Dock Strike Adjustment Factors for Major Categories of U.S. Imports and Exports, 1958-1974
K.7 | (#734 in RFD Series)
INTERNATIONAL FINANCE DISCUSSION PAPERS
DOCK STRIKE ADJUSTMENT FACTORS FOR MAJOR CATEGORIES OF U.S. IMPORTS AND EXPORTS, 1958-1974
by
Peter Isard
Discussion Paper No. 60, February 7, 1975
Division of International Finance Board of Governors of the Federal Reserve System
“ ¢
The analysis and conclusions of this paper represent the views of the author and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or its staff. Discus-
sion papers in many cases are circulated in preliminary form to stimulate discussion and comment and are not to be cited or quoted without the permission of the author.
Peter Isard*/ February 1975
Dock Strike Adjustment Factors for Major Categories of U.S. Imports
and Exports, 1958-1974
I. Introduction
During the past seventeen years, U.S. trade volumes have fluctuated | significantly in response to seven major shutdowns of U.S. ports; see Table 1. Together these dock strikes have curtailed trade volumes in 9 out of 68 quarters, while promoting trade, in anticipation of or recovery from the strikes, in approximately another 10 quarters. Thus, about 30 percent of the quarterly trade records since 1958 reflect the influence of strikes. Consequently, attempts to explain quarterly patterns of trade during this period must either discard a significant fraction of the sample or devise a suitable method for preventing dock strike fluctuations from biasing econometric estimates of the effects on trade volumes of income, prices and other explanatory variables. |
This paper constructs and tests quarterly dock-strike adjustment factors for a number of major categories of U.S. imports and exports for the 1958-1974 period 2/ Our dock-strike adjustment - factors are distinguished
*/ I am indebted for data and information provided by Mr. Harry Cohany of the Bureau of Labor Statistics, Captain John Haynes of the New York Shipping Association, Mr. Langston of the Waterfront Commission of New York, and Mr. H.F. Sickinger of the Maritime Administration. I am also particularly grateful to Daniél Roxon, Barbara Lowrey and P.A.V.B. Swamy for providing early direction, to Peter Hooper for very helpful criticism and his assistance in defining the commodity coverage, and to Andreas Duus III for considerable assistance in the early stages of my research.
1/ Import categories are: (1) foods, feeds, and beverages, (2) consumer goods (excluding foods and automotive products), (3) consumer nondurables (excluding foods), (4) consumer durables (excluding automotive products),
(5) industrial supplies and materials (excluding fuels and lubricants), (6) capital goods (excluding automotive products) and (7) all items. Export categories are: (8) agricultural products, (9) capital goods (excluding aircraft and automotive products), (10) consumer goods (excluding foods and automotive products), (11) consumer nondurables (excluding foods), (12) consumer durables (excluding automotive products), (13) industrial supplies
and materials (excluding agricultural products) and (14) all items.
0 ere eee
-2-
2/
from most others— in three important ways, each of which involves the
introduction of prior information. (1) We derive quarterly adjustment ‘Table 1: Major U.S, Dock Strikes Since 19582/
October 1 - 8, 1959 -- East and Gulf Coasts December 24, 1962 - January 25, 1963 -- East and Gulf Coasts January 1l - February 12, 19650/ -- East and Gulf Coasts
December 21, 1968 - February 14, 19692/ -- East and Gulf Coasts
July 1 - October 8, 1971 -- West Coast October 1 - November 28, 1971 -- East and Gulf Coasts January 17 - February 20, 1972 -- West Coast
a/ See Appendix A for a chronology of port closings. b/ Terminal date for the port of New York. Other ports returned to work
later; see Appendix A.
factors by aggregating an estimated series of daily strike impacts, where the day-to-day timing of the strike impacts is based on daily strike information. (2) The estimated magnitudes of our (daily) strike impacts are based on weekly information on longshore manhours, rather than monthly or quarterly information on trade volumes or manhours = And (3) we explicitly introduce estimated information on the shares of the various trade categories
2/ Other dock strike adjustment factors include those provided by the Labor Department report, Impact of Longshore Strikes on the National Economy (January 1970), and the various dock strike dummies employed in numerous . empirical models of U.S. imports and exports. For a description of an unpublished dock strike dummy constructed somewhat in the spirit of ours, see W. Takacs, "The U.S. Import Surcharge of 1971," International Monetary Fund memorandum DM/74/43, April 22, 1974.
3/ Data on trade volumes are not available for periods shorter than one month.
- 3 «
that are transported by ship through the striking regions.1/
Dock strikes do not coincide neatly with quarterly time periods, and there is much to be gained by first constructing strike adjustments on a daily basis and then obtaining quarterly adjustments as sums of the appropriate daily adjustments, Daily trade volumes: (a) rise sharply in anticipation of strikes during the weeks prior to the expiration of longshore contracts, and also during the final weeks of any Taft-Hartley injunction periods; (b) drop sharply during strike periods; and (c) typically are abnormally high during the weeks following a strike, reflecting attempts to recover the net loss of trade during the strike and anticipation periods. Although we do not have data on daily trade volumes or longshore manhours, the trade profile just described in evident in weekly manhour data, as | shown in Figure 1..
The problem of dock strike adjustment is essentially a problem of estimating the time paths of trade volumes -- or in our approach, of longshore manhours -- that would have prevailed in the absence of the strikes. Since trade volumes -- and hence longshore manhours -- are sensitive to income, prices, and other variables that fluctuate over time, it is invalid to assume that manhours would have been purely a function of time (or some extrapolation of historic manhours) in the absence of dock strikes, Nor is it desirable to judge a-priori the impacts on longshore manhours (or trade volumes) of fluctuations in income, prices, etc. It is true, however, that ~~ 4/ Our construction of separate dock strike adjustment factors for
different trade categories is a fourth distinction in comparison to many of the other dock strike dummies employed in econometric work.
eh a ce ner rca TSR
hE 0 RAN I ET Bn lee cel ee
Thousands of
Manhours 1100
990
880
770
660
- 3a - -
550
440
330
220
110
For week ending
7
MUN WS July
Source:
Figure 1: Weekly Longshore Manhours for
Contract Expiration
2-day Strike
UCBs I CiIrnunMNevenvigreir~As
Aug. Sept.
See Appendix B,
Oct,
Nov.
End of TpftzHgrtley
Dec,
the Port of New York
Fiscal Year 1969
nc enn ea
Jan.
THO e+ etme
SMe eee cee ae ow.
Feb,
Recovery
Mar.
April
9 te 12229 5 12 Wa 2 7 1 23 2 6 AS 30 & 13 BO oP Ww ast
May
8 is aa 2 June
Day — Month
-4-
if we had data on sufficiently many strikes we could invoke the law of large numbers, knowing that fluctuations about the seasonally-adjusted time trends of explanatory variables would wash out on the average and have no net effects on manhours, and thus accepting as the basis for our strike adjustments some average or representative profile of the deviation between the paths of actual and seasonally-adjusted time-trend manhours during individual strike episodes,
Unfortunately, we do not have good data on a large number of strike episodes. Of the seven major dock strikes during our data period, the 1959 and 1962-63 East and Gulf Coasts strikes predate our weekly longshore manhour data, while the 1971 and 1972 West Coast strikes and the 1971 East and cult Coasts strike had coincident impacts that are difficult to separate. Only for the 1964-65 and 1968-69 East and Gulf Coasts strikes can we use weekly data to estimate the time paths of deviations between actual and seasonally-adjusted trend manhours. For lack of an alternative, however, we have essentially chosen to apply the law of large numbers to these two strike episodes, adopting as our representative profile of dock strike impacts the profile of average percentage deviations of actual from seasonallyadjusted trend manhours during these two experiences.
Section 2 presents a detailed derivation of our representative profile of the impacts of dock strikes on longshore manhours. In Section 3 we combine this representative profile of the magnitude and duration of
strike impacts with information from strike chronologies on the day-to-day
timing of these impacts during individual strike episodes. For each strike episode we aggregate the daily strike impacts into quarterly strike impacts, and percentage impacts on longshore manhours are translated into assumed percentage impacts on trade by ship through the striking region. The assumptions and calculations of this section lead to general formulas (one for each quarter affected by strikes) for the impact of dock strikes on the quarterly volumes of total U.S. imports or exports of any category, where these general formulas are stated in terms of the shares of these trade volumes that are transported by ship through the striking regions.
Since we do not have data on the shares of total U.S. imports or exports of individual commodity groups that are transported by ship through particular regions, we are forced to estimate these shares on the basis of reasonable assumptions. Section 4 discusses these assumptions and presents our share estimates and the implied numerical dock strike adjustment factors for each of the 14 trade categories. The major assumption on which our share estimates depend is the assumption that within each commodity category the share of West Coast trade transported by ship is the eame as the share of East and Gulf Coasts trade transported by ship.
In constructing our strike adjustment factors we make no allowance for substitution, during dock strike episodes, between trade by ship and trade by other methods of transportation, or between trade through the
striking regions and trade through other regions. Figure 2 seems to show
-6-
evidence of such substitution,2! but we do not feel comfortable with any method that might have been used to make crude allowances for these effects.
The empirical performance of our dock strike adjustment factors is analyzed in Section 5. These factors (D) have been constructed in a manner designed to yield an estimated c-coefficient of one when log (D) is used as a dummy variable in the regression equations:
log (TVAL) | =a+ bt +c log 0). + a. where for each commodity group of imports or exports TVAL = value of trade in current dollars t = index of time a,b,c = coefficients to be estimated uo = stochastic error term
Although for most of the 14 trade categories the estimates of c-1 differ significantly from zero on the basis of the t-test, the 14 estimates of c all lie within the range between .39 and 1.4. For each strike-affected quarter the signs of residuals greater than one standard error of regression (hereafter referred to as "large residuals") are tabulated and "aggregated" over the cross-section of commodity groups in an informal search for residual patterns that are similar in a large number of trade categories and might conceivably be attributed to deficiencies in the general strike-impact
5/ During January 1969, when East and Gulf Coasts longshoremen were on strike, there was a jump in the volume of West Coast ship handling; and perhaps s-ne of the permanent jump in East and Gulf Coast imports by other methods or transportation, starting in October 1968, was related to the prospect of strike and/or to higher ship handling costs under a new longshore contract. Figure 2 also shows evidence of substitution between alternative methods of transportation during the 1971-72 strikes.
