's bargaining unit description.) But that title change
was the only change in the position. Everything else
about it—duties, responsibilities, pay, ad the persons
holding the job—remained the same. 161 Since bending
coordinators filled precisely the same role as bending
crew chiefs, and since bending crew chiefs were ex-
cluded from the bargaining unit by virtue of the Board's
exclusion of "fabricating shop leadmen," I can only con-
clude that the bending coordinators have never been part
of the unit and are not entitled to backpay.
Bending Coordinators as Supervisors. Section 2(11) of
the Act defines -the term "supervisor" as— _ _
. . . any individual having authority, in the interest
of the employer, to hire, transfer, suspend, lay off,
recall, promote, discharge, assign, reward, or disci-
pline other employees, or responsibly to direct
them, or to-adjust their grievances, or effectively to
recommend such action, if in connection with the
foregoing the exercise of such authority is not of a
merely routine or clerical nature, but requires the
use of independent judgment.
As touched on above, bending coordinators do
"assign" employees to particular tasks. But the record is
too sketchy to permit any determination of whether that
function "requires the use of independent judgment."
And bending coordinators, recommend the hiring and
discharging of employees. But again, the record is insuf-
ficient to permit a determination of whether such recom-
mendations are "effective" and are based on independent
judgment. As for the various secondary indicia of super-
visory status, by and large they cut both ways.162
Given the state of the record, it might well be that fur-
ther information that is wholly consistent with the exist-
ing evidence on point would make it clear that bending
coordinators are "employees" see, e.g , Bodolay Packag-
ing Machinery, 263 NLRB 320 (1982); Vapor Corp., 242
NLRB 776," 787 (1979); Commercial Movers, 240 NLRB
288 (1979). But it equally might be that further informa-
tion (again, wholly consistent with existing evidence)
would prove that bending coordinators are "supervi-
sors": see NLRB v. Porta Systems, 625 F.2d 399 (2d Cir.
1980); Abilene Steel Metal v. NLRB, 619 F.2d 332 (5th
Cir. 1980); Conair Corp., 261 NLRB 1189, 1203-04 fn. 9
(1982). And what all that adds up to is that the General
161 Chastain and Newell were the 'bending crew chiefs at the time of
the change, and they remained on as bending coordinators
" 2 For example on the one hand (1) bending c000rdinators are paid
on an hourly basis and receive the fringe benefits of hourly employees,
while all persons other than bending coordinators claimed by FSC to be
supervisors are salaried and receive different kinds of fringe benefits, (2)
during the Steelworkers' organizing efforts the bending coordinators did
not attend management meetings on union matters, and at least one bend-
ing coordinator became a member of an antiunion "employee" commit-
tee, and (3) bending coordinators wore red hardhats, not the white ones
worn by all other persons FSC claims to be supervisors On the other
hand (1) everyone (except the bending coordinators) who the General
Counsel claims to have been, "employees" wore yellow hardhats (1 e,
bending coordinators were the only personnel in the Charlotte facility to
wear red hardhats), and (2) the bending coordinators park their cars in
the area of the plant reserved for supervisors
Counsel has not met his burden of proving that the bend-
ing coordinators are employees, not supervisors.
D. Medical Insurance Premiums
On May 5,- 1974, FSC implemented in its employee
group medical insurance plan and, at the same time, in-
creased the insurance contributions ("premiums") paid by
participating employees. FSC made the changes at all of
its plants except Charlotte. Since FSC excluded the
Charlotte plant from the changes because of an antiunion
discriminatory intent, the Board ordered the Company to
retroactively grant the medical insurance benefits to the
employees in the Charlotte bargaining unit.
Some of the Charlotte employees did incur medical ex-
penses during the post-May 1974 period which were not
covered by the medical insurance plan then in effect in
Charlotte but which would have been covered had the
May 1974 changes been implemented at Charlotte. (I will
refer to these previously nonreimbursable expenses as
"additional medical claims.") FSC has reimbursed the
employees for their additional medical claims -
The Parties' Positions The General Counsel's concern
is that FSC has required all persons who were in the
Charlotte bargaining unit during the period at issue to
pay to FSC amounts equal to the higher premiums they
would have had to have paid if FSC had in fact imple-
mented the medical plan changes in May 1974. (I will
use.the term "additional premiums" to refer to the differ-
ence between: (a) the total amount of premiums an em-
ployee would have to have paid had the new plan been
implemented, and (b) the premiums actually paid by an
employee during that period.) 163 According to the Gen-
eral Counsel, FSC is entitled to the additional premiums
only in limited respects. To more fully describe the Gen-
eral Counsel's position, it is best to divide the employees .
into three categories.
Category 1: Employees whose additional medical claims
exceeded their additional premiums. The General Counsel
agrees that FSC is entitled to collect the additional pre-
miums from these employees.
