Room air conditioners : a manufacturing opportunity in Georgia

Project B-140-9 R 0 0 M A I R C 0 ND I T I 0 NE R S
A Manufacturing Opportunity in Georgia
Prepared for The Georgia Department of Commerce
Scott Candler, Secretary
Eugene Queen Research A~sistant
Industrial Development Branch Engineering Experiment Station Georgia Institute of Technology
December, 1958

Foreword This report offers new insights into an important manufacturing opportunity for Georgia. The factors upon which the study focuses--climate and income--have long been considered of "obvious" importance to the sale of room air conditioners. However, there have been to date but limited efforts to determine statistically what the precise effects of these key factors might be. By providing statistical measures of the degree of influence exerted by each factor, Mr. Queen's analysis suggests a new and valuable basis for forecasting sales of room units. As the report shows, this information casts a new light on questions important to the location of new manufacturing plants. Comments or questions regarding the analysis are invited. More detailed information regarding specific location possibilities within the area recommended for a room air conditioner manufacturing plant will be provided on request.
Kenneth C. Wagner, Head Industrial Development Branch
-i-

Acknowledgments The author wishes to express appreciation to all those who gave of their time and special knowledge in the preparation of this report, especially Mr. Everett L. Rudeseal, Georgia Power Company, for making available data on the number of domestic customers of electric utilities; Mr. James M. Van Buren of LIFE, for the results of the LIFE Study of Consumer Expenditures; Mr. Charles L. Skinner, Georgia Motor Trucking Association, for timely information about Atlanta's trucking services; Mrs. Mildred T. Wilson, Southern Technical Institute, for statistics on that school's graduates; and to several members of the Industrial Development staff for advice and editorial services: Mr. Robert Bullock, Research Assistant; Dr. Ernst W. Swanson, Senior Research Economist; Dr. Kenneth C. Wagner, Head; and Mrs. Annie F. Edwards and Mrs. Betty Jaffe for their preparation of the final report,
-ii-

Table of Contents

Foreword

i

Acknowledgments

ii

summary

1

I. Introduction

3

II. The Market Analysis

4

The Regions

6

The Analysis

8

III. A Market Forecast

12

IV. The Comparative Location Study

17

Selection of Distribution Centers

17

v. Atlanta as a Location for the Room Air Conditioner

Industry

29

The Labor Market

29

Labor Costs

30

Technical Training

30

Proximity to Markets

31

Appendix

I. Estimate of Sales

32

II. Forecast Methodology

37

Maps and Graphs

Map 1. Major Regional Markets for Room Air Conditioners

2

Map 2, Room Air Conditioner Sales--Percentage Distribution

by Regions

5

Map 3. Rank of States in Sales of Room Air Conditioners, 1957

9

Figure 1. U, S. Room Air Conditioner Production, 1947-1957

14

Figure 2, U, S. Room Air Conditioner Production, 1947-1957,

Two Year Moving Average

15

Figure 3, Modified Growth Curve

16

-iii-

SUMMARY
This study is concerned with the spatial distribution of the market for room air conditioners, both present and future, and with the implication of the findings for plant location decisions.
Briefly stated, the findings are these:
1. There are at least two major regional markets which may be differentiated on the basis of the importance of factors which influence purchase decisions: the desire for comfort and the ability to pay for it. 2. One of these regions, corresponding roughly to the South Atlantic, East South Central, and West South Central states, has a greater potential for market growth than the other, consisting roughly of the New
1 England, ~iddle Atlantic, East North Central and West North Central
States. 3. As a consequence, the national market center, which is now in the vicinity of Louisville, Kentucky, may be expected to shift southward, and in turn, the plant locations which would provide the maximum effectiveness in national market penetration may be expected to lie south of Louisville. 4. Specialization in a regional market could be pursued most effectively in the southern states.
Given the unique importance of purchasing power in the southern market, as developed by the analysis, and the assumption of continued income growth in that region, it is reasoned that the market growth in the south is causing a continuing shift southward of the market center.
Predicated on the thesis that the market center is in general an optimum location for manufacturing facilities, and on the inference of a continuing southward shift from the results of the market analysis, three cities are identified as worthy of consideration in future plant location decisions.
One of these cities, Atlanta, a major distribution center for the South, offers excellent transportation facilities, established marketing channels, labor with varied degrees of skill, and other advantages which make it one of the most favored areas for plant location in the South. Further and more detailed consideration of this metropolitan area is recommended as a prerequisite to a location decision.
2:-1/ Readers not familiar with Census definitions of regions may refer to
Map
-1-

r

...

MAP 1

MAJOR REGIONAL MARKETS FOR ROOM AIR CONDITIONERS

I N I
~SALES OF ROOM AIR CONDITIONER ~UNITS
PRESENT LOCATIONS OF MAJOR MANUFACTURES OF ROOM UNITS

I. INTRODUCTION
Many of the conclusions drawn in this report are based on the effects of climatic and income differentials on the geographic distribution of room air conditioner sales. That such effects do exist is nothing new; certainly there is no intent here to belabor such an obvious point. Nevertheless, further understanding of the market for this product can be gained by a reexamination of these effects. Measurement of the extent of the interrelationships among sales, income, and climate, in particular, can lead to more definite and more fruitful conclusions than those intuitively accepted as "obvious."
Mere intuition is often sufficient to describe and predict the behavior of an individual; indeed, it is sometimes more appropriate than an analytical tool that presupposes rational behavior. But in dealing with mass consumer behavior with characteristics that differ among the various regions, a more objective approach is needed. Such an approach is taken here, in that the "obvious" is treated as an hypothesis.
The Second Section demonstrates that there is an empirical basis for defining regional markets for room air conditioners. Statistical treatment of sales data on a regional basis confirms the theoretical argument that income and climate are among the major determinants of sales, and measures the degree of influence exerted by these factors within the three regions defined in Section II.
A forecast of 1959-60 production of room units is made in Section III, with a very brief consideration of the southern region's probable share of that market.
In Section IV, a simple scheme is constructed for the purpose of locating the market center, based on the 1957 distribution of sales.
Section V devotes attention to the merits of the general vicinity of Atlanta as regards plant location factors. More detailed information will be provided as desired for any firm seeking a location meeting a particular set of requirements.
-3-

II. THE MARKET ANALYSIS

Sample sales data obtained from a large number of utilities, were used to estimate sales of room air conditioners by states.l/ Map 2, which follows, shows the distribution of sales by Census regions. It is questionable, however, whether this definition of regions is appropriate for room air conditioner markets. It is apparent that there is considerable variation among the states in sales volume. A critical question, therefore is: What are the factors which determine sales volume?
The answer could encompass many particulars, but it will be simplified in the present case to consider only certain measurable factors. Certainly the desire for comfort, in so far as a room air conditioner can provide it, and the necessary purchasing power are two pertinent factors. Admittedly, the conditions necessary for human comfort are complex, but relief from high temperatures and excessive humidity are primary considerations.
Bosen and Thorn of the U.S. Weather Bureau have made considerable progress in developing measures of the need for summer cooling. Thorn has recently published values termed "cooling degree days" for a number of cities throughout
the United States.~/ The cooling degree days variable is used in this study as
one of the factors influencing purchase decisions. Not all states could be

11 Appendix I sets forth in detail the methodology used for these
estimates.
~/ J. F. Bosen, Office of Climatology, U.S. Weather Bureau, Washington,
D. c., has developed two linear equations which provide a Discomfort Index
appropriate to the need for summer cooling. The first equation involves dry bulb and wet bulb temperatures; the second, dry bulb temperature and dew point temperature.

Earl C. Thorn, Meteorologist, U.S. Weather Bureau, Washington, D. C., has proposed the Discomfort Index as a basis for measuring cooling degree days. A base figure of 60 is subtracted from the average Discomfort Index for each day, and the remaining values are accumulated into monthly and annual totals of cooling degree days.

The equations are:

(1) DI = 0.4(td + t w) + 15

(2) DI

0.55td + 0.2tdp + 17.5

= Discomfort Index

= Dew point temperature

= Dry bulb temperature

t values are simultaneous

= Wet bulb temperature

For further discussion, see "Cooling Degree Days," E. c. Thorn, July 1958,
p. 65ff., The Industrial Press, New York.

