1905703

8/10/2019 1905703

1/33

Location of Industry and Regional Patterns of Business-Cycle BehaviorAuthor(s): Rutledge ViningSource: Econometrica, Vol. 14, No. 1 (Jan., 1946), pp. 37-68Published by: The Econometric SocietyStable 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 [email protected].

.

The Econometric Societyis collaborating with JSTOR to digitize, preserve and extend access toEconometrica.

http://www.jstor.org

This content downloaded from 132.248.147.90 on Mon, 3 Nov 2014 12:53:41 PMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=econosochttp://www.jstor.org/stable/1905703?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/1905703?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=econosoc

8/10/2019 1905703

2/33

LOCATION OF

INDUSTRY

AND REGIONAL

PATTERNS

OF

BUSINESS-CYCLE

BEHAVIOR

By

RUTLEDGE VINING

In a recent

paper' we presented

he first tage

ofa statistical nquiry

into the nature

of the abstraction nvolved

n the use of

nationalseries

for

the

description

nd

analysis

of business cycles.

The results were

given of

an analysis of one

aspect

of

the cyclical

behavior

of

national

income-its percentage

rate ofchange.

This rate of change

of

the na-

tional

total was

interpreted

s

a

parameter

of

the frequency

istribu-

tion of the respectiverates of

change of the component

parts of the

largerarea, and statisticsdescribing

certain

pattern of behavior

of

these distributionswere shown. The distributions

ppear

to have a

characteristic hape approximatingogarithmicnormality, nd there

is a suggestion

f a systematic

evelopment s the different hases

of

thebusinesscycle unfold.

The

skews

of the distributions ppear

to

be

positive

during he period

of

the expansion

when the rate of growth

s

increasing.When the rate ofgrowth

egins to decline,

t was tentatively

suggested

that

the

skew

shifts

o

a

negative

and remains

a

negative

through

he

absolute turning

oint of

national

income

and during

he

period

when the rate

of

contraction

s

increasing.

When this rate

of

contraction eginsto decline,the skewof thedistribution gain shifts

to

the

positive.

An

attempt

was made to rationalize

his

cyclical

evolu-

tion

of

shape

of

these

distributions. t was further oted

that

the

ex-

treme

movementsfromyear to year are found

generally

mong the

same

set

of

states,

there

being

a

tendency vident

for

he more

parsely

settled

tates that

are highly

pecialized

n raw-material

roduction

o

cluster

n

the ends

of the

distribution

nd,

in

years

of marked

change,

in the

same

end

of the distribution.

It is proposed n the presentpaper to discuss n more detail the geo-

graphical

make-up of these annual frequency istributions

f regional

rates of businesschange. We think

hat certain

generalizationsmay be

made

regarding

regionalpattern

f short-run usinesschange,

and we

should like to

analyze possible

factors hat

might

ccount

for

the ob-

servable imilarities nd

differences. ur previousdiscussion

of

the

fac-

tors determining

he

cyclicalresponsiveness

f a

given region

aid

em-

phasis upon industrial ocation

or

specialization

and

the

institutional

and physicalfactors nfluencinghe region's ommercial ies. The ques-

tion

we raise

now has

to

do

with certain attributes

of

industrial

location that

we regardas particularly

elevantfor regionalbusiness-

cycle analysis.

I

"Regional

Variationn

Cyclical

Fluctuation

Viewedas

a Frequency

Distribu-

tion,"

ECONOMETRICA,

Vol. 13,

July,

1945,

pp.

183-213.

37

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp

8/10/2019 1905703

3/33

38

RUTLEDGE VINING

1. REGIONAL ECONOMIES IN INTRANATIONAL

BUSINESS-CYCLE

ANALYSIS

In the presentpaper, as in the papercited,we shall workwithstate

data, but it may be indicated here

that states

are not what we

shall

regard n later work s "regionaleconomies."

A state generallywill be

found o includeparts of everal "regions" uch

as

we

should ike to

con-

sider

as

units

for

regionalbusiness-cycle

nalysis.

It

is

the "natural"

trade area

familiar o

marketing pecialists

that would seem

to

be the

more practicable representation f our unit. It is not surprising hat

buying

habits and commercial ontacts develop

that tie the business

units of an area to a particulartradingcenter. There must be an ag-

glomeration endencyhere ust as there

s in manufacturingndustry.

In the latter field, repair and machine-shop

facilities, echnological

skills, ommunication

nd

transportation

acilities hat

develop

to serve

one firm r industry

n

a given ocalityare available for

other

firms

r

industries. his is part of the working

f

the

principle

f

external cono-

mies. That is, many such facilities

re "lumpy" and discontinuous

n

size,

and

the

unit

of

optimum

ize

for

fficient

peration

s

larger

than

required

o

meetthe demands

of

one

firm

r

ndustry.

t becomes

easier

and less costly o establish firm r ndustrynan area after ther irms

and

industries ave been established.

Thus itmustbe withcommercial

and

marketing acilities.Banking,

brokerage,wholesaling,

nd

other

business services agglomerate

n

a

tradingcenter,through

which the

"export"commodities f

an area

are passed

to the "outside"

world nd

throughwhich the "imported" commodities-the

"exports"

of

other

regions-are drawn fordistribution

within he area. This is ilotto say

that

no interregional rade goeson except through

he

trading enters,

and we do not object to our boundariesbeing zones rather han lines.

We are saying merely hat the great

bulk of the products of certain

types of ndustry f a particular rea

are channeledthrough popula-

tion

concentration

hat

s regarded

y the nhabitants f the

area

as

the

trading nd financial enter.Of particular mportance re the banking

connectionswithinthese units. The

ebb and flow

of a

region's iquid

funds s

reflected n the reserve xperience f particularbanks,

whose

spatial sphere

of

operations

s

fairly

efinite.

Now if tweretrue thatall economic nterprisesf oneof these units

could be

neatlyclassed

nto those

employments

hoseproducts

re sold

and

consumed ocally

and

those whose

products re exported o other

areas,

a

relatively imple analysis ofregionaldifferences

n

cyclical

re-

sponsivenesswould be suggested. f it were found that each region's

''export' employment

was

confined

o

the

production

of

one

product

whose consumptionmarketencompassesthe entiregreater

rea

made

up

of

the

interlocking egions,

he

analysis

would

be

yet simpler.

rom


8/10/2019 1905703

4/33

REGIONAL PATTERNS OF BUSINESS-CYCLE

BEHAVIOR 39

one pointofview,the

home-market mploymentould be thought f s

employment hat s

auxiliary

o and a part ofthe export ndustry. ust

as independentmachine

shops in areas specializing

n lumberproduc-

tion could be considered s a part of the lumber ndustry, o mightwe

consider he bakeryestablishments nd retail

marketing ervices o be

auxiliary employment

o the lumber industry.

The existence of the

lumber ndustry n that

area would implytheexistenceof these other

enterprises, ut theconverseof this statement

would obviouslynot be

true.From this point

ofview,we could regard

hiswholeclass ofhome-

market ndustries r employments s unique

in that a home-market

industrymay be physically

ut not economically uite the same from

regionto region,beingclosely tied in each regionto the "export" in-

dustry. Thus, while

an export industryor

employmentmight show

regionaluniformityn

its behavior,being affected

imilarly egardless

oflocation,the cyclicalbehavior of home-market

mploymentwould

show more or

less

wide regionaldifferences,uch industries

n each re-

gion assuming certain

of the cyclical characteristics-in

more or less

diluted form-of the

export ndustry f that

region.The feature hat

would make for regional

differencesn business-cycle

ehavior would

consist argely ofthe varying ncomeelasticitiesofthe differentex-

port" products.