Category 2: Employees who had additional medical
claims but whose additional premiums exceeded the amount
of their additional claims. The General Counsel's position
is that FSC is entitled to the additional premiums from
this category of employees only up to the amount of
each of the employee's additional claims. FSC insists it is
entitled to the full amount of additional premiums.
Category 3: Employees who had no additional medical
claims. The General Counsel urges that FSC be preclud-
ed from collecting any additional premiums from this
category of employees. FSC, again, claims that it is.enti-
tled to the full amount of additional premiums from such
employees
Medical Insurance premiums—Conclusion. The lines are
drawn most clearly in respect to Category 3 employ-
ees—those who had no additional medical claims.' My
conclusion is that FSC is not entitled to recover the ad-
ditional premiums from such employees. One' basis for
i63 October 1976 FSC again modified the medical Insurance plan; at
all locations except Charlotte, by eliminating the employees' group Insur-
ance premiums Thus the period under consideration here is limited to
May 1974 to October 1976
FLORIDA STEEL CORP 921
that conclusion is precedent: Rice Lake Creamery Co.,
151 NLRB 1113, 1129-31 (1965), enf. denied 365 F.2d
888 (D.C. Cir. 1966). But beyond that, the only bases for
charging the employees for the additional premiums are
that: (1) the employees would be unjustly enriched by
having "had the benefit of the increased coverage with-
out the corresponding detriment of higher premi-
ums";164 and (2) only by that approach would the em-
ployees be in the same position they would have been in
absent FSC's discrimination. .
The problem with those arguments is that neither takes
into account that insurance, by its nature, is forward
looking and involves protection against possible future
loss. Had FSC implemented the insurance changes at
Charlotte in 1974, each premium payment by an employ-
ee would have bought something- "security against an-
ticipated loss." 165 But for those employees with no addi-
tional medical claims for the period May 1974-October
1976, payment of the additional premiums at this junc-
ture buys nothing. For those employees. (1) Requiring
them to pay the additional premiums does not put them
in the same position that they would have been in but for
FSC's discrimination There is simply no way to retroac-
tively give them the additional security they would have
had if FSC had implemented the medical plan nondiscri-
minatorily. (2) It is hardly unjustly enriching the employ-
ees to refrain from requiring them to make the additional
premium payments. Rather, requiring them to do so
would be requiring them to pay something for nothing.
I recognize that requiring the extra premiums would
ease FSC's burden and would put FSC more closely in
the position it 'would have been in but for the discrimina-
tion But the object of a Board order in an 8(a)(3) case is
to make the discriminated-against parties whole, as
nearly as possible That rarely means restoring things to
exactly how they would have been but for the discrimi-
nation And sometimes the order to make whole will be
costly to the respondent While that Ought not to be the
object of a Board backpay order,' 66 individual employ-
ees ought not to be asked to pay for benefits they did not
buy simply in order to reduce that costliness.
That leaves for consideration the Category 2 employ-
ees—those whose additional medical claims are less than
their additional premiums. In Rice Lake, three employees
incurred additional medical expenses during the period of
their unjust discharge by their employer The additional
medical claims of two of the three totaled less than the
premiums they would have paid had they remained em-
ployed. As to those two employees (Category 2 in the
above listing), the Board held that: (1) they should pay
no additional premiums; and (2) they should receive no
compensation for their additional medical claims: 151
NLRB at 1130. While the bases for the Board's , decision
on this point are not spelled out in detail, the logic is
clear. For an employee to be entitled to reimbursement
for any additional medical claims, he ought to be treated
as though he chose the revised medical coverage, includ-
164 FSC Br 107
' 65 Metropolitan Police Retiring Assoc v Tobriner, 306 F 2d 775 (DC
Cir 1962)
166 E g, Republic Steel Corp v NLRB, 311 US 7, 10, 12 (1940)
ing all the additional premiums that went with it. Since
no one would choose such retroactive ,coverage if it re-
sulted in a net loss, the only answer is to disallow both
the employees' claims for reimbursement of their addi-
tional medical claims and the employer's demand for re-
imbursement for the additional premiums.
Under the circumstances I conclude that: (1) FSC may
not recover any additional premiums from employees
who had no additional medical claims; (2) FSC may not
recover any additional premiums from those employees
who did have additional medical claims if the amount of
those claims was less than their additional premiums; and
(3) FSC is not required to pay the additional 'medical
claims of any employees whose additional premiums ex-
ceeded their additional medical claims, and • FSC may
take as an offset against any remaining backpay due such
employees the amounts already paid such employees as
reimbursement for such additional medical claims.
E Vacation Benefits.