-4-


MAP 2 ROOM AIR CONDITIONER SALES* (Percentage Distribution By Regions)
I lll I
WEST SOUTH CENTRAL 23.7%
*(Based on estimated national total of 1,304,000 units, which includes an estimate for those states and the District of Columbia not included in Appendix Table 1).

considered in the analysis, since cooling degree data for some of them were not available. Per capita income in the various states was used as a measure of purchasing power.
The analysis which follows is a study of the relationships between the two factors--"climate" and income--and sales per thousand domestic customers of the electric utilities. The method employed to measure these relationships is multiple correlation, and the hypothesis is that sales are "dependent" upon income and the climatic factor.
The basic data are given in Table I. The results of the analysis are arranged in tabular form in Table II. Contrary to expectation, it will be noted that in the aggregate, sales are negatively correlated with cooling degree days. As will be seen, the aggregative analysis is somewhat deceiving, in that the coefficients reflect implicitly the fact that most of the high income states are in the North, and the states where air conditioning is most needed or desirable are generally in the low-income group.l/
The aggregative analysis also suggests that a basis for developing more appropriate regional definitions is needed for the purposes of this market study. The basic concept in regional grouping is dual. First, the region must be unbroken and continuous. Second, it must be relatively uniform in climate or income. The question is whether regions can be defined in terms of geographic areas differing in income and climate characteristics,
The Regions
The various states were ranked according to per capita income and number of cooling degree days, and then compared for similarities of state groupings. There are patterns in these listings, although some slight modification is necessary to preserve geographic grouping.
Two groups of states are well defined. The first consists of Arkansas, Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Tennessee and Texas. The second group is composed of Illinois, Massachusetts, Michigan, Minnesota, Missouri, Nebraska, New York, North Dakota, Ohio
l/ When this is taken into account in the second order coefficients, the
Tntahenegeafotiiunvsealyl c, roerresrfeullleatctit(oRsn1ab23ecto)w,nesweindheicsrhaabletlasekeadnsedgbreionetchoomffae c"it(mor 1rps2r'oLv3ne) mtobeenccto"omneosvisdeoremrtaehwteihoanltowssmiemraulollr-edre. r
Coefficients, i.e., it tends to better agreement with the hypothesis. These results are not particularly enlightening, except to serve as a contrast to the results obtained when the same methods are applied to the same data grouped as Various regions.
-6-

'
TABLE I
SALES, PER CAPITA INCOME, AND COOLING DEGREE DAYS FOR SELECTED STATES, 1957

State
Massachusetts New York Ohio Illinois Michigan
Minnesota Missouri North Dakota Nebraska North Carolina
South Carolina Georgia Florida Tennessee Alabama
Mississippi Arkansas Louisiana Oklahoma Texas
Montana Wyoming Colorado New Mexico Arizona
Utah Nevada Washington California

Sales (per 1000 Customers)
8 28 16 21 4
19 36 25 31 27
80 80 42 77 37
21 44 102 36 82
3 12 4 14 24
5 25 4 13

Per Capita Income
$ 2, 335 2,578 2,255 2,447 2,141
1,850 1,940 1,435 1,818 1,317
1,180 1,431 1,836 1,383 1,324
958 1,151 1,566 1,619 1,791
1,896 2,038 1,996 1,686 1,750
1,694 2,423 2,128 2,523

Annual Cooling Degree
Days
1020 1056 1342 1195
311
954 1756
745 972 2182
2549 2168 3763 2119 2755
2583 2302 3026 1905 2812
606 405 556 951 2227
767 990 247 1245

Source: Income data from Survey of Current Business, August 1958. Cooling degree data are from Thorn's article in Air Conditioning, Heating, and Venti!!ting, July 1958, pp. 68-72.

-7-

and Oklahoma. A third group which is less well defined comprises Arizona, California, Colorado, Montana, Nevada, New Mexico, Utah, Washington and Wyoming.
A third ranking of states in order of total sales volume was used in part as confirmation of these conclusions, and to aid in assigning states for which no cooling degree data were available. Much the same pattern emerges if states are checked off on a map of the United States as they appear in this last ranking. See Map 3.
Thus, the two major regional markets seem to consist of one group of states ranging eastward from the Great Plains through the Middle Atlantic states up to New England, and a second group extending southward from the Middle Atlantic states and over the Gulf Coast into Texas.
The Analysis
As a test of these regional definitions, correlation techniques may be applied to the data for these regions, as was done for the national data. The results appear in Table II. The differences from the overall or aggregative analysis are of considerable importance.
First, the fact that differences do exist proves that the joint influence of climatic factors and income is of a different nature among the regions. This establishes the case for regional market differentiation. The general improvement of the correlation coefficients resulting from the market breakdown demonstrates the validity of the groupings, granted the ~priori hypothesis that income and the climatic factor are major determinants of sales.
Second, the nature of the differences has specific implications for future market growth.
In the "South," income is clearly the dominant factor in determining sales. The climatic factor does not specifically enter in, except in the s~nse that the climate is uniformly such that a room air conditioner is desirable.
In the other major market area, climate is dominant. Income is available, provided that climatic factors make an air conditioner sufficiently desirable to warrant its purchase. Thus, occasional cool summers may depress considerably the sales of room units in this region.
In the region composed of the Mountain and Pacific states, both factors are considerations, but the climatic factor is more important.
-8-

,
MAP 3 RANK OF STATES IN SALES OF ROOM AIR CONDITIONERS, 1957
(Thirty States Comprising 97.3 Per Cent Of The Market)

I 1.0 I

18 STATES ILLUSTRATING



~~:Eor~i~~NR~F REGIONAL

TABLE II RESULTS OF CORRELATION ANALYSIS

coeffici!?t Symbols-
r12 rl3 rl2.3 rl3.2 Rl.23

Aggregati:ve

Correlation Coefficients

"South"

"North"

-0.46
o. 73
-0.12 0.64 0. 73

0.42 -0.003 0.54 -0.37
0.54

-0.41
o. 70
-0.47 0.72 0.78

"West"
0.22 0. 73 0.44
o. 77
0.79

l/ For the
iinscotmhee reafsoxre2,

thaaenndaclcyoosroriseli,lnagptieodrnecgbareepetiwtadeeanyssal"eapsserwxc3asa pdiTetahse"ignsinaattleeedsrpxarne1dt, atppieeorrn

capita coafpirta12

income. The notation is per capita sales and per

standar capita

din,corm12e;

3wbiethingtheth

e o

correlation ther factor,

between cooling

degree days, multiple cor

held con relation,

stant stat a measure

i

sti of

cally. the va

r

iaRo1

i

i213

t

is y

i

th n

e c per

oefficient capita sa

l

of es

associated with the variability in both of the other variables.

-10-

With the growth of room air conditioner sales over the past six years, shifts have occurred in the spatial distribution of the market. Indeed, much of this growth is attributable to a shift; namely, increased penetration of the southern market. From 1952 to 1957, the average retail price of room units declined. During this same period, per capita income in the Southeast, for example, increased from $1,194 to $1,427. Considering the relative importance of purchasing power in the "South," it is almost certain that the increasing income and declining price combined to increase sales to the extent that total market growth is in large measure attributable to growth in the southern regional market.
The implication is that sales in the southern region may be expected to increase over time with income growth. Sales in the other regions, however, are more subject to the vagaries of year to year weather conditions. Thus the real growth!/ in the room air conditioner market will occur in the "South." In other words, the market center of the nation will continue to shift southwards.
By real growth is meant increased sales relative to consumer popui.e., an increased rate of buying not attributable to population
-11-