The demand for he "export"product of some regions

would be little affected y the business cycle

so that the purchasing

power

would be sustained with

ittle

decline during national

business

depression. The demand

for the "export"

products of other regions

would be very drastically

ffected y

the business cycle so

that

pur-

chasing power

in

such

regions

would

fall

greatly

n

national business

depressions

nd

rise greatly

n national

business

expansions.The

re-

gions that would be affectedmost would be thosewhose "imports"are

"necessities"that

would decline ittle

n

use

in

a depression nd whose

"exports"

are

durable,

storable commodities

whose demand would

fall

to

a low level n

a

depression.Under

these

conditions,moneypayments

to the rest of the nation would not be easily adjusted

to the

falling

money receipts

from he rest of the nation,

ocal banks would rapidly

lose reserves o other

regions, nd local purchasing

power would con-

tinue

to

decline until ultimately purchases

are

choked

off

to

equal

money receipts. f the reverse of these conditionsweresatisfied n a

given region,

hat

region would

show

relatively

mall

declines

n

na-

tional

depressions

and

would tend

to

gain banking

reserves

during

those times.

The above

is

a

sketch of a conceptual

framework

hat

has given

a

semblance of direction

o our statistical nquiries.

t is

obviously

not

true

that

a

particular

rea

specializes

n

but one

"export" product,

ut

by

and

large

t

will

be found

that

the

specialization

s

frequently uite


8/10/2019 1905703

5/33

40

RUTLEDGE

VINING

marked. t is

also

not true n general

that

"export"

productshave na-

tion-wide

markets,

ut

it

is

the

case

in

many

nstances,

nd

at least

the

structure

f these

regional

nterlocks

may

be investigated.

Finally,

as

was indicated,a state is not whatwe shouldregardas an "economic

region"

and will

generally

ncludeparts

or the

whole of

several

inte-

grated

tradeareas; and

the

behavior

n

the

businesscycle of

state

fig-

ures will

be a

sort

of

average

of

the

behaviorof the

figures f ts

com-

ponent

regions.

tates

comprised

f

economically imilar

egions

hould

show a

similarity f

behavior

n

the

business

cycle. Statistics

descrip-

tive

of industrial

ocation

by states

should

indicate how a

particular

state

mightbe

expected

to respond

n

a

bu.siness

ycle

and

should

help

explainobservablegeographical atterns fbusiness-cycleesponse.

2.

CERTAIN

RELEVANT

FEATURES OF

INDUSTRIAL

LOCATION

In

our study of

industrial ocation

we

have

made

an

adaptation

of

certain

echniquesof

Professor

argant

Florence.

On the basis of

occu-

pations statistics

or

GreatBritainand

the United

States,Florence

has

set

apart

a

class of

industry

hat

he

designates

as

"residentiary"

or

"ubiquitous"

industry2-residentiaryn the sense that

they

typically

residewhere the consumingpopulationresides.With reference o a

givenregion, he

industries

therthan

the

residentiary

nesare

called

the

"primary" ndustries.

ndustries

re

classified

nto one or the

other

of these

types

on

the

basis

of

a

measuredevised

by

Florence

nd

called

the

coefficient

f

localization.'

For

each

region

a

"location factor" s

computedfor

ach

industry y

obtaining he ratio of the

percentage f

total

employment

n

the

given

region

ound

n the

given

ndustry o

the

same

percentage

for

the nation

as a

whole.

If

the

industry

were

per-

fectly niformlyistributedmongall theregions ll theregional oca-

tion factorswould

be

unity.

f

the

industrywere

highly ocalized

in

a

very few

regions,

hese

few

regionswould

have

high location

factors

for

that

industry,

nd the

rest

of

the

regionswould show

location fac-

tors that are small or

zero.

The

coefficient

f localizationfor a

given

industry

s obtained

by

computing

the

weighted

avergagedeviation

from

nity

of the

ocationfactors

or

ll

the

regions,

he

weight or

ach

region

being

the

proportion f total

national

employment

ound

n

that

2

A. J. Wensley nd P. SargantFlorence, Recent IndustrialConcentration,

Especially nthe

Midlands,"

Review

f

Economic

tudies,

Vol. 7, June,

1940,

pp.

139-158;

P. Sargant

Florence,

Memorandum

n

Long

Range

Planning," Mim-

eographedRelease

of

the

National

Resources

Planning

Board, 1940,

pp.

14-30;

P. Sargant

Florence,

Geographical

Distribution f

Economic

Activity

mong

Broad

Industry

Groups,"

Mimeographed

Release of the

National

Resources

Planning

Board,1940,pp.

1-24.

By

"industry" s

meant

nyspecific

ypeofeco-

nomic

production o

thatthe term

ncludes

griculture

nd trade.

3Florence, "Memorandum

n

LongRange

Planning,"pp.

19-25.


8/10/2019 1905703

6/33


BEHAVIOR 41

region.This measure divided by two varies

between zero and unity.4

The coefficient f localization measures

complete uniformity f geo-

graphicaldistribution s zero As greater nd

greater eographical on-

centration s encountered, he measureapproaches unity. The coeffi-

cient for

automobile repair shops

in

1930 as

computed by Professor

Florence using state data, for example, was

0.068. The coefficientor

automobile manufacturewas 0.597.

Automobilerepair shops are rela-

tivelyuniformly istributed ver the nation.Automobilemanufacture

is highly ocalized.

I

Identical

results re obtainedby "adding

the plus differencesor, since

they

total the

same, the minus

differences)

f the local percentage

f

workers

n

the

given ndustry rom he ocal percentage f workersn all industry."bid., p. 19.

These measures

may be

presented ymbolically s follows:

Let E=Total

gainfully mployed n Nation,

Ea= Total employedn Industry

in

the

Nation,

ei=Total

employedn Region

1,

ela=Total

employed

n Region 1 in Industry

.

The location

factor f

ndustry in Region 1 is

defined s

eia

el

E.

E

The coefficient

f ocalization s computed

s follows:

(1)

(2)

(3)

Location

Weighted

Deviation

Factor

from

Unity

ela

ela

Ea

el

el

ei ela

1 _

E-

1 - _ = -

ei

E Ea E E Ea

**

........... ..

...............................................

**

.

.

. . .

.

.

. .

.

.

. .

.

.

. .

.

. .

. .

.

. . . .

.

(

ee

en en

ena

ena

El 01

E-

E

E

a

n

en

E E

CoefficientfLocalization- . . . . . . .

nI

e; eia

E

Ea

The algebraic

um is obviously

ero, ndicating he

equality of the

positive nd

negative

deviations.The absolute sum,

disregarding

igns, variesfrom ero

to

two. f in

each region

he two

percentages

re

identical-i.e.,

if

the ndustry

were

perfectly

niformlyistributed-then

he sum of the

deviations s zero. If all

of

the

given ndustry

s

located

n

a single

egion, nd

if

regions

nd industries

re so

defined

s to be ofnegligible

ize relative

o the total economy, hen

the absolute

sum

of the

deviationswill

approachtwoas a limit.