On January 1, 1974, FSC changed- its vacation policy
in the following ways, at all locations other than Char-
lotte. No changes in vacation policy were implemented
at the Charlotte facility. •
(1) Employees who had been with PSC- for at least 10
years became eligible for 3 weeks of vacation. Previously
employees had to have 12 years' service to be eligible for
3 weeks' vacation.
(2) Employees who had been with the Company for
20 years became eligible for 4 weeks' vacation.
(3) Employees with less than 1, year's service no
longer were entitled to a vacation. Previously they were.
In sum, the changed vacation policy benefited employ-
ees with 10 or 11 years' service and those with more
than 20 years' service. It adversely affected employees
with less than 1 year's service.
The Board concluded that FSC had excluded the
Charlotte employees from the "improved vacation
policy" 67 for antiunion reasons. FSC accordingly was
ordered to grant to the Charlotte employees the vacation
benefits that FSC had granted elsewhere, retroactive to
January 1, 1974. FSC has agreed to make the ordered
'vacation benefit payments to employees. But FSC con-
tends that it is entitled to be reimbursed by those Char-
lotte employees who, subsequent to January 1, 1974, re-
ceived vacation benefits when they had less than 1 year's
service
FSC's contention is erroneous The adverse changes
that FSC made in the first year employees' vacation
rights are readily separable from the improvements in the
vacation rights that FSC granted to its long-service em-
ployees. And the Board ordered only that F'SC grant the
withheld fringe "benefits" to the Charlotte employees.
The Order says nothing about FSC having to implement
at Charlotte the reductions in new employees' vacation
rights that FSC made elsewhere. The language used in
the Order was not inadvertent: see, e.g., NLRB v. Key-
stone Steel & Wire, 653 F.2d 304, 308 (7th Cir 1981).
Since the Order does not require, either expressly or im-
167 220 NLRB at 264
922 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
plicitly, that FSC implement any 1974 vacation policy
changes that adversely affected the Charlotte employees,
FSC is not entitled to be compensated by any employee
based on a retroactive implementation of those adverse
changes.
F Cutoff Date of the Backpay Period for Clerical
, Employees
In the period 1977-1978 there were about 150 employ-
ees in the Charlotte bargaining unit All but 4 of those
150 were paid on an hourly basis. The other four—
"clerks" and "recorders"—were salaried. (I will refer to
those four as "bargaining unit clencalst") 168 Until the ar-
nval of the Steelworkers at Charlotte, FSC kept the sala-
ries of the bargaining unit clericals equivalent to its cleri-
cal employees of the same grade- who were not in the
bargaining unit. But as in the case of the hourly employ-
ees, after the arrival of the Steelworkers FSC stopped
granting periodic increases to the bargaining unit cleri-
cals even though FSC continued to grant such increases
to employees who were not in the bargaining unit. The
Board concluded that_FSC had thereby violated the Act.
The only backpay issue involving the bargaining unit
clericals relates to the cutoff date from backpay liability.
According to FSC's answer to the General Counsel's
backpay specification: "the backpay period for [bargain-
ing unit clericals] ended November 20, 1977 when the
Company offered in negotiations with the Union to im-
plement on November 20" the salary increases imple-
mented on that day' for clericals not in the bargaining
unit.
The facts are that in early December 1977, in the
course of a meeting with union representatives, FSC rep-
resentatives proposed salary 'increases for the bargaining
unit clericals that were identical to the increases that the
nonbargaining unit clericals, had started getting on No-
vember 20, 1977 The union representatives neither ac-
cepted the proposal nor objected to it, and the meeting
moved on to other subjects. The matter of clerical sala-
ries was not raised again in FSC-uniOn meetings until
September 1978. (At that time union officials, upset at
learning that the bargaining unit clericals had missed the
pay increase that FSC had earlier proposed, quickly
agreed to an increase for the clericals. FSC implemented
the increase a few days later.)
At no time has the Union ever told FSC that it did not
want FSC to implement a proposed pay increase for the
clericals. Nor has there been any agreement between the
Company and the Union regarding the appropriate pay
scale for the bargaining unit clericals.
As discussed in section I,B, above, the import of the
various underlying Board orders in this proceeding is
(among other things) that FSC had been under an obliga-
tion to pay to the Charlotte bargaining unit employees
increases that, under all the relevant circumstances, are
analogous in timing and amount to those received by
168 The Company refers to all nonsupervisory, salaned, clerical, and
"technical" employees as "salaried non-exempt" employees ("Non-
exempt", refers to the applicability to them of the Fair Labor Standards
Act "Non-exempt" thus distinguishes them from management personnel,
who are also salaned, but who are "exempt ") The pleadings and the evi-
dence in this proceeding track company usage
other FSC employees The only limitation on that obli-
gation is that such increases are not to be put into effect
if "the Union as bargaining agent objects " 169 The
Union did not "object" to implementation of the 1977 in-
crease for the clericals. It certainly did not do so express-
ly. And even if silence by the Steelworkers could under
some circumstances be interpreted as an objection, that
was not the case with regard to the 1977 increase . The
Company raised the matter only once (prior to Septem-
ber 1978) And it did so in the course of a meeting that
dealt with numerous matters of far greater importance—
in dollars-and-cents terms—than small raises for four em-
ployees. The only implication that FSC could reasonably
have drawn in those circumstances from the Steelwork-
ers' silence was that the_ Union overlooked the subject,
not that it opposed the increase.
