Falling Profits, Rising Profit Margins, and the Full

CHARLES L. SCHULTZE
BrookingsInstitution
Falling
Profit
the
Profit
Profits,
Margins,
Rising
and
Full-Employment
Rate
SINCE EARLY1973 the major price aggregateshave risen far more than
can be explainedby the directeffectsof risingunit laborcosts and higher
prices for fossil fuels and other importedcommodities.The margin of
pricesover standardunit laborcosts for privatenonfarmdomesticoutput
widenednot only duringthe last phasesof expansionin 1972-73but even
moresharplyin the recessionthat followed.At today'sprice-costrelationships, full-employmentlevels of output and associatedlevels of productivity would generatevery large profits.The full-employmentprofit rate
(firstcousinto the full-employmentsurplus)has risenvery rapidlyduring
the past two years,accountingfor an importantpart of the inflationand
perhaps-like its budgetarycousin-for an importantpartof the recession.
The behaviorof two majorpriceindexes-the deflatorsfor privatenonfarm domesticoutput and for the gross productof privatenonfinancial
corporations-was comparedto variousmeasuresof unit laborcosts. But
first,each priceindexwas adjustedto excludethe effectof the relativeincreasein domesticfossil-fuelpricessincethe onsetof the embargoin October 1973.1After the adjustment,each deflatorexcluded,at least concepNote: I am gratefulto James Beckerand LeonardHerk for researchassistance,and
to membersof the Brookingspanel for commentsthat improvedthe quality,even if they
added to the length, of this sector report.
1. GeorgePerryhas calculatedthe dollarrevenuesof domesticoil producersquarterly
since 1973:2; see EdwardR. Fried and CharlesL. Schultze,eds., HigherOil Prices and
Problem(Brookings Institution, 1975), chap. 2,
the WorldEconomy:The Adjustmenit
449
450
BrookingsPaperson EconomicActivity,2:1975
tually,the directeffect of increasesin (1) importprices;(2) farm prices;
and(3) domesticfossil-fuelprices.In additionthe deflatorfor the outputof
privatenonfinancialcorporationsexcludesthe pricesof sectorswhoseoutput is particularlyhard to measure,such as households,financialinstitutions,andmost of the residential-rental
sector,withits low laborintensity.
Pricesin Relationto StandardCosts
Table 1 showsthe ratio of the adjustedpricedeflatorto estimatedstandard unit labor costs (SULC) for each of the two sectors.The standard
unitlaborcosts werederived,in a more or less simple-mindedway, by fitting productivityin each sector(quarterly)to a time trendplus a cyclical
term,the ratioof actualto potentialgrossnationalproductin eachquarter.
In the privatenonfarmsector two time trendsare incorporated,one for
1954-64and one for 1965-75,but the cyclicaladjustmentis constrainedto
be the samein both periods.2This formulationis clearlymisspecifiedwith
respectto veryshort-runmovements,sinceactualproductivityalwaysfalls
below trendin the earlystages of a recessionby more than the equation
predicts.But for movementsof morethana few quartersthe equationcapturesthe cyclicalswings(see the appendix).Standardunit labor costs are
then derivedby settingthe ratio of actualto potentialGNP at 0.97, and
allowingonlythe trendmovementin productivityto affectunitlaborcosts;
an indexof the ratio of pricesto standardcosts, so defined,for the private
nonfarmsector,is depictedin figure1.
As table1 shows,pricesin bothsectorsfell graduallyrelativeto standard
unit laborcosts duringthe periodprecedingthe introductionof wageand
table 2-4. A calculation was made of the excess of the increase in these revenues,
comparedwith what would have occurredhad oil prices risen at the same rate as the
privatenonfarm deflator.An estimatedincrement,graduallyincreasingover time, was
added to take account of rising pricesfor coal and naturalgas; the incrementequaled
30 percentby early 1975. The resultingdollar value of increasedrevenuesaccruingto
producersof domesticfossil fuel was dividedby the current-dollarvalue of the private
nonfarmGNP to yield the price adjustment.A correctionwas made to eliminateestimated revenues of noncorporatefossil-fuel producers,and a similar calculation was
carriedout for the privatenonfinancialcorporatesector. By early 1975 the adjustment
lowered the private nonfarm deflatorby 1.5 percent and the deflator for the private
nonfinancialcorporatesector by 1.8 percent.
2. The data for privatenonfinancialcorporationsare fitted from 1958 on, since the
data are availableonly since then.
INr
.4.X.2
%W~
~ ~ ~ ~~~~~~~~~~~~~~~~~~~0
X?-~~~~~~~~~~~~~~
f
l
i
l
I..
U,
.2
l
0~~~~~~~~~~~~~0
iz
.4b
0:?
?
%ss
Brookings Papers on Economic Activity, 2:1975
452
Table 1. Ratio of Adjusted Prices to Standard Unit Labor Costs,
Private Nonfarm and Nonfinancial Corporate Sectors, 1965-75a
Sector
Year
and
quarter
1965
1966
1967
1968
1969
1970
1971:1
2
3
4
1972:1
2
3
4
1973:1
2
3
4
1974:1
2
3
4
1975:1
2
Private
nonfarm
100.1
98.6
98.7
97.0
96.8
97.0
97.1
96.8
96.6
96.5
95.6
95.3
95.0
94.6
93.4
93.8
94.0
94.2
95.0
96.1
97.3
98.6
99.3
99.0
Private
nionifintancial
corporate
99.9
99.2
99.9
98.4
97.2
97.6
97.4
97.3
97.0
96.9
96.0
96.0
96.0
95.9
94.5
94.8
95.1
95.0
95.8
96.6
97.7
98.6
98.6
Sources: Underlying data on prices, compensation per manhour, and productivity from U.S. Bureau of
Labor Statistics, Chartbookon Prices, Wages, and Productivity, various issues. Standard unit labor costs
were derived as explained in the text. Also see the text for a description of the adjustment to the price
indexes.