8/10/2019 1905703

7/33

8/10/2019 1905703

8/33


BEHAVIOR

43

thesefeatures f ndustrialocation. or thispurpose,we arrayed or

each ndustriallassificationhe tatefiguresor hepercentagef

otal

employmentoundn that ndustry.6or eachof hese rrays herange

limits f ach quintilenterval ere ocated, nd these ange imits nd

the median igure ere hendividedby the ppropriate ational er-

centage. he results fthisprocedurere shown n Table 1. This

table

provides ertain oints n the rray f tate ocation actors or ach n-

dustrial

lassification,nd these atter re arrayed rom op

to

bottom,

in theorder fwhat ppears o be thedegree f ocalization. he table

shows pproximatelyhat which lorence's oefficientf ocalization

shows n a muchmore ondensed orm,nd for ts awkward ulkiness

it may possibly fferompensationn theform f an enhanced isual

impressionfthecharacter f ndustrialocalization.

For ourpurposes, hesedata are fraught ith hortcomings,nd it

may be instructiveo refer o the mplicationsf a fewof thesede-

ficiencies.

f

the

subareasunder

tudy

were

approximately

f

equal

"economicize" and small elative o the otal rea,

f

he

tandards

f

living nd consumerasteswere pproximatelyniforms between e-

gions, f ndustriallassificationserefine nough rawn n order o

provide nternal niformity,henwe might ossibly ationalizen ex-

pectation or certain eneral attern o be exhibited y

the

columns

in this able, he ndustrieseing rrayedn order fdegree f ocaliza-

tion. The

market reas of differentroductswould obviously

how

widedifferences.upposeweshouldmeasurehe ypical

market

rea

of

a

product y some veragedistance etween he point

t

which

he

value s added and thepoint ffinal onsumption. frequencyistri-

bution f

hese

ariates, ypical

istance or ach

defined

roduct,

ould

beconstructed,ut twould edifficultoanticipatehegeneral hape

of his

istribution.

f

t were

U-shaped

t

would

mean hat

relatively

large roportionfproductsre sold t a point lose o their espective

points

f

production,relativelymallproportion

t intermediateis-

tancesfrom hese

points, nd another elativelyargeproportion

old

in all

corners fthenational rea. If t were single-humpedistribu-

tion,

t

wouldmean hat here sa concentrationoint or

market

rea

distances nd

that

extreme istances

n

either ide of

the

concen-

trat

n become rogressivelyess frequent. hose products,

r

value-

additions,whose market reas encompass he entirenationmust

of

ourse e the

highlyocalized

ndustries-some

harp roduction

d-

vantage eing ound t certain oints

r centers ithin

he

arger

rea.

Around nd

amongthose primary lusters

will have

been

built

up

industries ith rade reasof ntermediateength,nd with

ach

popu-

6

The data analyzed

werefrom

he Sixteenth

ensus

oftheUnited

tates:1940.

The Labor

Force-Employment,

Unemployment,

ccupation,

ncome,by States,

Vol. IV.

Washington,

overnment rinting

Office, 941.


8/10/2019 1905703

9/33

44

RUTLEDGE

VINING

TABLE

1

RATIOS

OF

PERCENTAGE

EMPLOYED

IN

STATE

TO

PERCENTAGE

EMPLOYED

IN

NATION

BY

INDUSTRIES

IN

1940

20%

40%

50%

60%

80

%

All

All

No

Three

of

of

of

of

of

But

But

Industrial

Classification

States

States

the

the

the

the

the

Three

One

Less

Less

States

States

States

States

States

States

State

Than

Than

Less

Less

Less

Less

Less

Less

Less

Than

Than

Than

Than

Than

Than

Than

1.

Autos

and

Auto

Equipment

........

.....................

0.00

0.00

0.00

0.08

0.08

0.15

0.36

0.62

15.00

2.

Coal

Mining

.

..........................................

0.00

0.00

0.00

0.00

0.00

0.33

1.25

4.00

17.00

3.

Crude

Oil

and

Natural

Gas

Products

......

...............

0.00

0.00

0.00

0.00

0.00

0.25

2.50

5.00

11.25

4.

Leather

and

Leather

Products

.......

...................

0.00

0.00

0.00

0.13

0.13

0.38

1.13

3.25

15.38

5.

Transportation

Equipment

(Except

Autos)

.....

...........

0.00

0.00

0.00

0.14

0.29

0.43

1.14

3.00

3.57

6.

Non-ferrous

Metals

and

Products

......

..................

0.00

0.00

0.00

0.33

0.50

0.83

1.50

3.00

10.33

7.

Iron

and

Steel

and

Their

Products

.......................

0.00

0.04

0.07

0.21

0.25

0.50

1.25

2.25

3.14

8.

Machinery

Mfg

........................................

0.04

0.04

0.08

0.21

0.21

0.50

1.29

2.08

3.04

9.

Textile

Products

and

Apparel

Mfg

........................

0.00

0.00

0.07

0.23

0.28

0.51

1.70

2.42

5.28

10.

Paper

and

Allied

Products.

.............................

0.00

0.00

0.14

0.43

0.57

0.71

1.29

2.71

7.14

11.

Chemicals

and

Allied

Mfg

...............................

0.00

0.00

0.20

0.40

0.50

0.80

1.20

2.20

9.00

12.

Petroleum

and

Coal

Products

.

..........................

0.00

0.00

0.25

0.50

0.50

1.00

2.25

3.25

4.25

13.

Logging,

Sawmils,

and

Planing

Mills

......................00

0.08

0.31

0.62

0.62

1.39

2.39

2.92

8.69


8/10/2019 1905703

10/33

REGIONAL

PATTERNS OF BUSINESS-CYCLE

BEHAVIOR 45

TABLE

1

(concluded)

14.

Stone,

Clay

and

Glass

Products

........

..................

0.00

0.14

0.43

0.57

0.57

0.71

0.86

2.29

4.71

15.

Other

Mines

and

Quarries

..........

.....................

0.00

0.25

0.50

0.75

0.75

1.00

3.00

13.25

37.50

16.

Furniture

and

Misc.

Wooden

Goods

......................

0.00

0.13

0.38

0.75

0.88

1.00

1.50

2.13

2.75

17.

Agriculture

...

........................................

0.01

0.17

0.57

0.92

1.28

1.61

1.94

2.60

3.12

18.

Printing

and

Publishing

and

Allied

Mfg

...................

0.21

0.29

0.50

0.64

0.64

0.86

1.00

1.36

2.21

19.

Finance,

Insurance,

and

Real

Estate

......

...............

0.25

0.41

0.53

0.63

0.75

0.84

1.06

1.47

1.88

20.

Domestic

Personal

Service

..............................

0.39

0.46

0.69

0.79

0.87

0.92

1.35

1.69

2.08

21.

Amusement,

Recreation,

etc

.............................

0.33

0.56

0.67

0.78

0.78

0.89

1.00

1.33

3.00

22.

Food

and

Kindred

Products

Mfg

.........................

0.33

0.46

0.50

0.75

0.83

1.04

1.29

1.38

1.67

23.

Communications

...............

........................

0.33

0.44

0.67

0.89

0.89

1.00

1.11

1.33

1.56

24.

Government

...........................................

0.47

0.67

0.83

0.97

0.97

1.08

1.33

1.86

2.44

25.

Wholesale

Trade

................

.......................

0.44

0.56

0.67

0.82

0.85

1.04

1.22

1.37

1.63

26.

Utilities

..............................................

0.42

0.50

0.75

0.92

0.92

1.08

1.25

1.33

1.50

27.

Construction

...........................................

0.44

0.70

0.87

0.96

0.98

1.07

1.15

1.39

1.61

28.

Transportation

.

........................................

0.44

0.56

0.79

1.00

1.04

1.10

1.23

1.38

2.02

29.

