The Econometric Society

Location of Industry and Regional Patterns of Business-Cycle Behavior
Author(s): Rutledge Vining

Source: Econometrica, Vol. 14, No. 1 (Jan., 1946), pp. 37-68
Published by: The Econometric Society
Stable URL: http://www.jstor.org/stable/1905703 .
Accessed: 03/11/2014 12:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
The Econometric Society is collaborating with JSTOR to digitize, preserve and extend access to Econometrica.
http://www.jstor.org
This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

All use subject to JSTOR Terms and Conditions
LOCATION OF INDUSTRY AND REGIONAL PATTERNS

BEHAVIOR
BUSINESS-CYCLE
By
RUTLEDGE
OF
VINING
In a recentpaper' we presentedthe firststage of a statisticalinquiry

into the natureof the abstractioninvolvedin the use of national series
for the descriptionand analysis of business cycles. The results were
given of an analysis of one aspect of the cyclical behavior of national
income-its percentagerate of change. This rate of change of the national total was interpretedas a parameterof the frequencydistribution of the respectiverates of change of the componentparts of the
largerarea, and statisticsdescribinga certain pattern of behavior of
these distributionswere shown. The distributionsappear to have a
characteristicshape approximatinglogarithmicnormality,and there
phases of
is a suggestionof a systematicdevelopmentas the different
the businesscycle unfold.The skews of the distributionsappear to be
positiveduringthe period of the expansionwhen the rate of growthis
increasing.When the rate ofgrowthbegins to decline,it was tentatively
suggestedthat the skew shiftsto a negative and remains a negative
throughthe absolute turningpoint of national income and duringthe
period when the rate of contractionis increasing.When this rate of
contractionbegins to decline,the skew of the distributionagain shifts
to the positive.An attemptwas made to rationalizethis cyclicalevolution of shape of these distributions.It was furthernoted that the extreme movementsfromyear to year are found generallyamong the
same set ofstates,therebeinga tendencyevidentforthe moresparsely
settledstates that are highlyspecialized in raw-materialproductionto
clusterin the ends of the distributionand, in years of markedchange,
in the same end of the distribution.
It is proposedin the presentpaper to discuss in moredetail the geographicalmake-up of these annual frequencydistributionsof regional
rates ofbusinesschange. We thinkthat certaingeneralizationsmay be
made regardinga regionalpatternofshort-runbusinesschange,and we
should like to analyze possible factorsthat mightaccount for the obOur previousdiscussionof the facservablesimilaritiesand differences.
tors determiningthe cyclical responsivenessof a given regionlaid emphasis upon industriallocation or specialization and the institutional
and physicalfactorsinfluencing
the region'scommercialties. The question we raise now has to do with certain attributes of industrial
location that we regardas particularlyrelevantfor regionalbusinesscycle analysis.
I "Regional Variationin CyclicalFluctuationViewedas a FrequencyDistribution," ECONOMETRICA,Vol. 13, July,1945, pp. 183-213.
37

38
RUTLEDGE
1. REGIONAL
ECONOMIES
VINING
IN INTRANATIONAL
BUSINESS-CYCLE
ANALYSIS
In the presentpaper, as in the paper cited, we shall workwith state

data, but it may be indicated here that states are not what we shall
regardin later workas "regionaleconomies." A state generallywill be
foundto includepartsofseveral "regions"such as we shouldlike to consider as units for regionalbusiness-cycleanalysis. It is the "natural"
trade area familiarto marketingspecialiststhat would seem to be the
more practicable representationof our unit. It is not surprisingthat
buying habits and commercialcontacts develop that tie the business
units of an area to a particulartradingcenter. There must be an agglomerationtendencyhere just as thereis in manufacturingindustry.
In the latter field,repair and machine-shopfacilities,technological
facilitiesthat develop to serve
skills,communicationand transportation
one firmor industryin a given localityare available forotherfirmsor
industries.This is part ofthe workingofthe principleofexternaleconomies. That is, many such facilitiesare "lumpy" and discontinuousin
operationis largerthan
size, and the unit of optimumsize forefficient
requiredto meetthe demandsofone firmor industry.It becomeseasier
and less costlyto establisha firmor industryin an area afterotherfirms
and industrieshave been established.Thus it must be withcommercial
and marketingfacilities.Banking, brokerage,wholesaling,and other
business services agglomeratein a tradingcenter,throughwhich the
"export" commoditiesofan area are passed to the "outside" worldand
throughwhich the "imported" commodities-the "exports" of other
regions-are drawn fordistributionwithinthe area. This is ilot to say
that no interregionaltrade goes on except throughthe tradingcenters,
and we do not object to our boundariesbeing zones ratherthan lines.
We are saying merelythat the great bulk of the products of certain
types of industryof a particulararea are channeledthrougha populationconcentrationthat is regardedby the inhabitantsofthe area as the
tradingand financialcenter.Of particularimportanceare the banking
connectionswithinthese units. The ebb and flowof a region'sliquid
fundsis reflectedin the reserveexperienceof particularbanks, whose
spatial sphereof operationsis fairlydefinite.
Now ifit weretruethat all economicenterprisesofone ofthese units
could be neatlyclassed intothoseemploymentswhoseproductsare sold
and consumedlocally and those whose productsare exportedto other
in cyclical reareas, a relativelysimple analysis of regionaldifferences
sponsivenesswould be suggested.If it were found that each region's
''export' employmentwas confinedto the productionof one product
whose consumptionmarketencompassesthe entiregreaterarea made
up of the interlocking
regions,the analysis would be yet simpler.From

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
39
one pointofview,the home-marketemploymentcould be thoughtofas

employmentthat is auxiliaryto and a part of the exportindustry.Just
as independentmachine shops in areas specializingin lumberproduction could be consideredas a part of the lumberindustry,so mightwe
considerthe bakeryestablishmentsand retailmarketingservicesto be
auxiliary employmentto the lumber industry.The existence of the
lumberindustryin that area would implythe existenceof these other
enterprises,but the converseof this statementwould obviouslynot be
true.From thispointofview,we could regardthiswholeclass ofhomemarketindustriesor employmentsas unique in that a home-market
industrymay be physicallybut not economicallyquite the same from
regionto region,being closely tied in each regionto the "export" industry. Thus, while an export industryor employmentmight show
regionaluniformity
in its behavior,being affectedsimilarlyregardless
of location, the cyclical behavior of home-marketemploymentwould
show more or less wide regionaldifferences,
such industriesin each region assuming certain of the cyclical characteristics-in more or less
diluted form-of the exportindustryof that region.The featurethat
in business-cyclebehavior would
would make for regionaldifferences
consistlargely of the varyingincome elasticitiesof the different"export" products.The demand forthe "export" productof some regions
would be little affectedby the business cycle so that the purchasing
powerwould be sustainedwithlittledeclineduringa national business
depression. The demand for the "export" products of other regions
would be very drasticallyaffectedby the business cycle so that purchasing power in such regionswould fall greatlyin national business
depressionsand rise greatlyin national business expansions. The regionsthat would be affectedmost would be those whose "imports"are
"necessities"that would declinelittlein use in a depressionand whose
"exports" are durable, storablecommoditieswhose demand would fall
to a low level in a depression.Under theseconditions,moneypayments
to the rest of the nation would not be easily adjusted to the falling
moneyreceiptsfromthe rest of the nation, local banks would rapidly
lose reservesto otherregions,and local purchasingpower would continue to decline until ultimatelypurchases are choked offto equal
money receipts.If the reverseof these conditionswere satisfiedin a
given region,that regionwould show relativelysmall declines in national depressionsand would tend to gain banking reserves during
those times.
The above is a sketch of a conceptual frameworkthat has given a
semblance of directionto our statisticalinquiries. It is obviouslynot
true that a particulararea specializesin but one "export" product,but
by and large it will be foundthat the specializationis frequentlyquite

40
RUTLEDGE
VINING
marked.It is also not true in generalthat "export" productshave nation-widemarkets,but it is the case in manyinstances,and at least the
structureof these regionalinterlocksmay be investigated.Finally, as
was indicated,a state is not what we should regardas an "economic
region" and will generallyinclude parts or the whole of several integratedtrade areas; and the behavior in the businesscycle of state figures will be a sort of average of the behaviorof the figuresof its componentregions.States comprisedofeconomicallysimilarregionsshould
show a similarityof behavior in the business cycle. Statisticsdescriptive of industriallocation by states should indicate how a particular
state mightbe expectedto respondin a bu.sinesscycle and should help
explainobservablegeographicalpatternsofbusiness-cycleresponse.
2. CERTAIN
RELEVANT
FEATURES
OF INDUSTRIAL
LOCATION
In our study of industriallocation we have made an adaptation of

certaintechniquesof ProfessorSargant Florence. On the basis of occupations statisticsforGreat Britainand the United States, Florencehas
set apart a class of industrythat he designates as "residentiary"or
"ubiquitous" industry2-residentiary
in the sense that they typically
reside where the consumingpopulation resides. With referenceto a
given region,the industriesotherthan the residentiaryones are called
the "primary"industries.Industriesare classifiedinto one or the other
of thesetypeson the basis ofa measuredevised by Florenceand called
the coefficient
of localization.' For each regiona "location factor" is
computedforeach industryby obtainingthe ratio of the percentageof
total employmentin the givenregionfoundin the givenindustryto the
same percentagefor the nation as a whole. If the industrywere perfectlyuniformly
distributedamong all the regionsall the regionallocation factorswould be unity. If the industrywere highlylocalized in a
very few regions,these few regionswould have high location factors
forthat industry,and the rest of the regionswould show location factors that are small or zero. The coefficient
of localizationfor a given
industryis obtained by computingthe weightedavergage deviation
fromunityofthe locationfactorsforall the regions,the weightforeach
regionbeingthe proportionoftotal national employmentfoundin that
2 A. J. Wensleyand P. Sargant Florence,"Recent IndustrialConcentration,
Especiallyin the Midlands," ReviewofEconomicStudies,Vol. 7, June,1940,pp.
139-158; P. SargantFlorence,"Memorandumon Long Range Planning,"MimeographedRelease of the National ResourcesPlanningBoard, 1940, pp. 14-30;
P. Sargant Florence,"Geographical Distributionof Economic Activityamong
Broad Industry Groups," MimeographedRelease of the National Resources
PlanningBoard, 1940,pp. 1-24. By "industry"is meantany specifictypeofeconomicproductionso that the termincludesagricultureand trade.
3Florence, "Memorandumon Long Range Planning,"pp. 19-25.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
41
BEHAVIOR
region.This measure divided by two varies between zero and unity.4