III. A MARKET FORECAST
One of the basic assumptions of this study is that the market for room air conditioners will continue to grow. At the same time, substantial future growth of the national market is a major inference from the analysis, based on the increasingly important role played by southern markets, and the evidence that enlargement of these markets is closely allied to income growth. The question of how much growth may be expected in the national and regional markets is obviously pertinent to location decisions. Therefore, an attempt is made here to answer that question.
Two methods of forecasting were used, with results in close agreement. The first method is simply a statistical examination of production data, in a search for consistent patterns of growth behavior over the past 11 years (1947 to 1957). Statistically speaking, the second method is slightly more sophisticated, in that it utilizes the relationship between income and the level of production. A brief discussion of these techniques may be found in Appendix II.
A logarithmic graph of the adjusted production data (Figure 1) shows clearly a marked acceleration in growth from 1951-52 to 1953-54 (a reflection of the sharp rise in the actual data from 1952 forward), followed by a period of lesser but more steady growth (which is not evident in an arithmetic graph of the data).l/
Figure 2 shows no marked tendency to regularity in production growth. A more nearly linear path would provide a much better basis for forecasting, and a statistical transformation designed to reveal such hidden tendencies was applied.
The result is the interesting curve in Figure 3. An extrapolation of the nearly linear growth of the last four years, after correction for the transformation applied earlier, gives a production forecast of about 2,250,000 units for 1959-60.
1/ A transformation explained in Appendix II clearly indicates a subltantial upward shift of supply and, possibly, demand. The upward shift was
due to expansion of production facilities, and entrance of new producers into the market. There is a strong suggestion that the present growth rate in
production (and therefore sales) is about the same as in earlier years, but a much higher level. Compare Figures 1, 2, and 3.
-12-

r
As an alternative, the relationship between production and U.S. total personal income was formulated empirically, and used as a means for forecasting production based on a forecast of income. After determining the nature of these various relationships, a forecast of 2,280,000 units was obtained for 1959-60,
Each of the two methods rest upon assumptions about stability of the economic factors involved, but in view of the short range of the forecasts, they are believed to be not only reliable but probably conservative. Forecasts for the more distant future would become increasingly speculative, due to the shortness (10 years) of the series used as a basis.
The future magnitude of the southern market cannot be described with precision. At the time of this writing, historical data similar to estimates for 1957 contained in this report have not been developed, and an elaborate analysis cannot be justified.
It is appropriate to consider what the "South's" share of the national market might be by 1960. A conservative estimate would be 55 per cent, or approximately 1,245,000 units; a more optimistic estimate of 60 per cent would mean 1,359,000 units, Either of these estimates is greater than total national production in any year prior to 1954.
Market growth is to be expected in other regions, of course, but in view of the analysis, growth elsewhere will not be as great, absolutely or relatively, as in the South. The market expansion in other regions will be tied to such factors as population increase and family formation; in the South, the additional powerful influence of income growth will dominate.
-13-

,
FIGURE 1 U.S. ROOM AIR CONDITIONER PRODUCTION, 1947-1957
(Original Data, In Thousands Of Units) 2,000 , . - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - .
1,750
1,500
1,250 iii'
z1-
:;:)
u.
0 II) 1,000
0 z
~
:;:)
0 J: .1._-
750
500
Census of Manufactures, 1954, 1955. Facts for Industry, 19471957. 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
-14-

FIGURE 2 U.S. ROOM AIR CONDITIONER PRODUCTION, 1947-1957
(Two Year Moving Average)
2,000 , - - - - - - - - - - - - - - - - - - - - - - - - - - - - - . . . . . . ,
1,000 900 800 700 600 500 400
en
z1-
:::1 300 ::E
0 0
~
u.. 200
0 en
0 z
<(
en
:::1 0 J:
1- 100 90 80 70 60 50
46-47 47-48 48-49 49-50 50-51 51-52 52-53 53-54 54-55 55-56 56-57
-15-

r

10,000 9,000 8,000 7,000 6,000 5,000
4,000

3,000
~
~

Vl
::::>

2,000

.e.J:.

z.V...l.
::::>

:l:
0 1,000
~ 900

11. 800

0

Vl 700

0

~ 600

Vl

::::> 0

500

.J...:. 400

300

200

FIGURE 3 MODIFIED GROWTH CURVE*
(Two Year Moving Average)
____ ..... __ ......
... ............
... -----------------------
---PLOT OF ACTUAL OATA PLUS "K" ---PROJECTION OF 194752 GROWTH
RATE FOR COMPARISON -----FORECAST PROJECTION

46-47 4748 48-49 4950 50-51 5152 52-53 5354 54-55 5556 56-57 5758 5859 5960 *K = 547, See Appendix 2.

-16-

IV. THE COMPARATIVE LOCATION STUDY
The cost of shipping an assembled room air conditioner includes freight on sheet metal and hardware of the kind available almost everywhere. In addition, the unit's bulk includes empty space which, although necessary in air flow design, is costly to transport, As a consequence, market orientation of manufacturing plants affords an opportunity to reduce distribution costs of the assembled unit.
Clearly, there must be an optimum plant location with respect to costs of distribution, The total cost of distribution for a product manufactured in a given location depends on the volume shipped to the various markets served. Thus this cost is a function of the distance from the manufacturing site to the various markets, weighted by the volume of units shipped to those markets. If the volume shipped to each distribution point were known, a manufacturing site could be chosen in such a manner as to minimize the cost of distribution.
Appendix I sets forth estimates of sales by states. If these estimates could be allocated to more specific locations, then comparisons could be made between the location advantages of various manufacturing sites with respect to market penetration (in terms of access).
Data of the kind and extent suitable for such comparisons are not available, but after certain simplifying assumptions, approximations may be obtained. State sales data as such are not useful for comparative purposes, as it would be difficult to select a single point within a state from which distances to manufacturing sites would be representative statistically of the whole state, An alternative is to select from each state major distribution centers, allocate state sales proportionally to these centers, and compare their distances from the various cities with plants. This is the method used here,
Selection of Distribution Centers
The problem becomes one of selecting the distribution centers and developing a suitable method of proportional allocation. In general, major distribution centers are also major population centers. From each state those metropolitan areas were selected which account for at least 50 per cent of the total metropolitan area population in that state, The 50 per cent level was chosen simply to reduce the number of cities that would be involved, and consequently reduce the amount of computation.
-17-

,

For t y-t hree c1 t1 es1-I were se 1ecte d and , 1 nso f ar as poss1 bl e, d1 stances

from these major distribution centers to cities with room air conditioner

plants were obtained. In some cases, present sites are in cities for which

distance tables would be very difficult to construct, and nearby major cities

were substituted. Highway mileages were used in this study primarily because

a considerable portion of room unit output is transported by truck. A mile-

age table was constructed as shown in Table III. The column headings are

cities with plants or cities near plants of some of the major manufacturers

of room units, plus certain other cities used for comparative purposes. The

cities heading the rows are the selected distribution centers. The column

totals are the total mileage between plant sites, actual or hypothetical, and

the major markets in each state. If sales were the same in each of the

selected cities, the most favored site would obviously be the one with the

smallest column total (Louisville, Kentucky). Since sales are not uniformly

distributed, the matter is in doubt until the mileages in the body of the

table are weighted by the volume of shipments to each destination. The

"weights" are an approximation of sales in the various metropolitan areas.

They are derived by application of ratios to the estimates of state sales as

derived in Appendix I. The ratios are simply the percentages of the states'

total metropolitan area wholesale sales accounted for by the individual met-

ropolitan areas.

For example, the "weights" for the Miami and Tampa metropolitan areas

were derived as follows:

Estimated Unit Sales ("Weight")

Sales of Merchant Wholesflers (Thousands of Dollars)-

Florida

52,600

Miami

(22 ,550)

Tampa-St. Pete

(12,566)

Other metropolitan areas

$1,323,972 567,534 316,265 440,173

Source: Census of Business, 1954, U.S. Department of Commerce

l/ The metropolitan areas of Baltimore, Maryland, Norfolk-Portsmouth and
Richmond, Virginia, and the District of Columbia which should be included by the method of selection, were not used. Sufficient sample data for state sales estimates were not available.