Other

Personal

Service

(Laundries,

Hotels,

etc.)

.

..........

0.57

0.70

0.78

0.89

0.92

1.00

1.14

1.38

1.70

30.

Business

Serv.

and

Repair

Serv.

(Including

Auto

Repair).

.

0.47

0.58

0.84

1.00

1.00

1.11

1.16

1.21

1.42

31.

Professional

and

Related

Services

.......

.................

0.58

0.69

0.89

1.00

1.01

1.07

1.16

1.25

1.33

32.

Retail

Trade

.

.........................................

O.50

0.64

0.79

0.94

0.99

1.03

1.09

1.14

1.25


8/10/2019 1905703

11/33

46

RUTLEDGE VINING

lation aggregate

will

be

the

service and

other

ndustries

with

the

ex-

tremely

hort rade

radii. Thus,

an ideal Table

1,

t

might

e

supposzd,

would

show the most

highly

ocalized ndustries s

lacking ltogether

n

many regions.Onlya fewregionswould show as greatrelative mploy-

ment

n

these

ndustries s

does the

nation,

nd

a

very

few

would

show

a

relativelyhigh proportion

f their

employment

n these

industries.

The products

of these

ndustrieswould be import

products

for

nearly

all

of the

regions

f the nation. As the industries ecome less

localized

(as

the

typical

trade radii

become

smaller),

fewer

regions

would

lack

the

industries

ltogether,

more

regions

would move into

the

exporting

class,and the ocation factors n thespecializing egionswould become

smaller.The "ideal" Table 1, that is, might uggesta surface.Thus,

starting

with he

upper

eft-hand

orner

with

he

most

highly

ocalized

industries)

nd

movinghorizontally cross the cumulativedistribution

of

regions,

his surfacewould move

along

the zero line

leaving

t

rela-

tively

ate but

finally limbingvery rapidly

to

a

height

sufficient

o

make

the area underthe

traced-out ine

equal

to

unity7-the

atter

on-

dition

lways being

the case.

For

the next most

highly

ocalized

indus-

try,

the

surface would leave

the zero

line

slightly

arlier and

climb

somewhat ess rapidlyto a heightsomewhat ower than the industiy

just preceding.

At

length,

we

should exhaust the list

of

ndustries hat

are entirely acking

n

at

least

one region,

nd as we

move

down the

array

of

ndustrieswe should

approach

that ist

of

ndustries hat

would

be

represented y

a

height

of

unity

over the

entire

range

of

regions.

That

is to

say,

the

first olumns would show

a

continuous

rise

to

the

neighborhood

f

unity.

Then for

a while the columns would rise to

a

maximum

nd

then

decline

to

the

neighborhood

f

unity. Finally,

the

columnswoulddeclinecontinuouslynd approachunity.

The

conditions o

which the data in the actual Table

1

are

subject

diverge

n

great

measure from hose

outlined

n the

second

sentence

f

the

preceding aragraph.

The states

showvast

differences

n

size

so that

industries

ighly

ocalized

n

large states

show

much smallermaximum

locationfactors

han

ndustries qually highly

ocalized

in

small states

Average income, iving standards, and consumerhabits differmpor-

tantly

from

region

to

region

so

that

employment

hat

is

necessarily

residentiarys of varying relative importancefromregion to region.

The industrial lassificationsre verybroad and obviously ack internal

homogeneity

o

that in

certaincases the

location factors re rendered

quite ambiguousby

the

nclusion f uniformly istributed

mployment

7

If

the

regions

were

ofequal size

in

terms f

employment,

he

simplemean o-

cation

factor

would

be

uniity.

hat

is,

the mean

ordinate

nder

he

curvewould

be

one,

and the base

of the

figure

would

extend

from ero to 100

per

cent.

The

mean

ordinate

would

also

be

unity

f

the regionswere

not of

equal size, provided

there

s

zero correlation

etween ize

and the

magnitude

f

the

ocation

factor.


8/10/2019 1905703

12/33

REGIONAL PATTERNS

OF BUSINESS-CYCLE

BEHAVIOR 47

and highly ocalizedemployment

within he

same industrial

lassifica-

tion. Nearly

all

of our-ndustrial lassifications

ould show n

some de-

gree this ambiguity.

But the industries anked

fourteenth,

ifteenth,

seventeenth,nd twenty-secondn Table 1 are especially outstanding

instances of internalheterogeneity.

lay and

brick-making

material

forregions

s

large

as states

is almost "ubiquitous"

materialand the

heavy transportation

harges

result in certainsubclasses of

"Stone,

Clay,

and

Glass Products" manufacturing eing

residentiary.

n the

other

hand

glass

manufacturing ppears to be relatively

highly o-

calized.

Nonferrous-metals ining

s

very

highly ocalized,but other

subclasses

of

"Other Mines

and

Quarries"are residentiary.

Agricul-

ture" is no more a single ndustry han is "Manufacturing," nd the

production

of certain extremely mportant

agricultural products

is

highly ocalized,

while certain

other subclassesare residentiary.

uch

"Food

and Kindred Products"

manufacturies

s bakeries and ice

plants are residentiary,

ut other

ubclasses showhigh ocalization.

Nevertheless,

able

1

showsroughly

he characteristics f

our hypo-

thetical surface.

Were we to omit the ambiguous

classifications

oted

above,

the

similarity

would be almost

striking, onsidering

he

other

discrepanciesn the underlying onditions.The degreeoflocalization

shades off

gradually

from he

employments

hat are

virtually

nde-

pendent

of ocal

demand,

through

hose

that

produce exportable

prod-

ucts for

which the

local

demand

is

not

negligible,

o those

finally

hat

are

virtually

whollydependentupon

local demand. From

this

table a

rough

lassificationmay

be

made and a

fairly ood conception

may

be

obtained

of

the relative

mportance

f the "carrier" ndustries

within

given

state. Referring

o

the

table,

we

may

set

apart

the

following

n-

dustries s the "passive" industries.At therightof each industry he

median

state

percentage

f total

employment

s

given:

Retail

Trade ..............

.

.................

13.8%

Professional

nd Related

Services

..............

7.4

BusinessService

and

Repair

Services

...........

1.9

Domestic

Personal

Service

....................

4.5

Other

Personal Services

Laundry,

Hotels, etc.).

.

.

3.4

Transportation

...........................

5.0

Construction

................................

4.5

Utilities..........1 ..

1

Wholesale

Trade

..........

2.3

Government

................................

3.5

Communications

..........

0.8

Food and

Kindred

Manufacturing

.............

2.0

Amusement,

Recreation,

tc

...................

0.7

Finance,

Insurance

and

Real Estate

............ 2.4

Printing

nd

Publishing

......................