The coefficientof localization measures complete uniformityof geographicaldistributionas zero As greaterand greatergeographicalconcentrationis encountered,the measure approaches unity. The coefficient for automobile repair shops in 1930 as computed by Professor
Florence using state data, forexample, was 0.068. The coefficient
for
automobilemanufacturewas 0.597. Automobilerepairshops are relativelyuniformly
distributedover the nation. Automobilemanufacture
is highlylocalized.
I Identicalresultsare obtainedby "adding the plus differences
(or, since they
of the local percentageof workersin the
total the same, the minusdifferences)
givenindustryfromthe local percentageofworkersin all industry."Ibid., p. 19.
These measuresmay be presentedsymbolicallyas follows:
Let E=Total gainfullyemployedin Nation,
Ea= Total employedin Industrya in the Nation,
ei=Total employedin Region 1,
ela=Total employedin Region 1 in Industrya.
The locationfactorof Industrya in Region 1 is definedas
eia
el
E.
E
of localizationis computedas follows:
The coefficient
(3)
WeightedDeviation
fromUnity
(2)
Location
Factor
(1)
ela
ela
**
..............
Ea
E-
ei
**
.
.
...............................................
.
ena
en
of LocalizationCoefficient
El
E
. . . . .
01
el
el
Ea
1 -
ei
ela
Ea
ee
en
en
E-
E
ea
ena
E
nI
e;
eia
E
Ea
The algebraicsum is obviouslyzero,indicatingthe equality of the positiveand
negativedeviations.The absolute sum, disregardingsigns,varies fromzero to
two. If in each regionthe two percentagesare identical-i.e., ifthe industrywere
distributed-thenthe sum of the deviationsis zero. If all of
perfectlyuniformly
the givenindustryis located in a singleregion,and ifregionsand industriesare so
definedas to be ofnegligiblesize relativeto the total economy,thenthe absolute
sum of the deviationswillapproachtwo as a limit.

42
RUTLEDGE
VINING
The residentiaryindustriesare those with "low" coefficients

of localization-that is, those that are uniformlydistributedaccordingto
populationand that depend upon local marketsforthe disposal oftheir
products.The residentiaryservice industriesinclude gas, water, and
electricutilities,transport,trade,and finance,professionalservices,entertainment,
and personalservices.In addition,"many manufacturing
industriesshow extremelylow coefficients
and resemblein thisthe residentiaryserviceindustries."5The workof Florence evidencesthe relative stability,as between regions,of the proportionof total employmentfoundin these residentiaryindustries.Obviously,the magnitude
of this proportiondepends in some measureupon the principleaccording to whichregionsare delineated,forthe largerthe area representing
a "region" the broader will be the categoryreferring
to residentiary
inanufacturingindustries.But fora definedset of areas, the location
factorsofcertainemploymentswillshowa strongconcentrationaround
unity,and the percentageof total employmentfoundin thislist of employmentswill vary fromregionto regionwithina relativelynarrow
range.
It is the characterof the restof the employmentthat would seem to
differentiate
regionsor states with respectto theircyclical responsiveness. The residentiaryindustrieswe may call the "passive" industries,
forthese industriestypicallyplay a passive role in the short-runprocess of businesschange in a given region.Local income constitutesthe
marketfortheirproducts,and they move along withthe generalbusiness activity of the region. The primaryindustrieswe may call the
"carrier" industries-by analogy fromthe fields of mechanics or of
chemistryor of medicine.These industriesare those that typicallydepend fortheirmarketsupon the income of other regionsand are the
media by whicheconomicchange is transmittedfromregionto region.
Economic change capable of affectingnational income initially impingesupon certainindustrieslocated in certain regions.The changes
that are broughtabout in these industriesare propagated to otherregions and diffusedover the entire national area (or over the entire
economicallyrelatedarea) by way of the effectof these initialchanges
upon the expenditureson productsof the primaryindustriesof other
regions.The primaryindustriesare the connectinglinks; the residentiaryindustriesare passive in this process of interregionaldiffusionof
economicchange.
Using quite simple methodsof analyzingthe occupational statistics
of the states of this nation, we may presentconcretelysome idea of
5 "Instancesare building,baking,repairwork,job-printing,
and it is clearthat
such productionhas to be located wherethe populationis, because of heavy
transportcost in relationto value of commodity,because of its perishability,
because ofneed forclose communication
withthe customer."Ibid., p. 20.

REGIONAL PATTERNS OF BUSINESS-CYCLE
BEHAVIOR
43
thesefeaturesofindustrial
location.For thispurpose,we arrayedfor
each industrial
classification
thestatefigures
forthepercentage
oftotal
employment
foundinthatindustry.6
For eachofthesearraystherange
limitsofeach quintileintervalwerelocated,andtheserangelimitsand
the medianfigurewerethendividedby theappropriate
nationalpercentage.The resultsofthisprocedure
are showninTable 1. This table
providescertainpointsonthearrayofstatelocationfactors
foreachindustrialclassification,
and theselatterarearrayedfromtopto bottom,
in theorderofwhatappearsto be thedegreeoflocalization.The table
showsapproximately
that whichFlorence'scoefficient
of localization
showsin a muchmorecondensedform,and forits awkwardbulkiness
it maypossiblyoffer
compensation
in theformofan enhancedvisual
impression
ofthecharacterofindustrial
localization.
For ourpurposes,thesedata are fraught
withshortcomings,
and it
maybe instructive
to referto the implications
of a fewof thesedeIf the subareasunderstudywereapproximately
of equal
ficiencies.
"economicsize" and smallrelativeto thetotalarea,ifthestandardsof
livingand consumer
tasteswereapproximately
uniform
as betweenregions,if industrial
classifications
werefineenoughdrawnin orderto
provideinternaluniformity,
thenwe mightpossiblyrationalizean expectationfora certaingeneralpatternto be exhibitedby thecolumns
in thistable,theindustries
beingarrayedin orderofdegreeoflocalization. The marketareas of different
productswould obviouslyshow
widedifferences.
Supposeweshouldmeasurethetypicalmarketarea of
a productby someaveragedistancebetweenthe pointat whichthe
valueis addedand thepointoffinalconsumption.
A frequency
districould
butionofthesevariates,typicaldistanceforeachdefined
product,
be constructed,
butit wouldbe difficult
to anticipatethegeneralshape
ofthisdistribution.
If it wereU-shapedit wouldmeanthata relatively
largeproportion
ofproducts
are soldat a pointcloseto theirrespective
at intermediate
dispointsofproduction,
a relatively
smallproportion
sold
tancesfromthesepoints,and anotherrelatively
largeproportion
in all cornersofthenationalarea. If it werea single-humped
distribution,it wouldmeanthatthereis a concentration
pointformarketarea
distancesand that extremedistanceson eitherside of the concentraton becomeprogressively
less frequent.Those products,or valueadditions,whose marketareas encompassthe entirenation must
adofcoursebe thehighlylocalizedindustries-some
sharpproduction
vantagebeingfoundat certainpointsorcenterswithinthelargerarea.
Aroundand amongthose primaryclusterswill have been built up
industries
withtradeareasofintermediate
length,
and witheachpopu6 The data analyzed werefromthe Sixteenth
CensusoftheUnitedStates: 1940.
Occupation,Income, by States,
The Labor Force-Employment,Unemployment,
Vol. IV. Washington,GovernmentPrintingOffice,1941.

44
RUTLEDGE
VINING
13.12.11.10.9. 8. 7. 6. 5. 4. 3. 2. 1.
Iron
Coal
Autos
Paper
Crude
Textile
Logging,
and Leather
Oil
Chemicals
and Machinery
Petroleum
and
Non-ferrous
and Mining
and
Steel
and
and
RATIOS
Mfg Transportation
Auto
Products
Allied
OF
Sawmils,
and
Metals
Coal
Leather
and
Allied
Natural
and
and
Their
Mfg
Products
Equipment
.
Gas EquipmentIndustrial
Products
Apparel
Products
.
PERCENTAGE
Planing
Products
MfgProducts
........
(Except
Products
Mills
.......
.
......
Classification
......
EMPLOYED
Autos)
IN
.....
STATE
TO
...........
..........................
...................
..........................................
..................
.............................
...............
.....................
...............................
........................
........................................
.......................
......................00
0.00
0.00
0.00
0.04
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
LessNo
Than
States
0.00
0.08
0.00
0.00
0.04
0.00
0.04
0.00
0.00
0.00
0.00
0.00
0.00
LessThree
Than
States
of
0.31
0.20
0.25
0.07
0.08
0.07
Less the
0.14
0.00
0.00
0.00
0.00
0.00
0.00 Than
States 20%
of
0.40
0.21
0.62
0.21
0.50
0.23
0.43
0.33
Less the
0.13
0.00
0.14
0.00
0.08 Than
States40%
of
0.21
0.62
0.50
0.50
0.28
0.25
0.57
0.50
0.29
0.13
0.00
Less the
0.00
0.08 Than
States50%
TABLE
1
PERCENTAGE
EMPLOYED
IN
NATION
BY
of
0.51
0.80
0.71
1.39
0.50
0.50
1.00
0.83
0.43
Less the
0.25
0.33
0.38
0.15 Than
States60%
of80
1.20
2.39
1.70
1.29
1.25
2.25
1.29
1.50
1.14
1.13
1.25
2.50
Less the
0.36 Than
States%
All
2.20
2.71
2.42
2.25
But
2.92
2.08
3.25
3.00
3.00
3.25
5.00
4.00
Less Three
0.62 Than
States
INDUSTRIES
IN
1940
All
9.00
8.69
One
But
5.28
3.04
3.14
4.25
7.14
10.33
Less
3.57
15.38
11.25
17.00
15.00Than
State