The cutoff date of the backpay period for the bargain-
ing unit clericals accordingly is the same as for the other
bargaining unit emloyees (see part I,C, above).
G Merit Pay
At all material times FSC employees have been peri-
odically evaluated—generally every 6 months—by their
supervisors in "performance reviews " 170 Until January
14, 1973, the grades received by FSC's ,Charlotte em-
ployees as a result of these performance reviews affected
their pay: Employees who received grades above "satis-
factory" (or its equivalent) were paid more than employ-
ees graded satisfactory or worse The differential was
known as "merit pay."
Complications arose if an employee who had been re-
ceiving merit pay subsequently received a grade of satis-
factory or worse. FSC did not reduce employee's pay.
Rather, it "red-circled" the employee. When a general
wage increase was thereafter implemented, the red-cir-
cled employee received the greater of: (a) 50 percent of
the increase; or (b) the amount that would put him on a
par with Other employees of his grade with satisfactory
or worse performance evaluations. Of course, once the
employee's pay was at the proper, nonment, level, he
was no longer red-circled.
In January 1973 FSC ended the merit pay system at
Charlotte. Thereafter employees who were being paid at
the regular, nonment, , level in January 1973 received no
merit pay notwithstanding evaluations of better than sat-
isfactory. And employees who in January 1973 were re-
ceiving merit pay kept getting the differential even if
their performance review grades fell to satisfactory or
worse. The Board held that PSC's post-January 1973
failure to grant merit pay, to employees who had quali-
fied for the additional pay violated the Act. It according-
- ly ordered FSC to make such employees whole "by
paying each of them a sum of money equal to that which
he would have earned . . . if Respondent had not with-
held his merit increase."71
226 NLRB at 124
170 See, in this connection, the discussion above of Robert Early and
other FSC crane operators
171 220 NLRB at 268 The Steelworkers and FSC agreed to the termi-
nation of the merit pay system as of April 24, 1977 (See part VII,
above ) Thus FSC has Incurred no backpay liability for its failure to pay
merit pay subsequent to that date
.FLORIDA ; STEEL CORP 923
Only two issues regarding inent pay are outstanding.
First, would the wage controls imposed by Phase III of
the Economic Stabilization' Act have caused FSC to re-
frain from granting merit pay increases during the period
that it was in effect. And second, some of the employees
who became entitled -to merit pay after January 1973
subsequently received evaluation's of satisfactory or
worse. They accordingly would have been red-circled
had FSC actually. had its merit pay system then in effect.
Should the backpay owed to those employees reflect
those-below-merit evaluations.
The Impact of the Economic Stabilization Act. This issue
has already been litigated and decided. Administrative
Law, Judge Wilson heard evidence on this precise sub-
ject. He found that: "Originally in 1973 at Charlotte this
merit plan was interrupted by the wage guidelines estab-
lished by the Cost of Living Council, but that interrup-
tion was removed on' August 13, 1973. 172 That finding
is conclusive. Employees who became entitled to merit
pay in the period , January through August 1973. would
not have started receiving merit pay- until August 13,
1973. ,FSC's backpay obligations should .reflect that cir-
cumstance.
:The Impact of FSC's Red-circled Procedures. It is evi-
dent that had FSC continued its pre-1973 merit pay prac-
tices at Charlotte, employees there whose performances
deteriorated to below-merit levels thereby would have
ceased qualifying for merit pay. Their subsequent pay in-
creases would accordingly have been affected by FSC's
red-circle procedures. All that is as FSC contends. The
only matter of concern is the nature of the evidence sub-
mitted by FSC purporting to show which employees
would have been red-circled.
At various times in the period at issue employees were
graded on a letter or integer basis. The record "is clear
that for employees graded on that basis, "4" or "5" or
"A" or "B" meant above satisfactory and qualified the
employees for merit pay "C," "D," or "E," or "1," "2,"
or "3," meant satisfactory or worse. 173 Thus _C or 3 (or
worse) was not high enough for merit pay.