0.9

54.2%


8/10/2019 1905703

13/33

48

RUTLEDGE

VINING

TABLE

2

Q

UINTILE

INTERVALS*

IN

THE

ANNUAL

ARRAYS

OF

ANNUAL

STATE

INCOME

PAYMENTS

EXPRESSED

AS

PERCENTAGES

OF

YEAR

PREVIOUS

Yer

1st

Quintile

2d

Quintile

3d

Quintile

4th

Quintile

5th

Quintile

Year

interval

interval

interval

interval

interval

1930

69.03-

82.64%

82.90-

86.27%

86.35-

90.38%

90.94-

92.96%

93.38-101.24%

1931

66.81-

81.23

81.51-

83.38

83.66-

85.33

85.45-

88.23

88.50-

95.79

1932

59.91-

73.75

74.19-

75.85

76.26-

77.53

77.71-

80.76

80.82-

88.62

1933

89.90-

96.30

96.44-

97.38

97.71-100.61

100.67-103.44

104.32-113.77

1934

92.80-111.29

111.59-112.76

112.98-119.35

119.45-124.21

124.31-136.53

1935

105.88-109.04

109.61-110.31

110.42-112.46

112.81-114.53

115.43-145.81

1936

104.95-113.11

113.24-115.53

115.59-117.58

117.82-119.84

120.41-124.65

1937

96.18-103.14

103.25-104.72

105.24-106.93

107.00-108.46

109.04-115.31

1938

83.15-

89.26

89.55-

91.38

91.51-

93.75

93.93-

95.49

95.63-101.40

1939

99.85-104.99

105.12-106.46

106.50-107.34

107.41-108.87

109.13-117.57

1940

102.73-105.57

106.18-106.87

107.12-109.07

109.24-110.53

110.89-121.13

1941

113.42-116.29

116.93-119.38

119.53-122.23

122.68-126.41

128.35-140.24

1942

108.58-116.84

120.11-127.48

127.87-130.36

131.79-142.81

146.16-166.72

*

These

are

not

in

all

strictness

quintile

intervals.

Actually,

the

object

was

to

divide

the

states

of

each

annual

array

into

five

equal

groups,

and

these

intervals

represent

the

limits

shown

by

these

groups.

Some

discretion

was

exercised

where

the

variation

between

states

at

these

limits

was

so

insignificantly

small

as

to

render

arbitrary

the

placing

of

a

state

in

one

"quintile

interval"

rather

than

in

the

following

one.

For

example,

of

the

65

groups,

7

contain

but

8

states

and

7

contain

12.

One

group

contains

but

7

states.


8/10/2019 1905703

14/33

REGIONAL

PATTERNS OF BUSINESS-CYCLE

BEHAVIOR

49

The

percentages

hown

represent, f

course,

the central

tendency

around

which

the

figures

or

he

various states are

dispersed.

f

a

given

state's

dispersion

round these central tendencies s

random,

then

we

may say

that in the

neighborhood

f

55

per

cent of a

state's total

em-

ployment

will be

found

n

these

industries hat are

pretty

definitely

residentiary

r

passive.

Actually,

there s

good

reason

to think

that a

state's deviation fromthe central

tendency

s not

random and

that

those states

with the

higherper

capita

income

tend

to fall

to the

right

of the central

tendency,

o that the

aggregate

percentage

n

this

ist

of

employment

would bear a direct relation

with the standard

of

living.

As

the

per

capita

incomeof a

region

risesthere

eems to be a

tendency

to allocate more of its resources o service ndustries.But the55-per-

cent

figure

ives

an

approximate

onception

f the

relative

mportance

of

these

residentiary

ndustries n a

state's total

employment.

o

this

figure

hould be added

a certain

proportion

f total

employment hat

is

found

n classifications

ominally

primary

ut

that

include

employ-

ment

that

produces

for ocal demand.8

For

example,

high

proportion

of

agricultural

ncome comes from

dairyproducts,

poultry, ruck,

nd

livestock

sold for ocal

consumption.

Certain

lumber

products

are lo-

cally consumed, nd it would doubtless be found that a considerable

proportion

f

a state's

manufacturing

s

for trade

area

that

is

essen-

tially

ocal.

Suppose,

as a

roughguess,

we

should

say

that

one-half

he

employment

n

agriculture,

n

textiles nd

apparel,

in

lumber,

nd

in

furniture

nd miscellaneouswood

products

are

engaged

in

producing

for uch

a

local

trade area. If we take the

median

percentage

mployed

in

these

industries

s

the

typical

state

employment,

his

would

add

nearly

15

per

cent to

the 54

per

cent

and

would

indicate

that

typically

about 70 percent of theemploymentna statewould be ofthepassive

type

while

about 30

per

cent

would

be

primary

r

active

or

"carrier"

8

It should

be remarked

hat

in

the

cases

of

some states

much

of

the

employ-

ment

n

industrial lassificationshat are

nominally assive

s in

fact

dependent

upon

"external"

demand. For

example,

ertainof

the

sparsely

populated moun-

tain

states,

such as

Nevada,

Montana,

Wyoming,

nd

Utah,

have

a

quite

high

proportion

f heir

mployment

n

transportation.

his,

presumably,

s

because

of

the trunk

ast-west

ransportation

ines

passing

through

hese

states.

Also,

the

highproportionfemploymentn retailtrade,amusements, inance nd insur-

ance, hotels,

t

cetera,

n such

places

as

New York and

Massachusetts

s

in

con-

siderable

measure

dependent

pon

ncome

hanges

xternal

o these

areas.

These classifications re

ambiguous

in

another

respect. Some

industries,

though

bearing

he same

name n

the various

places,

will

n

fact be

partsof

dif-

ferent

ndustries,

nd an

industry

hat

is

passive

in some

regions

may

be in

an-

other

egion

n

auxiliary

f

a

primaryndustry.

or

example,

good portion f

the

utilities n

Michigan

s an

auxiliary

f

the

automobile

ndustry.


8/10/2019 1905703

15/33

50

RUTLEDGE

VINING

in

the sense that the demand for ts products s external o the

state.9

Thus, were ll

states "typical" in thisrespect,

henatureofthe

products

of around one-third

f the employment

f a statewould determine

he

sensitivity fthat state to nationalcyclicalforces.

With

the above generalizations

egardingndustrial ocation

n

mind,

we may approach our principal

problem. Our

questions now

are: Do

states fall systematicallynto

groups of states

showing significantly

160-

120-

110

100 /_

60-

70-

60-

50-

1930

1931

1932

1933

19'3'

1935 1956 1937 1936

1T99 19'0

1911

1912

FIGURE 1.-Ranges

of quintile ntervals n the annual

arraysof annual

state

incomepayments xpressed

s percentages

f year previous.

9

Florence

estimates

smaller

figure or the residentiary

omponent

f em-

ployment.Calculations

for he whole

UnitedStates ofthe

occupiedpresons

hat

must ocate

close to the

consumer,

ncluding onstructionnd

some manufactur-

ing, show a proportion

f 48% of

the total of occupied persons.

But

for im-

ited areas the necessaryproportions undoubtedly ower.... We may take

35% of

all

workers s

a reasonable

minimum or service

workers.

Adding

3%

forbuildingworkers

s a similarminimum, he

minimum esidentiary

ompon-

ent of all

workers ppears

as 38%."

Op.

cit.,

p. 16. In this

instance

Florence

is dealing

withan area

smaller hana state, and,

of more mportance,

isprob-

lem is differentrom

hat with

which we are

concerned.He is attempting

o

anticipate heminimum

number f

erviceworkers

equired or he establishment

of

a

given

primary ndustry.We

are attempting

o estimate he typical propor-

tion

of

employment roducing

oods

sold

in

the

ocal

market.


8/10/2019 1905703

16/33

REGIONAL

PATTERNS OF BUSINESS-CYCLE BEHAVIOR 51

similar

behavior

with

respect

o rates ofchange of ncome? fso, do the

primary ndustries f. he

states within given groupprovidea ration-

ale

forthis similar

behavior?

3.

REGIONAL

PATTERNS OF CYCLICAL FLUCTUATIONS

This problem f

classifying egions nto groups howing nternaluni-

formity

f

cyclical

behavior

s,

so

far

as we

know,

a

problem

hat

has

not been studied

extensively.'0 t is

a

problem,however,

that would

seem to introduce nteresting nalogies with other fieldsof

analysis.