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
32.31.30.29.28.27.26.25.24.23.22.21.20.19.18.17.16.15.14.
Food
Other
Stone,
Retail Other
Finance,
BusinessUtilities and Domestic
Printing
Furniture
Wholesale
Agriculture
Clay
Professional
Amusement,
Government
and Mines
Trade
Construction
... and
Serv.
Personal
Transportation
and
and
Communications and
Trade
Kindred
Personal
and
Insurance,
Misc.
.
Glass
Service
Publishing
and
Recreation,
Related
Quarries
Repair
Service
Products
etc
and Wooden
Real
Products
...............
Serv.
................
Mfg
Services
(Laundries,
Allied
..........
Estate Goods
........
....... .
Mfg
......
Hotels,
(Including
etc.)
Auto
.
TABLE
Repair).
.
1
..................
.....................
.................
...............
........................
.........................................
........................................
.......................
...........................................
..............................
...........................................
.............................
...................
........................................
......................
. ..........
..............................................
.........................
0.21
0.01
0.33
0.33
0.47
0.25
0.00
0.00
0.33
0.39
0.00
0.57
O.50
0.44
0.44
0.42
0.44
0.47
0.58
0.46
0.56
0.41
0.13
0.58
0.29
0.17
0.46
0.25
0.70
0.56
0.64
0.70
0.50
0.56
0.44
0.69
0.67
0.14
0.50
0.50
0.67
0.57
0.84
0.75
0.53
0.38
0.43
0.83
0.69
0.79
0.89
0.79
0.87
0.67
0.67
0.50
0.78
0.75
1.00
0.64
0.75
0.78
0.63
0.92
1.00
0.89
1.00
0.96
0.92
0.82
0.89
0.79
0.75
0.94
0.97
0.57
1.00
0.83
1.01
0.78
0.64
1.28
0.85
0.75
0.88
0.92
1.04
0.92
0.97
0.89
0.87
0.75
0.99
0.98
0.57
1.11
1.61
0.71
1.04
0.89
0.86
1.00
1.00
1.10
1.00
0.92
1.03
1.07
1.04
1.00
1.07
1.08
1.08
0.84
1.11
1.00
1.16
1.29
1.00
1.94
1.50
1.23
1.15
1.25
1.33
1.35
1.16
1.14
1.22
1.06
1.09
3.00
0.86
1.21
2.60
1.33
2.13
1.38
1.36
1.33
1.25
1.33
1.86
1.14
1.39
1.37
1.69
1.47
13.25
1.38
1.38
2.29
2.21
1.61
3.00
2.75
4.71
1.42
1.67
3.12
1.25
1.33
1.70
1.50
1.63
2.02
2.44
1.56
37.50
2.08
1.88

(concluded)
45
46
RUTLEDGE
VINING
lation aggregatewill be the service and other industrieswith the extremelyshorttrade radii. Thus, an ideal Table 1, it mightbe supposzd,
would showthe mosthighlylocalized industriesas lackingaltogetherin
manyregions.Only a fewregionswould show as greatrelativeemploymentin theseindustriesas does the nation,and a veryfewwould show
a relativelyhigh proportionof their employmentin these industries.
The productsof these industrieswould be importproductsfornearly
all of the regionsof the nation. As the industriesbecome less localized
(as the typical trade radii become smaller), fewerregionswould lack
the industriesaltogether,more regionswould move into the exporting
class, and the locationfactorsin the specializingregionswould become
smaller. The "ideal" Table 1, that is, mightsuggesta surface.Thus,
startingwiththe upperleft-handcorner(withthe mosthighlylocalized
industries)and movinghorizontallyacross the cumulativedistribution
of regions,this surfacewould move along the zero line leaving it relatively late but finallyclimbingvery rapidly to a height sufficientto
make thearea underthe traced-outline equal to unity7-the latterconditionalways being the case. For the next most highlylocalized industry, the surface would leave the zero line slightlyearlier and climb
somewhatless rapidlyto a heightsomewhatlower than the industiy
just preceding.At length,we should exhaust the list of industriesthat
are entirelylacking in at least one region,and as we move down the
arrayofindustrieswe shouldapproach that list ofindustriesthat would
be representedby a height of unity over the entirerange of regions.
That is to say, the firstcolumnswould show a continuousrise to the
neighborhoodof unity. Then fora while the columns would rise to a
maximumand then decline to the neighborhoodof unity.Finally, the
columnswould declinecontinuouslyand approach unity.
The conditionsto which the data in the actual Table 1 are subject
divergein great measurefromthose outlinedin the second sentenceof
theprecedingparagraph.The states showvast differences
in size so that
industrieshighlylocalized in large states show much smallermaximum
locationfactorsthan industriesequally highlylocalized in small states
Average income,living standards,and consumerhabits differimportantly fromregion to region so that employmentthat is necessarily
residentiaryis of varying relative importancefromregion to region.
The industrialclassificationsare verybroad and obviouslylack internal
homogeneityso that in certaincases the location factorsare rendered
quite ambiguousby the inclusionofuniformly
distributedemployment
7 If the regionswereofequal size in termsofemployment,
the simplemeanlocation factorwould be uniity.That is, the mean ordinateunderthe curvewould
be one, and the base of the figurewould extendfromzero to 100 per cent. The
mean ordinatewouldalso be unityifthe regionswerenot ofequal size, provided
thereis zerocorrelationbetweensize and the magnitudeofthelocationfactor.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
47
BEHAVIOR
and highlylocalized employmentwithinthe same industrialclassification. Nearly all of our-industrialclassificationswould show in some degree this ambiguity.But the industriesranked fourteenth,fifteenth,
seventeenth,and twenty-secondin Table 1 are especially outstanding
instances of internalheterogeneity.Clay and brick-makingmaterial
forregionsas large as states is almost "ubiquitous" materialand the
heavy transportationcharges result in certain subclasses of "Stone,
Clay, and Glass Products" manufacturingbeing residentiary.On the
other hand glass manufacturingappears to be relativelyhighly lominingis very highlylocalized, but other
calized. Nonferrous-metals
subclasses of "Other Mines and Quarries" are residentiary."Agriculture" is no more a single industrythan is "Manufacturing,"and the
production of certain extremelyimportantagriculturalproducts is
highlylocalized, while certain other subclasses are residentiary.Such
"Food and Kindred Products" manufacturiesas bakeries and ice
but othersubclasses show highlocalization.
plants are residentiary,
Nevertheless,Table 1 showsroughlythe characteristicsof our hypothetical surface.Were we to omit the ambiguous classificationsnoted
above, the similaritywould be almost striking,consideringthe other
discrepanciesin the underlyingconditions.The degree of localization
shades offgradually fromthe employmentsthat are virtually independentof local demand,throughthose that produce exportableproducts forwhichthe local demand is not negligible,to those finallythat
are virtuallywhollydependentupon local demand. From this table a
roughclassificationmay be made and a fairlygood conceptionmay be
obtained of the relativeimportanceof the "carrier"industrieswithina
given state. Referringto the table, we may set apart the followingindustriesas the "passive" industries.At the rightof each industrythe
medianstate percentageof total employmentis given:
Retail Trade ...............
. .................
Professionaland Related Services...............
Business Service and Repair Services............
Domestic Personal Service .....................
OtherPersonal Services (Laundry,Hotels, etc.). . .
Transportation............................
Construction.................................
..
Utilities...........1
Wholesale Trade ...........
Government.................................
Communications...........
..............
Food and Kindred Manufacturing
Amusement,Recreation,etc....................
Finance, Insurance and Real Estate .............
Printingand Publishing.......................
13.8%
7.4
1.9
4.5
3.4
5.0
4.5
2.3
3.5
0.8
2.0
0.7
2.4
0.9
54.2%

48
RUTLEDGE
in
VINING
thestates
groups,
at
These
and
1931
1930
1932
1933
1935
1934
1941
1936
1942
1940
1939
1937
1938
these are
following
these
not
one.
limitsin
Year
Yer
Q
all
Forwas
intervals
so
UINTILE
1st
69.0366.8159.9189.9083.15strictness
example,
interval
Quintile
represent
of
INTERVALS*
104.95-113.11
108.58-116.84
102.73-105.57
73.75
81.23
92.80-111.29
96.18-103.14
96.30
99.85-104.99
89.26
105.88-109.04
113.42-116.29
82.64%
the thequintile
IN
insignificantly
65
limits
small
intervals.
as
groups,
7 to shown
THE
2d
89.55-
AS
81.5182.9074.1996.44-
ANNUAL
interval
Quintile
by
109.61-110.31
103.25-104.72
113.24-115.53
111.59-112.76
75.85
120.11-127.48
91.38
97.38
83.38
106.18-106.87
105.12-106.46
Actually,
116.93-119.38
render
contain
86.27%
these
the
but
8
arbitrary
groups.
object
ARRAYS
PERCENTAGES
OF OF TABLE
3d
YEAR
the was
states
ANNUAL
83.6676.2686.3591.51Some
to
and
interval
Quintile
7
STATE
97.71-100.61
112.98-119.35
127.87-130.36
105.24-106.93
77.53
85.33
119.53-122.23
placing
93.75
110.42-112.46
106.50-107.34
107.12-109.07
115.59-117.58
PREVIOUS
90.38%
of divide
a discretion
contain the
INCOME
was
12.state
4th
states
in
of
One
85.4593.9377.7190.94one
interval PAYMENTS
exercised
each
Quintile
group
119.45-124.21
131.79-142.81
122.68-126.41
107.00-108.46
109.24-110.53
95.49
112.81-114.53
80.76
88.23
100.67-103.44
107.41-108.87
117.82-119.84
where
"quintile
annual
the
contains
array
but
7 interval"
into
variation
92.96%
EXPRESSED
5th
80.8288.50-
interval
five
Quintile
rather
states.
110.89-121.13
109.04-115.31
115.43-145.81
146.16-166.72
95.63-101.40
120.41-124.65
109.13-117.57
128.35-140.24
124.31-136.53
88.62
104.32-113.77
95.79
93.38-101.24%
than
equal
between