But during the period 1973 through 1975 employees
sometimes received noninteger numerical grades (like 2.8
or 3.4) 174 During the 1973-1975 period in which those
noninteger numerical . grades were given, no employee
went on to merit, pay or came off it (because of FSC's
wrongful cessation of the merit system). Accordingly the
question is this: Had FSC been on the merit pay system
in 1973-1975, how would FSC have dealt, in terms of
merit pay, with noninteger grades—grades that were
never in fact considered for merit pay purposes. One
possibility is that FSC would have granted merit pay to
any employee who received a grade at all better than 3.0
("satisfactory"). Under that assumption a 3.1 grade
would entitle employees to merit pay. Another possibili-
ty is that FSC would have insisted that employees re-
ceive at least a 4.0 ("above satisfactory") grade. Under
172 220 NLRB at 265
173 See Tr 52-53 (witness Rogers) and GCX-7
174 The grades were sometimes listed as a total of 10 separate grades
for individual aspects of the employee's performance Thus an employee's
grade might be "28" or '34" But for all intents and purposes those
would be equivalent to 2 8 or 3 4 See GCX-7 and Tr 358
that approach an employee receiving merit pay would be
red-circled if he received an evaluation of 3.9 or lower.
FSC proposed a middle ground. -Its answer to the
specifications assumes that had merit pay been effect in
the 1973-1975 period, employees receiving merit pay
would have continued to do so if they were graded 3.5
or higher. Such employees would have been red-circled,
under this . assumption, with grades of 3.4 or lower 175
FSC's approach is a reasonable and fair one. (Indeed,
the General Counsel seems to interpolate the grades the
same way FSC does, referring to a , "34" (i.e., 3.4) as
"satisfactory" and "36" as !`above satisfactory.") 1. 76 I Ac-
cordingly conclude that the backpay due FSC employees
because of FSC's unlawful cessation of its merit pay plan
should reflect FSC's red-circle procedures as applied to
the evaluations set forth in Respondent's Exhibit 21.'77
IX. CONCLUSION
Interest Rate. All of the Board orders underlying this
backpay proceeding provide that FSC shall pay interest
on the backpay amounts at 6 percent per annum. The 6-
percent figure was in keeping with the Board's invariable
practice at the time. In' August 1977 the Board changed
its approach to determining the interest rate to be im-
posed on respondents. 178 And since then the interest rate
levied on backpay has been considerably higher than 6
percent. —
The Steelworkers argues that the new higher rates
ought to be applied to the backpay FSC owes to interest
rates "is not applicable in, cases in which an earlier order
of the Board providing for a different interest rate has
been enforced by a court of appeals." S. E. Nichols of
Ohio, Inc., 258 NLRB 1, 1-2 (1981). Accord: Iron Work-
ers- Local 378 (Judson Steel Corp.), 262 NLRB '421 (1982).
Since all the underlying , orders in this proceeding were
enforced by courts of appeals (see fn. 1, above), I am re-
quired to leave, the 6-percent figure intact
Computation of Backpay. FSC's Charlotte and Indian-
town plants have each employed well over 100 bargain-
ing unit personnel throughout the backpay period. And
all such employees are owed backpay. Moreover, the
backpay period is a lengthy one. Because of that, because
there are no 'issues outstanding on matters such as interim
earnings or numbers of ,hours each - of the various em-
ployees worked, and because the parties , are in a better
position than is the Board to do the .arithmetic computa-
tion necessary to arrive at backpay -amounts, the parties
'stipulated that "computation of individual employee
backpay amounts owing will ,not be - sought from the Ad-
ministrative Law 'Judge _or. .• any' reviewing authori-
nes." 179 In view of that stipulation, the accompanying
175 FSC witness Rogers was frank to say that the 3 5 cutoff that the
Company is arguing for is solely one of "Judgment" Tr 358
176 G C Br at 41
177 The General Counsel raised several arguments not specifically
dealt with above in opposition to FSC's red-circle contentions I consid-
ered those arguments, but found none to be persuasive
' 75 The case happened to Involve FSC 231 NLRB 651
175 I am including the stipulation in the record as AU-S.
924 DECISIONS OF NATIONAL-LABOR RELATIONS BOARD
Order will simply require FSC . to pay the employees
named in the backpay specifications amounts reflecting
the conclusions of this decision.
. „
.ORDER
The Respondent Florida Steel Corporation, Charlotte,
North Carolina- , and Indiantown, FlOrida, its officers,
agents, successors, and assigns, shall, within 30 days of
the date of this Order,- make whole the employees named
in the backpay specifications in this proceeding by
paying to those employees amounts reflecting the various'
conclusions set forth in the foregoingdecision.