a. The ratios are derived from underlying indexes for which 1967 = 100.
pricecontrolsin August 1971. Marginswere then squeezedmore tightly
until early 1973, when Phase III was introduced.In the two subsequent
years,however,the marginof pricesover standardunit laborcosts recoverednot only the losses imposedby pricecontrolsbut all of those suffered
in the sevenyearssincethe mid-1960s.A substantialpart of the pricerise
from 1973:1 to 1975:1 is attributableto the increasein this margin:3
3. The changeattributableto the grossincreasein the marginis the differencebetween
the actual price change and the price change with a constant percentage markup over
standardunit labor costs.
CharlesL. Schultze
453
Private
Private nonfinancial
nonfarm corporate
sector
sector
Increasein percentagemarkupover standard
unit labor costs
Standardunit labor costs
Domestic energyrents
Total percentchange
7.2
13.6
1.7
22.5
4.9
12.3
2.1
19.3
Most priceequationsthat relatepricesto standardunit laborcosts also
assigna modestpricingeffectto deviationsof actualfrom standardcosts.
To what extentcould the rise in the ratio of priceto standardunit labor
costs be due to this effect?For each of the two sectorsthe followingequation was fittedfor quarterlypercentchangesin prices:
t-3
(1)
P,
=
t-1
ao + E ali SULCi + E
i=t
i=t
a2i
(AULCi - SULCQ)+
Ut,
where
P=
price deflator
SULC = standard unit labor costs
A ULC = actual unit labor costs
u = error term.
All variablesare expressedas quarterlychangesin logarithms;the private
nonfarmsector was fitted from 1954:1 to 1971:3, and the privatenonfinancialcorporatesectorfrom 1958:1 to 1971:3.
The resultsare shownin table2. The purposeof this particularequation
wasnot to determineeitherthe best structuralor the most usefulpredictive
relationshipbut to capturethe historicalbehaviorof pricesrelativeto unit
laborcosts. The datawerefittedonly throughthe thirdquarterof 1971to
avoidincorporatingthe effecteitherof pricecontrolsor of the increasein
the grossmarginsincetheirremoval.
In table 3, the movementof pricessince the quarterimmediatelyprecedingpricecontrols(1971:2) is comparedto an extrapolationbased on
the equationson unit labor cost. (The actualpriceshave again been adjusted to removethe effect of increasesin the relativeprice of domestic
fossil fuels.)
Brookings Papers on Economic Activity, 2:1975
454
Table2. Estimatesof Equation(1), PrivateNonfarmand
Nonfinancial
CorporateSectorsa
Sector
Variableb
and
summarystatistic
Private
nonfinancial
corporate
Private
nonfarm
Parameter
Coefficient t-statistic Coefficient t-statistic
Variable
Constant
ao
0.126
...
0.074
...
SULC
a, (t)
(t-1)
(t-2)
(t-3)
Mali
0.231
0.202
0.173
0.145
0.751
4.6
8.5
6.8
2.7
10.1
0.312
0.231
0.150
0.069
0.762
5.2
8.4
5.1
1.1
9.1
(t)
0.141
0.049
0.190
3.8
1.3
0.149
0.079
0.228
3.9
2.2
...
(AULC-SULC)
a2
(t-1)
Za2i
Summarystatistic
RI
Durbin-Watson
...
0.66
2.036
0.68
2.011
Source: Text equation (1).
a. Period of fit: nonfarm, 1954:1-1971:3; nonfinancial corporate, 1958:1-1971:3.
b. SULC = standard unit labor costs; AULC = average unit labor costs. The variables are expressed
as quarterlychanges in logarithms.
Duringthe periodof effectivepricecontrols,between1971:2and 1973:1,
the price of value added had risen in the privatenonfarmeconomy by
about 1?/2percentless, and in the privatenonfinancialcorporatesectorby
about 1 percentless, than could have been predictedon the basis of the
behaviorof unit labor costs. By the end of 1973 that shortfallhad been
wipedout in both sectors.Overthe two-yearperiodfromthe end of effective pricecontrolsto early1975,pricesfor nonfarmproductsrose20.7 percentcomparedwitha riseof only 12.6percentpredictedby the equationon
unit labor costs. In the nonfinancialcorporatesector the actual increase
was 17.2percentcomparedwith the predicted11.5percentrise.
Reasonsfor RecentMovements
Nothingin the analysisitselfsuggeststhe reasonsfor the recentbehavior
of pricesrelativeto unit labor costs. Duringthe period of effectiveprice
455
Charles L. Schultze
Table 3. Comparison of Actual Ratio of Prices to Standard Unit Labor
Costs with Estimates from Equation (1), Private Nonfarm and
Nonfinancial Corporate Sectors, Quarterly, 1971-75
Sector
Privatenonfinancial
corporate
Privatenonfarm
Period
Actual
indexes
Estimated
inidexes
Actual
inidexes
Estimated
indexes
Price indexes(1971:2 = 100)
1971:2
3
4
1972:1
2
3
4
1973:1
2
3
4
1974:1
2
3
4
1975:1
100.0
100.7
100.8
101.9
102.2
102.6
103.4
104.3
105.7
107.0
108.9
111.7
115.5
119.0
122.8
125.9
100.0
100.8
101.6
102.6
103.4
104.1
105.0
106.0
107.2
108.5
110.0
111.6
113.4
115.3
117.4
119.4
4.3
7.1
20.7
6.0
5.3
12.6
100.0
100.5
100.6
101.6
101.8
102.2
102.8
103.5
104.7
105.9
107.1
109.3
112.5
115.6
118.7
121.3
100.0
100.7
101.2
102.0
102.6
103.1
103.6
104.4
105.4
106.6
107.8
109.3
110.9
112.7
114.6
116.4
Percentchanges
1971:2-1973:1
1973:1-1974:1
1973:1-1975:1
3.5
5.6
17.2
4.4
4.7
11.5
Source: Actual, table 1; estimated, text equation (1).