Millions of current dollars not seasonally adjusted 450
2450
2100 | |
East and Gulf Coasts, by ship
1750 - of oy 1 1750 { er" a “A ‘ | ‘ An ‘ Va ' / ‘ : , : \ H ' t See \ 1 . . \; ‘ ? 1400 rd) A 1’ 1400 ' *~ t \ \ e t oar ‘ e \ : "Ny \ , A n . A / \ A ' \ : \ | ; ‘ / ‘ i /\ / ¥ , ‘e a | | me see “4 ‘ y . / ’ \ ? we \ i 1050 , | *-d , ! La 1050 ' ‘, ’ N / VI e \ ; ‘y y " . East and Gulf Coasts, transportation by other ' methods 700 N 700 West Coast, by ship 350 B50
a West Coast, transportation by other methods
° tame ots ¢ bd e o e. otal eta, *. ee ee eee -*s,, gn cee nee Oe Nae Oe ite La i . eed taal ad
ee eS SS | 1968 1969 1970 1971 1972 SOURCE: Department of Commerce, Bureau of the Census. U.S. Foreign Trade: Highlights of Exports and Imports (FT990 publicstions). Various issues, West Coast Imports represent imports through the Los Angeles and San Francisco . Customs Regions; East and Gulf Coaste Imports are the residual,
@-7-
formulas. On the basis of this analysis we re-examined the strike chronologies and found good reasons to revise the general formulas for
two strike episodes, as described in detail below.£/
For regressions using the revised adjustment factors, in 3 out of 19 strike-affected quarters
at least half of the trade categories show large residuals of the same sign; but these residual patterns may be caused by factors other than errors in the strike-adjustment variables and we do not have good reasons to further revise the general strike-impact formulas for these quarters. During strikeaffected quarters the average frequency of large residuals for the crosssection of commodity groups is 1.3 times as large as it is during
other quarters.
We conclude that our empirical tests do not suggest any obvious deficiences that might easily be remedied. On the other hand, it is obvious that significant deficiencies, which are difficult to remedy, are inherent in the numerous simplifying assumptions that we are forced to make in | constructing the strike adjustment factors. We do not really need empirical tests to tell that we are stretching the truth. Nevertheless, we feel that the prior-information content of our strike adjustment factors far exceeds that of any other set of dock strike dumnies.
Before proceeding to the details of our calculations, the reader should be aware that East and Gulf Coasts strikes refer to strikes of the
International Longshoremen's Association (ILA), whereas West Coast Strikes
6/ The above discussion of c-coefficients and t-tests refers to the revised adjustment factors.
-8-
refer to those of the International Longshoremen's and Warehousemen's Union (ILWU). The trade data we use in this paper are disaggregated by 9 customs regions, on the basis of which we define the Los Angeles and San Francisco regions as the West Coast, and the Boston, New York, Baltimore, Miami, New Orleans, Houston and Chicago regions as the East and Gulf Coasts. Our treatment of the Chicago region (and the Boston region to a minor extent) is inaccurate -- the Great Lakes Districts of the ILA bargain separately and do not strike with the East and Gulf Coasts. However, our dock strike adjustment factors appear to be affected only marginally by this simplification.’
Information on strike chronologies is presented in Appendix A. Appendix B tabulates the data we use on longshore manhours for the Port of New York. Our dock strike adjustment factors are listed in Tables 5 and 6
on pages 24b and c.
2. A Representative Profile of the Impacts of Dock Strikes on Longshore a eeeeeeeeeeeesSC‘(Mamhours
In this section we examine weekly data on longshore manhours for the Port of New York and compare the behavior of actual manhours during
the 1964-65 and 1968-69 dock strikes with a definition of seasona Lly-
i/ This assertion is based on the facts that the Chicago region accounts for less than 10 percent of both imports and exports by ship of the seven non-West-Coast regions, while trade by ship through Great Lakes Districts within the Boston Region is negligible. The errors due to the inaccurate treatment of the Great Lakes Districts are not dampened by the fact that Great Lakes shipping was closed for the Winter during most of the East and Gulf Coast shutdowns. Moreover, we have failed to consider three important Great Lakes shutdowns: (1) the Lake-wide Longshore strike during 1960; (2) the strike of Canadian Longshoremen during 1961; and (3) the closing of the
19740 Canal and the subsequent strike by Canadian Longshoremen in September
adjusted trend manhours during the same period .2/ From these comparisons we define a representative profile of the impact of dock strikes on longshore manhours.
We define seasonally-adjusted trend manhours for any week ending on Sunday as average manhours per week during the two-year period of interest (796.91 thousand for 1964-65; 695.09 thousand for 1968-69) multiplied by a seasonal adjustment factor for the month which includes the Sunday on which the week ends. The seasonal adjustment factors are based on data for 1960,61, 66 and 70, since seasonal patterns in all other years covered by our manhour data were either affected significantly by dock strikes or, for 1967, by the disruption of trade during the months following the War in the
Middle East.2!
8/ As noted above, these are the only two strike episodes for which such comparison is both feasible and appropriate.
9/ For each month in each of these 4 years we calculated the ratio of average weekly manhours (for weeks ending on Sundays) during the month to the average level of manhours per week for the year as a whole. The 4 ratios corresponding to the same month in different years were then averaged, for each month, to yield the following seasonal factors:
Jan. = 1.0224 May = .9948 Sept. = .9528 Feb. = 1.0467 June = .9792 Oct. = .9852 Mar. = 1.0872 July = .9624 Nov. = .9828 Apr. = 1.0344. Aug. = = 9792
9720 Dec.
-9-
For both the 1964-65 and 1968-69 strike episodes, a short shutdown began when contracts expired on October 1 and ended two days later under an 80-day Taft-Hartley restraining order. In both cases the strikes were resumed after the 80-day injunctions expired. As shown for 1968-69 in Figure 1 above, there were anticipatory build-ups of trade prior to both the contract expiration date and the end of the injunction period, and a recovery of trade after the strike period. (Throughout this paper we distinguish semantically between strike episodes and strike periods; the term strike episode refers to the sequence of anticipation periods, strike period, and recovery or readjustment period.)
On the basis of Figure 1 and other impressions, we assume that our representative manhours profile follows the pattern shown in Figure 3. This, of course, must be adjusted for cases in which Taft-Hartley injunctions are not applicable, for cases in which the strike is not resumed immediately following the Taft-Hartley expiration date, and so forth. The major assumptions embodied in Figure 3 are that anticipation periodslast 3 weeks, recovery periods last 12 weeks, and during both types of periods, trade volumes change from day to day in the illustrated linear patterns. Obviously, strike episodes are not this regular; but we relax these assumptions only when we have information that suggests an appropriate modification. It should be noted that we have assumed the recovery period to begin in full force on the day after the strike ends; this may be particularly unrealistic for exports, which in many cases cannot resume until imports have been unloaded.
Under the above assumptions regarding the durations and shapes of
the anticipation and recovery periods, together with the definition of
Figure 3: Representative Manhour Profile
Deviations of Actual Manhours from Seasonally-Adjusted Trend Manhours
i Contract 12 ~ week «0 3-week Expiration 3-week ” Anticipa- Date Anticipa- Recovery Period ' tion tion Period O Period es ee ee ae _. > Time 80-day Taft-Hartley
Injunction Period
Strike Period
wee cr
- 10 -
seasonal ly-adjusted trend manhours, we now consider the actual manhour data for the 1964-65 and 1968-69 strikes, with the objective of estimating the average magnitudes of the anticipatory buildups, strike losses, and recoveries. Table 2 summarizes these data. (It may be noted that in the 1964-65 case, the strike was postponed following the expiration of the Taft-Hartley period, and actual manhours were assumed to be normal during the interval between the Taft-Hartley expiration date and the resumption of the strike.) Differences between the 1964-65 and 1968-69 profiles are attributed to different cyclical movements of income, relative prices, and other explanatory variables; for example, the apparent 100.5 percent readjustment after the earlier strike is evidently overstated due to cyclical phenomena. The rationale for taking the average profile as representative is the hope that the effects of these
10/
cyclical movements will wash out on the average. As tabulated, the profile we assume to be representative is a 29.5 percent increase in average weekly manhours during the three weeks prior to contract expiration, a 41.7 percent increase during the three weeks prior to the Taft-Hartley expiration date,
a 93.5 percent reduction in average weekly manhours during the strike period, and in the 12 weeks following the strike, a 78.5 percent recovery of the net manhours lost during the strike and anticipation periods combined.
10/- It may be noted that the estimated 100.5 percent readjustment after the 1964-65 strike and-56.5 percent readjustment after the 1968-69 strike seem to reflect cyclical phenomena associated with the fact that real income grew faster between the first and second quarters of 1965 than between the
corresponding quarters of 1969. These estimates are also consistent wit.’ possible anticipations of the boom in third-quarter 1965.
-ll-
Table 2: Manhour Profiles
All manhour data are in units of one thousand. See Appendix B for actual manhour data.
1964-65 Strike First anticipation period, September 10-30, 1964
Average weekly manhours = gsga/ 13.2% above normal2/ level of 759. Total anticipatory buildup = 300 manhours.
Second anticipation period, November 30-December 20, 1964
Average weekly manhours = 1142¢/ , 46.4% above normal®/ level of 780. Total anticipatory buildup = 1086 manhours.
Strike period, January 11-February 12, 1965
Average weekly manhours = 629/, norma1>/ = 822. Total strike loss = 3583 manhours Percentage reduction in manhours = 92.5%
Readjustment period, February 13-May 7, 1965
Average weekly manhours = 10238/; normal®/ = 839,
Total readjustment = 2208 manhours.
Net manhours lost during strike and anticipation periods combined = 2197.
Percentage readjustment = 100.5%
1968-69 Strike
First anticipation period, September 10-30, 1968
Average weekly manhours = 965a/ 45.8% above normal= b/ievel of 662. Total anticipatory buildup = 909 manhours.
Second anticipation period, November 30-December 20, 1968
Average weekly manhours = 933¢/, 37.0% above normal)/ level of 681. Total anticipatory buildup = 756 manhours.
' Strike period, December 21, 1964 - February 14, 1965 Average weekly manhours = 39£/; normall/ = 709.
Total strike loss = 5360 manhours . Percentage reduction in manhours = 94.5 percent
(continued)
- lla - |
Table 2: (continued)
Readjustment period, February 15-May 9, 1965
Average weekly manhours = 9058/; norma1}/ = 731.
Total readjustment = 2088 manhours
Net manhours lost during strike and anticipation periods combined = 3695.