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
49
The percentagesshown represent,of course, the central tendency

around whichthe figuresforthe various states are dispersed.If a given
state's dispersionaround these central tendenciesis random,then we
may say that in the neighborhoodof 55 per cent of a state's total employmentwill be found in these industriesthat are pretty definitely
residentiaryor passive. Actually,thereis good reason to thinkthat a
state's deviation fromthe central tendencyis not random and that
those states withthe higherper capita incometend to fall to the right
of the centraltendency,so that the aggregatepercentagein this list of
employmentwould bear a direct relationwith the standard of living.
As the per capita incomeof a regionrisesthereseems to be a tendency
to allocate more of its resourcesto service industries.But the 55-percent figuregives an approximateconceptionof the relativeimportance
of these residentiaryindustriesin a state's total employment.To this
figureshould be added a certainproportionof total employmentthat
is foundin classificationsnominallyprimarybut that include employmentthat producesforlocal demand.8For example,a highproportion
of agriculturalincomecomes fromdairy products,poultry,truck,and
livestocksold forlocal consumption.Certain lumberproducts are locally consumed,and it would doubtless be found that a considerable
proportionof a state's manufacturingis fora trade area that is essentiallylocal. Suppose, as a roughguess, we should say that one-halfthe
employmentin agriculture,in textilesand apparel, in lumber,and in
furnitureand miscellaneouswood products are engaged in producing
forsuch a local trade area. If we take the median percentageemployed
in these industriesas the typical state employment,this would add
nearly15 per cent to the 54 per cent and would indicate that typically
about 70 per cent of the employmentin a state would be of the passive
type while about 30 per cent would be primaryor active or "carrier"
8 It should be remarkedthat in the cases of some states
muchof the employthat are nominallypassive is in factdependent
mentin industrialclassifications
upon "external"demand. For example,certainof the sparselypopulatedmountain states,such as Nevada, Montana, Wyoming,and Utah, have a quite high
in transportation.
proportionoftheiremployment
This,presumably,is because of
lines passingthroughthese states. Also, the
the trunkeast-westtransportation
highproportionof employmentin retail trade, amusements,financeand insurance, hotels,et cetera,in suchplaces as New York and Massachusettsis in considerablemeasuredependentupon incomechangesexternalto theseareas.
These classificationsare ambiguous in another respect. Some industries,
thoughbearingthe same name in the variousplaces, will in factbe parts of differentindustries,and an industrythat is passive in some regionsmay be in anotherregionan auxiliaryof a primaryindustry.For example,a good portionof
the utilitiesin Michiganis an auxiliaryof the automobileindustry.

50
RUTLEDGE
VINING
in the sense that the demand forits productsis externalto the state.9
Thus, wereall states "typical" in thisrespect,thenatureoftheproducts
of around one-thirdof the employmentof a state would determinethe
sensitivityof that state to national cyclicalforces.
Withtheabove generalizationsregardingindustriallocationin mind,
we may approach our principal problem. Our questions now are: Do
states fall systematicallyinto groups of states showing significantly
160-
120110
100 /_
60706050-
1930 1931 1932 1933 19'3' 1935 1956 1937 1936 1T99 19'0 1911 1912
FIGURE
1.-Ranges of quintileintervalsin the annual arraysof annual state

incomepaymentsexpressedas percentagesof year previous.
9 Florence estimatesa smaller figureforthe residentiarycomponentof employment."Calculationsforthe wholeUnitedStates ofthe occupiedpresonsthat

and some manufacturmustlocate close to the consumer,includingconstruction
ing, show a proportionof 48% of the total of occupied persons.But forlimited areas the necessaryproportionis undoubtedlylower.... We may take
35% of all workersas a reasonable minimumforservice workers.Adding 3%
forbuildingworkersas a similarminimum,the minimumresidentiarycomponent of all workersappears as 38%." Op. cit., p. 16. In this instance Florence
is dealingwithan area smallerthan a state, and, of moreimportance,his probfromthat with which we are concerned.He is attemptingto
lem is different
anticipatethe minimum numberofserviceworkersrequiredfortheestablishment
of a givenprimaryindustry.We are attemptingto estimatethe typical proportion of employmentproducinggoods sold in the local market.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
51
similarbehaviorwithrespectto rates ofchange ofincome?If so, do the

primaryindustriesof.the states withina given group providea rationale forthis similarbehavior?
3. REGIONAL
PATTERNS
OF CYCLICAL
FLUCTUATIONS
This problemof classifyingregionsinto groupsshowinginternaluniformityof cyclical behavior is, so far as we know, a problemthat has
not been studied extensively.'0It is a problem,however,that would
seem to introduceinterestinganalogies with other fieldsof analysis.
Just as in all problemsof the analysis of variance, we have here a set
of variates that may in principlebe classifiedin accordance withcertain attributes.On the basis ofsuch classifications,
the variationwithin
groups of regionscould be compared with the variation between the
groupssuch that patternsof cyclicalbehaviormay be discernedif they
exist.The systemsofclassificationthenmay yieldadditionalinsightinto
the processinvolvedin the interregional
diffusionoflocalized economic
shock.
As a matterof fact,the methodof studythat we have adopted up to
the presentdevelopmentofour inquiryis onlyanalogous to an analysis
ofvariancein a quite roughand backhanded manner.The data that we
have now at hand would hardlyjustifyan elaborateanalysis,but a simpler method can at least suggestan outline of furtherinquiry. Table
2 and Figure 1 show by years the quintile intervalsof state income
paymentsexpressedas percentagesof the precedingyear." We have
firstendeavoredto findout if the shiftsamong the occupants of these
quintileintervalsgive evidence of a regionalpatternof behavioras the
different
phases of the business cycle develop. These intervalsrepresent segmentsof the horizontalscales of our frequencydistributionsof
state percentagechanges,and if the individualstates appear to be distributedat randomfromyear to year withinthese rangesthenwe may
10 Withoutmakinga pretenseof documentingthe discussionof this problem
of regionalvariation of economic fluctuations,we should at least like to call
attentionto two sets of stimulatingarticlesin this field:the papers by D. G.
Champernowne,"The Uneven Distributionof Unemploymentin the United
Kingdom, 1929-36," Reviewof EconomicStusdies,Vol. 5, February, 1938, pp.
93-106 and Vol. 6, February, 1939, pp. 111-124; and the articles by H. W.
Singer,"The Process of Unemploymentin the Depressed Areas (1935-1938),"
ibid., Vol. 6, June,1939, pp. 177-188, and "Regional Labour Markets and the
Process of Unemployment,"
ibid., Vol. 7, October,1939,pp. 42-58.
11The data used in the computationofthesepercentagesare the same as those
used in the analysis describedin the cited articlein the July,1945,ECONOMETRICA. The estimates of state income payments for 1929 through1939 are
those publishedand discussedin the Surveyof CurrentBusiness,Vol. 22, 1942,
pp. 18-26; the data for1940 through1942 are fromthe June,1943,issue,Vol. 23,
pp. 10-22.

52
RUTLEDGE
VINING
considerthat the data evidences no such patternas we have in mind.

But if states fromyear to year move in groups withinthe frequency
distribution,and if thereis evidence of characteristicgroupingswithin
the different
phases of the business cycle, then a patternwould be at
hand that would provide a basis for furtherstudy. Our method of
classifyingthe states was as follows:For each state the numberof the
quintile interval (lowest quintile designated as first,highestas fifth)
that included the percentagefigurefor that state was listed for each
year. States that year afteryear fell into the same or adjacent quintile intervalswere groupedtogether.Because of the fact that onlynine
or ten states can occupy any given quintileintervaland because of the
verynarrowlimitsofmany of the quintileintervals,certaindeviations
fromthe apparent group patternwere ignoredin groupingthe states.
The procedurewas firstto set apart certainstates that appeared to the
eye to conformmost consistentlyto a pattern and then to set apart
certainstates whose patternsbear a strongfamilyresemblanceto that
of the group except that occasional marked discrepanciesoccur. This
procedureis merelya shortand roughmethodof studyingthe matter
of whetherthe variance indicated by the variation withingroups is
smallerthan that indicated by the variationsbetweengroups.Should
a group of states fall in the same quintileintervalyear afteryear, it
would mean that one quintile intervalis sufficientto include all the
variationof this group of states.'2
In Group I we includeArkansas,Mississippi,and Alabama, and it is
clear that Tennessee and Georgia show strongfamily resemblances.
With somewhat more discrepanciesthe remainingstates of the Old
South with the exceptionof Florida show a familyresemblance.The
quintilepatternsforthis group are shown in Table 3. Consideringthe
relativelynarrowrangesof the quintiles,these state figuresshow a remarkableconformanceto a pattern.Amongthe firstthreestates,it can
be said that thereare fewsignificantdiscrepancies.In 1938, Mississippi
in the second quintileintervalappears relativelylow comparedto the
othertwo states whichare within3 percentagepoints of each otherin
the upper part of the array. Yet Mississippi in this year is within3
12
The range of the componentgroup comparedwith the range of the total
tells us not much,of course,unlesswe have some conceptionof what we should
expectit to be, assuminghomogeneityof our variates. But fromwhat is known
about the relation between range and size of sample drawn froma normal
population,the rangeexhibitedby the 48 states should not be much morethan
1.5 to 2 timesthe range withina componentgroup composedof from5 to 10
states. We make no effortat this time,however,to develop such a criterion.
A possible alternativesimple methodof classificationmightconsistof computingforeach year and foreach state the deviationsfromthe mean in termsof
standardunitsand arrangingthesedeviationsinto groups.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
53
BEHAVIOR
TABLE 3
QUINTILE
PATTERNS OF THE STATES WITHIN GROUP I
Year
Ark.
Miss.
1930
1931
1932
1933
1934
1935
1936
1937
1.938
1
2
5
3
4
4
5
1
5
3
1
5
5
1
1
5
3
5
3
5
1
2
5
1
5
5
1939
1940
1941
1942
Ala. Tenn.
1
1
4
3
5
3
5
1
3
2
4
5
4
Ga. Ken.
1
2
4
4
5
2
4
3
3
4
4
5
2
1
2
5
4
5
3
2
1
4
2
4
4
4
1
3
3
3
3
4
4
3
2
3
2
2
3
La.
N.C.
S. C.
2
4
4
2
4
1
3
3
5
2
1
3
3
1
3
5
5
4
1
1
3
4
1
4
5
5
5
1
4
2
4
5
5
3
4
1
4
4
pointsofAlabama. In the followingyear,1939,whileArkansasand Alabama are within1 percentagepoint of each other,Mississippi is about
5 pointsabove these othertwo. In 1940,Arkansasand Mississippistay
together,but Alabama is about 6 points above the others.The rest of
the states show discrepanciesfromtime to time, yet all are clearly
pulled toward the same general vicinitywithinthe array,and in the
cases of most of the discrepanciesreferenceto the percentagefigures
will show a close similarity.In 1931, the declinesregisteredby the first
threestates weresomewhatmoremarkedthan in the restof the group.
In 1933,Northand South Carolina werecompletelyout ofline showing
rises of some 12 per cent while the others of the group were showing
practicallyno change in income. The discrepanciesin 1935 are not so