controlsthe behaviorof pricesrelativeto unit laborcosts supportsRobert
Gordon's1973findingsthat PhaseII priceand wagecontrolsdid squeeze
margins.4It seemseminentlyreasonableto attributethe initialrisein prices
relativeto unit laborcosts during1973to the gradualdismantlingof controls.And some of the furtherexpansionin the marginduring1974might
be ascribedto fearsof renewedimpositionof priceand wagecontrols.But
the magnitudeof the expansion,and its continuationin the face of overwhelmingcongressionalhostilityto a reimpositionof controlsand of the
4. RobertJ. Gordon, "The Response of Wagesand Pricesto the FirstTwo Years of
Controls,"BPEA, 3:1973, pp. 765-78.
456
BrookingsPaperson EconomicActivity,2:1975
sharpestsagin aggregatedemandsincethe 1930s,suggeststhatmuchmore
was involvedthan the simplefear of futurecontrols.Severalhypotheses
can be offered.
INCREASES IN IMPORT PRICES
As Gordonpointsout, the structurallyexpectedeffectof increasesin import pricesin elicitingsympatheticrises in prices of competingdomestic
productscannotbe coaxedfromthe time-seriesdatapriorto 1973because
the variancein the relativepriceof importswas so smallduringthe period.
It is hard to believe,however,that the sharprise in importprices(apart
fromfood and fuel)beginningin 1973did not elicit significantpricegains
amongcompetingdomesticproducts.So long as the relativepriceof competitiveimportsdoesnot fall(throughdifferentialpricebehaviorabroador
dollarappreciation),and so long as unit labor costs do not rise to fill the
gap, a disequilibrium
remains,in the sensethat the full-employment
profit
ratewill remainabove"normal."(In a deeprecessionlike the currentone,
thisfactmaybe hiddenby the cyclicaldepressionof actualproductivityand
the correspondingrisein actualunit laborcosts-hence the kinshipof the
full-employment
profitrateandthe full-employment
budgetsurplus,noted
above.)
THE RISING COST OF CAPITAL GOODS
The priceequationsset out earlierincorporate,duringthe periodof fit,
a very slight downwardtrendin the ratio of priceto standardunit labor
costs. In the privatenonfarmsector,standardunit laborcosts represented
56.5 percentof pricein 1967.The price equationimpliesthat apartfrom
cyclicalvariations,priceswill riseby 0.5 percentper yearplus 0.75 percent
for each 1 percentrisein standardunit laborcosts. On a steady-statebasis
thismeansthatwheneverstandardunitlaborcosts riseby morethan2 percent per year,priceswill increaseby less than that rise.The "break-even"
point for the nonfinancialcorporatesectoris a 1.3 percentannualrise in
standardunit laborcosts. Obviously,this is not a well-specifiedstructural
relationship.In a sustainedperiodof verymuchlargerrisesin standardunit
labor costs, or when prices of capital goods rise significantlyfasterthan
standardunit labor costs, the relationshipsof priceto unit labor cost impliedin the equationsclearlywill not hold up.
CharlesL. Schultze
457
In the purelycompetitiveworld,a sustainedand rapidrise in pricesof
capitalgoods would tend to reducethe marginalefficiencyof investment
to the pointwherecapacityexpansionwas slowedrelativeto the growthof
demand(underconditionsof full employment).Priceswouldthenriserelative to standardunit labor costs by enoughto restorerates of returnon
pricingrules,in whichpricesare
investment.In a worldof administrative
increasedas standarddirectcostsrisebut someattentionis paidto a target
rateof return,an accelerationof inflationwouldlikelyinitiatethe following
sequence:
First,the old pricingrules,relatingpricesto standarddirectcosts,would
be followed.
Second,this practicewoulddepressthe marginof pricesrelativeto standardunit labor costs, and yield a decreasingrate of returnto new investment, sincecapital-goodspriceswouldbeginto rise fasterthan the rate of
increasein the nonlaborreturnsper unit of output.
Third,ratherthan waitingfor the classicallong-runadjustmentto take
place throughthe slowdown of capacityexpansion,price policy in the
sectorwould be adjustedto correctmargins
dominantadministered-price
for the inadequacyof historicalpricingrules.
Fourth, this developmentwould reversethe downdriftof the ratio of
priceto standardunit labor costs and restoreit to earlierlevels.
To get some impressionof the potentialmagnitudeof pricechangesrelated to capitalrecovery,I have constructeda price seriesfor the private
nonfinancialcorporatesectorbased on a set of pricingrules designedto
yield, at standardoutputvolune, an unchangedrate of returnto capital,
pricedat reproductioncosts.Thisserieswasconcocted,using1969as a base
year,by settingpricesso as (1) to recoverstandardunit laborcosts; (2) to
recoverthat fractionof the deviationof actualfrom standardunit labor
costs impliedin the priceequationdiscussedearlier;(3) to recoverindirect
taxesperunit of output(witha smalladjustmentto allowrecoveryof propertytaxes at standardvolumesof output);(4) to recover,at standardvolumes of output,the 1969capitalcosts per unit (net interest,depreciation,
and profits)extrapolatedby the price deflatorfor nonresidentialfixedinvestment.