-18-

For Miami, the estimate would be 567,534/1,323,972 x 52,600 or 22,550; Tampa, 316,265/1,323,972 x 52,600 or 12,566. These two estimates are then applied as "weights" to distances from Miami and Tampa to the cities heading the columns.
The entries in the distance table were multiplied by similarly derived weights to obtain Table IV and summed as before. In this case, the distribution of sales is such that the location most favored with respect to national market penetration (given the 1957 national sales distribution as estimated in Appendix I, Table I) is again Louisville, Kentucky. The rank of the first six of the cities examined, in ascending order of weighted distances, is as follows: Louisville, Kentucky; Indianapolis, Indiana; Birmingham, Alabama; Memphis, Tennessee; Cincinnati, Ohio; and Atlanta, Georgia.
The reliability of the method of allocating sales among selected metropolitan areas is supported to a considerable extent by the findings of the LIFE Study of Consumer Expenditures in 1956.1/ According to this survey, metropolitan area residents accounted for 79 per cent of the total air conditioner market; non-metropolitan area residents accounted for 21 per cent. The South was the only exception to this national pattern. In that region, sales were about equally divided between the two groups. This indicates that coverage of the southern markets involves a greater number of distribution points than in other regions, and that some advantage could be obtained by locating near these points. A further inference is that the results obtained in the location analysis possibly do not place the national market center as far south as it actually is. If more southern distribution centers had been included in the computations, the relative positions of the hypothetical southern plant locations would tend to improve, since the majority of the additional markets lie south of those chosen for the computations, and therefore farther from the present actual plant sites.
The implication of a southward shifting market center for future manufacturing plant location decisions is clear, If competitive advantages can be obtained by locating near the market center, then relocation or branch plant expansion of existing production facilities now elsewhere will result in a larger share of a growing market.
F.or practical purposes, the knowledge that the market center is shifting southward is sufficient to enable most manufacturers to improve significantly
1/ LIFE Study of Consumer Expenditures, TIME, Incorporated, 1957.
-19-

--,

TABLE III. MILEAGES

Locations Markets

tJl

tJl <I) ~ rl
0
::<:: <U
tJl ~
<I) 0 I=IH

tJl
0-rl bOO <U ~ 0-rl
rl ......
,.C:.--1 UH

.o...r.!.
0
g.<U
~ ~
rl rl "CI"CI ~ ~ HH

or!
.u

<U

~

~

rl

C) 0

~
url

o..cr:l

<I) ~ <I) or! ~ tJl
::l ~ <U 0
~ C) ...... tJl rl or!
::<::::;c

~~
:>0<:>0<
~ ~
z z <I) <I)

bO~
~ bO rl rl
tll..C:
~ C)
<U-r! ....:1::<:!

~

0 1-1

...... <U

:>0<

4-l

4-l ~

::l <I)

l'!:lZ

<I)
1-1"0
~ ~
rl ......
.u >-.
...... 1-1
<U <U I'J:l::<:!

...tcuJ:l
bO rl
<I)
:3:

Birmingham, Ala.

860

676

525

499

772

993

752

933

807

15' 104

Little Rock, Ark.

587

661

566

639

739 1,311

828 1,075 1,097

13' 231

Los Angeles, Cal.

1,805 2,108 2,095 2,294 2,149 2,875 2, 400 2,642 2,726

27,843

Bridgeport and

New Haven, Conn.

1,261

932

785

713 1,022

78

806

428

245

5,006

Hartford, Conn,

1,287

958

823

751 1,034

119

815

395

286

4,541

Miami, Fla.

1,573 1,396 1,222 1,157 1,475 1,346 1,440 1,498 1,155

22,550

Tampa, Fla.

1,370 1,185 1,019

961 1,272 1,197 1,228 1,349 1,006

12' 566

I N

Atlanta, Ga.

899

715

549

466

814

863

760

901

694

58,615

0
I

Chicago, Ill.

341

0

191

302

97

831

205

538

669

52,550

Indianapolis, Ind.

472

191

0

110

281

710

248

492

565

16,994

Cedar Rapids, Iowa

119

227

395

505

248 1,068

432

765

896

2,223

Des Moines, Iowa

0

341

472

582

364 1,165

558

871 1,019

8,376

Sioux City, Iowa

197

504

666

776

500 1,345

709 1,042 1,173

4,162

Kansas City, Kan.

206

504

484

599

572 1,223

694

981 1,065

23' 921

Wichita, Kan.

422

722

712

826

790 1,441

932 1,208 1,303

13' 361

Louisville, Ky.

585

311

118

109

399

771

344

547

617

10,677

New Orleans, La.

1,056

977

845

843 1,067 1,361 1,090 1,275 1' 173

62,249

Boston, Mass.

1,313

992

936

864 1,138

222

BOO

460

387

7,939

Detroit, Mich.

607

279

273

256

369

645

84

255

519

6,143

Minn.-St. Paul, Minn.

264

420

611

713

351 1,261

628

956 1,107

15,451

Jackson, Miss.

844

760

667

703

840 1,248

929 1,159 1,060

10,600

St. Louis, Mo.

368

291

238

346

388

969

485

732

816

24,208

Omaha, Neb.

140

486

592

700

502 1,295

698 1,011 1,148

11' 681

New York, N. Y.

1,165

831

710

647

931

0

711

373

184

140,395

Charlotte, N. C.

1,049

775

583

492

865

618

737

761

433

12,344

Greensboro-High Point

and Winston-Salem, N. C. 1,071

797

640

542

894

529

724

656

342

8,217

TABLE III, MILEAGES (Continued)

Locations Markets

Ol

..O..l
..-I

....

Q)
..s.:.:
0
::t: al
<ll !3:
Q) 0 .:::IH

<ll
0-rl 000 al s:: 0-..t
.... ..-I
,.d..-1
UH

0
l?w .~......~.
"tl"tl s:: s:: HH

+J
aal
....
0 0 S::...t
.u..t,o.d

QJ s:: Q)...t
~ ~
al 0
!3: 0
....-.I...O..l
::t: ;3: .

~~ >0<>0<
z!Q3): z!Q3):

00~
.s.:.:...0..0
Ol..d
s:: 0
jiJ

0~
..-10
~~ 4-I!J: ::l QJ
~z

Q)
1-r"t:l
~ ~
.... ..-I +JJ>., ..-I 1-r al al ~::t:

Ol +J
,.d
..0..0
QJ
;3:

I N t-'
I

Cincinnati, Ohio Cleveland, Ohio Columbus, Ohio Oklahoma City, Okla. Philadelphia, Pa, Pittsburg, Pa. Columbia, S. C.

582

302

110

0

392

647

277

436

500

678

335

306

242

439

488

223

192

333

645

313

173

108

402

537

236

332

392

563

850

797

895

914 1,526 1,038 1,284 1,368

1,083

773

652

577

863

88

634

364

102

789

461

356

284

550

364

341

222

209

1,111

828

622

536

894

703

837

862

517

8,058 14,924
5,396 13,396 34,916 15,440 27,266

Greenville, S. C. Knoxville, Tenn. Memphis, Tenn. Dallas, Texas Houston, Texas

1,027 868 637 709 954

730 573 562 955 1,102

538 385 453 920
1,041

440 281
500 981
1,090

827 666 652 1,054 1,180

718 726
1,170 1,632 1, 709

717 558 699 1,163 1,299

861 844 933 1,409 1.,525

533 541 958 1,441 1,484

15,216 6, 401
46,663 50,968 58,006

San Antonio, Texas

987 1,242 1,185 1,248 1,324 1,889 1,435 1, 660 1,684

15,549

Milwaukee, Wise.

364

97

281

392

0

931

297

626

759

11 J 615

Totals

30,858 27,162 24,536 24,969 30,030 38,612 29,791 34,853 33,313

TABLE III. MILEAGES

Locations Markets

til

l=l

til til
-1-1-..l
l=l 00 til H ...... 0
.j..l ())
<C!l

~
..c
OOtll
~e.g~
oM ...... ~<

.-...~..

()) 0H

.j..l til

-I-IU

0

.--I..C

1-< -1-1

.u.tcil

1-<
z 0

Ul til til ...... til
.-I M
til ()) ClE-1

oM 0.. 0.. -~
l=l Ul 0 Ul Ul-M ~ Ul 0 til
til-~ ...,~

())
...... ...... :>-.
>-~ ~ 0 Ul ;:I
-~ .j..l
;:I l=l 0 ())
H::.<:

())
())
Ul Ul
.o.Mc U())l ~
()) ()) ~E-1

Ul
~ til
()) l=l ...... til 1-< ~ 0 til
oM ;3: ;:I ()) 0
ZH

Ul
.j..l
..c
00
-~ ())
~

Birmingham, Ala. Little Rock, Ark. Los Angeles, Cal.

158 562 2,289

0 395 2,122

418 780 2,460

672 340 1,429

254 269 1,869

399 552 2,127

255 139 1,824

365 461 1,937

15,104
13' 231 27,843

Bridgeport and

New Haven, Conn.