Just as

in all

problemsof

the analysis of variance, we have herea set

of variates that

may

in

principlebe classified

n

accordance

with

cer-

tainattributes.On thebasis of uch classifications,hevariationwithin

groups of regionscould

be compared with the variation

between the

groups uch that patterns

f cyclicalbehaviormay be discerned f they

exist.The

systems f

lassification

hen

may yield dditional

nsight

nto

the process nvolved n

the nterregional iffusion f ocalized economic

shock.

As a

matter f fact, he

methodof study

that

we

have

adopted up

to

the present evelopment f our nquiry s onlyanalogous to an

analysis

ofvariance na quite rough nd backhandedmanner.The data that we

have now at hand wouldhardly ustify n elaborate analysis,but a sim-

pler method can at least

suggest an outline of furthernquiry. Table

2 and

Figure

1

show by years the quintile

intervals

of

state

income

payments expressed s

percentagesof the precedingyear."

We have

first ndeavored to find

out if the shifts mong the

occupants of these

quintile

ntervals

give evidence

of

a

regionalpattern

of behavior

as

the

different hases of the

business cycle develop. These intervals repre-

sentsegments fthehorizontal cales of ourfrequency istributions f

state

percentage hanges,

nd

if

the

individual

states

appear

to

be

dis-

tributed

t

randomfrom

year to year

within

hese ranges

then

we

may

10

Withoutmaking pretense f documenting he discussion f this problem

of regional variation of economic fluctuations, e should at least like to call

attention o two sets of stimulating rticles n this field:the papers by

D.

G.

Champernowne,

The

Uneven Distribution

of

Unemployment

n the

United

Kingdom, 1929-36," Review f EconomicStusdies, ol. 5, February, 1938,

pp.

93-106 and Vol. 6, February, 1939, pp. 111-124; and the articles by H. W.

Singer, The Process of Unemploymentn the Depressed Areas (1935-1938),"

ibid., Vol. 6, June,1939, pp. 177-188, and "Regional Labour Markets and the

Process of Unemployment,"bid., Vol. 7, October,1939, pp. 42-58.

11

The data used in the computation fthesepercentages re the same as those

used in

the analysis described n the cited article n the July, 945,ECONOMET-

RICA. The estimates of state income payments for 1929 through 1939 are

those

published nd discussed n the Survey f Current usiness, Vol. 22, 1942,

pp. 18-26; the data for1940 through 942 are from he June, 943, ssue,Vol. 23,

pp. 10-22.


8/10/2019 1905703

17/33

52

RUTLEDGE

VINING

consider hat the

data

evidencesno such

pattern

s we have

in

mind.

But if states

from

year to

yearmove in

groups

withinthe

frequency

distribution,

nd ifthere s

evidence of

characteristic

roupings

within

thedifferenthases of thebusiness cycle,then a patternwouldbe at

hand

that would

provide a

basis

forfurther

tudy.

Our method of

classifying he

states was as

follows:

For each state the

number

of the

quintile nterval

lowestquintile

designated

as

first, ighest

s fifth)

that

included

the

percentage

figure or

that state was

listed

for each

year. States

that

year

afteryear

fell nto

the same

or

adjacent

quin-

tile ntervals

were

grouped ogether.

ecause ofthe fact that

only nine

or ten

states

can

occupyany given

quintile

nterval nd

because of

the

verynarrow imits f manyof thequintile ntervals, ertaindeviations

from

he

apparent

grouppattern

were

gnored n

grouping he

states.

The

procedurewas

first o

set apart certain

tates that

appeared

to the

eye

to conformmost

consistently o a

pattern

and

then to set

apart

certain

tates whose

patternsbear a

strong

amily

esemblance

o that

of

the

group

exceptthat

occasional

marked

discrepancies ccur.This

procedure s

merely

short nd

rough

methodof

studying he matter

of

whether

he

variance

indicatedby

the variation

within

groups is

smallerthan that indicatedby thevariationsbetweengroups. Should

a

group of states fall

in

the same

quintile nterval

year

after

year,

it

would

mean

that one

quintile

interval s sufficient

o

include all

the

variation

ofthis

group of

states.'2

In

Group

we include

Arkansas,

Mississippi, nd Alabama, and it

is

clear

that

Tennessee and Georgia

show

strong

family resemblances.

With

somewhat

more

discrepanciesthe

remaining

tates

of

the

Old

South

with the

exception of

Florida show

a

family

resemblance.

The

quintilepatternsforthisgroupare shown n Table 3. Considering he

relatively

narrow

ranges ofthe

quintiles,

hesestate figures how a

re-

markable

onformance o a

pattern.

Among

he

first

hree

tates,

t can

be said

that there re

few

ignificant

iscrepancies.

n

1938,

Mississippi

in

the

second

quintile

nterval ppears

relatively

ow compared

to

the

other

wo states

which re

within

percentagepoints

of each other

n

the

upper

part of

the

array. Yet

Mississippi

n

this

year

is within

3

12

The

range

of the

component

roupcompared

with

the

range

of

the

total

tellsus notmuch,ofcourse,unless we have someconception fwhat we should

expect t to

be,

assuming

homogeneity f our

variates.

But from

what s

known

about

the

relationbetween

range and

size of sample drawn

from normal

population, he

range

exhibited y the

48 states

should

not be

much more han

1.5 to 2

times

the

range within

component

roupcomposedof from to

10

states.

We

make no

effortt this time,

however,

o developsuch

a criterion.

A

possible

alternative imple

methodof

classification

might

onsist

of com-

puting or ach

year

and for

ach statethe

deviations rom he

mean

n

terms f

standard

unitsand

arranging

hese

deviations

nto

groups.


8/10/2019 1905703

18/33

REGIONAL

PATTERNS

OF

BUSINESS-CYCLE

BEHAVIOR

53

TABLE

3

QUINTILE PATTERNS

OF

THE STATES WITHIN GROUP

I

Year Ark. Miss. Ala. Tenn. Ga. Ken. La. N.C. S.C.

1930

1 1

1

1 1

1

2

1

1

1931

2

1

1 2

2

3 4 3

4

1932

5

5 4

4

5

3 4 5

5

1933

3

3

3 4

4

3

2

5

5

1934 4

5

5 5

5

3 4 4

5

1935

4

3 3

2

3 4 1

1 1

1936

5

5 5

4

2 4

3

1 4

1937 1

1 1

3

1

3 3 3

2

1.938

5

2

3 3

4

2 5 4 4

1939

3 5 2 4 2 3 2

4

5

1940

1 1

4 4 4

2

1 1

5

1941

5

5

5

5

4 2

3 4

3

1942

5 5

4

2

4

3 3 4

4

pointsof

Alabama. In

the

following

ear,

1939,whileArkansas nd Ala-

bama

are within1

percentage oint of

each

other,

Mississippi s

about

5 points

bove these

other wo.

In

1940,Arkansasand

Mississippi

tay

together, ut Alabama is about 6 pointsabove the others.The restof

the

states

show

discrepancies

fromtime

to

time, yet all

are

clearly

pulled toward

the same

general

vicinity

within

the

array,

nd

in

the

cases of

most

of the

discrepancies

reference o the

percentage

figures

will

show a

close

similarity. n

1931,the

declines

registered

y

the first

three tates

were

omewhat

more

marked han in

the

rest of the

group.