Percentage readjustment = 56.5%
Average Profile
29.5% additional average weekly manhours during first anticipation period (prior to contract expiration)
41.7% additional average weekly manhours during second anticipation period (prior to Taft-Hartley expiration)
93.5% reduction in average weekly manhours during strike period
78.5% readjustment of net manhours lost during strike and anticipation periods combined.
a/ Represents average weekly rate for September 9-30. September 30 manhours are assumed equal to one-sixth of manhours worked during week ending September 29.
b/ Normal = seasonally-adjusted trend manhours, as defined in the text.
c/ Represents average weekly rate for December 2-20. All manhours during the week ending Sunday December 22 are assumed to have been worked on or prior to December 20. During 1964, the port of New York was hit by a wildcat strike on December 21 and 22; during 1968, all East Coast ports were on strike during these two days.
d/ Represents average weekly rate for January 13-February 9.
e/ Represents average weekly rate for February 13-May 11. Assumes manhours at rate of 10 per day for February 10-12 strike period.
f/ Represents average weekly rate for December 23-February 9.
g/ Represents average weekly rate for February 15-May 11. Assumes manhours at rate of 7 per day for February 10-14 strike period.
- 12 -
3. Assumed Impacts of Dock Strikes on Quarterly Trade Volumes: General Formulas
In this section we apply the representative manhour profile to each of the seven major strike episodes of the 1958-74 period, adjusting the timing of the manhour profile to fit the specific chronologies of the individual episodes, and translating the impacts on manhours into impacts on trade-byvessel (i.e., ship) in the striking regions. The estimates of daily impacts on trade-by-vessel in striking regions are then aggregated over quarters to arrive at general formulas for quarterly impacts on trade by all methods of transportation through all regions.
The translation of manhour changes into changes in trade-by-vessel is based on two assumptions. First, it is assumed that trade-by-vessel is negligable during strike periods, so that the 93.5% representative reduction in manhours corresponds to a 100% reduction in trade-by-vessel., Next it is assumed that the elasticity of trade-by-vessel with respect to manhours is constant, and hence equal to 100/93.5 = 1.07. Thus the 29.5% and 41.7% representative increases in manhours during anticipation periods are assumed to correspond to 31.6% and 44.7% increases in trade-by-vessel, A strong implicit assumption is that the same percentage changes apply to all commodity groups. Following the representative Strike, there is a 78.5 % of recovery of both net manhours lost and net trade-by-vessel lost.
We now consider the major strike episodes individually, drawing heavily on the strike chronologies in Appendix A. Postponing any specific references to exports, imports, or particular commodity groups, we use the
notation:
- 13 - T = recorded or actual volume of trade through all regions 7 by all methods of transportation, per quarter T, = normal {i.e., seasonally-adjusted trend) volume of trade through all regions by all methods of transportation, per
quarter
s__ = share of Tp normally traded by vessel through the East and Gulf Coasts regions
3, * share of Th normally traded by vessel through the West Coast region
The latter three variables are assumed to be approximately constant during each individual strike episode, but will later be assumed to vary between strike episodes, as well as between different trade categories.
3a. East and Gulf Coasts Strike, 1959-60. Based on the strike chronology (see Appendix A) we assume a 21-day anticipation period ending on September 30, 1959, during which the daily volume of trade-by-vessel was 31.6% greater than normal on the average ==! Trade-by-vessel is assumed to have been negligable between October 1 and October 8 and normal during the Taft-Hartley injunction period from October 9 through November 30. The recovery period is assumed to have started on December 1, the day a new contract agreement was reached, and was intially assumed to last for 12 weeks. (A. revision of this latter assumption is reported at the end of this subsection.) The recovery is assumed to have been 78.5 percent of the net trade-volume lost during the
anticipation and strike periods combined, with 60.2 percent of the recovery
to have increased at a constant rate from a normal level 22 days before the contract deadline to 63.2 percent above normal .on the last day of September,
117 On the basis of Figure 3, the daily volume of trade-by-vessel is assumed
initially assumed to have occurred during the first 31 days (i.e., December) .22/
These assumptions imply:
. 21 for 1959 III (92 days): T= om + 9> -316 s_ T ¢ ey ) 92 * EG n
1+ .072 T
¢ : S56) a 8
for 1959 IV (92 days): T= TT - 92 S56 T
n
+ .602 x .785 x ( oF = .072) Se¢ Ta
= (1 - .080s T ¢ 2G be 8 for 1960 I: T=T + .39 7 —e-, or 196 n 8x | 85 x (32 072) 3.4 T
= (1 + .005 T ¢ Si¢) tn
12/ Consider the recovery trfangle (with height h and base of 12 weeks or 84 days) from Figure 3:
h AREAS . ARERS- 0 t ~~ 84
The fraction of the recovery that occurs during the first t days of the 84 day period is AREA / (AREA, + AREA >)
Since AREA. + AREA = —84h nce 1 9 5 and
= & (84-t) Gi eh ,
AREAI -1 AREA? - (Se AREA, +AREA, ~ AREA, + AREA,
2 For t = 31, this fraction is l Ay. = ,602
- 15 -
- After strike adjustment factors constructed from these assumptions were tested for a cross-section of trade categories, as discussed in Section 5 below, we conjectured (largely on the basis of the pattern of residuals) that it would be more appropriate to assume that the recovery was accelerated during the pre-Christmas season. Given that net trade losses were less than 8 day'sworth of shipping volumes, our revised assumption is that the entire recovery occurred during December. This leads to the following changes:
g T
for 1959 IV; _ revised T = Th - 30 S56 n
8 + .785 x ( 92 7 -072)s, TF Zz 7 7 T (1 0 58.) a
for 1960 I: revised T = T n
3b. East and Gulf Coasts Strike, 1962-63. Trade-by-vessel is assumed to have been 31.6 percent greater than normal during a 21-day anticipation period ending September 30.
T (621II) = T + 21 T n 99 x 316s, n
= (1+ 00728, T Trade-by-vessel was down 100 percent during the first 5 days of October, then normal until the last 3 weeks of the Taft-Hartley period, during which time (12/3 - 12/23) it was 44.7 percent above normal in anticipation of a strike resumption. Trade-by-vessel was dowm 100 percent during the last
8 days of December.
-16 -
8 T (62IVv) ~ a ~ 3 SEG Th + 3 * 447855 Tt ~ 92 EG qT,
= (1 - -039sEg) Te
The striking ports remained closed for the first 25 days of January. During the remaining 65 days of the first quarter, 94.9 percentls/ of the 78.5 percent recovery occurred; the final 5.1 percent of the recovery occurred
in the second quarter.
25 25
T (631) tT 90 SEG To + .949 x .785 x 0 9 86 =(1- -0958,,) Th
T (63II) = T + .051 x .785 x 224+ .039 - .072)s 7
= (1 + .010 T (2 So Ta
3c. East _and Gulf Coasts Strike, 1964-65. Trade-by-vessel is assumed to
have been 31.6 percent greater than normal during a 21 day anticipation period ending September 30,
21 T (64III) = Th + 92 x -31685¢ Th
= (1+ 7072854) Th
Trade-by-vessel was. down 100 percent on October 1 and 2, then normal until
the last 3 weeks (11/30 - 12/30) of the Taft-Hartley period, during which time it averaged 44.7 percent above normal. The December 20 strike deadline was extended (on December 16) for 20 days, during which time we assume that trade-by-vessel was normal. |
—— 13/ This, and similar calculations below, are based on the formula derived in footnote 12.
- 17 -
T (64IV) = T 2. s| T + 2 447s T n 2 EG on 92% ***" See ty
= + .080 (1 8 S30) TT
On January 11 the strike resumed, lasting 33 days. During the 47 remaining days of the quarter, 80.6 percent of the 78.5 percent recovery occurred; the remaining 19.4 percent occurred in the second quarter.
33
3 = - e e — * . 7 = 0 T (651) Th 30 + .806 x .785 x ( 072 80) Sec TO
3 Sac Tn 90 = (1 - .231s,,) T,
33 II) = + .19 78 - .072 = .080 T T (65II) TO 194 x 5 x (3B 86 n
= (1 + -0338,6)T,
3d. East and Gulf Coasts Strike, 1968-69. Trade-by-vessel is assumed to have been 31.6 percent greater than normal during a 21 day anticipation period ending September 30.
21
T(68III) =T +—-x .316s_ T n 92 EG n
= (1 + .072 S50) Th Trade-by-vessel was down 100 percent on October 1 and 2, then normal until the last 3 weeks (11/30 - 12/20) of the Taft-Hartley period, during which time it averaged 44.7 percent above normal. Trade-by-vessel was down 100 percent during the last 11 days of December.
. 2 21 1 TOV) = T= 9285 Ty + $2 * -447 Sug T - G8ug T,
The strike was initially treated as continuing through the first 45 days of
- 18 -
the first quarter (the back-to-work-date initially assumed here, but
revised below,is that for the port of New York; see chronology, Appendix A). Accordingly, it was initially assumed that 78.4 percent of the 78.5 percent recovery occurred during the remaining 45 days of the quarter,with the remaining 21.6 percent of the recovery occurring during the second quarter.
= 7 -43 45 T(69I) = T, -95 Sug Ty + -784 x .785 x (G5- .072 + 039)8 TO
= (1 - .213sgg) T,
T(69II) = T + .216 x .785 x (43- .072 + .039)s_ T = (1+ .079s )T ( 0) n
When strike adjustment factors constructed from these assumptions were tested for a cross-section of trade categories, as discussed in Section 5 below, we found that trade volumes tended to be overestimated in 1969 I and underestimated in 1969 II. Accordingly, we decided to refine our assumptions about the timing of the recovery, based on knowledge of the dates that individual ports returned to work, together with information regarding the importance of individual ports in East and Gulf Coasts trade by vessel. Specifically, given the back-to-work date for each major port, we first estimated the fraction of each port's recovery that occurred in 1969 I, and then computed a weighted average of these fractions using as weights each port's share in the 1968 plus 1969 imports by vessel of all the ports combined. This weighted average fraction of the recovery that
occurred in 1969 I is .670, rather than .784 which we initially. estimated
- 19 -
under the simplying assumption that all ports returned to wark.on the date that the port of New York actually returned. Using the same weights, the number of days on strike during 19691 was refined from 45 (for the port of
14/ New York) to a weighted averageof 52.73. = Thus:
for 1969 I: ised T=T 32.23 s__T or 3 revise = 1-90 Ec in
+ .670 x .785 (52.73 - .072 + 1039) T oo, 90 EG n
= (1-.295 T ( ete n
for 1969 II: revised T=T + .330 x . 785 x 52.73 = .072 +.039)s_, T n 90 EG ‘n
= (I+ -143)8_, Th |
3e. West Coast Strikes, 1971 and 1972. Since the West Coast Strike (7/1- 10/8) was the first major shutdown in the West in over 20 years, and ‘since informed opinion holds that there was little evidence of advance buying in
anticipation of the strike, we assume there was no anticipatory build up before
15 the strike. There was no West Coast trade-by-vessel during the third quarter:”
T
4%, (71 III) = - s, T,
The strike continued’ through the first 8 days of October, following which trade losses began to be recovered during a Taft-Hartley cooling-off period.