932 1,049

685 1,701 1,303

820 1,203 1,426

5,006

Hartford, Conn.

992 1,107

732 1,736 1,355

860 1,255 1,472

4,541

Miami, Fla.

672

783

751 1,374

952 1] 116 1,042

887

22,550

Tampa, Fla.

461

546

593 1,120

684

911

813

650

12,566

I N

Atlanta, Ga.

0

158

260

839

421

432

423

513

58,615

N I

Chicago, Ill. Indianapolis, Ind.

715

676

775

955

760

311

562

977

549

525

583

920

667

118

453

845

52,550 16,994

Cedar Rapids, Iowa

861

823

977

828

830

513

621 1,025

2,223

Des Moines, Iowa

899

860 1,049

709

844

585

637 1,056

8,376

Sioux City, Iowa

1,106 1,028 1,246

777

986

784

773 1,179

4,162

Kansas City, Kan.

805

726

984

498

681

526

468

878

23,921

Wichita, Kan.

1,038

849 1,153

391

756

744

561

842

13' 361

Louisville, Ky.

432

399

464

879

593

0

383

737

10,677

New Orleans, La.

513

365

783

498

195

737

410

0

62,249

Boston, Mass.

1,084 1,220

845 1,868 1,470

973 1,389 1,583

7,939

Detroit, Mich.

741

755

689 1,194

942

363

751 1,099

6,143

Minn.-St. Paul, Minn.

1,097 1,058 1,193

960 1,084

729

864 1,279

15,451

Jackson, Miss.

421

254

681

418

0

593

213

195

10,600

St. Louis, Mo.

550

511

720

657

518

267

303

724

24,208

Omaha, Neb.

1,014

937 1,174

682

890

710

680 1,086

11} 681

New York, N. Y.

863

993

618 1,649 1,248

771 1,170 1,361

140,395

Charlotte, N. C.

260

418

0 1,099

681

464

636

783

12,344

Greensboro-High Point

and Winston-Salem, N. C.

337

498

78 1,170

757

486

735

850

8,217

-,

TABLE III. MILEAGES (Continued)

Locations Markets

<1l

1::1

.....-.1..

..-1

<0

~

0
Q) 1-1

p.. p..

Q)
......

Q)

~ <1l

<1l <1l .j..l..-1
1::1 00
<1l 1-1
...... 0 .j..l Q)
<~

~<11
]~ 1::1 ~
..-1 ...... ~<

.j..l <1l .j.JtJ
0 .-t.d
1-1 .j..l
<1l 1-1
.ud z0

<0 <1l Ill ...... <1l
.-t M
<1l Q)
AE-<

..-1
1::1 <0 0 Ill Ill..-!
~Ill
tl Ill <ll..-1
...,~

.-t:>..
>..-~~
Ill t~l ..-l.j..l ~ 1::1 0 Q) ...:I~

Q) <0 <0
..-1 Ill .d Q)
'
Q) Q) ~E-<

~ ~
1-1..-1
0 <0
..-1 ~ ~
Q) 0 Z...:l

<0 .j..l
.d
00 ..-1 Q)
~

I

Cincinnati, Ohio

466

499

492

981

703

109

500

843

8,058

N
w
I

Cleveland, Ohio Columbus, Ohio

709

741

574 1,218

953

351

741 1,084

560

593

501 1,092

813

221

610

943

14,924 5,396

Oklahoma City, Okla.

912

745 1,126

214

602

814

487

683

13,396

Philadelphia, Pa.

776

893

528 1,561 1,163

689 1,076 1,276

34,916

Pittsburg, Pa.

741

812

535 1, 282

977

404

803 1,116

15,440

Columbia, S. c.

219

383

95 1,065

646

518

654

742

27,266

Greenville, S. C.

159

317

101

989

571

419

541

672

15,216

Knoxville, Tenn.

201

267

226

896

521

267

415

632

6,401

Memphis, Tenn.

423

255

636

470

213

383

0

410

46,663

Dallas, Texas

839

672 1,099

0

418

879

470

498

50,968

Houston, Texas

842

677 1,101

242

422

987

580

391

58,006

San Antonio, Texas

1,022

864 1,288

278

610 1,137

726

585

15,549

Milwaukee, Wise.

814

772

865 1,054

840

399

652 1,067

11 J 615

Totals

28,034 27,545 29,858 36,705 30,461 24,468 26,818 35,612

---------------------- -,

Locations Markets

<I)
QJ
..~..
0
~ til
<I) ;3: QJ 0 .::IH

TABLE IV. WEIGHTED MILEAGES

<I)
0..-1 000 til ~
.0........-..1..
...c: .......
UH

..<.I.)
.......
0
~til
..m....m..
""0""0
~ ~ HH

....

.j..l

til

~
..~..

0
~

..0..

.u..t..o.C:

QJ ~ QJ...t
~ ~
til 0 ;3: 0
........ ....... <I)
~;3:

..I<: ...I<:
1-< 1-< 0 0
:><:><
~ ~ zz

~
OOtll ~ 00
.... .-4
<ll...C: ~ 0 tll...t ...:I~

...~<:
0 1-< ....... 0
<1l :><
4-1 4-1 ;3:
p::l::jQzJ

QJ

1-<""0

s0 ~ til

.-4
.j..l

...>.... .

....... 1-<

til til

p::j~

Birmingham, Ala.

12,989 10,210 7,930 7,537 11' 660 14,998 11' 358 14,092 12,189

Little Rock, Ark.

7' 767

8,746 7,489 8,455

9,785 17,346 10,955 14,223 14,514

Los Angeles, Cal.

50,257 58,698 58,331 63,872 59,835 80,049 66,823 73' 561 75,900

Bridgeport and

New Haven, Conn.

6,313

4,666 3,930 3,569

5,116

390 4,035 2,143 1,226

Hartford, Conn.

5,844

4,350 3,737 3, 410

4,695

540 3,701 1,794 1,299

Miami, Fla.

35,471 31,480 27,556 26,090 33,261 30,352 32,472 33,780 26,045

I N

Tampa, Fla.

17,215 14,891 12,805 12,076 15,984 15,042 15,431 16,952 12,641

.p-

Atlanta, Ga.

52,695 41,910 32,180 27,315 47' 713 50,585 44,547 52,812 40,679

I

Chicago, Ill.

17' 920

0 10,037 15,870

5,097 43,669 10' 773 28,272 35,156

Indianapolis, Ind.

8,021

3,246

0 1,869

4, 775 12,066 4,215 8,361 9,602

Cedar Rapids, Iowa

265

505

878 1,123

551 2,374

960 1,701 1,992

Des Moines, Iowa

0

2,856 3,953 4,875

3,049 9,758 4,674 7,295 8,535

Sioux City, Iowa

820

2,098 2' 772 3,230

2,081 5,598 2,951 4,337 4,882

Kansas City, Kan.

4,928 12' 056 11,578 14,329 13' 683 29,255 16,601 23,467 25,476

Wichita, Kan.

5,638

9,647 9, 513 11' 036 10,555 19,253 12,452 16,140 17' 409

Louisville, Ky.

6,246

3,321 1,260 1,164

4, 260 8,232 3,673 5,840 6,588

New Orleans, La.

65,735 60,817 52,500 52,476 66,420 84,721 67,851 79,367 73,018

Boston, Mass.

10,424

7,875 7,431 6,859

9,035 1,762 6,351 3, 652 3,072

Detroit, Mich.

3, 729

1, 714 1, 677 1,573

2,267 3,962

516 1, 566 3,188

Minn.-St" Paul, Minn. Jackson, Miss.

4,079 8,946

6,489 8,056

9,441 11,017 7,070 7,451

5,423 19,484
a, 904 13,229

9,703 14, 771 17,104 9,847 12,285 11,236

S t . Louis, Mo .

8,909

7,045 5,762 8,376

9,393 23,458 11' 741 17' 720 19,754

Omaha, Neb.

1,635

5, 677 6,915 8,177

5,864 15,127 8,153 11' 809 13,410

New York, N. Y.

163,560 116,668 99,680 90,836 130' 708

0 99,821 52,367 25,833

Charlotte, N. C.