In

1933,North

and

South

Carolina

were

completely ut

of ine

showing

rises

of some 12

per cent while

the

others of the

group were

showing

practicallyno

change in

income. The discrepancies n 1935 are not so

TABLE 4

QUINTILE

PATTERNS OF

THE

STATES WITHIN

GROUP II

Year

N.Y. Mass.

R.I.

N.J. Vt.

N.H. Me. Md.

Mo.

1930

4

4

4

5 3

4

5

5

3

1931 4

5

5 4

4

5

5

5

3

1932

3

5

4 4

3 3

4

4

3

1933 1 1 1 1 1 3 3 2 2

1934 1

1 1 1 1

2

1 2 1

1935 1

1

1

1 2 1 2

1 3

1936

1

1

1 2

2 1

1

2

2

1937

2

2 2

3

1 2

1 4

2

1938 3

3

2

3

2

4 3 4

3

1939

2

2

4 4 4 1

3

3 3

1940

2

2

2

4

2

1

2

5

1

1941

1

2

4 2

1

1 3 4 4

1942 1

1 1

2

1 1 2

3

2


8/10/2019 1905703

19/33

54

RUTLEDGE

VINING

great as the quintilefigureswould

suggest, range of 5 per

cent being

sufficiento include all-the

tates of the

group. n 1936, the

first hree

states of the group

showed somewhat argergains than the

others.

n

1940, certain f thegroupgot out of ine,Mississippi ccounting or he

extreme ow of the

entire rray of states

and South Carolina beingal-

most the extremehigh.But even here,

range of 10 per

cent would in-

clude

all

the

states

and a

rangeof ess than 5

per cent would

nclude

ll

except

the

extremes.Withthese

qualifications,t can be said that this

group of states follows ngeneralthe

same pattern f

behavior.

In

Group

I are

placed New

York,Massachusetts,Rhode

Island,

and

New

Jersey,

nd

a

strong amily esemblance s

in

evidence n

the cases

of Maine, New Hampshire, and Vermont.The pattern of Maryland

in some

respectsresembles

he

patternof

this group, but it also

ap-

pears

to

relate tselfwith the

pattern

of

another

group

that will be

de-

scribed

ater.

Missouri

we

are

tentatively lacing

with this

group,

al-

though

he

basis forthis classifications

somewhat light.

The pattern

for

his

group

s shown

n

Table

4.

Here, too,

the

conformance

o a

pattern

ppears

to

be quite

evident.

In

the

case

of

the

first our tates there s

no significant

ivergence rom

thegeneralpattern. n 1932 New Yorkwas in the third uintile nterval

whileMassachusettswas

in

the

fifth,

ut New York was in the high

part

of the

third

while

Massachusetts was

in

the low

part of

the

fifth.

While

the

range

of the

rates of

change for ll states

in

this

year

was 28

per

cent, pproximately per

cent was

the difference

etween he

rates

of decline

of these two states.In

1939,

divergencies

were

apparent.

But

they

seem to

be

unimportant, he rate of increase of New York and

Massachusetts n the second

quintile

nterval differing y less than 3

percentagepointsfrom hat of Rhode Island and New Jersey n the

fourth.

n

1940,

New

Jersey

s

about

4

points above

the

other

three

states,

and

in

1941

Rhode Island

appears to have received

a

greater

impetus

than

the

other states.

Maine,

New

Hampshire,

nd

Vermont

how only lightdiscrepancies

from he group pattern.

Reference o

the percentagefigures or these

states indicates a rather

close

conformancewith the

group pattern

throughout

he

years.

The patternofMaryland conforms oughlywith that ofthisgroup,

but

it

shows evidence of

being pulled

in

the direction f the

pattern

of

the

group

to

be

next

described, speciallyfor helatterfive

years.

Missouri's s

a

more

tenuous

classification.When thepercentage ig-

ures forthis

state are

placed along side

of those of the

other states

of

this

group,

the

conformance

s

very close. But forreasons

that

will

be

presented t a later

point in our

discussion t appears likely that the

grouping

will

follow

geographical ineswhen such large

areas as states


8/10/2019 1905703

20/33

REGIONAL

PATTERNS OF

BUSINESS-CYCLE

BEHAVIOR

55

are

regarded

s the

"regions."

When the

behavior

ofmore

appropri-

ately

defined

regions"-is

nalyzed

there

s

evidence

that the

satisfac-

tionof

certain

conditions

will

result n

groupings

hatfollow

ndustrial

lineswith ess regardforgeographical ontiguity. n the case ofMis-

souri

the

conformance o

this

pattern

should

perhaps be

explained

on

special

grounds, nd

these

groundswill

be

presented

when the

attempt

is

made to

rationalize he

group

patterns.

In

Group

III we

place

Indiana,

Ohio,

Michigan,

llinois,and

Penn-

sylvania.

Connecticut

nd

Delaware bear

a

strong

amily

esemblance.

West

Virginia

and

Wisconsin

show

resemblances,but

marked dis-

crepancies are

in

evidence.

The

quintilepatterns

for

this group

are

shown nTable 5.

TABLE

5

QUINTILE PATTERNS OF

THE

STATES

WITHIN

GROUP III

Year

Ind. Ohio

Ill.

Mich. Pa.

Conri.

Del.

W.

Va.

Wis.

1930 2 2

2

2

3

4

2 2

2

1931

2

2

2

2

3 4 5 4

2

1932 1

1

1 2

2

3

3 3

2

1933 3 3 1 1 2 2 2 4 3

1934

3

3

3 5

3

2

3

4

3

1935 4

3 3

5

1

2 2 1

5

1936

4

4

3

4

3

3 4 4

3

1937 5 4

5

5

3 4

4

3

3

1938 1 1

1 1 1 1

1

2

2

1939

5

5 5

5 3 4

5

1

1

1940

3

3

3

5 3 4

5 3

3

1941

5

4

2

4

2

4

1

2

3

1942 2

2 1

2

1

2 1 2

2

A

general

pattern

s

in

clear

evidence

in

the

cases of

the first

ive

states. The

pattern

of

Pennsylvania,

however,

ppears

to be drawn

to

some extent in

the

direction

of that

of

Group

II. In

1930,

the

tendency

was f r

Group

II

states

to

be

in

the

upper quintile

interval

and

for

Group

III

states to

be

in the second.But

Pennsylvania

was the

up-per

extremestate

of

the third

quintile

interval. In

1935,

it

occupied

a

Group

II

position,

nd

from

1939 on it

coniformed

ith

the

pattern

of

Group II.

Michigan

shows

severalmarked

discrepancies

nd

is

obviously

the

morevolatile

of

this

volatile

group.

n

six of the

thirteen

years,

t

was

within

ur

states

of one ofthe extreme

nds

of the

array.

n

1933,

ts

rate of decline

was

greater

than the

other

declines

of

the

group,

and

in

1934,

its rate

of

rise was

considerably

greater.

After

1939

the

rises

registered

y

Michigan

were

considerably

greater

than

those of

the

other

tates

ofthe

group.


8/10/2019 1905703

21/33

56

RUTLEDGE VINING

These other states

of the

group are

not so farout of

line with

one

another n 1933

as the-quintile

ocationswould

suggest.

The rangeof

variation

for ll 48 stateswas 24 percentage

points,

whiletherangeof

variationforGroup II statesexceptingMichiganwas 5 per cent.