14/ It is important to note that the representative profile of strikeimpacts, which is partly based on this 1968-69 strike episode, is estimated from manhours data for the Port of New York in isolation, and is independent of the treatment of the strike termination dates for other Ports.
15/ We let AT and AT, denote the impacts on total trade (T) of the West Coast and East and Gulf Coasts strikes.
- 20 -
The entire 84-day recovery occurred during the fourth quarter. During the last 3 weeks of the 80-day Taft-Hartley period (12/4/-12/25), trade-by-vessel increased by 44.7 percent in anticipation of a strike resumption, which was
postponed, however, until January.
Sw 92° Won 92
= -868s,7 Trade-by-vessel was normal for the first 16 days of January, then down 100 percent during a 35 day strike. The last 40 days of the quarter brought 72.6 percent of the recovery from this second strike; the remaining 27.4 percent
occurred during the second quarter.
35. ; ~~ (35 21 T -25 T + .72 .7 22. £5 A - (72 I) O1 Ww 6 x .785 x Gi 447 x 92) Sw TO
= 722248, TH
35 21 «274 x .785 x 1 - .447 x90 ds, Th
AW (72 IT)
=. T 061s, a
3f. East and Gulf Coasts Strike, 1971-72. Because of the West Coast Strike
underway and the apparent reluctance of the Nixon Administration to invoke its powers under the Taft-Hartley Act, the anticipatory build-up of trade prior to the East and Gulf Coasts contract deadline is assumed to have been abnormally large. ‘Specifically, we assume a six week anticipation period, ending September 30, during which trade-by-vessel was 38.15 percent (the
average of the representative 31.6 and 44.7 percent anticipations) greater
than normal.
-21-
. 42 _ ATs (71 III) = 92 * 3815 Sp = .174 sc -
A 59 day strike began on October 1, ending under a Taft-Hartley injunction. An 84 day recovery period began when ports resumed work on November 29, with
63.1 percent of the recovery occurring during the final 33 days of the year.
"59 59 | AT, (71 V) = - 35 85, Ty + -631 x «785 cH 174)s_ 7
4108, T The remaining 36.9 percent of this recovery occurredduring the first 51 days of the first quarter. In addition, trade-by-vessel increased by 44.7 percent of normal volumes during the 3 weeks (1/26-2/16) preceding the end of the Taft-Hartley period, in anticipation of a strike resumption. However, workers remained on their jobs and after a new contract was ratified on March 8, it is assumed that trade volumes dipped below normal for 12 weeks to reduce importers' inventories by 78.5 percent of the previous anticipatory build up. 51.0 percent of this reduction occurred during the last 24 days of the quarter,
with the remaining 49.0 percent occurring in the second quarter.
A Teg (72 1 = -369 x .785 x 2 - .174)s, 7
n +3 447s T - 1510 x .785 St x .447s_, T 91 EG n 92° * EG = 198s, Th
21 T (7211) =- A ee ) 490 x .785 x 2 x .4h795,
EG n
- 22 -
3g. All Strikes, 1971 and 1972. When we combine the West Coast and East
and Gulf Coasts strikes, using the formula T = T tat tat we have: T (71 IV) = (+ -868s,, - -4108,,) Th T (721) = (le 22248, + .198 Sq) T
T (72 II) = (1 +. .061s,, - 041s, ,) T
4. Assumptions Regarding Estimated Trade Shares and Implied Dock Strike Adjustment Factors .
In this section we discuss the assumptions which underlie our estimates of the SEG and 5. shares, and then plug these estimates. into the formulas from the previous section to calculate implied values of T/T,» our dock-strike adjustment factors. Trade shares, which vary between strike episodes, are estimated for 14 categories of imports or exports (as listed in footnote 1); thus we estimate a different dock strike adjustment variable for each trade category. | |
Apart from total imports, total exports, and exports of agricultural products, the trade categories which interest us here are end-use commodity groups, Import and export data for end-use commodity groups --- or for selected Schedule A or Schedule B commodity groups --- are not broken down
jointly by method of transportation and Customs Regions. Data on trade by
either method of transportation or Customs Region are not readily available
- 23 -
for end-use commodity groups, although both sets of data awe readily available since 1967 for Schedule A import commodities and Schedule B export commodities.
Under these data constraints, we first match the end-use commodity groups of interest with similar groups of Schedule A import commodities or Schedule B export commodities. For the 1968-69 and 1971-72 strike periods,
we estimate the s G share (respectively, the sy share) for each Schedule A
E or B commodity group as the product of (1) the fraction of total U.S. imports or exports of that commodity group that are shipped by vessel during the two year period of interest (either 1968-69 or 1971-72) and (2) the fraction of total U.S. imports or exports of that commodity group that enter or leave the U.S. through East and Gulf Coast Customs Regions (respectively, through West Coast Customs Regions) during the two-year period of interest. In essence, this methodology assumes that within each commodity group, West Coast trade by vessel is the same share of total West Coast trade as East and Gulf Coasts trade by vessel is of total East and Gulf Coasts trade, although this share may vary between different commodity groups and over time. Table 3 presents these estimates, along with the definitions and sources of the import and export records on which the estimates are based.
For the strike episodes prior to 1967 we do not have data by commodity on shares of trade by vessel and shares of trade through different Customs
Regions. Apparently, the only useful information available is a time series
on the shares of imports and exports of all commodities that enter and leave
- 23a - ..
Table 3: Estimated Fractions of U.S. Trade Normally Transported by Vessel Through Striking Regions
East and Gulf Coasts?! East and Gulf Coastst/ West Coast?!
1968-69 1971-72 1971-72
IMPORTS : Foods, Feeds, Beverages c/ 722 -702 -118 Consumer Goods d/ -510 491 -179 Nondurables d/ -540 -512 121 Durables d/ 0475 461 .249 Industrial Supplies e/ 558 542 .087 Capital Goods £/ 425 387 114 All Items g/ 523 -495 -105
EXPORTS
Agricultural Products h/ 745 753 142 Capital Goods i/ -388 343 061 Consumer Goods j/ -416 -409 .071 Nondurables j/ 475 -428 -048 Durables j/ -333 -383 103 Industrial Supplies k/ -634 624 -104 All Items 1/ 479 451 -089
SOURCE: U.S. Department of Commerce, Bureau of the Census. JU.S. Foreign Trade: Highlights of Exports and Imports (FT990 publications), various issues, Tables I[-2, I-6, E-2, E-5. a/ Los Angeles plus San Francisco Customs Regions. b/ All Customs Regions except Los Angeles and San Francisco. c/ "Food and live animals" plus "beverages and tobacco." d/ Consumer nondurables represent "clothing, excl. footwear" plus "footwear-rubber, leather and other" plus "printed matter." Consumer durables represent "watches, clocks, and parts" plus "musical instruments and parts, incl. phonographs, tape recorders, phonograph records, etc." plus "toys, sporting goods, and amusement equipment" plus "artworx: collectors' items and antiques" plus "radio receiving sets" plus "television receiving sets. Consumer goods represent the sum of consumer nondurables plus consumer durables. Automoti-= products are excluded. e/ "Crude materials, except fuels --- inedible" plus "animal and vegetable oils and fats" plus "chemicals" plus "manufactured goods classified chiefly by material." Petroleum
products and gas are excluded.
f£/ "Machinery" plus "aircraft and parts (excl. tires, engines, and electrical)." Transport equipment other than aircraft are excluded.
g/ "Grand total" of Schedule A imports with no exclusions.
h/ “Agricultural commodities" in Tables E-5.
i/ "Machinery--electric and nonelectric" minus "household electrical appliances" plus "railway vehicles and parts." Transport equipment other than railway vehicles and parts are excluded. ; oO
j/ Consumer nondurables represent "clothing, excl. footwear" plus "books, periodicals and other printed matter" plus "photographic supplies -- sensitized film, paper, etc." Consumer durables represent “household electrical appliances" plus "muisical instruments and parts -- phonographs, tape recorders, phonograph records, etc." plus "toys, sporting goods, and amusement equipment." Consumer goods represent the sum of consumer nondurables plus consumer durables. Automotive products are excluded.
k/ "Crude materials, except fuel -- inedible" plus "mineral fuels, libricants, and related material" plus "chemicals" plus "manufactured goods classified chiefly by materials.
1/ "Grand total" of Schedule B exports with no exclusions.
11 ee te ae RT RE SR Line ee nei RI arnt vp pati
- 24 -
the United States by vessel, together with our estimates of trade shares
in the years subsequent to 1967. In most cases we therefore approximated the SEG shares for each strike episode prior to 1967 as (1) our estimates of the s__ shares for corresponding commodity groups during the 1968-69 years, multiplied by (2) the fraction of all U.S. imports or exports traded by vessel in the two years including the strike @isode, divided by (3) the fraction of all U.S. imports or exports traded by vessel during 1968 and 1969. That is, we simply scaled our estimated sc shares for 1968-69 to reflect trends in the fraction of total imports or exports traded by vessel. This procedure reflects the assumptions that prior to 1968-69: (a) each group of imports or exports showed the same trend as total imports or exports in the proportion traded by vessel; and (b) the fraction of each group traded through a given region was the same as in 1968-69.
In the cases of imports of foods, feeds and beverages for 1959-60 and exports of agricultural products for each of the three strike episodes prior to 1968-69, assumption (a) cannot be made, since it would imply that shares of trade by vessel exceeded one. For these cases we take the share of trade by vessel to equal one, and combine this with assumption (b) in constructing the SG shares.
Table 4 presents our estimates of the extrapolated sc shares for the strike episodes prior to 1968-69, along with details of their calculation.