12,949

9,567 7,197 6,073 10,678 7,.629 9,098 9,394 5,345

Greensboro-High Point

andWinston-Salem, N.C.

8,800

6,549 5,259 4,454

7,346 4,347 5,949 5,390 2,810

...,

TABLE IV. WEIGHTED MILEAGES (Continued)

Locations Markets

<ll Q)
l:l ..-1 0
:E til
<ll ~
Q) 0 QH

<ll 0..-l
tti>l OOs::
0..-l ..-1 ...... ,.d.--1
UH

<ll
....-.4..
0 0.. til til
~ ~
..-1 ..-1
'"l:l0'"s0::
HH

..-1
.j..l
gtil
..-1 0 0 l:l..-1
u o ..-l,.d

Q) l:l
Q)-,.l
s:: ..1<: <ll
;::J
til 0 ~ 0
...... <ll ..-l..-1
:E::s:

~~ >0<>0<
~ ~
z z Q) Q)

!)()~
s:: !)()
..-1 ..-1 <ll..d l:l 0 tll..-1
...:!:E

0~
...... 0
<II>'
4-1 4-1 ~ ;::J Q) IX1Z

Q)
1-<'"0
~ ~ .-4 ......
.j..l
'"' ..t..i.l. til
1X1:E

Cincinnati, Ohio

4,690

2,434

886

0 3,159 5,214 2,232 3,513 4,029

Cleveland, Ohio

10,118

5,000 4,567

3,612 6,552 7,283 3,328 2,865 4,970

I
N

Columbus, Ohio

3,480

1,689

934

583 2,169 2,898 1,273 1,791 2,115

Vl

Oklahoma City, Okla.

7,542 11,387 10,677 11' 989 12,244 20,442 13,905 17,200 18,326

Philadelphia, Pa.

37,814 26,990 22,765 20,147 30,133 3,073 22,137 12,709 3,561

Pittsburg, Pa.

12,182

7,118 5,497

4,385 8,492 5,620 5,265 3,428 3, 227

Columbia, S. C.

30,293 22,576 16,959 14,615 24,376 19,168 22' 822 23,503 14,097

Greenville, S. C.

15,627 11,108 8,186

6,695 12,583 10' 925 10,910 13 J 101 8,110

Knoxville, Tenn.

5,556

3,668 2,464

1,799 4,263 4,647 3,572 5,402 3,463

Memphis, Tenn.

29,724 26,225 21,138 23,332 30,424 54,596 32,617 43,537 44,703

Dallas, Texas

36,136 48,674 46,891 50,000 53,720 83,180 59,276 71,814 73,445

Houston, Texas

55,338 63,923 60,384 63,227 68,447 99' 132 75,350 88,459 86,081

San Antonio, Texas

15,347 19,312 18,426 19,405 20,587 29,372 22' 313 25 J 811 26,185

Milwaukee, Wise.

4,228

1,127 3,264

4,553

0 10,814 3,450 7, 271 8,816

Totals

789,230 690,368 620,019 627,454 765,287 869,590 753,101 833,495 770,031

--------------------------------------

TABLE IV. WEIGHTED MILEAGES

Locations Markets

l1l l1l
"s':-:'..-b<O
,..l.1.,l 10-1
"'-' Q) <(C!J

~ .a
bO<U
~ ~
..~...,.g...,
I'Q<(

,...sl.1..:..l:,

0 Q) 1-1

"'-' l1l
"'-'U 0
.-~.a

1-1"'-'

.ula1l

1-1
z 0

<I) l1l <I)
,,........,, >l1:l
l1l Q)
AH

.-l

0..

s::

..0....
<I)

0 <I)

<ll-M

~
u

<I) <I)

l1l ....
..,:::

,..Q..),

,...., :>,

..:>..

~
u

<I) ;j

-;Mj "'s-::'
0 Q) ....:!:><::

Q) Q) <I) <I)
.a.... <I) Q)
~
Q) Q)
::E!H

<sI:):

<1l Q)

<s1::l

,...., l1l

1-o-M
0 ..<.I.)

~ ;j
Q) 0 Z...:l

Birmingham, Ala. Little Rock, Ark. Los Angeles, Cal.

2,386 7,436 63,733

0 5,226 59,080

6,313 10,320 68,494

10,150 4,499
39,786

3,836 3,559 52,039

6,026 7,304 59,222

3,852 1,832 50,786

5,513 6,099 53,932

Bridgeport and

New Haven, Conn.

4,666

5,251

3,429

8,515

6,523

4,105

6,022

7,139

Hartford, Conn.

4,505

5,027

3,324

7,883

6,153

3,905

5,699

6, 684

Miami, Fla.

15,154

17,657 16,935

30,984

21,468

25,166

23,497

20,002

I

Tampa, Fla.

5,793

6,861

7,452

14,074

8,595

11' 448

10,216

8,168

N
Q"\

Atlanta, Ga.

0

9,261 15,240

49' 178

24,677

25,322

24,794

30,069

Chicago, Ill.

37,573

35,524 40,726

50,185

39,938

16,343

29,533

51,341

Indianapolis, Ind.

9,330

8,922

9,908

15,634

11,335

z,oos

7,698

14,360

Cedar Rapids, Iowa

1,914

1,830

2,172

1,841

1,845

1,140

1,380

2,279

Des Moines, Iowa

7,530

7,203

8,786

5,939

7,069

4,900

5,336

8,845

Sioux City, Iowa

4, 603

4,279

5,186

3,234

4,104

3,263

3, 217

4,907

Kansas City, Kan.

19,256

17,367 23,538

11' 913

16,290

12,582

11,195

21' 003

Wichita, Kan. Louisvi11e, Ky. New Orleans, La. Boston, Mass.

13' 869 4,612
31,934 8,606

11' 343 4,260
22' 720 9,686

15,405 4,954
48' 741 6,708

5,224 9,385 31,000 14,830

10,101 6,331
12' 139 11' 670

9,941 0
45,878
7' 725

7,496 4,089 25,522 11,027

11' 250 7,869 0
12,567

Detroit, Mich. Minn.-St. Paul, Minn. Jackson, Miss. St. Louis, Mo. Omaha, Neb. New York, N. Y. Charlotte, N. C.

4,552 16,950
4,463 13,314 11,845
121' 161 3,209

4,638 16,347
2,692 12,370 10,945 139,412
s, 160

4,233 18,433
7,219 17,430 13,713 86,764
0

7,335 14,833
4,431 15,905
7,966 231,511
13' 566

5,787 16,749
0
12,540 10,396 175,213
8,406

2,230
11' 264 6,286 6,464 8,294
108,245 5,728

4,613 13,350
2,258 7,335
7' 943 164,262
7,851

6, 751 19,762
2,067 17,527 12,686 191,078
9,665

Greensboro-High Point and Winston-Salem, N. C.

2,769

4,092

641

9,614

6,220

3,993

6,039

6,984

TABLE IV. WEIGHTED MILEAGES (Continued)

Locations Markets

aS aS -1..1..-1 !:lOll aS 1-1 .-10
<-1..1 t !Ql)

m
~aS
~ m
e~
.I.-Q1 <.C-I

aS l:l ..-1
.-I 0 Q) 1-1
-1..1 aS -1..10
.0-l..d
1-1 -1..1 aS 1-1
..d 0 tJZ

Ill aS Ill .-I aS
.-I M
aS Q) AE-1

..-1 p.. p.. ..-1
l:l Ill
0 Ill !ll..-1
u~Ill
Ill al..-1
.,~

Q)
.-I
.-I :>-.
>..-1~
u
Ill ;:l ..-1-1..1
;:l l:l
0 Q)
....:1:.0::

Q) Q)

Ill Ill

....-d1

Ill Q)

ft

Q) Q) ~E-1

aIll aS Q) l:l .-I aS 1-1..-1
0 Ill ..-1
~ ;:l Q) 0
Z....:l

Cincinnati, Ohio

3,755

4,021

3,965

7,905

5,665

878

4,029

6,793

I

Cleveland, Ohio

10,581

11,059

8,566

18,177

14,223

5,238

11,059

16,178

N -...J

Columbus, Ohio

3,022

3,200

2,703

5,892

4,387

1,193

3,292

5,088

Oklahoma City, Okla.

12,217

9,980 15,084

2,867

8,064

10' 904

6,524

9,149

Philadelphia, Pa.