Reference

o thepercentage

figures

orthe first ive tates

indicates

few otherdiscrepancies

of

importance. llinois

was low

in 1934, and

Illinois

and Pennsylvania

were somewhat

ow

in 1941. But it can be

TABLE 6

QIUINTILE

PATTERNS OF

THE STATES

WITHIN

GROUP

IV

Year Wash.

Ore.

Calif.

1930

3

2

5

1931

2 3

3

1932

2

2

3

1933

4

3

3

1934

3

4

3

1935

2 4

2

1936

5

5

5

1937 2

2

3

1938 4

4

4

1939 2 4 21940 4 3 4

1941 5

4

3

1942

5 4

3

said,

subject

to the qualifications

egarding

Michigan

in 1933

and

re-

garding

the

apparent

relationship

between

Pennsylvania

and Group

II, that

the five tates adhere

very closely

to a group pattern

nd

oc-

cupy

the

same vicinity f

thefrequency

istribution

ear

after

year.

The other states show certaindivergencies.Connecticutoccupieda

Group

II

positionduring

bout the

first alf of the period.

Since

1937,

it

has conformed

losely

with the

pattern

of

Group

III.

Delaware

also

is

attracted

to

the

Group

II

pattern.

West

Virginia

hows

marked

dis-

crepancies,

but

its percentagepattern

closely

resembles

hat of Penn-

sylvania.

Wisconsin shows

a moderate resemblance,

ut its

position,

like

Missouri's, ends

to remain n the

center f

the arrayor toward

the

lower

end

of the

array of the

absolute

rates of change.

In GroupIV are the Pacific Coast states-Washington, Oregon, nd

California.The

adherenceto

a

group

pattern

s somewhatcloser

than

the

quintile

nterval shown

in Table 6 would suggest.

Until 1941,

a

rangeof 8 points

would nclude

ll three

tates

n

any year,

nd except-

ing 1930

and 1934

a rangeof 4 points

would suffice.

n

1941

and

1942

Washington

nd

Oregon gained

at

a more

rapid

rate than did Califor-

nia. The

tendency

or his

group appears to

have been

for

relative

ta-

bility,

he

position

n the

array

tending

o be the

middle

ground.

The


8/10/2019 1905703

22/33

REGIONAL PATTERNS OF

BUSINESS-CYCLE

BEHAVIOR

57

exceptions re

1936 and the

two war yearswhen

the group

occupied

the

upper reachesof the

arrays.

In Group V are

placed

certain

political divisionsthat have

shown

a markedtendency orexpansion. The percentagefigureswhenplaced

alongside of each other do

not give as

clear an impression

of con-

sistency s do the

quintilefigures, ut

even here the

conformance o a

pattern s in

evidence. In

seven of the

years a range of 5

percentage

points

ncludes

ll

the states, n

three

moreyears

a

rangeof8

pointswill

suffice,nd

in

the

remaining hreeyears 14

points

separate the

lowest

fromthe highest

percentage

of the three.

Obviously,

expansionary

forcesof

an

unusual character

have been

in

operation

n

these

three

areas. It does notfollow, fcourse,thatthesimilarityn thereactions

TABLE 7

QUINTILE PATTERNS

OF

THE

STATES

WITHIN

GROUP V

Year

Dist. of

Colum. Va.

Fla.

1930

5

3

4

1931

5

5

4

1932

5

5

4

1933 1 2 2

1934

2 4 4

1935

4

2 4

1936

5

3 5

1937

2

2 4

1938

5

5

5

1939

1

4 5

1940

3

5

4

1941 2

5

2

1942 3

3 3

of these

reas is

due to the

similarity

n

economic tructure.

Noncyclical

factorswere doubtless

playing

a most

mportant

ole.

In

nine of

these

thirteen

ears

these

areas were well

nto

the

right

ail

of the

frequency

distribution.Virginia

did not

show

this

tendency

o

expand

so

much

as did

Florida,

and

the

District

of

Columbia

was somewhat

more er-

ratic. Mildercontractionsnd

strongerxpansions

wereshown

by

these

three reas

than

was

generally

he case.

Groups I, II, and III represent o us a fairly atisfactory lassifica-

tion of

cyclical

reaction

patterns

nd are

tentatively

atisfying

n

the

sense

that

certain tatistical

regularities

re

suggested

orwhich

n

ac-

counting

s called. More

interesting

lassification

may

be

made,

we

think,

when

more

appropriately

efined

egions

re used

in

the

analy-

sis,

but nevertheless he

state

groupings

aise certain

pertinent

ues-

tions.

Group

IV

is

not

quite

so

satisfactory

nd

Group

V

is

somewhat

less

so, for

we

feel

that what

similarity

s

suggestedmay

be

largely


8/10/2019 1905703

23/33

58

RUTLEDGE

VINING

TAB3LE

S

Q-UINTILE

PATTERNS

OF

THE

STATES

WITHIN

GROUP

VI

a

b

c

Y

e

a

r

-

_

_

_

-

_

_

_

-

_

_

_

_

_

_

-

_

_

_

_

_

_

-

_

_

_

-

_

_

_

-

_

_

_

_

_

_

_

_

_

-

_

_

_

_

_

_

_

_

_

Okla.

Tex.

Ariz.

Utah

N.M.

Colo.

Idaho

Mont.

N.D.

Nev.

Wyo.

Neb.

Iowa

S.D.

Kan.

Minn.

1930

1

2

2

3

3

4

3

1

2

5

4

5

5

5

5

4

1931

1

2

2

2

4

3

1

1

1

5

3

1

1

1

2

3

1932

3

4

1

1

2

3

2

1

3

4

1

1

1

1

1

2

1933

5

5

4

4

5

4

5

4

5

1

5

5

5

4

4

3

1934

1

3

4

3

4

2

5

5

1

4

4

1

1

5

3

2

1935

4

3

3

5

4

3

3

5

5

5

2

5

5

5

4

5

1936

2

3

4

3

5

4

5

1

1

2

2.

1

1

1

2

3

1937

5

5

5

5

4

5

5

3

5

4

1

1

5

1

4

3

1938

3

5

2

4

3

2

2

2

1

1

5

2

5

5

1

4

1939

1

1

2

1

4

2

1

3

4

5

3

1

5

3

1

2

1940

1

1

1

5

3

1

3

5

5

5

2

3

1

2

4

1

1941

1

4

3

3

1

1

3

2

5

1

2

1

4

4

5

1

1942

4

4

5

5

3

3

4

1

3

5

2

5

3

4

5

2


8/10/2019 1905703

24/33


BEHAVIOR

59

attributable

o

noncyclical

orces.But the rest

of the states-those

of

the Great Plains and Rocky

Mountain

regions-give even

less satis-

factory

esults.The methods

hat we are

usingdid not reveal

consistent

breakdowns nto groups showing nternaluniformityegarding ates

of change of income.Several

types of

factors ould

be responsible

or

this lack

of evidence of group

behavior

among these states.

In the

first

lace,

the data have

obvious shortcomings

romwhich

might

be

expected

considerablestatisticalerror.

The

bulk of the very

sparsely

populated

states

s included n the present

roup.

Of thesesixteen

tates

nine

are smaller

economicallythan

either Arkansas,

Mississippi,

or

Rhode Island. Any one of these

nine

accountsfor maller

ncome pay-

ments han either f these atter hree tates. Theseninestates-North

and South Dakota

and the

MountainStates excepting

Colorado-to-

gether

ccount

for

only about

2'

or 3 per cent

of

the national

income

payments,

which s less than

the incomepayments

of Texas

and about

the

same as those

of Missou

1905703

Documents

Transcript of 1905703