Assuminga constantratioof capitalstockto capacityoutputand a constant real interestrate, the resultantprice level would yield, at standard
outputvolume,the 1969rateof returnon incrementalinvestment,pricedat
Brookings Papers on Economic Activity, 2:1975
458
Table4. Actual,Fitted,andSimulatedPrices,Costs,andGrossMargins
perUnit of Output,PrivateNonfinancialCorporateSector,
SelectedPeriods,1969-75
Price component
Period and item
Unit
labor Indirect Gross
costs
taxes margins Total
Price
chanige Cyclical
from
absorption
previous incorporated
period
in gross
(percent) margin
1969 (base)
65.7
9.4
24.9
100.0
...
1971:2
Target-rate-of-return
simulation
Actual
Fitted
71.3
71.3
71.3
10.8
10.8
10.8
27.1
26.0
24.7
109.2
108.1
106.8
9.2
8.1
6.8
1973:1
Target-rate-of-return
simulation
Actual
Fitted
73.8
73.8
73.8
10.5
10.5
10.5
29.7
27.7
27.2
114.0
112.0
111.5
4.4
3.6
4.4
1975:1
Target-rate-of-return
simulation
Actual
Fitted
89.8
89.8
89.8
12.7
12.7
12.7
31.9
28.6
21.9
134.4
131.1l
124.4
17.8
17.1
11.6
-0.2
o0.4
-1.2
.
3.5
J
Sources: Actual, table 1; fitted, table 3; simulation, see text; all data converted to indexes, 1969 = 100.
a. The actual total price for 1975:1 was adjusted by 1.8 percent to exclude increases in energy prices.
reproductioncosts.5The realizedcapitalrecoveryperunit of outputwould
vary fromthe targetby the absorptionof cyclicalvariationsin unit labor
costs.
The resultsof this simulation-in termsof pricesand costs-are shown
in table4. Theyarealso comparedboth withactualpricesandmarginsand
withthoseimpliedin the pre-1971fittedrelationshipof pricesto unit labor
costs.
In the roughlytwo yearsbetween1969andthe quarterjust precedingthe
5. The price-costrelationshipswould generate(1) a returnon new investmentequal
to the 1969 returnon investedcapital; and (2) a series of "capitalgains" sufficientto
recoverthe differencebetweenrealand nominalratesif the excessof the nominalinterest
rate over the real rate correctlyanticipatedthe rate of inflation.
CharlesL. Schultze
459
impositionof pricecontrols,actualpricesrose more than the calculated
valuefromthe labor-costequationand somewhatless thanthe target-rateof-returnsimulation.After two years of controls, prices had risen by
slightlylessthaneitherthe labor-costequationor the target-ratesimulation
suggested.In the succeedingtwo years,however,the risesin the actualand
simulatedindexeswerevirtuallythe sameand muchlargerthan predicted
by the labor-costequation.
A REDUCTION IN EXPECTED OUTPUT LEVELS
To the extentthat pricingdecisionsaremadewitha targetrateof return
in mind,changesin expectationsaboutthe long-runaveragerateof capacity utilizationcould affectthe markupof pricesover standardunit labor
costs. If both average expectedoutputandpeak expectedoutputfall (and
by the sameamount),thereshouldbe no effecton margins,sincethe fall in
averageexpectedoutputwould be matchedby a fall in capacityrequirements.Butif averageexpectedoutputfallsby morethanpeakexpectedoutput, gross capitalreturnsper unit of outputwill have to rise to yield the
targetrate of returnon investment.
Figure2 tracesa twenty-quarter
movingaverageof the ratioof actualto
potentialGNP. While this is hardlythe most sophisticatedindicator,its
movementsshouldprovidesomemeasureof long-runchangesin expected
utilizationlevels. To the extent that rules of thumbin pricingpolicy are
influencedby the targetrate of return,an increasein expectedutilization
ratesshouldlower, and a decreaseshouldraise,the ratio of priceto standardunit laborcosts.
A Test of the Hypotheses
To the extentthat the hypothesesdiscussedabove are valid,the ratio of
price to standardunit labor costs should be influencedby the following
factors:The long-runfactorsare the relativepriceof capitalgoods (positive), and the long-runexpectedrate of capacityutilization(negative).The
short-runfactorsarethe deviationof actualfromstandardunitlaborcosts
(positive),andthe ratioof currentto long-runexpectedcapacityutilization
(positive,reflectingthe state of excessdemand).
ON
Q~~~~~~~~~~~~
z
ON
4-o
V-1
/
4t
_o
U
o*
L
o
X
oo
W
t
0
(
-
s
1)C
CharlesL. Schultze
461
The followingequationwas fitted to the quarterlyratio of prices to
standardunit labor costs for the privatenonfarmsector:
Pt_
(2)
SuLtC
=
aO + al
+ a3 (1)
+
(A ULC\
SULC)t
a4
+
a2
[y
[p-(yp (*
]
+ a5t + a6D + ut,
where
P = price deflator (1967 = 1.00)
A ULC = actual unit labor costs (1967
SULC = standard unit labor costs
t
=
time (1958:4 = 1)
y
=
yp
=
actual GNP
potential GNP
=
1.00)
(Y/YP)*= five-yearmovingaverageof y/yp
(PI/P)* = one-yearmovingaverageof the relativepriceof businessfixed
investment
D
u
=
=
price-control dummy (1971:3-1974:1
error term.