, APPEIsIDIX.D
Reconstructed Wage Rates, FSC's Charlotte Plant, as of
February 28, 19761
Grade,
,
Wage Rate
In
February
19732
Percent of
AHE in
February
19733
Recon-
structed
Wage
Rate as {
of
Febru-
ary 28,
'19764 •
1 RF & Mill $2 60: 81.25% $3.38
2 RF 2 75 85 94 3 58
3 RF 2 80 87 50 364
4 RF 2 85 89 06 3 70
5 RF /90 90 63 3 77
3 & 4 Mill 2.94 91.88 3 82-
5 Mill &-
6 RF & Mill 300 .
,
., 93 75 3 90
7 RF & Mill 3 17 99 06 4 12
8 RF & Mill 3 33 1.04 1 , 433
9 RF & Mill 3.50 109.4 4 55
10 RF & Mill 364 113 8 4.73
11 RF & Mill 3 83 , 119 7 4.98
12 Mill 4 05 126 6 5.27
13•Mill 4 27 133 4 5 55
14 Mill 449 140 3 5 84
15, Mill . . -4.71 147.2 6.12
17 Mill 5,16 161 3 671
1 February 28, 1976, is one of the dates on which FSC would
have granted a wage increase to its hourly Charlotte employees
but for the Company's unlawful change in wage policy
3 The last wage increase that FSC granted at Charlotte before
its change in wage policy was in,February 1973
3 Based on AHE of $3.20 (Charlotte SMSA primary metals
average AHE for the 12 months ended February 1973)
4 Based on AHE of $4 16 (Charlotte SMSA primary metals
average AHE for the 12 months ended February 1976)
APPENDIX E
Reconstructed Wage Rates, FSC's IndiantoWn Plant, is of
December 25, 19761
Grade Wage Rate asof Dec 19732
Percent of Reconstructed
AHE as of Wage as of '
Dec 19733 Dec. 25, 19764
1 $3 50 101 4 $4.42
2 380 110 1 480
3 400 115 9 505
4 4.30 124 6 5.43
5 470 136 2 594
6 515 149 3 651
7 575 166 7 727
9 615 178 3 777
1 December 25, 1976, is one -of the dates on which FSC
would, have granted a wage increase to its hourly Indiantown
employees but for the Company's unlawful change in wage,
policy
3 The last wage increase that FSC granted before its change
in wage policy was in April 1974. That increase was made
retroactive to December 1973 -
3 Based on AHE of $3 45 (Florida Statewide all manufactur-
ing average AHE for the 12 months ended December 1973)
Based on AHE of $4.36 (Florida Statewide all man-ufactur-
mg average AHE for the 12 months ended' December 1976).
APPENDIX F
Presumed Dates of Reconstructed Wage Increases'
Charlotte
Feb. 3, 1974 May 29, 1976
Nov. 30, 1974 Aug 28, 1976
Feb 22, 1975 Nov 27, 1976
May 24, 1975 Feb 26, 1977
Aug 30, 1975 Feb 26, 1978
Nov 29, 1975
Feb 28, 1976
Feb.
Feb.
25,
24,
1979
1980
1981, 1982, and subsequent years, if any last Sunday in
February
Indiantown
- Sept 28, 1974 June 26, 1976
Dec 21, 1974 Sept 25, 1976
Mar. 29, 1975 Dec 25, 1976
June 28, 1975 Dec. 25, 1977
Sept 27, 1975 Dec. 31, 1978
Dec 20,
Mar 27,
1975
1976
Dec.
Dec.
23,
28,
1979
1980
1981, 1982, and subsequent years, if any: last Sunday in
December
1 Source : GCX-97 and 100
V.
APPENDIX G
Table I Separate Regressions of FSCWages on AHE for EachCharlotteLabor Grade
Table 2—Continued
1980 Charlotte
Grade 11
Reconstructed Wage
1980 Indiantown
Grade 5
Reconstructed Wage
$670 $731
.814 8 64
$627
7 40
Table 3
FLORIDA STEEL CORP 925
Method-
ology
Adopted
by this
Decision
General-
Counsel's
Model
FSC's
Regres-
sion
Model
Labor
Grade
17
18
. -
Slope
1 722
1 823
Table 2
Base Period Wages at FSC's Charlotte Plant
May May May Jan"
1969 1970 1971 1972
1 RF & Mill $201 $2 14 $2.24 $2 35
Feb.