TABLE 4
QUINTILE
PATTERNS OF THE STATES WITHIN GROUP II
Year
N.Y.
Mass.
R.I.
1930
1931
1932
4
4
3
4
5
5
4
5
4
5
4
4
3
4
3
1934
1935
1936
1
1
1
1
1
1
1
1
1
1
1
2
1
2
2
4
2
2
2
1
1
1933
1937
1938
1940
1941
1942
2
1
1
1939
N.J. Vt.
3
2
2
1
2
2
4
1
N.H.
Me.
Md.
Mo.
4
5
3
5
5
4
5
5
4
3
3
3
2
1
2
1
3
2
2
1
1
1
2
1
1
1
1
2
3
2

5
4
3
3
3
1
4
2
54
RUTLEDGE
VINING
great as the quintilefigureswould suggest,a range of 5 per cent being

sufficient
to include all-thestates of the group. In 1936, the firstthree
states of the group showed somewhatlargergains than the others.In
1940,certainofthe groupgot out of line,Mississippiaccountingforthe
extremelow of the entirearray of states and South Carolina being almostthe extremehigh.But even here,a rangeof 10 per cent would include all the states and a rangeof less than 5 per cent would includeall
except the extremes.With these qualifications,it can be said that this
groupof states followsin generalthe same patternof behavior.
In Group II are placed New York, Massachusetts,Rhode Island, and
New Jersey,and a strongfamilyresemblanceis in evidencein the cases
of Maine, New Hampshire, and Vermont.The pattern of Maryland
in some respectsresemblesthe pattern of this group, but it also appears to relate itselfwiththe patternof anothergroup that will be describedlater. Missouri we are tentativelyplacing with this group,althoughthe basis forthis classificationis somewhatslight.The pattern
forthisgroupis shownin Table 4.
Here, too, the conformanceto a patternappears to be quite evident.
In the case ofthe firstfourstates thereis no significant
divergencefrom
the generalpattern.In 1932 New York was in the thirdquintileinterval
while Massachusetts was in the fifth,but New York was in the high
part of the thirdwhile Massachusetts was in the low part of the fifth.
While the range of the rates of change forall states in this year was 28
per cent,approximately3 per cent was the difference
betweenthe rates
of declineof thesetwo states. In 1939,divergencieswereapparent.But
they seem to be unimportant,the rate of increase of New York and
Massachusetts in the second quintile interval differing
by less than 3
percentagepoints fromthat of Rhode Island and New Jerseyin the
fourth.In 1940, New Jerseyis about 4 points above the other three
states, and in 1941 Rhode Island appears to have received a greater
impetusthan the other states.
Maine, New Hampshire,and Vermontshow onlyslightdiscrepancies
fromthe group pattern. Referenceto the percentagefiguresforthese
states indicates a rather close conformancewith the group pattern
throughoutthe years.
The patternof Maryland conformsroughlywith that of this group,
but it shows evidence of being pulled in the directionof the patternof
the groupto be nextdescribed,especiallyforthe latterfiveyears.
Missouri's is a moretenuous classification.When the percentagefigures forthis state are placed along side of those of the otherstates of
this group,the conformanceis very close. But forreasons that will be
presentedat a later point in our discussionit appears likelythat the
groupingwill followgeographicallines when such large areas as states

55
REGIONALPATTERNSOF BUSINESS-CYCLE BEHAVIOR
are regardedas the "regions." When the behavior of more appropriately defined"regions"-isanalyzed thereis evidence that the satisfaction of certainconditionswill resultin groupingsthat followindustrial
lines with less regardforgeographicalcontiguity.In the case of Missouri the conformanceto this patternshould perhaps be explained on
special grounds,and these groundswill be presentedwhen the attempt
is made to rationalizethe grouppatterns.
In Group III we place Indiana, Ohio, Michigan, Illinois,and Pennsylvania. Connecticutand Delaware bear a strongfamilyresemblance.
West Virginia and Wisconsin show resemblances,but marked discrepancies are in evidence. The quintile patterns for this group are
shownin Table 5.
TABLE 5
QUINTILE
PATTERNS OF THE STATES WITHIN GROUP III
Year
Ind.
Ohio
Ill.
Mich.
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
2
2
1
3
3
4
4
5
1
5
3
5
2
2
2
1
3
3
3
4
4
1
5
3
4
2
2
2
1
1
3
3
3
5
1
5
3
2
1
2
2
2
1
5
5
4
5
1
5
5
4
2
Pa. Conri. Del. W.Va. Wis.

3
3
2
2
3
1
3
3
1
3
3
2
1
4
4
3
2
2
2
3
4
1
4
4
4
2
2
5
3
2
3
2
4
4
1
5
5
1
1
2
4
3
4
4
1
4
3
2
1
3
2
2
2
2
2
3
3
5
3
3
2
1
3
3
2
A generalpattern is in clear evidence in the cases of the firstfive

states. The patternof Pennsylvania,however,appears to be drawn to
some extent in the direction of that of Group II. In 1930,the tendency
was for Group II states to be in the upper quintile interval and for
Group III states to be in the second. But Pennsylvaniawas the up-per
extremestate of the third quintile interval. In 1935, it occupied a
with the patternof
Group II position,and from1939 on it coniformed
Group II.
Michigan shows several marked discrepanciesand is obviously the
morevolatile ofthisvolatile group.In six of the thirteenyears,it was
withinfour states of one of the extremeends of the array. In 1933, its
rate of decline was greaterthan the otherdeclines of the group, and
in 1934, its rate of rise was considerablygreater.After1939 the rises
registeredby Michigan were considerablygreater than those of the
otherstatesofthegroup.

56
VINING
RUTLEDGE
These other states of the group are not so far out of line with one
anotherin 1933 as the-quintilelocations would suggest. The range of
variationforall 48 states was 24 percentagepoints,while the range of
variationforGroup III states exceptingMichiganwas 5 per cent.
Referenceto the percentagefiguresforthe firstfivestates indicates
few other discrepanciesof importance.Illinois was low in 1934, and
Illinois and Pennsylvania were somewhatlow in 1941. But it can be
TABLE
QIUINTILE
PATTERNS
OF THE
STATES
WITHIN
GROUP
IV
Year
Wash.
Ore.
Calif.
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
3
2
2
4
3
2
5
2
4
2
4
5
5
2
3
2
3
4
4
5
2
4
4
3
4
4
5
3
3
3
3
2
5
3
4
2
4
3
3
said, subject to the qualificationsregardingMichigan in 1933 and regardingthe apparent relationshipbetween Pennsylvania and Group
II, that the fivestates adhere very closely to a group patternand occupy the same vicinityof the frequencydistributionyear afteryear.
The otherstates show certaindivergencies.Connecticutoccupied a
Group II positionduringabout the firsthalf of the period.Since 1937,
it has conformedcloselywiththe patternof Group III. Delaware also
is attractedto the Group II pattern.West Virginiashows markeddiscrepancies,but its percentagepatternclosely resemblesthat of Pennsylvania. Wisconsin shows a moderate resemblance,but its position,
like Missouri's,tendsto remainin the centerof the arrayor towardthe
lowerend of the array of the absolute rates of change.
In Group IV are the PacificCoast states-Washington, Oregon,and
California.The adherenceto a group patternis somewhatcloser than
the quintile interval shown in Table 6 would suggest. Until 1941, a
rangeof8 pointswould includeall threestates in any year,and excepting 1930 and 1934 a range of 4 points would suffice.In 1941 and 1942
Washingtonand Oregongained at a more rapid rate than did California. The tendencyforthis groupappears to have been forrelativestability,the positionin the array tendingto be the middle ground.The

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
57
BEHAVIOR
exceptionsare 1936 and the two war years when the group occupied
the upper reachesof the arrays.
In Group V are placed certain political divisions that have shown
a markedtendencyforexpansion.The percentagefigureswhen placed
alongside of each other do not give as clear an impressionof consistencyas do the quintilefigures,but even here the conformanceto a
patternis in evidence. In seven of the years a range of 5 percentage
pointsincludesall the states,in threemoreyearsa rangeof8 pointswill
suffice,and in the remainingthreeyears 14 points separate the lowest
from the highest percentage of the three. Obviously, expansionary
forcesof an unusual characterhave been in operationin these three
areas. It does not follow,of course,that the similarityin the reactions
TABLE 7
QUINTILE
PATTERNS OF THE STATES WITHIN GROUP V
Year
Dist. of Colum.
Va.
Fla.
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
5
5
5
1
2
4
5
2
5
1
3
2
3
3
5
5
2
4
2
3
2
5
4
5
5
3
4
4
4
2
4
4
5
4
5
5
4
2
3
oftheseareas is due to the similarityin economicstructure.Noncyclical

factorswere doubtlessplayinga most importantrole. In nine of these
thirteenyears these areas were well into the righttail of the frequency
distribution.Virginia did not show this tendencyto expand so much
as did Florida, and the District of Columbia was somewhatmore erratic.Milder contractionsand strongerexpansionswereshownby these
threeareas than was generallythe case.
Groups I, II, and III representto us a fairlysatisfactoryclassification of cyclical reactionpatternsand are tentativelysatisfyingin the
sense that certainstatisticalregularitiesare suggestedforwhichan accountingis called. More interestingclassificationmay be made, we
think,when moreappropriatelydefinedregionsare used in the analysis, but neverthelessthe state groupingsraise certain pertinentquestions. Group IV is not quite so satisfactoryand Group V is somewhat
less so, for we feel that what similarityis suggestedmay be largely