27,095

31,180 18,436

54,504

40,607

24,057

37,570

44,553

Pittsburg, Pa.

11,441

12,537

8,260

19,794

15,085

6,238

12,398

17,231

Columbia, S. C.

5, 971

10,443

2,590

29,038

17,614

14,124

17' 832

20,231

Greenville, S. C.

2,419

4,823

1,537

15,049

8,688

6,376

8,232

10,225

Knoxville, Tenn.

1,287

1,709

1,447

5,735

3,335

1,709

2,656

4,045

Memphis, Tenn.

19,738

11,899 29,678

21,932

9,939

17,872

0

19' 132

Dallas, Texas

42,762

34,250 56,014

0

21,305

44,801

23,955

25,382

Houston, Texas

48,841

39,270 63,865

14,037

24,479

57,252

33,643

22,680

San Antonio, Texas

15,891

13,434 20,027

4,323

9,485

17,680

11,288

9,096

Milwaukee, Wise.

9z455

8z967 10Z047

12z242

9z757

4z634

7Z573 _5393

Totals

635,638 623,925 688,283 830,910 675,616 611' 735 626,893 760,723

their competitive positions in either the national or the southern regional market. In the process they would gain any advantages to be offered by newer production equipment and layout resulting from relocation or expansion of production facilities.
The structure of freight rates, which often involves zones of equal cost, makes it unnecessary to locate in a mathematically determined position to obtain the desired transportation advantages. From the more practical point of view, it is sufficient to choose a location near the market center which has other desirable location advantages, such as good transportation facilities and a plentiful labor supply with the necessary skills or trainability.
A manufacturer who wishes to concentrate primarily on regional sales, should choose a southern location for much the same reasons. A well chosen southern location would obtain regional market advantages, and would be near the area toward which the national market center is moving. Such a location would enable a producer to improve national penetration over time.
Regional specialization is not feasible unless the regional market in question is sufficient to absorb a major part of the manufacturer's output, and has enough growth potential to support planned increases in output. Of the two regions which meet the first requisite, the South is undoubtedly in a more favored position, as it has the greater growth potential. If regional specialization is at all desirable, there is little doubt of the choice between regions.
The remainder of this study is a consideration of the vicinity of Atlanta, Georgia, as a manufacturing location for the room air conditioner industry.
-28-

V. ATLANTA AS A LOCATION FOR THE ROOM AIR CONDITIONER INDUSTRY

Of the first six cities in rank of nearness to the national market center,1- / only three lie south of Louisville: Birmingham, Memphis, and Atlanta. These are the cities which will improve their positions as the market center shifts southward. Of these three, Atlanta has the advantage of being approximately equidistant from the East Texas-Gulf Coast markets and the Middle Atlantic-Eastern Seaboard markets, a fact of some importance in terms of transport time. Also, Atlanta has better developed transportdistribution facilities. Of course, a realistic plant location decision would have to be based on additional factors, such as desirable site availability and a study of the actual dollar costs of distribution for each of the three cities. As a preliminary step in that direction, certain general factors will be given attention in this report. The Labor Market

Among the many factors to be considered in a location decision, the supply of suitable labor is perhaps one of the most important. The Atlanta labor supply is favorable for industry.
The standard definition of the Atlanta labor market area includes Fulton,
DeKalb, Cobb and Clayton~1 counties, although employers in the area actually
draw on a much larger labor market area. The population in these four counties has increased nearly one-third since the 1950 Census. The trends by county are shown in Table V.
TABLE V GROWTH OF METROPOLITAN ATLANTA POPULATION
1950-1958

County Fulton DeKalb Cobb Clayton
Totals

PoEulation

1950

1958

473,572

564,500

136,395

222,000

61,830

95,500

22,872 694,669

33 1000 915,000

Per Cent Increase
19.2 62.8 54.5 44.3 31.7

l/ See page 19.
2/ After this study was completed Gwinnett County was added to the Atlan~ labor market area.

-29-

Some of this increase may be attributed to the employment opportunities

in the transportation industry in these counties, but more fundamental is the

general economic evolution from agriculture to industry, improved agricul-

tural technology, and a tendency for displaced farm labor to migrate to urban

areas with better job opportunities.

Commuting is also an important factor in the labor supply. A recent

study of commutingl/ shows that 29 per cent of employees in the four-county

area do not reside in the county in which they work. Further, 15 per cent of

Metropolitan Atlanta workers do not reside within the four-county area. It is

also noted that:

The effect of prestige firms with high wage scales on the

relative number of workers from outside counties can be

determined from statistics on the aircraft and automobile

assembly operations, involving several different establish-

ments in the Greater Atlanta area. The data indicate that

these plants obtained 54 per cent of counties other than the county where

their workers the plant is

lofrcoamted,2-1

Labor Costs
Much has been said and written about the higher productivity of southern workers. This subject received national attention in 1954 in an article by Robock and Peterson in the Harvard Business Review.l/ As pointed out in this article, the preponderance of case studies indicates lower labor costs, not merely lower wage rates, than in other parts of the country. This is accounted for by the large reserve of labor that new industry can draw upon, which enables an employer to be extremely selective in hiring applicants,
It is doubtful that wage differentials which may now exist can be expected to endure indefinitely. However, as the differentials decrease, the more attractive wage rates will cause more workers to enter the market for industrial labor, maintaining the advantages of selectivity.
Technical Training
Many manufacturers moving into a new location bring a cadre of trained to act in a supervisory role, particularly during the training
1/ "Analysis of Intercounty Commuting of Workers in Georgia," John L. Fulmer, Industrial Development Bran~h, Engineering Experiment Station, Georgia Institute of Technology, Atlanta, Georgia, August 1958.
~/ Ibid., p. 12
l l "Fact and Fiction About Southern Labor," Stefan H. Robock and John M.
Peterson, Harvard Business Review, March-April 1954.

-30-

period. For some types of operations, employers prefer to provide all of the training required, selecting applicants on the basis of aptitude rather than acquired skills or special technical knowledge. This may be especially desirable on assembly line operations, such as are found in room air conditioner plants. Much of the available labor in the Atlanta labor market will be of the unskilled, untrained type.
There will be an eventual, if not initial, need for some persons possessing skill and training in refrigeration mechanics. This need can also be met in the Atlanta labor market. The Southern Technical Institute, a unit of the Georgia Institute of Technology, is located in Chamblee, Georgia, approximately 13 miles northeast of Atlanta. "Southern Tech" offers an accredited Associate in Science degree in 11 technical fields, including heating and air conditioning. The number of graduates in heating and air conditioning technology is indicated in the following tabulation:

Year 1954 1955 1956 1957 1958

Graduates 17 22 32 34 27

Another advantage of an Atlanta location, in terms of educational facilities, is found in the proximity of the Georgia Institute of Technology and Emory University in Atlanta, and the University of Georgia in Athens. Georgia Tech is the site of the largest enginee~ing research facility in the South.

Proximity to Markets

Atlanta's proximity to the center of the room air conditioner market has already been developed in terms of geographical location. The advantage that would be obtained by locating near Atlanta is further enhanced by the ease of accessibility to the entire market area.
Long the distributim, center for the South, Atlanta has well developed transporLation facilities for shipment to any part of the region or nation. There are 13 main lines of 7 railway systems radiating from Atlanta. More than 250 merchandise and package cars originate in and move out from the city daily, in addition to regular car lots. Through express car service is operated between Atlanta and Boston, New York, Chicago, Cincinnati, St. Louis,

-31-

r

l

l

Los Angeles, Jacksonville, Miami, New Orleans, and other cities. At the time of this writing there are 65 regularly scheduled, inter-

state, general commodity truck lines serving Atlanta, along with approxi-

mately 30 contract carriers.

It is worthy of note that Atlanta will be a point of intersection for three of the interstate highways. Thus six super-highways will radiate from

Atlanta to all sections of the nation; and doubtless will increase the city's advantages as a transportation-distribution center.