=
1.0; 0 otherwise)
Two formsof the equationweretried,one usingthe laggedvalueof the
dependentvariable,the other using independentlag structuresand a rho
correction.In addition,both sets of equationswerefittedfirstto the period
1958:4 to 1971:2 and then to the longer period 1958:4 to 1974:1. The
basic results are shown in table 5. The coefficient on the relative price of
capitalgoods was in no case significant,and was of the wrongsign. All of
the equationsin table 5 wererefittedwithoutit.
Variouslag structuresweretriedfor alternativeB, the equationswithout
the lagged dependentvariable.In all cases the coefficientson the lagged
values of deviationsof actual from standardunit labor costs and on the
lagged values of the demandterm were highly insignificant,and clearly
"interfered"
with each other.The measureof expectedlong-termcapacity
utilization(y/yp)* is itself a twenty-quarter
movingaverageand needsno
lag. Onlythe price-controldummyseemedto requirea lag structure.Since
the margin-squeezing
effectof pricecontrolsoperatedthroughdelaysand
refusalsto allow a full cost passthrough,a constant degree of control
wouldresultin a gradualsqueezingof margins,at leastfor a while,as costs
rose. Relaxingthe severityof controlscould imply a maintenanceof the
squeezeon marginsat a constantlevel (thatis, the generallyacknowledged
462
BrookingsPaperson EconomicActivity,2:1975
easingof controlsthat accompaniedthe introductionof PhaseIII in early
1973neednot haveimplieda wideningof margins,only a cessationof the
previouslytighteningsqueeze).Finally, abandonmentof controlsshould
have led to a gradualrestorationof margins.A four-quarter,first-order
Almonlag was chosen, and, as expected,the negativecoefficientson the
constantdummytermincreasedin value over the four quarters.
The lowerbank of data in table5 comparesthe steady-statecoefficients
fromalternativeA with the coefficientvaluesand the sum of the dummy
coefficientsfromalternativeB. Exceptfor the versionof alternativeA fitted
through1974:1, the effectsof deviationsin actualunit labor costs and of
long-termchangesin capacityutilizationare virtuallythe same.About 60
percentof any deviationin actualfromstandardunit laborcosts is passed
throughinto prices. (Unit labor costs are approximately57 percentof
prices;0.35/0.57 = 0.61.)Thiseffectof deviationsin actualfromstandard
unit laborcosts is somewhathigherthan those observedby others,which
tendto take on a valueof 0.2 ratherthanthe 0.35 in the currentequations.6
But, as shown below, the currentestimateholds up well when projected
past the fittingperiod.
In the equationwiththelaggeddependentvariablefittedthrough1974:1,
the price-controldummyprobablyforces the coefficienton the lagged
dependentvariableto an artificiallyhigh level, given the hypothesissuggestedabovethat the initialeffectof pricecontrolswas to squeezemargins
In turnthis blowsup the steady-statecoefficienton the ratioof
gradually.7
AULCto SULCto an excessivelyhigh number.(Sincethe (y/yp)* term is
a smoothlong-laggedvariable,its raw coefficientvalue adjuststo the rise
in the coefficienton the laggeddependentvariable,so its steady-statevalue
is not so much distorted.)
Thecoefficienton long-termcapacityutilizationimpliesthat a 1 percent
reductionin long-termexpectationsabout the utilizationrate resultsin a
0.35to 0.40 percentpriceincrease.This resultis slightlyabovewhatwould
be predictedby a target-rate-of-return
costs,
hypothesis.Capital-recovery
in normaltimes,are27 percentof pricein the corporatesector;a 1 percent
"permanent"
drop in capacityutilizationwould thereforerequirea 0.27
percentchangein priceto maintainthe rate of returnon investedcapital
unchanged,comparedwith the 0.35 to 0.40 percentchangeimpliedby the
equations.
6. See Gordon, "Responseof Wages and Prices."
7. If the dummyis constant, its gradualimpact on marginscan be reflectedonly by
an increasein the value of the coefficienton P/SULC_i.
ChlarlesL. Schultze
463
Table 5. Results of AlternativeEstimates of Equation (2)
AlternativeAb
Equationwithlagged
dependentvariable
Variablea
and
summary
statistic
Fitted
through
1974:1
Fitted
through
1971:2
AlternativeBb
Equationwithoutlagged
dependentvariable
Fitted
through
1974:1
Fitted
through
1971:2
Coefficientand t-statistic
Variable
Constant
0.298
AULC/SULC
0.224
[y/yp - (y/yp)*] 0.046
(y/yp)*
-0.127
t
-0.0003
...
T2Dt
(t)
(t - 1)
(t-2)
(t- 3)
(P/SULC)_1
Rho
-0.007
...
...
...
1.6
3.1
1.0
2.0
2.9
2.2
0.606 6.5
...
0.712 3.3
0.242 3.3
0.061 1.4
-0.264 3.6
-0.0004 4.3
...-0.023
...
...
...
...
0.309 2.5
...
1.044
0.345
0.033
-0.390
-0.0006
7.2
3.8
0.5
4.3
5.2
4.9
-0.004
-0.005
-0.006
-0.008
1.3
3.3
4.3
2.6
...
0.642
1.050
0.347
0.073
-0.399
-0.0006
...
8.0
3.8
1.3
5.6
6.0
...
...
...
...
...
0.507
Steady-statevalueof coefficient
AULC/SULC
[y/yp-(y/yp)*]
(y/yp)*
t
0.569
0.117
-0.322
-0.0008
D
-0.0178
Summarystatistic
R2
Durbin-Watson
Standarderror
0.975
2.036
0.004
0.350
0.088
-0.382
-0.0006
...