1973
$260
2 RF 207 2 20 2 30 2.46 2 75
Separate Regressions of FSCWages on AHE for Each 3 RF 207 2 23 2 35 2 51 2 80
CharlotteLabor Grade 4 RF 2 13 2.30 242 2 58 2 85
5 RF 221 237 250 266 290
Labor . 3 Mill 2 30 243 2 53 2.67 2 94SlopeGrade 4 Mill 2 30 2 43 2 53 2 67 2 94
1 0 877 5 Mill 2-30 245 2 55 2.72 3 00
2 1 033 6 RF & Mill 231 2.47 2 59 275 300
3 1.110 7 RF & Mill 242 260 272 284 317
4 1 094 8 RF & Mill 252 270 2 84 3 01 3 33
5 1 956 9 RF & Mill . . 263 2 82 297 3.14 3 50
6 0 965 10 RF & Mill 275 294 310 .327 3.64
7 1 064 11 RF & Mill 284 308 325 352 383
8 1 055 12 Mill 301 3 25 343 3 61 405
9 1.052 - 13 Mill 3 17 3.41 3.61 3 80 427
10 1 141 14 Mill 3 33 3 57 3 79 3.98 449
11 1 328 15 Mill 359 373 398 417 471
12 1 329 16 Mill — — — . —
13 1 363 17 Mill . — — — 453 516
14 1 440 Average Hourly
15 1 523 Wage, All Grades $2 65 $2 84 $2 99 $3 21 $3 57
16 1.625
926 DECISIONS OF NATIONAL LABOR RELATIONS BOARD
Table.4 -Continued
- Table 4
Charlotte Lower Grades
February
Charlotte Lower Grades
Reconstructed Wage
Grade 1 RF and Mill
% of AHE
February
1973
Reconstructed Wage
Charlotte Upper Grades
'7 12F. and Mill
., 3 17
% of AHE
99%
1973 2.60 81% 1974 3 69 105%
1974 , 2.93 84% 1976 463 111%
1976 3 59 86% 1979 6.32 121%
1979 4.75 91% 1981 '. 774 124%
1981 5 78
Grade 6 Mill
92% 1973
1974
15 Mill
4.71
' 496
147%
142%
1973 . -3.00 -. 94% 1976 5.90 142%
1974 3.30 94% 1979 7.59 145%
1976 3.96 95% 1981 9.01 144%
1979 5.12 98%
1981 6 15 98%
Table 5A
Comparison of Tampa Wage Rate Reconstruction with Actual Tampa Rates
4/76 7/76 10/76 .•-- 5/774/78 4/79 4/80
Labor .
Grade Predict- Predict- Predict- Predict-Actual Actual Predict- Predict- Predict-Actual Actual r Actual Actualed ed ed ed Actual eded ed i
. 1 '$4.01 $3 97 $406 ' $4.04 $4 18' $4 14 '. $4 39 $4 37 $477 $4.86 $5 31 $5 44 $5.92 $6.08
2 4.23 • 4.19 428 ' 4.26 4.40 436 ' ' • 461 4.59 499 5.08 5 53 5.66 6 14 630
.: 3 * 4.49 . 4.46 4 54 4 53 466 4.63' 4 87 4 86
.
5 25 5 35 5 79 5 93 640 6 57
4 471 468 476 475 488 4.85. 509 508 547 557 601 615 662 679
5 4.98 495 503 5.02 5 15 5 12 5.36 5 35 5 74 5 84 . 628 642 6 89 706
6 5.84 , '582 589 589 601 599 .621 6.22 660 6 71 7.14 729 775 793
7 6.21 620 6 26 - 627 638 637' .659 6.60 697 7 09 . 751 767 812 831
.8 642 641 647 6.48 6 59 6.58 ' 680 681 7 18 7 30 772 7 88 8.33 8 52
'9 -6.64 663 669 670 6.81 680 701 703 740 752 791 810 852 874
10 6 85 6 85 6 90 6 92 7.02 - 702 • 7 23 7 25 7 61 N/R 8 15 N/R 8 76 N/R
Based on the preliminary AHE figure for April 1980
Table 6
Table 5B 5/69 5/70 5/71 1/72 2/73
Reconstructed FSC Charlotte Wages, Illustrative Grades and
Years
Labor Grade 1 (RF and Mill)
FSC Regression General Counsel
Model Models
1974 $2 86

$293
1980 4.36

5 07
Labor Grade 14 (Mill)
1974 $4.83

$478
1980 7.20

7 86
AHE, Primary
Metals,
Charlotte SMSA
Moving
Average
Average FSC
Charlotte Wage,
All Grades
Dollar Difference
Between
Average
Charlotte Wage
and AHE
2.57 272 2.85 299 320
265 284 299 321 357
$ 08 $ 12 $ 14
$ 22
$ 37
Average Charlotte
Wage as % of
AHE

103 1% 104.4% 104.9% 107.4% 111.6%
FLORIDA STEEL CORP 927
Table 10
Table 7
Reconstructed Indiantown Wage Rates
Reconstructed Charlotte Wage Rates, 1980
Grade 1 Grade 9
Dec 1974
FSC Regression General Counsel
Model Regression,
Labor Grade 1
$3 69 $3 96
FSC Proportional Index $479 $7 90 Dec 1980 - 486 744
FSC Combination Index 5 12 7 55 Labor Grade 4
FSC Across-the-Bd-Index 5.28 7 38 Dec. 1974 $465 $476
FSC Regression Model 4 36 7.60 Dec 1980 6 76 8 24
Labor Grade 9
Dec 1974 $6 60 $ 661
Dec. 1980 9 30 10 09
Table 8
Base Penod Wages at FSC's Indtantown Plant
Labor Sept. Dec. Oct Feb. Dec.