-32-

Appendix I
ESTIMATES OF SALES
The sales estimates are based on sample sales data published in
Electrical Merchandisingl/ and Electric Light and Power.~/ Basically, the
data consists of reports from electric utilities which estimated the number of unit sales of room air conditioners for 1957 and the number of domestic customers on residential rates in their territories. In the case of Electrical Merchandising, some 246 utilities, serving approximately 85 per cent of the nation's domestic customers, cooperated in the survey. One hundred and fifty-four utilities, serving approximately 80 per cent of the nation's domes~ic customers, contributed to the survey reported in Electric Light and Power, This later survey duplicated extensively the coverage of the first survey, and was used to amend the earlier estimates of sales and number of customers wherever possible.
The total number of customers in each state was obtained from the Statistical Bulletin, Electric Utilit~ Industry in the United States, published by the Edison Electric Institute. Since these data are reported as of the end of the year, the 1956 and 1957 data were averaged to obtain a result more representative of the sales period,
From these data, unit sales and number of domestic customers, the rate of buying for the year 1957 could be determined for the sample, on the state level. The method used to estimate total sales for each state was to assume that the rate of buying in the sample is applicable for the total number of customers in each state. The estimates are displayed in Appendix I.
As a check on the result for the nation's total sales, estimates of beginning and ending inventories and actual production were used. According to Electrical Merchandising, inventories at the end of 1956 and 1957 were approximately 450,000 and 750,000 units, respectively, Production in 1957 as reported
by the U.S. Department of Commerce~/ was 1,586,094 units. If the inventory
are assumed correct to the nearer 50,000 units, then the allowable range
l/ Statistical and Marketing Issue, McGraw-Hill Publishing Company, New York, January, 1958.
Twenty-Eighth Annual Major Appliance Survey," Haywood Publishing of Delaware, Chicago, Illinois, July 15, 1958, p. 66 ff. ~/ Facts for Industry series
-33-

for total estimated sales is from 1,236,000 to 1,336,000 units.l/ The total state sales imputed from the sample, as described previously, are 1,240,500 units, well within the allowable range, considering that this total does not include estimates for Maryland, Virginia, and the District of Columbia.
Since the sample data are not random, there is no precise measure of the accuracy of individual state estimates. Obviously, the larger the sample for a state the better the estimate of the state total. Some indication of relative accuracy among the states is found in the per cent of the total customers included in the sample for each state as indicated in Appendix Table II.

l/ Derived as follows:
Beginning inventory Add production Total available stock Less ending inventory
Total sales

425,000 to 475,000 units

1,586,000

units

2,011,000 to 2,061,000 units

725,000 to 775,000 units

1,236,000 to 1,336,000 units

-34-

Appendix Table I
ESTIMATED SALES OF ROOM AIR CONDITIONERS BY STATE, 1957

Region or State
New England
Maine New Hampshire Vermont Massachusetts Rhode Island Connecticut
Middle Atlantic
New York New Jersey Pennsylvania
East North Central
Ohio Indiana Illinois Michigan Wisconsin
West North Central
Minnesota Iowa Missouri North Dakota South Dakota Nebraska Kansas
South Atlantic
Delaware Maryland Washington, D. C.

No. of Units
28,700
900 1,500
100 11,000
600 14,600
222,000
130,300 27,900 63,800
151! 900
41,700 31,600 54,400 8,500 15,700
123,100
17,400 20,300 42,400 4,100
1,500 12,300 25,300
208 ' oool1
1,800

Region or State

No. of Units

South Atlantic (Contd.)
Virginia West Virginia North Carolina South Carolina Georgia Florida

2,200 28,700 43,800 78,900
52,6~0

East South Central
Kentucky Tennessee Alabama Mississippi

122,800
13,100 72' 900 26,200 10,600

West South Central
Arkansas Louisiana Oklahoma Texas

309,600
17,000 81,000 22,900 188,700

Mountain
Montana Ideh.o Wyoming Colorado New Mexico Arizona Utah Nevada

17,000
500 1,700 1,000 2,000 2,800 6,300 1,000 1,700

Pacific
Washington Oregon Calitornia
Total

57,400
2,800 2,800 51,800
1,240,500

l/ Regional total does not include estimates for Maryland, Virginia,
and Washington, D. C.

-35-

Appendix Table II
PER CENT OF TOTAL RESIDENTIAL UTILITY CUSTOMERS INCLUDED IN THE SAMPLE

Region or State
New England
Maine New Hampshire Vermont Massachusetts Rhode Island Connecticut
Middle Atlantic
New York New Jersey Pennsylvania
-Ea-st -N-or-th Central
Ohio Indiana Illinois Michigan Wisconsin
-We-st -N-ort-h Central
Minnesota Iowa Missouri North Dakota South Dakota Nebraska Kansas
~ Atlantic
Delaware Maryland Washington, D. C. Virginia West Virginia

Per Cent
71.4 14.4 72.1
so. 8
86.6 86.0 90.3
77 .o
84.3 27.2 93.6
65.8 66.4 19.2 74.2 84.3 60.8
44.9 63.3 30.5 53.4 4.9 30.0 49.1 30.8
58 .ll/
64.4
46.8

Region or State
~ Atlantic
North Carolina South Carolina Georgia Florida
~~Central Kentucky Tennessee Alabama Mississippi
- - West South Central Arkansas Louisiana Oklahoma Texas
Mountain
Montana Idaho Wyoming Colorado New Mexico Arizona Utah Nevada
Pacific
Washington Oregon California
Total U.S.

-Pe-r C-en-t
87.3 27.4 55.8 30.0
48.8
36.9 41.6 72.8 45.4
67.3
65.3 63.7 76.2 66.3
59.2
72.5 60.0 14.5 70.5 4.3 68.1 88.5 33.1
79.7
16.1 48.3 95.9
65.711

l/ Totals do not include data for Maryland, Virginia, and Washington, D. c.

-36-

Appendix II FORECAST METHODOLOGY

An arithmetic graph of production data from 1947 to 1957 reveals quite clearly that after the tremendous expansion of output from 1952 to 1954, the industry was faced with an inventory problem. The recessions of 1954 and late 1957 contributed to this problem and incidentally lend an element of conservatism to the forecasts.
To overcome the effects of inventory adjustments, two-year moving averages of the production data were used as the basis of the forecasts. This adjustment of the-data also has a tendency to bring production more in line with actual sales, so that although technically production is being forecast, the results should be reasonably close to sales.
The Modified Exponential Method
After the adjustment, the data still show no obvious tendency to regularity of growth (Figure 2). It will be noted that the logarithmic graph passes through three cycles of these orders of magnitude: 10 1 , 102 , and 10 3 . This characteristic can obscure regularity in growth simply through differences in magnitude of the data. A transformation was applied to reveal any such hidden tendency.
To each production datum, a constant factor (K) of 547l1was added. This
transformation has the effect of giving more emphasis to the increases from year to year. The transformed data is graphed in Figure 3.
By a linear extrapolation of the last section of the curve in Figure 3, and subtracting K, the first forecast of 2,250,000 units in 1959-60 was obtained. For short term forecasts this method, termed ''fitting a modified exponential curve," may be quite satisfactory.

1/ Derived by grouping the data into three parts: 1947-48, 1948-49,

1949-so, 1950-51; 1950-51, 1951-52, 1952-53, 1953-54; 1953-54, 1954-55, 1955-56,

)J -:- 1956-57. Summing and averaging, the mean of part 1 (designated as

124.75,

M2

= 609.25,

M3 = 1,443.00.
K = [M; - (M1M3

Then,
~Ml +

M3)

-

2M2]

= 547.

M1)

is

The K is added algebraically, i.e., a negative K would be subtracted.

-37-

Correlation Method

As a check on the results of the first method, U.S. total production and personal income were tested for degree of correlation for the purpose of using an income forecast as a basis for a production forecast. The advantage of such a method is that the wide range of factors determining income provides a much more stable base, and consequently greater reliability, than assumptions that might be made for the future of a particular product.
Two adjustments were made in the income data. The data were deflated to reflect price level changes, and converted to a two-year moving average series to increase the time-comparability of the two sets of data.
By least squares regression the following equations were specified, with the indicated correlation coefficients:

P = -4,313 + 21.22158(Y)

(r = 0.96)

Y = 193.85 + 9.7212(T)

(r 0.99)

where P = Production, Y = Income, T =Time, origin at 1947-48.

The forecast obtained for 1959-60 is 2,280,000 units, which closely agrees

with the former forecast of 2,250,000 units.

-38-