0.345
0.033
-0.390
-0.0006
-0.023
Valueof summnary
measuire
0.964
0.891c
1.742
1.950
0.004
0.004
0.347
0.073
-0.399
-0.0006
...
0.942c
1.877
0.003
a. AULC = actual unit labor costs; SULC = standard unit labor costs; y/yp = ratio of actual to
potential GNP, and * signifies five-year moving average; t = time; D = price-control dummy; P = price
deflator.
b. The fitting period for all four versions of the equation begins in 1958:4.
c. Dynamic k2.
The demandterm is small and in no case significant,althoughit does
carrythe "right"sign. Quitepossibly,however,a gradualdemandeffectis
beingroughlyoffsetby a growingeffecton priceof the A ULC/SULC term.
Sincemovementsin A ULC/SULC are dominatedby cyclicalproductivity
changes,which in turn are negativelycorrelatedwith demand,the two
effectstend to cancelout.8
8. But see the discussionof productivityin the appendix.
Brookings Papers on Economic Activity, 2:1975
464
Table6. ActualandPredictedValuesof the Ratioof Pricesto Standard
Unit LaborCosts,PrivateNonfarmSector,Quarterly,
1974:2to 1975:2
Yearantd
quarter
Actual
(1)
Predicted
(2)
Error
(1) - (2)
1974:2
3
4
1975:1
2
96.1
97.3
98.6
99.3
99.0
96.4
97.0
98.0
99.0
99.3
-0.3
0.3
0.6
0.3
-0.3
Sources: Actual values, table 1; predicted values, alternative B of text equation (2) fitted through 1974: 1.
Table6 comparesthe actualvaluesof P/SULC during1974:2to 1975:2
with values predictedfrom the alternativeB equation, fitted through
1974:1. The equationpredictsthe upsurgein marginsvery well. And althoughthe removalof pricecontrolscontributedabout 40 percentof the
rise (whichany equationwith a laggeddummytermmighthave approximated),changesin severalof the othervariablesare quiteimportant.
Table7 showsthe factorsthat contributedto the rise in P/SULC over
the past two years-from its low point in 1973:2to 1975:2.9
The findingsmightnow be summarizedas follows:
1. Long-termchangesin expectedcapacityutilizationaffect prices by
slightlymore than wouldbe expectedif pricingbased on a targetrate of
returnwerecommon.
2. About 60 percentof cost changesstemmingfromcyclicalchangesin
productivityare passedon in prices.
3. Changesin the degreeof excesscapacityhave littleeffecton margins
(whateverthey may do to wages).The close relationshipbetweencyclical
changesin actual unit labor costs and in excess capacity,however,may
obscuretwo distinctbut offsettingeffects:laggedcyclicalchangesin actual
unitlaborcoststendingto raisepricesand laggeddemandchangestending
to lowerthem.
4. Price controlsappearto have graduallyreducedpricesby 2 to 21/2
percent.
About40 percentof the sharprisein pricesrelativeto standardunitlabor
costs overthe past two yearsappearsto have been due to the removalof
9. Actually,there is a sudden one-quarterdip in P/SULC in 1973:1 not picked up
in the equation(see figure 1). This dip was ignoredin choosing the low point.
CharlesL. Schultze
465
Table 7. Factors Contributingto the Rise in the Ratio of Prices
to StandardUnit Labor Costs, Private Nonfarm Sector,
1973:2 to 1975:2
Percentagepoints
Componenit
of rise
Total rise in P/S ULC
Deviation in actual unit labor costs
Removal of pricecontrols
Reduction in long-termcapacity utilization
Time trend
Falling demand
Residualand rho correction
Rise
5.21
2.48
2.30
1.08
-0.48
-0.36
0.19
Source: Estimated from alternative B of text equation (2), fitted through 1974:1.
controls.Almost all of the remainder,paradoxically,stems from the net
impactof sharplydeclininglevels of economicactivity,throughthe effect
of cyclicaldeclinesin productivityand a fall in expectationsabout longtermcapacityutilization.Howeversuccessfulit maybe in moderatingwage
inflation,depressionof aggregatedemand-especiallyif continuedfor long
in its effectson price-wagemarperiods-appearsto be counterproductive
effectsof reducedaggregatedemandarenot
gins.The largemargin-raising
apparentto the naked eye in prior recessionsprincipallybecausethose
recessionswereneitheras sharpnor as long as the currentone.
Could otherpotentialculpritsbe responsiblefor the recentrise in price
margins-the competitiveeffects of rising import prices,the fear of the
reimpositionof controls,and escalatingexpectationsof long-terminflation? Have one or more of the independentvariablesused in equation(2)
recentlytaken on values out of line with past experience,therebyserving
as surrogatesfor a dummyvariablethat picksup one of the factorscited
above?The fact that the versionof the equationfittedthrough1971:2 has
virtuallythe same coefficientsas the versionfittedthrough1974:1 means
that the structurecapturedin that equationcould fully predictthe recent
rise in margins (after adjustment for the removal of price controls). Hence
the explanationrelyingon a surrogatedummyvariablewillnot hold.To the
extentthat some factorotherthanthose includedin equation(2) droveup
price margins,some offsettingchangein the earlierstructuremust have
occurredsimultaneously.On balanceit seemsmore probablethat the extent and depthof the recentdepressionin aggregatedemand,operatingon
an unchangedstructure,have been responsiblefor the marginincrease..