The final step in our calculationof dock strike adjustment factors
was to combine the estimated S56 and - shares with the expressions, from
- 2ha + ae
Table 4: Extrapolated Estimates of Fractions of U.S. Trade Normally Transporte by Vessel Through East and Gulf Coasts Regions a/
1959-60 1962-63 1964-65 IMPORTS . . Foods, Feeds, Beverages -855 b/ -841 825 Consumer Goods 611 594 -583 Nondurables 647 629 617 Durables . 569 553 543 Industrial Supplies 668 «650 638 Capital Goods -509, 495 -486 All Items 627 609 598 EXPORTS - Agricultural Products 814 b/ -825 b/ -836 b/ Capital Goods 444 452 450° Consumer Goods 476 485 482 Nondurables 543 554 551 Durables 381 388 386 Industrial Supplies 725 739 735 All Items 548 2559 555
SOURCES: (i) Table 3 and (ii) U.S..Department of Commerce, Bureau of the Census. U.Ss. Foreign Trade: Waterborne Exports and General Imports (formerly U.S. Waterbc-- Foreign Trade), FT985 publications, various issues, Tables I~-1 and E-l. a/ All notes to Table 3 apply. In-transit imports and exports were excluded in using Source (ii) to compute the fractions of total U.S. imports or exports traded by vessel. b/ Share of trade by vessel assumed equal to one; see text.
ee eee
- 2b -
Table 5. Dock Strike Adjustment Factors: IMPORTS a/
Foods, Feeds Consumer Consumer Consumer Industrial Capital All
and Beverages Goods Non- Durables Supplies Goods Items durables REVISED FACTORS
1959 III . 1.0616 1.0440 1.0466 1.0410 1.0481 1.0367 1.0451 IV -9358 29542 9515 9573 9499 -9618 = =.953C¢
1960 I 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1962 IIT 1.0606 1.0428 1.0453 1.0398 1.0468 1.0356 1.043° IV 9672 9768 9755 29784 9746 - 9807 .9762 1963 =I 29201 9436 9402 9474 - 9382 -9530 =. 9421 II 1.0084 1.0059 1.0063 1.0055 1.0065 1.0050 1.00é1 1964 III 1.0594 1.0420 1.0444 1.0391 1.0459 1.0350 1.043¢ IV 1.0660 1.0466 1.0494 1.0434 1.0510 1.0389 1.0475 1965 I 8094 8653 8574 8746 8527 8878 ~8615 II 1.0272 1.0192 1.0204 1.0179 1.0210 1.0160 1.019° 1968 IIT 1.0520 1.0367 1.0389 1.0342 1.0402 1.0306 1.0377 IV 9718 9801 9789 9815 9782 -9834 § .9796 1969 I 7870 8496 8407 8599 8354 -8746 = .8457 II 1.1032 1.0729 1.0772 1.0679 1.0798 1.0608 1.0745 1971 III 1.0041 9064 9681 8312 1.0073 .9533 9812 Iv. 8146 29541 8951 1.0271 8533 -9403 8882 1972 =I 1.1126 ~ 1.0571 1.0743 1.0355 1.0878 1.0511 1.0745 II 9784 -9908 9864 9963 -9831 -991L =.986i
INITIAL ESTIMATES?!
1959 IV 9316 9511 9482 9545 9465 9593 9496 1960 I 1.0043 1.0031 1.0032 1.0028 1.0033 1.0025 1.003: 1969 =I 28332 8822 8753 8903 8711 -9018 = .875- 1969 II 1.0570 1.0403 1.0427 1.0375 1.0441 1.0336 1.041-
— a/ Numbers represent estimated ratios of actual to normal trade volumes; for all other quarters these ratios are one. In general, these factors must be transformed before they are appropriate for use as dummy variables in trade equations; see Section 5. b/ Initial estimates equal revised factors in all but the four listed quarters, Recall Sections 3a and 3d for distinction between initial and revised factors.
- 24c - |
Table 6: Dock Strike Adjustment Factors: EXPORTS a/
Agricultural Capital Consumer Consumer Consumer Industrial All
Products Goods Goods Non- Durables Supplies Items . durables REVISED FACTORS
1959 III 1.0586 1.0320 1.0343 1.0391 1.0274 1.0522 1.0395 IV 9390 9667 9643 9592 9714 9456 - 9589
1960 I 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1962 III 1.0594 1.0326 1.0349 1.0399 1.0280 1.0532 1.0402 IV 9678 9824 -9811 9784 - 9849 9712 9782 1963 =I 9216 -9570 9539 ~9474 9631 -9298 9469 II 1.0083 1.0045 1.0049 1.0055 1.0039 1.0074 1.0056 1964 III 1.0602 1.0324 1.0347 1.0396 1.0278 1.0529 1.0400 IV 1.0669 1.0360 1.0386 1.0440 1.0309 1.0588 1.0444 1965 I -8069 -8961 -8886 8728 -9108 8303 -8718 II 1.0276 1.0148 1.0159 1.0182 1.0127 1.0242 1.0183 1968 III 1.0536 1.0279 1.0300 1.0342 1.0240 1.0456 1.0345 IV 9709 9849 - 9838 9815 -9870 9753 9813 1969 I - 7802 8855 8773 ~8599 -9018 8130 8587 II 1.1065 1.0555 1.0595 1.0679 1.0476 1.0907 1.0685 1971 III 9890 9987 1.0002 1.0265 9636 1.0046 9895 IV 8145 9123 8939 -8662 9324 8344 8923 1972 I 1.1173 1.0542 1.0651 1.0740 1.0528 1.1003 1.0694 Il ©9778 9897 9876 9854 9906 - 9808 9869
INITIAL ESTIMATES»/
1959 IV 29349 9645 9619 9565 9695 -9420 9562 1960 I 1.0041 1.0022 1.0024 1.0027 1.0019 1.0036 1.0027 1969 I 8297 9104 9039 -8903 9231 8535 8894 1969 II 1.0589 1.0307 1.0329 1.0375 1.0263 1.0501 1.0378
a/ Numbers represent estimated ratios of actual to normal trade volumes; for all other quarters these ratios are one. In general, these factors must be transformed before they are appropriate for use as dummy variables in trade equations; see Section 5. b/ Initial estimates equal revised factors in all but the four listed quarters. Recall Sections 3a and 3d for distinction between initial and revised factors.
- 25 -
Section 3, for the ratios of actual trade (T) to normal trade qa during quarters affected by strikes. Tables 5 and 6 present the resulting numerical estimates of the ratio T/T for the various groups of imports and
exports.
5. Empirical Performance of Dock Strike Adjustment Factors
Our dock strike adjustment factors (D) have been constructed as estimates, in each strike-affected quarter, of the ratios of actual trade volumes (T) to the "normal" trade volumes (T ) that would have occurred in
n
the absence of strikeeto If Th is explained by some "true" function T = £ (income, prices, other explanatory variables) (5.1) n
then ; T = (T/T) £ =D.£ (5.2) n
and log T = log f + log D © (5.3) -
On the basis of condition (5.3), we can test the appropriateness of our construction of D by using logD as a dummy variable in a regression explaining logT and examining whether the estimated coefficient on logD
17 ; _is close to one, 22/ Moreover, on the assumption that log D is uncorrelated
16/ We are assuming that unit values (prices) of tradeables are unaffected by dock strikes, Strictly speaking, the D factors have been constructed as estimates of the ratios of actual trade values to "normal" trade values.
17/ It is important to note that D must be transformed before it becomes an appropriate dummy variable. If the independent variable in our trade equation was T instead of log T, the appropriate analog of (5.3) would be T= f£+D> where D°’=T-T = F(1-YD. Thus, the appropriate dock strike dummy in such cases is n constructed from both the dock strike adjustment factors and observations on the dependent variable.
- 26 -
with the variables that explain log f, we do not need to know the "true" forr 2f the trade equation to test the coefficient on log D.
Our tests of the appropriateness of D are based primarily on the simple specification hypothesis: .
log TVAL, =a + be +c log De tu, . (5.4) where
TVAL = value of trade in current dollars
€ = index of time (t=1 for 1958 I, t = 67 for 1974 III) a,b,c = parameters to be estimated u = stochastic error term
' This choice of dependent variable avoids the problem of choosing a deflator, or obversely, of limiting the sample to a period for which an "appropriate" deflator is available. (For most of our trade categories, unit value indexes are not available on a quarterly basis before 1967.) The time index is intended as a proxy for trends in omitted independent variables, such as income and prices. On the presumption that either (i) trends in these omitted variables shift over time,or (ii) the impact of given trends on log TVAL shifts over time, due perhaps to a constant impact on trade volumes combined with a shifting rate of inflation in the prices of tradeables, the error terms are very likely to exhibit positive first-order serial correlation. Ordinary least-squares regressions in fact show very low Durbin-Watson Statistics, so we have adopted the Cochrane-Orcutt procedure to adjust for serial correlation.
It should be emphasized that our tests rely critically on the
assumption that log D is uncorrelated with the omitted price deflators and the
- 27 -
omitted cyclical components of the variables that enter the "true" relationship (5. 1)28/ If log D is highly correlated with an omitted variable, then we are not justified in expecting the estimates of c (from 5.4) to equal one (the coefficient of log D in 5. 3) 22f It should also be stressed that despite the impressive t-statistics associated with the estimates of a and b (reported below), model (5.4) is not appropriately designed for forecasting purposes; if it is desired to forecast TVAL on the basis of time and dock strikes alone, attempts should at least be made to pick up shifts in the b-coefficients over tine, |
Tables 7 and 8 present the regression results based on our initial specification of the dock strike adjustment factors. Table 9 compares the c-coefficient estimates and their t-values: (i) for estimates of (5.4) over the full sample period using a Cochrane-Orcutt correction for first-order serial correlation; (ii) for ordinary least squares estimates of (5.4), un-= adjusted for serial correlation, over the full sample periods and (iii) for specification (5.4) estimated with a Cochrane-Orcutt first-order correlation over roughly half of the sample period. (The low Durbin-Watson statistics mentioned above are reported in column 6 of Table 9.) ~ 187 If P is the omitted price deflator and TVAL = P.T is the "normal"
value of trade, then TVAL
TVAL = P* f —_— = PEC * D ( )° VAL )
and log TVAL = log P + log f + log D
19/ For a discussion of the omitted variable problem see H. Theil (1971),
Principles of Econometrics, New York : John Wiley and Sons, pp 548-50.
- 27a -
Table 7: Results for Regressions (5.4) based on Initial Formulas for D: IMPORTS
estimated coefficients 2 * Commodity group a b c R DW. RHO Foods, feeds, beverages (#0) 6.37 -0204 1.14 2972 2.32 858 (55.8) (7.95) (12.1) Consumer goods (#4) 5.59 -0390 =.593 996 1.91 869 (57.0) (18.0) (4.97) Consumer nondurables (#40) 4.58 -0390 = .536 994 2.02 -800 (55.7) (20.4) (3.65) Consumer durables (#41) 4.85 -0419 =, 683 996 1.82 886 (39.5) (15.7) (5.88) Industrial supplies (#1) 7.52 .00503 376 984 1.71 1.09% (55.7) (1.06) (3.33) Capital goods (#2) 4.42 -0495 .903 993 2.28 -831 - (34.4) (16.9) (3.58) All items 5.98 -.027 7.31 996 1.66 1.02*
(11.2) (-1.76) (8.43)
Sample period is 1958 I - 1974 III. tevalues in parentheses.
Import data are seasonally-adjusted and correspond to the Census end-use numbers in the parentheses following commodity-group titles, in millions of dollars. Imports of "2: items" are seasonally-adjusted balance-of-payments basis, in millions of dollars.