58
RUTLEDGE
VINING
1941
1931
1940
1933
1942
1939
1932
1930
1938
1937
1936
1935
1934
4 1 1 1 3 5 24 1 5 3 1 1
4 4 1 1 5 5 3 3 3 5 4 2 2
Y
e
a
r
Okla.
Tex.
-
5 3 1 2 2 5 4 3 4 4 1 2 2
_
_
_
Ariz.
-
5 3 5 1 4 5 35 3 4 1 2 3
_
_
_
Utah
-
3 1 3 4 3 4 5 4 4 5 2 4 3
_
_
_
N.M.
Q-UINTILE
3 1 1 2 2 5 4 3 2 4 3 3 4
_
_
Colo.
-
4 3 3 1 2 5 5 3 5 5 2 1 3
1 2 5 3 2 3 1 5 5 4 1 1 1
3 5 5 4 1 5 1 5 1 5 3 1 2
5 1 5 5 1 4 25 4 1 4 5 5
PATTERNS
_
_
_
Idaho
THE
TAB3LE
2.
2 4 5 1 3 4
5 1 3 1 2 1
5 1 5 1 1 5
3 4 1 5 5 5
5 1 5 1 1 5
_
Mont.
_
STATES
N.D.
_
b
_
WITHIN
Nev.
_
_
2 2 2 3 5 1
OF
GROUP
_
Wyo.
VI
_
_
Neb.
_
_
_
_
Iowa
_
4 4 2 3 5 1
5 5 4 1 1 5
S.D.
_
5 5 4 1 1 4 2 4 3 4 1 2 5
_
Kan.
_
2 1 1 2 4 3 3 5 2 3 2 3 4
Minn.
_

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
59
attributableto noncyclicalforces.But the rest of the states-those of

the Great Plains and Rocky Mountain regions-give even less satisfactoryresults.The methodsthat we are usingdid not reveal consistent
breakdownsinto groups showinginternaluniformityregardingrates
of change of income. Several types of factorscould be responsiblefor
this lack of evidence of group behavior among these states. In the
firstplace, the data have obvious shortcomingsfromwhich mightbe
expected considerablestatistical error.The bulk of the very sparsely
populatedstates is includedin the presentgroup.Of thesesixteenstates
nine are smaller economicallythan either Arkansas, Mississippi, or
Rhode Island. Any one of these nine accounts forsmallerincomepaymentsthan eitheroftheselatterthreestates. These ninestates-North
and South Dakota and the Mountain States exceptingColorado-togetheraccount for only about 2' or 3 per cent of the national income
payments,whichis less than the incomepaymentsof Texas and about
the same as those of Missouri. The estimates that are available are
roundedto the nearestmillion,and this mere roundingcould resultin
an "error" of from1 to 2 per cent in the cases of the smller statesthat is, the numberof digits in our computed percentagechanges is
more than is warrantedby the data. Of more importanceperhaps are
the errorsof statisticalestimate.Revisions of these figuresfrequently
involve changesof substantialamounts,and in the cases of the smaller
states the effectupon the percentagechanges may be quite marked.)3
But aside fromthe shortcomingsof the data,'4 thereare "real" factors
displayed by these state figthat may underliethe lack of uniformity
ures duringthis period. Withinthisdecade underreview,certain"noneconomic"forcesof great importanceaffectedthe incomesof the Great
regardto economic structure.The unprecedented
Plains states Mwithout
drouthsin this regionand the influenceof governmentalintervention
13 Regardingthe revisionsof theseestimatesof state incomepayments,it may
be in pointto call attentionto what was stated in footnote15, page 201 and 203
of the articlepreviouslycited in ECONOMETRICA, Vol. 13, July,1945. The same
analysis describedin the presentpaper was firstapplied to figurespublishedin
the SurveyofCurrentBusinessforOctober,1940. Althoughthe revisionsappearingin the July,1942,Surveywereconsiderable,even involvinga changein definition, the two sets of data yielded the same conclusionsregardipgthe cyclical
behaviorof rates of change.
characterof an
14 We are aware of the arbitrariness
and the question-begging
argumentthat accepts any supportthat a set of data may providefora hypothof these data. For
esis but explainslack of supportin termsof the shortcomings
all that we may say, the support that the data providemay be attributableto
the "shortcomings"of the data. This amounts to saying that at this stage of
the inquiryand withthe data that are available statisticalresultsmust be regardedas steppingstonesforintuition,and a priorireasoningmustbe the guide
in the tentativeinferencesdrawnfromthe data.

60
RUTLEDGE
VINING
could have affectedthese percentagechangessufficiently

to confusethe
quintile intervalpatterns.
The above factorsmentionedare such as would muddy up the statisticsin such a way that a patternthat mightactually existwould not
be revealed.There are otherconsiderationsthat would possiblysupport
doubts as to the existenceofa consistentpatternofcyclicalreactionsof
these statefigures.In previousparagraphswe spoke of a state as containingparts or the wholesof several economicregionsand of a state's
cyclical reactionpatternas being merelyan average of the respective
patternsof the componentregions.Where a state is populous and intherewould be a relativelylarge "sample" of redustriallydiversified,
gions representedby the state figures,and the quintile positionof the
state would not likely be affectedby erraticmovementsof single industriesor regions.Or wherethe regionsof a state or a group of states
are all specialistsin the same single "carrier" or "export" industry,a
consistentpatternofbehaviormay existeven thoughthe patternitself
shows erraticmovementsin time. But wherestates are sparselypopulated so that state figuresrepresentan average of a small numberof regions, and where these componentregionsof a state show wide economic diversity,then erratic behavior of a single industryor region
may each year determinethe quintile position of a state, and state
figuresmay reveal no group pattern.
However,forthe sake ofcompletenessthese sixteenstates are placed
in Group VI with threesubclassifications.Table 8 shows the positions
of the respectivestates in the annual frequencydistributions.In the
year 1933 the entiregroupshoweduniformreaction.These states practically occupied the top 40 per cent of the frequencydistribution.In
1930, 1931, 1933, 1935, 1937, 1939, 1940, and 1942 Group VIa showed
uniformity.Disregarding extremes,we may say that Group VIb
showeduniformreactionfrom1933 through1936 and again in 1939 and
1940. The patternswithinGroup VIc were similarfrom1930 through
1933 and again in 1935, 1936, and 1942. Borderlinestates appear to be
pulled in the directionof adjacent groups.Thus, Oklahoma and Texas
appear to be pulled in the directionof Group I. The patternsof Louisiana and Texas are quite similar.Minnesota is pulled in the directionof
Group III, and Wyomingin the directionof Group IV. In at least one
case geographicalcontinuityis broken,Nevada having a patternvery
similarto that of Connecticut.The states directlywest ofTexas appear
to be pulled in the directionof the MIountainstates farthernorth,and
these in turn are pulled in the directionof the northernGreat Plains
states.
Consideringthe patternsofthe six groups,then,it appears to be true
that whenareas as large as states are studiedcertaingeographicalpatternsofbusiness-cycleresponsemay be observed.Groups ofcontiguous

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
61
states respondsimilarlyin the business cycle,and the patternsof the

borderstates of these groupsappear to be "pulled" in the directionof
the neighboringgroup. Certain groups of states typicallyshow more
violentchangesthan do othergroups.A businessexpansionor contraction worksout over the entirenational area, but ratesof expansionor
contractiondifferin a typical manneras betweenareas.
The states of Group VI are among those whose income payments
fluctuaterelativelygreatly. The figuresfor the rates of change for
Oklahoma, Nebraska, Iowa, South Dakota, North Dakota, Montana,
Idaho, and Nevada werelocated in the extremequintileintervalsmore
than two-thirdsofthe time. It is rarethat these states are foundin the
middle quintile intervals of the frequencydistributions.This tendency to fall in one of the tails of the distributionis also true of Group I
states. In the patternsof Arkansas, Mississippi,and Alabama, the extreme quintile intervalsappear about two-thirdsof the time. One is
temptedto generalizethat such states show morerapid risesin years of
risingincomeand more rapid declinesin years offallingincome.Many
exceptionsappear in the data analyzed-too many, one may feel,for
the data to be regardedas supportingthe generalization.In working
with the data, however,a strongimpressionis gottenthat the exceptions may be largely attributableto differences
in the timing of the
turningpointsand that an analysis of quarterlydata willbe muchmore
revealing,particularlyforthese periods in the vicinityof the cyclical
turningpoints.
The converseof the above generalizationmay be stated with some
confidenceregardingthe behavior of Group II states. These states are
also foundin the extremequintile intervals,but they are in the high
quintileintervalswhen income is fallingand in the low quintile intervals when incomeis rising.Had rates of change been arrayed without
regardto sign,these states would have been, in general,the occupants
of the lower reaches of the array. There is one apparent exception.In
1933 these states were among those showingthe greaterdeclines. But
thisyear would be a case in point forthe observationmade above with
regardto the inadequacy of annual data in the analysis of the turningpointperiods.In an importantsense,1933 could be consideredas a year
of incomerise,the bulk of the forty-eight
states showingeitherincome
rises or else no appreciable changes in income,and the more volatile
raw-material-producing
states showingquite substantial gains. When
rates of change are arrayed,the Group II states (exceptingMissouri)
are foundin the lowest quintile intervalforty-four
times. Were each
ofthe forty-eight
statesequally likelyto fall in this interval,we should
expectto findGroup II statestheretwenty-onetimesduringthe period.
We are not in a positionto formulatewith confidencea generalization regardingthe timingof business change as between different
re-