466
BrookingsPaperson EconomicActivity,2:1975
Implications
for the Future
Whatdoes the foregoingimplyfor the movementof the ratio of priceto
standardunit labor costs if a rapidrecoverytakes place duringthe next
eighteenmonths?Using the alternativeB equationfittedthrough1974:1,
I have made a projectionbased on severalsimple assumptions:(1) real
GNP risesat a steady8 percentannualratestartingin the thirdquarterof
1975;(2) compensationper manhourin the privatenonfarmsector also
rises at an 8 percentannualrate; and (3) productivitybehavesin accordancewiththe equationdiscussedearlier(in the secondquarterof 1975the
residualfromthat equationwas virtuallyzero). Given these assumptions,
the ratio of pricesto standardunit laborcosts remainsvirtuallyconstant,
in the neighborhoodof 99.0 throughthe end of 1976.The favorableeffect
of the declinein actualrelativeto standardunitlaborcostsis fullyoffsetby
the continuedfall in the long-termmovingaverageof the capacity-utilizaterm.
tion variableand the rise in the currentcapacity-utilization
Severaladditionalpoints should be noted. First, the trend rate of increasein productivityincorporatedin the SULCvariableis only 2 percent
Hence,an 8 percentrisein
per year,somewhatless thanusuallyassumed.10
compensationpermanhourtranslatesinto a 6 percentrisein standardunit
labor costs. Second,the cyclicalproductivityadjustmenttends systematicallyto understatethe fallin productivityduringthe earlydownturnandto
overstatethe riseduringthe earlyupturn.The dropin A ULC/SULCin the
projectionmay, therefore,be understatedand P/SULC slightlyoverestimatedduringthe next four quarters.
The centralpoint of the projectionsis that the nonfarmprice deflator
(excludingdomesticenergyrents)is likelyto be determinedmainlyby wage
changesduringthe early phases of the recovery.While this is hardly a
10. The growth rate of productivitybetween 1955:4 and 1965:4, in both of which
y/yp was the same, was 2.61 percent;the equation shows a 2.55 percenttrend for this
period. Productivitygrowth between 1965:2 and 1973:1 (also periods with the same
y/yp) was 2.15 percent; the equation has a 1.95 percent trend for this period. The
differencebetween the peak-to-peakgrowth in productivityis almost exactly the same
as the differencebetween the two trendsin the equation.An equation fitted with only
one trend yields a 2.2 percenttrend for the entire period, and a slightly largercyclical
adjustmentterm (0.47 instead of 0.42). An equation using two time trends, and, as a
cyclical variable,lagged changes in output relative to trend, also yields a 2.0 percent
productivitytrend for the recent period(see the appendix).
467
CharlesL. Schultze
startlingconclusion,it impliesa sharpdeparturefrom the pricebehavior
of the past two years.
One of the majorfeaturesof the 1973-74downturnwas the erosionin
realwages,whichled to a sharpfallin realdisposablepersonalincome.The
role in this phenomenonof rising prices for oil and other importshas
long beenrecognized.But the rise in the ratio of pricesto wageshas been
muchlargerthan can be accountedfor by these developments.To the extent that the steady-statecoefficientson A ULC/SULCand on long-term
expectationsabout utilizationderivedearlierapproximatestructuralreality, sharpandprolongedrecessionstendto raisepricessignificantlyrelative
to standardunitlaborcosts.Whateveran extendeddepressionof aggregate
demanddoes to wage rates,it is not an effectiveway of squeezingpricewagemargins.
APPENDIX
Derivationof TrendProductivityIncrease
to CalculateStandardUnit Labor Costs
FOR BOTHsectors,privatenonfarmand privatenonfinancialcorporate,
output per employee hour (Q/H) was fitted to a time trend and a cyclical
adjustmentterm(the ratio of actualto potentialGNP, y/yp). Productivity
data are indexedto 1967 = 100.
The regressionfor privatenonfarmproductivityis
(A-1)
In (Q) = 4.281 D1+ 4.339 D2+ 0.00639 D1(t)
\Ht
(1212)
(332)
(58)
+ 0.00487 D2(t) + 0.419 ln(2)
(24)
(10)
R2
= 0.994; Durbin-Watsonstatistic = 0.339.
where
D, = 1, 1954: 1-1965:4; 0 otherwise
D2= 1, 1966:1-1975:2; 0 otherwise.
BrookingsPaperson EconomicActivity,2:1975
468
Theregressionfor productivityin the privatenonfinancialcorporatesector
is
4.300 + 0.00725(t) + 0.503 In(2).
(12)
(1222) (100)
In (
(A-2)
R1 = 0.994; Durbin-Watsonstatistic = 0.367.
The numbersin parenthesesare t-statistics.
Standardunitlaborcostsareestimatedby settingy/yp at a constant0.97,
calculatinga standardproductivity,and dividingthat into the actualcompensationper hour. Sincetwo time trends(with differentintercepts)were
usedin the privatenonfarmsector,a discontinuityin the measurementof
standardproductivitywas avoidedby locatingthe intersectionof the two
productivitytrends(1963:2) and switchingfrom one trendto the otherat
that point.
An alternativeproductivityequationwas tried for the privatenonfarm
sector,in whichquarterlychangesin productivitywerefittedagainsttwo
time trendsand a laggedseriesof quarterlychangesin actualoutputrelative to potential:
(A-3)
A ln
(H) =
+
aoD, + a1D2+ a2AIn (2L)
a3iA In
+ ut,
where
D, = 1, 1956-65; 0 otherwise
D2
=
1, 1966-75;0 otherwise.
The resultsof (A-3) are given in table A-1; the coefficientsaO and a, are
expressedas annualratesof change,in percent,andthe weightingstructure
for the sum of the a3sis from a second-orderAlmon lag.