*RHO is the coefficient of first-order serial correlation in the residuals, estimated by the Cochrane-Orcutt procedure. For all cases in which RHO > 1, the Hildreth-Lu procedure was also used to scan the range. .8< RHOS1. In each case this procedure selected RHO=l and changed the estimated c-coefficients by less than -001, although changes in the estimated a and b coefficients were very large.
- 27b - Table 8: Results for Regressions (5.4) based on Initial Formulas for D: EXPORTS
estimated coefficients -2
Commodity group a b c R D.W. RHO*
Agricultural products 5.79 .039 1.26 956 2.05 947 (8.26) (.312) (9.03) a
Capital goods (#2) 4.70 .0558 .935 -996 1.97 976 (6.47) (6.00) (8.96)
Consumer goods (#4) 3.34 -0531 1.34 -988 2.34 0975 (3.75) (4.49) (10.5)
Consumer nondurables (#41) 4.68 028 1.25 - 984 2.34 927 (22.6) (6.76) (11.1)
Consumer durables (#400) -7.61 119 1.59 984 2.29 -990 (4.64) (3.36) (7.65)
Industrial supplies (#1) -16.3 -.146 1.23 .979 1.58 1.01* (21.88) (-2.33) (12.7)
All items 7,38 -.022 1.13 993 1.68 1.02* (27.2) (-1.61) (12.6)
Sample period is 1960 I - 1974 TIT for agricultural products and all items, 1958 I - 1974 III for all other commodity groups.
t-values in parentheses
Export data are seasonally-adjusted and correspond to the Census end-use numbers in the Parentheses following the commodity-groups titles, in millions of dollars. Exports of "all items" are seasonally-adjusted balance-of-payments basis, in millions of dollars. Exports of agricultural products are seasonally-adjusted Survey of Current Business data, in millions of dollars. OO
*RHO is the coefficient of first-order serial correlation in the residuals, estimated by the Cochrane-Orcutt procedure. For all cases in which RHO & 1, the Hildreth-Lu Procedure was also used to scan the range .8 = RHOG&1. In each case this procedure selected RHO=1 and changed the estimated c-coefficients by less than .001, although changes in the estimated a and b coefficients were very large.
aw
- 27ce -
fable 4. autimated c-cociltese ius bate MW tabbed foo mit bor
Subsample estimates Full-sample with Full-sample with Cochrane-
Cochrane-Orcutt correction Ordinary Least Squares Orcutt correction t- value t-value ' tevalue t-value associated associated associated associated c with c_ with c-1 c with c D. W. c with c¢ IMPORTS Food, feeds, beverages 1,14 12.1 1.49 1,24 4.28 .274 1.36 b/ 9.96 Consumer goods 593 4.97 -3.41 -672 2.00 0277 1.06 b/ 7.25 Nondurables -536 3.65 -3.15 512, 1.45 393 .852-b/ 5.33 Durables -683 5.88 -2.73 0779 =—-2,.25 2255 91. b/ 6.75 Industrial supplies 0376 3,33 #5 .53 -657, 1.10 e141 471 c/ 2.62 Capital goods .903 3,58 ~.385 1.23 1.96 .370 1.71 c/ 6.49 All items 731 8.43 -3.10 -938 2.17 0141 -989 b/ 8.89 EXPORTS Agricultural products 1.26 9,03 1.86 1.65 2.79 .163 1.39 b/ 6.92 Capital goods 935 8.96 -.623 1.18 2.65 176 . 786 c/ 5.81 Consumer goods 1.34 10.5 2.66 1.60 2.67 - 138 1.29 c/ 5.60 Nondurables 1,25 11.1 2.22 1.38 ‘3.50 231 1.35 c/ 6.24 i Durables 1.59 7.65 2.84 2.18 2.14 126 1.22 c/ 3.40 Industrial supplies 1.23 12.7 2.37 1.42 3.42 .169 1.43 c/ 10.9 All items 1.13 12.6 1.45 1.43 2.78 112 1,22 b/ 9.66
a/ Full-sample periods and commodity groups are defined in Tables 7 and 8, b/ Subsample period is 1967 I - 1974 II, chosen for comparison with Hooper and Morisse, "Forecasting Techniques and the December Quarterly Trade Forecasts," Federal Reserve Board memorandum, December 13, 1974. c/ Subsample period. is 1968 I - 1974 II,chosen for comparison with Hooper and Morisse, ibid.
-28 -
On the basis of the t-values in the second colum of Table 9, all of the c-coefficients estimated in the first column may be regarded as greater than zero with a very high degree of confidence. On the basis of the t-values in the third column of Table 9, however, a majority of c-coefficients may also be regarded as different from one (since c-l may be regarded as different from zero) with a very high degree of confidence.
Nevertheless, the casual impression conveyed by a comparison of the full sample and subsample estimates of the c-coefficients is that these _ parameters are not as precisely estimated as the high t-values (in columns 2 and 3) suggest. In any case, it does not seem appropriate <o reject these dock strike dummies flatly on the grounds that the t-tests indicate the estimated c-coefficients to be significantly different from one (given the fact that all coefficients in column are exceed -37 and are less than 1.6). ‘In the first place, we can never be very confident that our specification
20 form satisfies the conditions required for the t-test to be valid 22/ Moreover, it may be that the dock strike dummies are slightly correlated with omitted variables, in which case we cannot expect the c-coefficients to precisely 2l/ |
equal one.
20/7 In particular, the t-tests are not valid #f the Cochrane-Orcutt procedure does not succeed in completely eliminating the serial dependence of the residuals; see J. Johnston, Econometric Methods, second edition, New York: McGraw-Hill, 1972, pp.246-9.
21/ If we could be confident that the full-sample results with Cochrane- Orcutt correction indeed reflected the "true model,'' Table 9 would show an interesting counterexample to the presumption (see Johnston, ibid., p249) that ordinary least squares is likely to understate the sampling variances in the presenceof serial correlation. The t-values associated with the Cochrane-Orcutt correction exceed those associated with the ordinary
least squares procedure in all cases, even though the estimated c-coefficients are lower with the Cochrane-Orcutt correction in all cases but one.
- 29 -
The dock strike dummies should also be judged on the basis of the patterns of residuals that emerge from regressions (5.4). ‘Table 10 describes these residuals for the ten non-overlapping commodity groups. (Imports and exports of consumer goods and all items are excluded to avoid double counting.) The table shows that the average frequency of residuals larger than one standard devation is one and one-half times greater during. strike episodes (38 percent) than during other quarters (25 percent). The last two columns also suggest that in the 1959-60, 1968-69 and 1971-72 strikes, better assumptions might be made about the timing of dock strike impacts (leaving aside the magnitudes of these impacts), since for each of these episodes the table shows a large number of large positive residuals in one quarter balanced by a large number of large negative residuals in another quarter. It seems unlikely that this pattern of residuals could be due to fluctuations in omitted income variables, which tend to exhibit cycles longer than the cycles in the signs of these residuals, although the pattern could be due to fluctuations in omitted price variables.
On the basis of these residual patterns we decided to re-examine the assumptions which underlie the general strike-impact formulas that were initially adopted in Section 3. Although this is a procedure of seeking rationalizations for. revising the general formulas ex-post, it is apparent from the patterns of residuals that our ex-ante formulas very probably are at least slightly misspecified for the 1959-60, 1968-69 and 1971-72 strike episodes. Thus, if convincing rationalization for revision can be found, we feel that a modification of the general strike-impact formulas is
warranted. We have tried to err in the direction of rejecting ex-post
Table 10: Signs of Residuals Greater Than One Standard Error of Regression, Based on
Initial Formulas for D.
Total Total
MFFB MCND MCD MIS MKG XAG xXKG XCND XCD _ XIS + - 59 IIL + l 0 . C
IV +2 + + oe + +2_5(7) 0
~ E
39 60 I = - Oo 8 - - + 1 4 62 III - - = 22 - - 0 6(7 IV + - 1 1 63 #1 + + 2 0 II - - 0 2
64 III 22 Oo. 12 Iv : - | _0 2
65 IL + +2 + + - - 4(5 2 II - +2 + 23 1
68 III 0 0 IV =. =2 i¢} 233
69 2 . 2 - -2 + - 1 46) “Ir + + + + 42 © + 6(7) 1
71 111 + ee OO (> ee IV =2 - - - <3 “2 <-2 #O 7(12)
72 #7 + +2 - + - - - 3(4) 4 Il + + + + 4 0
Percent of Residuals Greater than Average
one Standard Error of Regression During Percent
Strike
Quarters 37 26 32 42 47 #38 42 32 = «42 42 38
Other
Quarters 23 26 30 26 21 #21 30 26 «17 26 25
A SS SS SASS
Residuals are from regressions using the Cochrane-Orcutt procedure to adjust for first-order serial correlation, and are computed as actual minus estimated values of the dependent variable.
Blank squares indicate residuals less than one standard deviation; signs of residuals greater than two or three standard deviations are followed by the numbers 2 or 3, respectively.
The symbol headings on the first ten columns respectively represent imports of foods ~feeds-and-beverages, consumer nondurables, consumer durables, industrial supplies, and capital goods, and exports of agricultural products, capital goods,
consumer nondurables, consumer durables and induscrial supplies. arn erence ere ene _ ee Pr. . . 7
230 - |
rationalization, however, on the recognition that an improvement in the residuals may either reflect a true improvement in the accuracy of the strike adjustment factors or any of the apparent but erroneous improvements that are bound to emerge randomly in data-mining experiments. In particular, given that many important variables have been omitted from our regression equations, the rationalizations for adjustment and the implied changes in parameter values should be quite plausible if we are to have a high degree of subjective confidence that any improvements in the residuals are due to
a washing-out of true specification errors in the strike adjustment factors rather than a random washing-out of other specification errors.
In the end our reexamination led us to revise the general strikeimpact fdrmulas for the 1959-60 and 1968-69 episodes, leaving unchanged the formulas for 1971-72. For the 1968-69 case the revision can be defended ‘strongly; we abandoned the simplifying assumption that all ports returned to work when the Port of New York re-opened, and based our revised formulas on accurate information about the sequence of dates on which individual ports re-opened (see Section 3d). In the 1959-60 case we are considering a very short 8-day strike with a recovery period beginning on December 1; and it seemed reasonable to adopt the revised assumption that the recovery from such a short strike occurred entirely during the pre-Christmas season, instead of being spread out over a 12-week period (see Section 3a). For the 1971-72 case our best rationalizations for revision were not as convincing, so we abandoned them.