Table 11
Grade 1970 1971 1972 1973 1973 a Sept Dec Oct. Feb Dec
1 $229 $250 $2.75 $310 $350 1970 1971 1972 1973 1973
2 2 59 2 80 3 00 340 3.80 AHE All Manufacturing
3 2 88 3 11 3 32 3 60 400 Florida Statewide
4 317 3 39 3.63 3 90 430 Moving Average 283 304 3.16 3 23 345
5 3.47 3.70 3 95 430 4 70 AVERAGE FSC
6 3 76 3.99 425 4.75 5 15 Indiantown Wage, All
7 406 430 4.60 5.35 575 Grades 331 354 378 421 461
8 * * * Difference Between
9 4.37 461 493 5 75 6 15 Average Indiantown
Average Hourly Wage, All Grades Wage And AHE $ 48 $ 50 $ .62 $ 98 $1.16
3.31 3 54 3 78 4.21 461 Average Indiantown
Wages as % of AHE 117% 116% 120% 130% 134%
a This wage increase was agreed to dunng April 1974 but it
was made retroactive to December 1973
Table 9
General Counsel's Indiantown Model
December Reconstructed Wage Reconstructed WageAs Percent of AHE
Grade 1
1973 $3 50 101%
1974 3 96 105%
Table 12
Reconstructed Indiantown Wage Rates for 1980
Grade 1 Grade 9
FSC Percentage Index $521 $957
FSC Combination Index 612 934
FSC Across-the-Board Index 657 922
FSC Regression 486 930
General Counsel Regression 744 10 09
1976 490 112%
1979 6.66 122%
Grade 5
1973
1974
$470
5.16
136%
137% Table 13
1976 6.10 140% Spread Between Charlotte Grade 1
1979 786 143%
Grade 9 (RF and Mill) and Charlotte Grade 6 (RF)
1973 $6.15 178%
1974 6.61 176% 1969 1970 1971 1972 1973
1976 755 173% Grade 1 Wage 201 2 14 2.24 2.35 2 60
1979 931 170% Grade 6 Wage 2.31 247 2 59 2 75 3.00
Dollar Spread .30 33 35 40 40
Percent Spread 149% 154% 156% 170% 154%
Grade 7 Wage 242 2 60 2 72 2 84 317
Grade 15 Wage 3 59 3 73 3 98 4 17 4 17
$208
211
2 18
2 65
2 65
91%
84%
79%
85%
76%
1970
1971
1972
1973 (Feb))
1973 (Dec ),
Table 15
Table 18
, Year
1970
1971
1973
1980
Grade 9 as a Percentage of Grade 1,
incliantown Survey
149%
216%
191%
200%
928 DECISIONS OF 'NATIONAL LABOR RELATIONS BOARD
Table 13—Continued
Spread Between Charlotte Grade 1
(RF and Mill) and Charlotte Grade 6 (RF)
Table 16—Continued
Reconstructed Charlotte Wage Rates,
Gr.
Or
7

Gr
1

(RF

14
(RF)

(Mill)
Mill)
FSC's Regression Model
436 517 ;720
1969 1970 1971 1972 1973
Dollar Spread 117 113 126 133 154
Percent Spread 483% 435% 463% 468% 486%
Table 14
Average CPI Grade 9 as a Percent ofYear Average CPI Grade 1
1970 166%
1971 172%
1972 182%
1973 181%
Table]?
Base Period Relationships between FSC Indiantown Grades 9
and 1
Percentage Difference
, Year Between Grades 9 and
Dollar Difference
Between Grades 9 and
1
Average -CPI Grade 9 as a Percent ofYear Average CPI Grade 1 •
1974 191%
1975 180%
1976 , 203%
1977 191%
1978 184%
1979 181%
1980 172%
Table 16
Reconstructed Charlotte Wage Rates,
Gr
1
Or.
7
(RF
--Gr
14
Table 19
Reconstructed Indlintown Wage Rates
December 1974(RF) &
Mill)
(Mill) As Recommended by this Decision $381 $5.12 $670
General Counsel's Regression Models 396 5.16 661
February 1974 FSC's Regression Model 369 5.09 660
As Adopted by this Decision
General Counsel's Regression Models
FSC's Regression Model
$2 84
2 93
2 86
$347
3 69
3 42
$491
4.78
4 83
December 1980
As Recommended by this Decision $606 $8 14 $10.66
February 1980 General Counsel's Regression Models 744 8.64 10 09
As Adopted by this Decision $4 55 $5 55 $7 86 FSC's Regression Model 4.86 740 930
General Counsel's Regression Models 5 07 6 77 8 04