For purposesof this paperthe importantpointis the confirmationof the
2 percentvalue for the recenttime trend.The resultsare also interesting
for whatthey say about cyclicalproductivitychange.For each 1 percent
dropin outputbelow trend,productivityat firstdrops0.6 percent.If outof the proput thenresumesa movementparallelto trend,three-quarters
ductivityshortfallwill be madeup aftersix quarters,in whatis virtuallya
declininggeometricpattern.(Sincea second-orderAlmonlag wasimposed,
a geometricdeclineis not forceduponthe weightingstructure.)A freelyfit-
Charles L. Schultze
469
ted lag structuregave virtually the same coefficients for the first quarterand
the sum of the remaining five quarters (0.62; -0.46). An eight-quarter
Almon lag, for quarters after the current one, also gave similar results
(0.59; -0.46), but the negative coefficients deteriorated rapidly in significance after the fourth quarter, and the sum of the negative weights was no
greater than that in the five-quarter structure.
Table A-1. Results of Estimates of Equation (A-3)
Coefficient and
summary statistic
Value
t-statistic
ao
2.59
8.0
a,
2.02
0.613
-0.452
-0.165
-0.117
-0.080
-0.053
-0.037
5.7
10.5
4.6
2.7
4.2
2.1
1.9
0.7
Coefficient
a2
Maus
(a3, -1)
(a3, -2)
(a3, -3)
(a3, -4)
(a3, -5)
Summarystatistic
P2
Durbin-Watson
0.622
1.875
..
...
Discussion
Most of the discussion focused on the interpretation of the recent behavior of the ratio of price to standard unit labor cost (P/SULC). R. A.
Gordon introduced the possibility that the observed widening of the pricecost margin since early 1973 could reflect a shift in thinking on the part of
firms from historical to anticipated standard unit labor costs. In an inflationary period, the ratio of price to historical cost would rise even though
the ratio of price to anticipated cost was constant. William Nordhaus suspected that the recent rise in P/SULC might be overstated by Schultze
470
BrookingsPaperson EconomicActivity,2:1975
as the resultof an insufficientadjustmentfor increasesin energyprices.
Schultze'sadjustmentseemeddistinctlysmallerthan similarestimatesby
John Shoven and Nordhaus, which had indicateda marked-uprise of
roughly$2 in finalenergy-product
pricesfor each $1 increasein the cost of
crudeoil. Any understatement
of the energyeffectwouldoverstatethe rise
of P/SULC to be explainedby othercauses.Nordhausalso notedthat the
turnaroundin the margincoincidedapproximatelywith the 1973 Mideast war.
Moregenerally,Nordhauswas uncomfortablewiththe implicitassumption of the paperthatpricingbehavioris basedsolelyon laborcosts.What
appearsas an enlargedmarkupof value-addedpricesoverlaborcostssince
1973in realitymight be a constantmarkupon both labor and materials
costs, followinga majorrelativeincreasein materialscosts. Robert Hall
elaboratedon that criticism.He notedthat the pricemeasureof the paper,
the privatenonfarmdeflator,was a value-addedpriceindex whichnecessarilyabstractedfromtwo majornonwagesourcesof the recentinflation,
farmand importprices.He wouldhavepreferreda measureof the priceof
finalgoods ratherthan of value added.
Schultzerepliedthat his analysiswas basedimplicitlyon a model with
a long-runtargetrateof return,for which the adjusteddeflatorwas the
appropriateprice measure.A more inclusiveprice measurewould raise
aggregationproblemsin interpretingpricingbehavioras aimingat a target
rateof returnbasedon a markupoverall costs. He thoughtit unlikelythat
pricingrulesat the initialstage of processingwouldmarkup veryvolatile
pricesof raw materials;at more advancedstages of processing,the rawmaterialscomponentof costs mightbe incorporatedin a pricingrule,but
wouldbe less volatile.A value-addeddeflatorfinessessome of theseproblems. GeorgePerryviewedthe aggregationproblemas less importantfor
the broad privatenonfarmsector, which acquiresmost of its purchased
inputs from industrieswithin the sector, than it would be for a smaller
sectoror a particularindustry.
WilliamFeilnersuggestedthat, if, in fact, firmswere passing on their
tax burdenfrom FIFO inventoryaccounting,that would show up as a
wideningof marginsthat the equationmightwronglyattributeto the gap
betweenstandardand actualproductivity.Robert Solow called attention
to Schultze'stwo-partexplanationof the drop in the P/SULC margin
between1960 and 1971, whichinvokeda negative-timetrend and an increasein the expectedlong-runrate of capacityutilization.Solow sus-
CharlesL. Schultze
471
effectmightin fact reflectthe influence
pectedthat the largenegative-trend
of some omittedstructuralvariable.
ChristopherSims expresseddoubts that Schultze'sequationcould be
confidentlyinterpretedas reflectingpricing behavior. It could instead
describewage behavior,since the estimatedequationapproximatesa relationshipinvolvingthe reciprocalsof the realwage,of outputper worker,
and of output.It could be sayingsomethingabouttheproductionfunction,
sincetransposingoutputper workerto the left-handside of the equation
and the real wage to the right yields an equationthat could be used to
estimatethe elasticityof substitution.Finally,it couldbe sayingsomething
about the stability of relative shares-that the ratio of total wages to
nominalincometends to be stablein the long run. All in all, the a priori
informationseemed inadequateto Sims to identify the equation as an
explanationof markupbehaviorratherthan as a descriptionof the behaviorof otherelementsthat enterinto the determinationof the markup.