Tables 11-14, based on revised adjustment factors, are the analogs
of Tables 7-10. Comparisorn$of Tables 7 and 11 and Tables 8 and 12 reveal
a a ;
* Table 11: Results for Regressions (5.4) with Revised Strike Adjustments: IMPORTS \
Estimated Coefficients
Commodity Group 3a >. R2 D.W. RHO* Food, feeds, beverages (#0) 6.37. .0203 1.05 972 2.34 .857
(55.9) (7.95) (12.0)
Consumer goods (#4). 5.59 .0391 .555 997 1.86 «874 (55.7) (17.7) (5.30)
Consumer nondurables (#40) 4.58 .0390 .515 994 1.99 .805 : (55.0) (20.1) (3.93)
Consumer durables (#41) 4.85 .0419 644 -996 1.79 - 888 (39.0) (15.6) (6.08)
Industrial supplies (#1) 7.52 .00588 .395 -985 1.63 1.09% (59.5) (1.35) (3.92)
Capital goods (#2) 4.41 .0496 .884 993 2.25 -838 (33.4) (16.5) (3.99)
All items 7.03 -.0144 .696 -996 1.47 1.02% (26.3) (-1.28) (9.37)
Sample period is 1958 I - 1974 III.
_t-values in parentheses.
Import data are seasonally-adjusted and correspond to theCensus end-use numbers in the parentheses following cemmodity-group titles, in millions of dollars. Imports of "all items" are seasonally-adjusted balance-of-payments basis, in million of dollars,
*RHO is the coefficient of first-order serial correlation in the residuals, estimated by the Cochrane-Orcutt procedure. For all cases in which RHO > 1, the Hildreth-Lu procedure was also used to scan the range .8< RHOS1. In each case this procedure selected RHO=] and changed the estimated c-coefficients by less than .001, although changes in the estimated a and b coefficients were very Large.
- 30b - -
Table 12: Results for Regressions (5.4) with Revised Strike Adjustments: EXPORTS
Estimated Coefficients -2 Commodity Group a b ¢ R DW. RHO*¥ Agricultural products 5.39 00442 1.21 -963 1.82 .957
(6.13) (3.13) (10.5)
Capital goods (#2) ‘5.61 0470 = .817 -995 2.11 .967 . (12.1) (6.75) (8.09)
Consumer goods (#4) 4.10 0454 1.19 987 2.50 .967 .. (6.47) (4.75) (9.52)
Consumer Nondurables (#41) 4.75 -0265 1.13 0982 2.47 3 .915 ; (26.4) (7.14) (10.2)
Consumer Durables (#400) -2.57 0934 1.40 -983 2.40 986 (-.898) (3.47) (7.17)
Industrial Supplies (#1) 691 -.0753 1.14 -980 1.57 1.01% (4.299) 9 (+2.17) (13.2)
All items 7.40 -.0220 1.02 -993 1.74 1.02* (27.8) (+1.63) (12.7)
Sample period is 1969 I - 1974 III for agricultural products and all items, 1958 I 1974 III for all other commodity groups.
tevalues in parentheses ;
Export data sre seasonally-adjusted and correspond to the Census end-use numbers in the partheneses following the commodity-groups titles, in millions of dollars. Exports of "all items" are seasonally-adjusted balance-of-payments basis, in millions of dollars. Exports of agricultural products are seasonally-adjusted Survey of Current Business data, in millions of dollars.
*RHO is the coefficient of first-order serial correlation in the residuals, estimated by the Cochrane-Orcutt procedure. For all cases in which RHOS1, the Hildreth-Lu procedure was also used to scan the range .8 <= RHOS1. In each case this procedure selected RHO=1 and changed the estimated c-coefficients by less than .001, although changes in the estimated a and b coefficients were very large.
- 30c - |
. . . | ; Table 13: Estimated c-coefficients and t-values with Revised Strike Adjustment s~
Full-sample with Cochrane-Orcutt correction
t-value t-value associated associated ¢. with c _with c-l_ IMPORTS - Foods, feeds, beverages’ 1.05 12.0 71 Consumer goods 555 5.30 -4.25 Nondurables 0515 3.93 -3.70 Durables 644 6.08 -3 36 Industrial supplies 395 3.92 -6.00 Capital goods 2884 3.99 -.524 All items 696 9.37 -4.09 EXPORTS Agricultural products 1.21 10.5 1.82 Capital goods 817 8.09 1.81 Consumer goods 1.19 9.52 1.52 Nondurables 1.13 20.2 1.17 Durables 1.40 7.17... 2.05 Industrial supplies 1.14 13.2 1.62 All items 1.02 12.7 .249
Subsample estimates with Cochrane- Orcutt correction
c
1.15 b/
.818 b/ .707 b/ .805 b/ -473 c/ 1.37 c/ .847 b/
t-value associated
—with c__
a/ Full sample periods and commodity groups are defined in Tables 7 and 8. b/ Subsample period is 1967 I - 1974 II, chosen for comparison with Hooper and Morisse, "Forecasting Techniques and the December Quarterly Trade Forecasts,’ Federal Reserve Board memorandum, December 13, 1974. c¢/ Subsample period is 1968 I - 1974 II, chosen-for comparison with Hooper and
Morisse, ibid.
- 30d -
Por ck)
Table 14: Signs of Residuals Greater Than One Standard Error of Regression Based on Revised Strike Adjustment Factors Total Total MPFR MCND MCD MIS MKG XAG XKG XCND_XCD_XIS + = 59 III + 1 0 Iv +2 + + EI qg_* +2. _5(7) 0 oa 60 - - 3 g + 1 2 62 IIL : : bad - bal ie] 5 Iv a 1 1 63 I + + 2 0 It bal - 0 2 64 III <2 (t) 1(2) IV - 0 1 65 I + + + - - 3 2 Il +2 + 2(3) 0 68 III (¢) ce] Iv - -2 0 2(3) 69 I - -2 + + + - 3 3(4) II - + + + - - - 3 4 71 IIl + - + +2 3(4) 2 Iv -3 - - - -3 =2 =-2 0 7(13) 72 ~«#6«f + +2. - + - 3(4) 2 II + + + 3 0 Percent of Residuals Greater Than Average One Standard Error of Regression During Percent Strike ‘ . Quarters 32 26 21 42 #37 «#38 «#842 ~=«=832~ =«42 32 34 Other Quarters 23 28 34 28 #21 28 #28 = «= 23~=619 28 26
Residuals are from regressions using the Cochrane-Orcutt procedure to adjust for first-order serial correlation, and are computed as actual minus estimated values of the dependent variable.
Blank squares indicate residuals less than one standard deviation; signs of residuals greater than two or three standard deviations are followed by the numbers 2 or 3, respectively.
The symbol headings on the first ten columns respectively represent imports of foods ~ feeds-and-beverages, consumer nondurables, consumer durables, industrial supplies, and capital goods, and exports of agricultural products, capital goods, consumer nondurables, consumer durable and industrial supplies.
- 31 -
strikingly similar estimates of the a and b coefficients in at least half the cases, although in some cases the estimates of these coefficients are affected considerably. In columm three of Table 13, the critical tvalue is 1.67 for the one-tailed 95-percent confidence test and 1.30 for the 90-percent confidence test; and on this basis most of the c-coefficients may still be regarded as different from one with a very high degree of confidence, even though all lie within the range between .39 and 1.4. This suggests the unsurprising conclusion that our construction of strike adjustment factors still involves significant errors.=2
Comparison of Tables 10 and 14 shows that the suspicious cyclical patterns of residuals have been partially eliminated by the revisions. There remain three strike-affected quarters (1959 IV, 1962 III, and 1971 IV) in which estimated trade values contain large errors of the same sign for at least half of the commodity groups; but these large residuals could be caused by factors other than errors in the strike adjustment variables. During strike-affected quarters the average frequency of large residuals for the cross-section of commodity groups is 1.3 times as large as it is during other quarters, based on the revised strike adjustment factors.
We conclude on the note that our empirical tests of the revised strike adjustment factors do not suggest any obvious deficiences that might easily be remedied. On the other hand, it is obvious that significant deficiences,which are difficult to remedy, are inherent in the numerous simplifying assumptions that underlie the strike adjustment calculations. ~~ 227 As noted above, this conclusion could be challenged by asserting
either that serial correlation invalidates the t-tests or that the dock strike dummies are correlated with omitted variables.
We do not really need formal t-tests to tell that we have been stretching
the truth. Nevertheless, we feel that the prior-information content of
our strike adjustment factors far exceeds that of any other set of dock
strike dummies,
- 33 - -~
Appendix A:_ Chronology of Major U.S. Dock Strikes Since 195823/
1.) East_and Gulf Coast Ports, 1959. Contract expired 12:01 AM on October 1. Shutdown began immediately, surprising union leaders, employers and mediators, who thought they had arranged a 15-day extension. Taft-Hartley was invoked. Ports reported back to work on October 9. Three-year contract was reached on December 1 and ratified by rank and file on December 10.
2. East and Gulf Coast Ports, 1962-63. Strike began October 1,
1962 upon expiration of work contracts. As in the four previous Atlantic
and Gulf Coasts contract negotiations, a temporary restraining order (Taft- Hartley Injunction) was issued on October 4, with return to work on October 6. Strike resumed at 5 p.m. on Sunday December 23 at the end of the 80-day injunction period, All ports returned to work on January 26.
3. East and Gulf Coasts Ports, 1964-65. Strike began at one
minute after midnight, October 1, 1964, and was ended the same day at 8 p,m. by a 10-day restraining order. Return to work began at 7 p.m., October 2. The injunction was extended through the remaining 70 days of the TaftrHartley "cooling off" period. A master contract was outlined on December 16, with
agreement to continue work 20 days past the December 20 strike deadline
—— 23/ Principal sources of the information provided in this section are:
U.S. Department of Labor, Bureau of Labor Statistics. "Fact Sheet on Collective Bargaining in the Longshore Industry, 1958-70." News Release 71-470, September 10, 1971 (plus updated information provided by Mr. Harry Cohaney of B.L.S.).
U.S. Department of Labor. Impact of Longshore Strikes on the National
Economy. January 1970
The New York Times, various issues.
- 34 -
in order to allow time for union membership to ratify the new contract. —¢ However, port activity was below normal during this 20-day interval. Due partly to confusion over new contract terms, over half of New York piers were closedby a 2-day wildcat strike on December 21 and 22. Baltimore was also shut down from December 21 through the Christmas holidays, and some South Atlantic and Gulf ports were operating below normal: On January 8 the New York rank and file rejected a revised master contract, to the shock of I.L.A. President Gleason, who blamed rejection on a lack of information and understanding by the union membership. A1IJ ports from Maine to Texas were struck at 12:01 a.m. on January 11. On January 20, the New York membership accepted the same contract, after "an intensive educational campaign." On February 10 all ports were still on strike (settlements had not been reached in the Galveston and Miami areas}2/and President Johnson named an informal committee to catalyze a return to work. A return to work order, effective February 13, reopened all ports north of Norfolk, along with Mobile and New Orleans. The strike continued in other South Atlantic. and Gulf ports, which began returning to work on March 6. .