62
RUTLEDGE
VINING
gions.At one timetherewas a fairlyprevalentnotionthat a westward

lag of businessfluctuationsexistsin the United States-declines or recoveriesin businessbeing firstnoticeablein the East, then in the Middle West, finallyin the Pacific Northwest.As more data have been
made available, this view has not been substantiated,15
although the
fact that currentdata, reflectingthe behavior of the economic structure that now prevails, does not indicate such a lag is not evidence
againstthe view that such a patterndid existwhenthe economicstructureofthe countrywas different.
If employmentin certainregionswere
concentratedin industriesproducingforinventorydemand, fordurable-consumer-goods
demand, and for plant and equipment demand,
then these regionswould be likelyto be affectedsomewhatearlierin a
recoveryor decline. But it seems clear that the location of industryis
not suggestiveofa situationthat would give this westwardlag to business fluctuations.As a matterof fact, our impression,as pointed out
above, is that furtherstudy will substantiatethe hypothesisthat the
raw-materials-producing
areas, the demand for whose "export" products is so subject to "inventory"decisions,will typicallyshow the initial developmentsof cyclical change.'6But whatevermay be said now
regardingthe timingof the turningpoints, there is clear evidence of
typical differences
in the degree to which regionsare affectedby the
business cycle.
The finalpointto be raisedin the presentpaper involvesthe question
ofthe extentto whichour observationson the regionalnatureofindustrial location may throwlight upon the factorsthat would determine
the sensitivityofa givenregionto the cyclicalforcesoperatingwithina
larger area.
4. FACTORS
UNDERLYING
AND AMPLITUDE
REGIONAL
OF CYCLICAL
VARIATION
IN TIMING
FLUCTUATION
In Section 2 a procedureof Sargant Florence was adapted, and the

various industriesof a region were classifiedinto "residentiary"or
"passive' industrieson the one hand and "primary"or 'carrier" in15 Joseph Demmery and Fred Ritchie, "The Westward Lag in Business,"
JournalofBusiness,Vol. 7, April,1934,pp. 173-181.
16 Cf. pp. 207-208 of the articlecited in ECONOMETRICA,July,1945. If space
permittedit would perhapsbe of interestto comparecertaintentativeconclusions reached froma study of internationaldata in Sir William Beveridge's
Full Employment
in a FreeSociety,New York, 1945,withthe presentconclusions
inferredfromthe behaviorof regionalseriesrepresenting
the businessbehavior
forthe componentparts of a singlenationalarea. Referenceis had hereparticularlyto AppendixA, paragraphs33, 35-42. There are suggestionsof these conclusions,it may be recalled,in Keynes's briefdiscussionof the role of agricultural and mineralproductionin the businesscycle, The GeneralTheoryof Em-
ployment,Interestand Money, New York, 1936, pp. 329-331.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
63
dustrieson the other,the characterof the latter type determininga

region'sresponsivenessto business-cycleforces.We indicatedtherethe
basis for our rough estimate that typicallyabout two-thirdsof a region's employmentproducesforlocal consumption,this portionof the
employmentthus beingpassive in the interregionalpropagationof economicshock,and that about one-thirddepends upon externaldemand,
being consequentlythe active carrierin the transmissionof business
perturbations.Let us now give a sketch, rough though it is, of the
groups of states. For the
characterof the employmentof our different
presentpurposelet us simplyrecordcertainobservationsregardingthe
states that formthe more consistentpatterns.In Group I these states
are Arkansas,Alabama, and Mississippi.In Group II, New York, New
Jersey,Massachusetts,and Rhode Island are the states more consistentlymovingtogether.In Group III, thesestates are Illinois,Ohio, Indiana, Michigan,and Pennsylvania.Certaiinstates of Group VI, while
not forminga consistentpattern,fall fairlyregularlyin the tails of the
frequencydistributions,and in this respectthey share a characteristic
of the Group I states. These states are Oklahoma, Iowa, Nebraska,
South Dakota, Montana, North Dakota, Idaho, and Nevada. Finally,
Missouriis ofsome interestbecause ofits conformancewiththepattern
ofthe Group II states.
From the various census reportsof 1940-Occupations, Agriculture,
and Manufactures-the followinginferencesmay be drawn: Arkansas
and Mississippiare highlyspecialized in agriculture,and that agriculture is in turn highlyconcentratedin cotton production.Typically,
cottonaccounts formorethan half of the total farmincome,and more
than half of the employedpopulation is engaged in agriculture.Alabama is also a heavy cotton specialist,nearlyhalf of tne farmincome
being derived fromcotton production.In the case of Arkansas and
Mississippi,cotton and lumberproductionconstitutethe bulk of the
"primary"or "carrier"employment,although,of course,certainparts
of these states depend upon other industriesfor "exchange" on the
"outside" economy.In Alabama, whereagricultureis not quite so important,two otherindustriesare added. In coal miningand iron and
steel manufacturingtogetherthere are approximately6 per cent of
Alabama's employment.In the textileindustrythereis an additional 6
per cent. These two industries,along withcottonand lumber,appear to
relate Alabama's business fluctuationsto "external" regions.It would
seem that in the case of these threestates,the "carrier"industriesconsist of a few industries,each producingproductshaving high income
elasticities.This conditionprovides an easy rationale for the violent
fluctuationsof incomein these states,and the generalsimilarityof the
main carrierindustriesis in conformancewith the uniformityof the
rates of business change in these states.

64
RUTLEDGE
VINING
In the case of the Group II states the carrieror primaryindustries

seem to be mainly among the processorsof durable and semidurable
importancein'the four
consumergoods. Agricultureis of insignificant
states underreview,therebeing only3 or 4 per cent ofthe employment
engagedin agricultureand that in dairyingand truckfarmingforlocal
consumption.Textiles and apparel is the largestsinglecarrierindustry
employing8 per centin New York and 23 per centin Rhode Island. The
other employmentrelatingthese states to "external" regionsis well
diversified.If foreach industrytherewerearrayedthe state figuresfor
percentageof total employmentfound in that industry,these states
wouldbe foundin thehighquintileintervalsin practicallyall the manufacturingindustries.In orderto account forthe 25 to 35 per cent ofemploymentthat is estimated to be engaged in producingfor external
markets,at least seven or eight broad industrialclassificationswould
have to be included.Even in such industriesas would ordinarilybe regarded as primaryindustries,local marketsin the cases of these states
must take a high proportionof such production.Thus, it would seem
that these populous areas are economicallyrelativelydiversifiedand
and produceexternallymarketedproductshavingincome
self-sufficing,
elasticitiesthat are low relative to the elasticitiescharacterizingthe
agriculturaland mineralraw materials.All of thismeans a milderbusiness fluctuation.The agriculture,lumber,paper, shoes, and textilesof
Maine, New Hampshire,and Vermontdepend presumablyupon the
marketsof the metropolitanareas of this region,and thus the economies of these states are tied to the patterns of these more populous
areas even thoughtheeconomicstructuresof these smallerstates bear
littleresemblanceto the otherstates ofthegroup.In Connecticut,some
20 per cent of the employmentis in those industriescharacterizingthe
industryof the states of Group III, and as a consequence, it would
seem,the patternof this state appears to be a combinationof Group II
and Group III.
Missouri'spattern,it has been noted,is verysimilarto that of Group
II even thoughgreat distances separate these areas. There is littlebasis, however,forattributingthis similarityto a similarityof economic
structure.Even thoughthe economicstructureswereidentical,it would
be extremelyunlikelythat uniformityof cyclical behavior would be
maintainedwhenthe states are geographicallyso farapart. It has been
indicatedthat the "passiveness" or "localness" of a givenindustryis a
matterthat depends in some considerabledegreeupon the definitionof
theregion.The marketareas ofmanyofa state's residentiaryor passive
industrieswill reach out into contiguousstates so that ordinarilythe
experienceof surroundingregionswill have its effectupon a given state
regardlessof the main outlinesof its economic structure.In the cases

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
65
of economicallysmall, highlyspecialized states or of smaller regions

specializingin one or a fewproductsmarketedin centralmarkets,this
relationshipwithsurrounding
areas may not be important;but forareas
as large and as diversifiedas the largerstates, geographicalcontiguity
seems practicallyessential for consistentpattern behavior. The uniformityof behavior in the present case may be explained on other
grounds. It will be noticed that the patternsof the Group II states
show relativelysmall rates of increase in years of expansionand relatively small rates of decline in years of contraction. The quintile
patternofany state showing this same relative stability would conformin general to that of the Group II states. It will be found that
Missouri'seconomyis perhapsas diversifiedas that of any state in the
employmentsconformquite
country.The percentagesin the different
closely to the national percentages-indicating, that is, a self-sufficiencyand diversification
that wouldbe foundin fewotherstates.'7No
one or several industriesdominatesthe economy.Its farmingis diverspecialtythis specializationis
sified,and if thereis any manufacturing
and
in food processingand shoe manufacturing.This self-sufficiency
diversification
would indicatea relativestabilitythat would renderthe
patternsimilarto those of the Group II states.
The Group III states-Indiana, Ohio, Michigan,Illinois,and Pennsylvania-are dominatedby the heavy industries.In these states, the
metal-productsindustries,machineryproduction,automobile production,and the auxiliaryindustriessuch as coal mining,account forfrom
15 per cent to 25 per cent of the total employment.The urban areas of
thisregiondepend to a considerableextentupon thesefewtypesof industry,and the farmingareas produce in the main corn and hogs and
dairy products,the latterforlocal metropolitanconsumption.In general, these stages show more violent changes than do the Group II
states, and the specializationin durable manufacturesforinvestment
demand would be in accordance with this greateramplitude of shortrun businesschange.
The statesbelongingto Group VI that we are discussing-Oklahoma,
Iowa, Nebraska, South Dakota, Montana, North Dakota, Idaho, and
Nevada-are ofinterestat thistimeby virtueoftheirfairlyregularappearance duringthisthirteen-year
periodin the extremetails ofthe frequency distributions.Special and randomcircumstancescould doubtless account formuch of this extremebehavior,particularlyin the case
of the Great Plains agriculturalstates. But it should be noted that all
17 Sargant Florence concluded that Missouri was the most economically
diversifiedof all the states. Cf. his "The Measurementof Economic Balance
and Diversification,"Mimeographedreleaseofthe National ResourcesPlanning
Board, 1940.