4. East and Gulf Coasts, 1968-69. Strike began on October 1, 1968 with cargo operations. resumed in all ports on October 3, following a Taft- Hartley return-to-work order, ‘Strike resumed on December 21 after the
24] The master contract, negotiated in New York, came to terms on 5 main items: wages, duration of contract, hours of work, pensions and welfare. These terms applied as well to other Atlantic ports as far south as Norfolk, with local problems to be negotiated separately. South Atlantic and Gulf Ports bargained separately on all issues,
25/ At the time, the ILA followed the rule that all ports must settle
before any port returned to work.
- 35 -
injunction period éxpired. New York returned to work on February 15;
New Orleans, Baltimore, Norfolk, Miami, and Bridgeport on February 22; Tampa, Charleston, Savvannah and Wilmington on February 24; Philad. _phia and Mobile on February 25; Portland on March 17; Boston and the last of the Atlantic ports, Galveston and Houston on April 2; and the lastof the Gulf ports on April 13.26/ .
| 3. East, Gulf and West Coasts, 1971-72. West Coast ports struck on July 1, 1971 upon expiration of work contracts. This was the first major West Coast strike since 1948, with little evidence of advance buying in anticipation of the strike” On September 1, 9 of Chicago's 10 Largest grain elevators were closed by a strike on Grain Workers Local 418 (an
affiliate of the I.L.A.), tying up exports of corn and soybeans. On October
1, 45,000 East and Gulf Coast Longshoremen struck, joining 15,000 West Coast
longshoremen. A back-to-work order for West Coast longshoremen and Chicago grain handlers was obtained on October 6, 28/ with work resuming on October 9. East and Gulf Coast longshoremen returned to work on November 29 (negligible
numbers on November 28) under a Taft-Hartley injunction, but New York docks
26/ More detailed information on back-to-work dates is provided in a Maritime Administration memorandum, "Longshore Strike- Atlantic and Gulf Ports: December 21, 1968 - April 13, 1969," dated April 15, 1969.
27/ In addition to the history of West Coast settlement without strikes, advance buying may have been discouraged by the fact that each of the last 7 East and Gulf Coast strikes had been followed by 80-day work, resumptions under Taft-Hartley injunctions,
28/ At the time, the East and Gulf Coast strike was not judged to be critical, partly because shippers had greatly accelerated trade in recent months in anticipation on the strike,
- 36 -
were "comparatively inactive" during the first week of return. West Coast cooling off period ended December 25, but ILWU President Bridges declared
that workers would not resume their strike until at least January 10.
West Coast strike resumed January 17, followed by a settlement on February 19 and return to work beginning February 21. Cooling-eff period for East
and Gulf Coasts expired February 16, but workers remain on their jobs. On March 8, the union memberships in most East and Gulf Coast ports ratified
a new contract, with Philadelphia becoming the last port to reach settle-
ment on March 29.
- 37 «
Appendix B: Hours Worked by Members of the International Longshoremen's Association in the Port of New York
Table B.1l Average I.L.A. Weekly Hours per Month, 1960-63
in units of one thousand hours
1960 1961 1962 1963 January 783 810 800 166 February 836 797 817 1,092 March 906 - 871 887 967 April 872 778 845 857 May 849 722 802 788 June 861 698 786 782 July 787 765 777 781 August 824 757 769 769 September 803 767 800 760 October 815 771 718 796 November 803 800 78L =o 795 December 817 785 961 821 Total hours 43, 734 40,529 41,641 40,903
worked per year
Table B.2 Thousands of I.L.A. Ho
Week Ending ©
January 4 11 18 25
February 3 10 17 24 March 3 10 17 24 31 April 7 14 21 28
June 2
Hours
697 772 924 863
850 845 793 866
- 38 -
urs Worked Per Week, 1964
Week Ending
July 7 August 4
September 1 8
1
22
29 October 6
November 3 December 1
29
Total
Hours
757 730 785 764. 746 726 758 720 766 688 826 782 931 568 748 808 802 746 826 761 845 895 948
1,198 952 602
42,026
Table B.3 Thousands of I.L.A. Hours Worked Per Week, 1965
Week Ending Hours Week Ending Hours January 5 637 July 6 663 12 610 13 721
19 56 20 718
26 70 27 680 February 2 64 August 3 806 9 56 . 10 740
16 679 ; 17 713
23 1,198 24 727 March 2 1, 232 31 793 9 1,172 September 7 795
16 1,076 14 916
23 994 21 880
30 985 28 836
April 6 1,038 October 5 852 13 953 12 757
20 816 19 844
. 27 984 26. 828 May 4 : 901 November 2 774 11 863 , .@9 892
~ 18 834 . 16 833
-. 2s 814 ; 23 874 June 1 725 30 860 8 846 December 7 898
15 © 861 | 14 858
22 CO; 735 21 924
29 708 28 736
Total 40,853
Table B.4 Thousands of I.L.A. Hours Worked Per Week, 1966
Week Ending Hours Week Ending Hours January 4 771 July 3 845 11 806 10 727
18 839 17 847
25 848 24 797 February 1 814 31 828 8 972 August 7 783
15 870 14 793
22 8357 21 818
March 1 907 28 775 8 960 September 4 774
15 948 1l 713
22 938 18 793
27 602 25 789
April 3 868 October 2 843 10 698 9 841
17 950 16 772
24 899 23 822
May 1 ; 889 30 . . 893 8 833 November 6 808
15 831 13 717
22 859 20 905
29 831 27 733
June 5 738 December 4 858 12 858 11 809
19 846 18 844
26 802 25 814
31 673
Total 43,651
Table B.5 Thousands of I.L.A. Hours Worked Per Week 1967 RRR al Ae
Week Ending Hours Week Ending Hours January 8 766 July 2 756 15 840 9 635
22 843 16 754
29 824 23 675 February 5 803 30 742 12 810 August 6 696
19 835 13 700
26 821 20 717
March 5 878 ; 27 735 12 881 September 3 740
19 «| 917 10. 599
26 857 17 698 April 2 944 24 678 9 860 October 1 758
16 762 8 680
23 812 . 15 672
30 835 22 640
May 7 812 29 489 14 784 November 5 634
21 747 12 883
28 . 7124 19 849
June 4 667 26 745 11 769 December 3 813
18 764 10 811
25 733 17 839
24 772
31 676
Total Hours 39,700
~- 42 -
Table B.6 Thousands of I.L.A. Hours Worked Per Week, 1968 en kann an MEA
Week Ending Hours Week Ending Hours January 7 704 July 7 643 14 748 14 793 21 863 21 762 28 881 28 799 February 4 853: August 4 821 11 821 1l 795 18 742 18 740 25 749 25 787 March 3 802 September 1 820 10 815 8 689 17 819 15 876 24 55 22 947 31 482 29 1,037 April 7 1,003 October 6 356 14 766 , 13 565 21 916 , 20 620 28 867 27 710 May 5 806 November 3 633 12 800 10 669 19 695 17 720 26 751 24 786 June 2 699 December 1 - 755 9 575 8 878 16 879 15 : 916 23 748 22 738
30
772 29 34
Total Hours: 38,522
- 43 -
Table B.7 Thousands of I.L.A. Hours Worked Per Week, 1969
Week Ending Hours Week Ending Hours January 5 41 July 6 614 12 41 13 712
19 36 - 20 720
26 37 27 644 February 2 40 Y August 3 755 9 44 10 726
16 240 17 684
23 937 24 678
March 2 957 31 705 9 977 September 7 606
16 970 14 721
23 997 21 665
30 926 28 671
April 6 701 October 5 661 13 927 12 675
20 = 909 19 583
27 920 oo 26 627
May 4 870 . November 2 - - 667 11 815 9 665
18 813 16 659
25 779 23 652
June 1 -675 30 583 8 719 December 7 691
15 756 14 649
22 664 21 721
29 728 28 490
Total 33,767
- 44 -
Table B.8 . Thousands of I.L.A. Hours Worked Per Week, 1970
Week Ending Hours Week Ending Hours
January 4 620 July 5 527 11 687 12 616
18 671 19 599
25 711 26 570
February 1 727 August 2 612 8 655 9 566
15 620 16 595
22 706 23 596
March 1 594 30 588 8 686 September 6 568
15 728 13 552
22 701 20 569
29 615 27 595
April 5 678 October 4 601 12 648 ll. 589
19 636 18 ° 552
26 636 25 614
May 3 657 November 1 632 10 634 8 563
17 592 15 573
24 626 22 642
. 31 572 29 478
June 7 572 December 6 - 582 14 625 13 454
21 607 20 688
28 659 27 489
Total 31,898
- 45 -
Table B.9 Thousands of I.L.A. Hours Worked Per Week, 1971
Week Ending
January 3 10 17 24 31 February 7
March 7
April 4
June 6
Hours
512 656 576 588 611
632 -
566 553 591 571 595 576 541 573 553 594 558 577 559 563 532 554 494 570 548 588
Week Ending July 4
August 1
September 5
October 3
November 7
December 5
Total
574 559 451
26,024
7 Table B.10 Thousands of I.L.A. Hours Worked Per Week, 1972
Week Ending Hours Week Ending Hours January 2 458 July 2 569 9 610 9 404 16 604 . 16 478 23 609 . 23 527 30 663 30 484 February 6 605 August 6 507 13 643 13 456 20 645 20 465 - 27 495 27 490 March 5 567 September 3 476 12 552 10 404 19 539 17 453 26 508 24 489 ‘April 2 418 October 1 475 9 533 8 506 16 516 15 465 23 510 22 500 30 478 29 430 May 7 489 November 5 500
14 490 12) «4427. 21 465 19 464 - 28 479 26 409 June 4 451 December 3 466 - A. 484 10 467 - 18 512 17 485 25 425 24. 474 3L 433
Total 26,477
Cite this document
Federal Reserve (1975, January 31). Dock Strike Adjustment Factors for Major Categories of U.S. Imports and Exports, 1958-1974. Ifdp, Federal Reserve. https://whenthefedspeaks.com/doc/ifdp_1975-60
@misc{wtfs_ifdp_1975_60,
author = {Federal Reserve},
title = {Dock Strike Adjustment Factors for Major Categories of U.S. Imports and Exports, 1958-1974},
year = {1975},
month = {Jan},
howpublished = {Ifdp, Federal Reserve},
url = {https://whenthefedspeaks.com/doc/ifdp_1975-60},
note = {Retrieved via When the Fed Speaks corpus}
}