66
RUTLEDGE
VINING
these states are highly specialized in producingraw materials. Oklahoma is dominatedby cotton and petroleumproductionand refining.
Iowa is predominantlyagriculturaland specializes in corn and livestock. Nebraska's concentrationis in corn,livestock,and wheat. South
Dakota is as predominantlyagriculturalas Arkansas and Mississippi,
depending,however,upon wheat, corn, and livestock. Montana and
North Dakota are dependentupon wheat and livestock. Montana, in
addition, has nearly 10 per cent of total employmentin miningand
metal refining,the principalmineralinvolved being copper. Idaho is
dominated by livestock agriculture,nonferrousmetals mining,and
lumber.Nevada is principallya miningstate, 15 per cent of its employmentbeing engaged in the miningof nonferrousmetals. Mining
employsas many people as does agriculturein Nevada, and what agriculturethereis, is mainlyrange livestock.
Thus, these states, like the Group I states, are specialistsin agricultural and the otherextractiveindustries.A relativelyhigh proportion
of the employmentof these states is dependentupon externaldemand,
and thislatterwould appear to have a highshort-runincomeelasticity.
The situationof these raw-material-producing
states would seem to be
somethinglike that of the marginalfacilitiesof a manufacturing
firm.
For much of theirproduction,demands are only forthcoming
at high
levels of national incomeand high levels of national investmentactivity. When investmentactivity slackens, the use of the raw-materials
output of these mountain,plains, and southeast states declines drastically. This decline in the primaryindustriesis then transmittedto
those residentiaryindustriesproducingforlocal demand.
Two observationsof Keynes in his General Theoryare of especial
pertinenceregardingthe ratherexceptionalviolenceof cyclicalfluctuation in these areas in which the primaryindustriesare predominantly
agriculturaland mineral.Keynes raises, in order to explain away, an
apparent paradox. He says:
We have seen above that the greaterthe marginalpropensityto consume,the
greaterthe multiplier,and hence the greaterthe disturbanceto employment
to a given changein investment.This mightseem to lead to the
corresponding
paradoxicalconclusionthat a poor communityin whichsaving is a very small
proportionofincomewill be moresubject to violentfluctuations
than a wealthy
communitywheresaving is a larger proportionof income and the multiplier
consequentlysmaller.
He thendissolves the paradox and concludes that

whilstthe multiplieris largerin a poor community,
the effecton employmentof
fluctuationsin investmentwill be much greater in a wealthy community,
assumingthat in the lattercurrentinvestmentrepresentsa muchlargerproportion of currentoutput.

REGIONAL
PATTERNS
OF BUSINESS-CYCLE
BEHAVIOR
67
He indicates symbolicallyhis conclusions,setting up an expression

representing
the relativechangein incomefora givenrelativechangein
investment,
1 - C/Y
dY/Y
dI/I
dC/dY
"As wealthincreasesdC/dY diminishes,but C/Y also diminishes."'

8
Now as amatter of fact,his original paradox substantiallydescribes
the facts so far as the regionswithinthis countryare concerned. The
poorerregionswithinthe nationmake up the bulk ofthe areas exhibiting the moreviolent cyclicalfluctuations.The reasonforthis,however,
is not hard to find.These less wealthyareas happen to be those great
"colonial" stretchesthat yield up the extractiveraw materialsthat are
processedin the manufacturing
areas. They are also highlyspecialized
in their productionso that importsand exports representrelatively
highproportionsof total income.If the average and marginalpropensities are definedas in the literatureon the "foreigntrade multiplier"'9that is, as propensitiesto consumelocal products-and if the states of
this nation were arrayed in accordance with wealth or per capita income, then we should undoubtedly find as wealth increases that
dC/dY willdiminishbut that C/Y willincrease.That is, the above ratio
ofthe proportionalchangein incometo a proportionalchangein investmentwould be lessenedas we move along the arrayfromthe poorto the
wealthyareas. dC/dY would diminishbecause importswould be relatively insensitiveto changes in income in the highlyspecialized rawmaterial-producing
areas and relativelysensitiveto changes in income
in the manufacturing
Northeast.The increaseof C! Y would reflectthe
increasedproportionofemploymentin thewealthierareas thatproduces
for "local" consumption.This perverse result, of course, would not
representa functionalrelationshipbetweenwealth and these propensities. Rather, it would be due to the negative correlation,in this case,
betweeneconomicself-sufficiency
and wealth and to the nature of the
economicspecializationsinvolved.
The otherobservationof Keynes that may be noted here in connection withthe observableregionaldifferences
in cyclicalfluctuationshas
to do withhis briefdiscussionof the role of agriculturein the business
cycle.20He interpretsa change in the carry-overof agriculturalproducts as a changein the currentrate ofinvestmentand pointsout that in
18 The GeneralTheoryof Employment,
Interestand Money, New York, 1936,
pp. 125-126.
19 GottfriedHaberler,Prosperity
and Depression,3rd ed. rev., Geneva, League
of Nations,1941,pp. 461-473.
20 Keynes,op. cit.,pp. 329-332.

68
RUTLEDGE VINING
industrysuch
whereagricultureis the predominant
a community
in
will
importance
be
of
overwhelming
changes carry-over
compared
fluctuations.
GroupsII and VI
withanyotherusualcauseofinvestment
thatare described
but the effects
are made up of suchcommunities,
forthesestates.A preby Keyneswouldbe even moreintensified
outputwouldrepdominant
partofthevalueoftheirentireagricultural
forthem(in the senseadoptedin
expenditures"
resent"investment
discussionsof the internationaltrade multiplier),and carry-over
changeswouldbe accompaniedby pricechanges.Keynesgoes on to
fluctuation,
exsuggestthatin thepastperhapsno causeofinvestment
ceptwar,
of agriwas in any way comparablein magnitudewithchangesin the carry-over
culturalproducts.Even today it is importantto pay close attentionto the part
played by changesin the stocksof raw materials,both agriculturaland mineral,
oftherateofcurrentinvestment.I shouldattributethe slow
in the determination
has been reached,mainly
rate of recoveryfroma slump,afterthe turning-point
to the deflationaryeffectof the reductionof redundantstocks to a normal
level.... Sometimes,indeed,the reductionof stocks may have to be virtually
completedbeforeany measurabledegreeof recoverycan be detected.
It is commonunderstanding
in tot%lconsumer
nowthata reduction
demanddoes nothave to take place in thelatterphaseofan upward
in orderthatthereshoulddevelopa reductionin
businessmovement
decision.All thatis required
demandforoutputsubjectto inventory
is thecompletion
depletedinventories
ofthebuildingup ofpreviously
ofconso thatthedemandforoutputis loweredto theneighborhood
sumertaking.Or again,it is not necessarythat consumerdemand,
thatis, nationalincome,risein thelatterphaseofa downwardmoveand otherdurableraw
mentin orderthatthedemandforagricultural
materialoutputshouldrise.It is onlynecessarythatthatportionof
in invenconsumption
thathas been providedfroma disinvestment
of these
A development
be reducedsufficiently.
toriesor carry-over
erratic
ofrelatively
mustsurelylead to an expectation
considerations
ofturning
points
andviolentfluctuations
and ofan initialdevelopment
withintheGroupI and GroupVI states.In fact,theworstofan assortseemsto be the
so faras thebusinesscycleis concerned,
mentofworlds,
lotoftheseareas:a lowaveragepropensity
to consumehomeproducts
and dependenceuponinterre(thatis, a highdegreeofspecialization
to consumehomeproducts
gionaltrade);a highmarginalpropensity
ofdemandforimports);a highlysensi(thatis,a lowincomeelasticity
multiplicand
(thatis, a highinviolentfluctuating
tiveand relatively
ofdemandfortheexportproducts).
comeelasticity
University
of Virginia


The Econometric Society

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Econometric Society

Uploaded by

Copyright:

Available Formats

Location of Industry and Regional Patterns of Business-Cycle Behavior

Author(s): Rutledge Vining

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

LOCATION OF INDUSTRY AND REGIONAL PATTERNS

In a recentpaper' we presentedthe firststage of a statisticalinquiry

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

In the presentpaper, as in the paper cited, we shall workwith state

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

one pointofview,the home-marketemploymentcould be thoughtofas

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

In our study of industriallocation we have made an adaptation of

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

region.This measure divided by two varies between zero and unity.4

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

The residentiaryindustriesare those with "low" coefficients

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

REGIONAL PATTERNS OF BUSINESS-CYCLE

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

The percentagesshown represent,of course, the central tendency

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

1.-Ranges of quintileintervalsin the annual arraysof annual state

9 Florence estimatesa smaller figureforthe residentiarycomponentof employment."Calculationsforthe wholeUnitedStates ofthe occupiedpresonsthat

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

similarbehaviorwithrespectto rates ofchange ofincome?If so, do the

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

considerthat the data evidences no such patternas we have in mind.

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

PATTERNS OF THE STATES WITHIN GROUP I

practicallyno change in income. The discrepanciesin 1935 are not so

PATTERNS OF THE STATES WITHIN GROUP II

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

great as the quintilefigureswould suggest,a range of 5 per cent being

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

REGIONALPATTERNSOF BUSINESS-CYCLE BEHAVIOR

PATTERNS OF THE STATES WITHIN GROUP III

Pa. Conri. Del. W.Va. Wis.

A generalpattern is in clear evidence in the cases of the firstfive

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

PATTERNS OF THE STATES WITHIN GROUP V

oftheseareas is due to the similarityin economicstructure.Noncyclical

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

attributableto noncyclicalforces.But the rest of the states-those of

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

could have affectedthese percentagechangessufficiently

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

states respondsimilarlyin the business cycle,and the patternsof the

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

gions.At one timetherewas a fairlyprevalentnotionthat a westward

In Section 2 a procedureof Sargant Florence was adapted, and the

ployment,Interestand Money, New York, 1936, pp. 329-331.

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

dustrieson the other,the characterof the latter type determininga

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

In the case of the Group II states the carrieror primaryindustries

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PM

of economicallysmall, highlyspecialized states or of smaller regions