Application of Input-Output Models to Agriculture

Author(s): Earl O. Heady and John A. Schnittker

Source: Journal of Farm Economics , Aug., 1957, Vol. 39, No. 3, Part 1 (Aug., 1957), pp.
745-758
Published by: Oxford University Press on behalf of the Agricultural & Applied
Economics Association

Stable URL: https://www.jstor.org/stable/1234429

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].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

and Oxford University Press are collaborating with JSTOR to digitize, preserve and extend
access to Journal of Farm Economics

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT
MODELS TO AGRICULTURE*

EARL O. HEADY AND JOHN A. SCHN1r'r.rsJi

Iowa State College

ROBLEMS of the agricultural economy and their solution

ingly spring from its interrelations with other sectors of the n
economy. Few attempts have been made to describe and pre
interrelationships. Input-output analysis is one simple descriptiv
can be used for this purpose.' Its usefulness as a predictive de
fully established.2 The technique probably has not been used w
sis on agriculture (a) because it appears to be a complicated
and (b) because it has, as do most techniques, limitations in pr
away from a given point in time. The purpose of this paper is
the simple logic and computational steps of input-output anal
results from a simple model are also presented. There are a
methods by which interrelations of various sectors of the eco
be studied.3 It is not the purpose of this paper to weigh the d
vantage of each. However, a brief review of the restrictions a
tions of input-output analysis is included.

Requirements and Nature of Analysis

Input-output analysis involves mathematics scarcely beyo
arithmetic. Any person interested in the method can unde
matrix algebra in a period of two or three days study. Mathem
not the greatest difficulty in input-output analysis. Final calcu
volving matrix inversion can be made in a very short time. The
tively inexpensive, even where a machine is hired for a problem
to 30 to 40 sectors. The larger cost, both in time and funds, i
in assembling data from the census and other sources. Usefuln
sults depends more on this unromantic process of assembling
on intricacies of mathematics. The main work is not unlike a
scriptive analysis of census data. As a guide for efficient cler
however, it is important that the model of the relevant secto
gregation procedures be set forth explicitly. Unfortunately, s
data for this and similar aggregative analyses is far from perfe

* Journal paper J-3157 of the Iowa Agricultural Experiment Station, pr

1 See, as a basic reference, W. W. Leontief, Structure of the American
1919-1929, 2nd Ed., New York: Oxford University Press, 1951.
2 See especially, National Bureau of Economic Research, Input-Outpu
An Appraisal, Princeton: Princeton University Press, 1955.
For example see, L. R. Klein, A Textbook of Econometrics, Evansto
Peterson, 1953.
745

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
746 EARL 0. HEADY AND JOHN A. SCHNI-rK;Kn

tive judgment supplements data at every step in deciding which inputs

and outputs can be meaningfully thrown into one category.

Clerical and Arithmetic Steps

If a preliminary model defining the sectors of the economy whose
transactions are to be analyzed is given, the clerical steps can be
initiated. These two steps are not, of course, independent: model build-
ing is a continuous process. The clerical opportunity for disaggregating
data of particular economic strata determines which unique sectors
can be meaningfully analyzed. The sectors finally selected for analysis
will define the clerical steps to be used in collecting and aggregating
the data from the various sources. After the sectors (industries or seg-
ments of industries) have been decided on, the first clerical step is as-
sembling the inputs (value of materials and services) that one particular
sector draws or uses from each other sector. Conversely, this process is
one of dividing the total output of one sector into the parts that go to
other sectors. Suppose that the sectors are designated by numbers and
that we have sectors 1, 2, 3, 4, . . . n; where 1 may be Midwest agri-
culture, 2 may be New England agriculture, 4 may be the farm machinery
industry, n may be the transportation industry, etc. We can then desig-
nate the output of these sectors as X,, X2, X, X4, . . . X, where X1 re-
fers to the total output (value of) Midwest agriculture, X2 refers to the
total output of New England agriculture, Xn refers to the value of serv-
ice output by the transportation industry, etc. In other words, clerical
operations first include computation of the total value of output, X1, for
Midwest agriculture. The same step is repeated for each other sector.
Next, clerical operations are used to determine what part of the output
of one industry or sector moves to (is an input for) each other industry or
sector. Thus we designate x12 as the amount of output from sector 1 used
by sector 2 (i.e., the absolute amount of X1 which goes to X2), x24 as the
amount of product from sector 2 used by sector 4, etc. In general, these
input quantities are designated as xij where i refers to the sector produc-
ing the input and j refers to the sector using it (i.e., the i represents the
row of the input-output table, system of equations or matrix and the j
representing the column). The next clerical step is to compute the abso-
lute amount of the product from each sector which moves into final con-
sumption (i.e., final demand or bill of goods) without moving through
the other industrial sectors. Final demand can include direct consumption
by households, government, foreign trade, or similar segments of the
economy which are defined to be autonomous. Thus we compute Y1, Y2,
Y3, Y4, ... Yn, with Yi representing the amount of Xi "consumed di-
rectly." A table of total inputs (flows) and outputs, such as the one below,
then exists:

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT MODELS 747

INTERSECTOR FLOWS OF GOODS AND NET Ou'TUTrS

Sectors using (Xj) Final Total

demand output
1 2 8 4 n (Yi) (Xi)
1 Xll X12 X13 X14 * * Xin Y1 Xl
2 x21 X22 X23 X24 .* * X2n Y2 X2
8 xi X32 X3a X34 . .? X3n Ya Xs
Sector pro- 4 x41 X42 X4 44 X4n Y4 X4
ducing (Xi)

n Xn l X n2 Xn3 Xn4 .* * Xn Yn Xn

In the main body are included the xij quantitie

top the (j-th) sector using the quantity and along
sector producing it. (The values xij for i = j will
put is defined on a net basis and an intrasector f
example, the number in the cell of the first row
is the quantity of output from sector 1 which is
next to last column, Y1 is the quantity of outpu
directly into consumption; X1 is the total output of
basic table can be establishing the system of e
X1 - xl2 - Xl3 - X14 - xln = Y1
-X21 + X2 - X23 - X24 * * * - X2n = Y2

--Xnl - Xn2 - Xn4 - * * + Xn = Yn

These flow equations involve elementary arithmetic. The first one states
that the amount of sector 1 moving into direct consumption (Yi) is equal
to the output of sector 1 (X1) minus the sum of the amounts (Ixlj) from
sector 1 used by other intermediate sectors. The total flows in tabular
form are of some interest in themselves and may add about as much to
knowledge as the interdependence coefficients (elements of the inverse
matrix) to be computed later.
The next clerical step, one accomplished with a desk calculator, is
computation of the technical input-output coefficients. An input-output
coefficient (aij), not to be confused with the inter-sector flow (xij) de-
scribed previously, shows the amount of output from one sector (the i-th
producing sector) required per unit of output of another sector (the j-th
using sector). These technical input-output coefficients, aij, are computed
as in (2) where the amount of output from sector i, used by sector j, is
divided by the total output of sector j:
xij
(2) aij =
Xj

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
748 EARL 0. HEADY AND JOHN A. SCHNITTRE

From these calculations, it is evident that xlj = aijXj. In other words,

x12 = a12X2, meaning that the total quantity of output from sector 1 used
by sector 2 is equal to the amount of sector 1 used per unit of output
from sector 2, multiplied by the total output of sector 2. Calculation of the
input-output coefficients makes possible the substitution of the aijXj
quantities for the xij quantities in the equations of (1). The result is the
system of equations in (3) which involve only elementary arithmetic.
Again, they say: the amount of product from one (the i-th) sector moving
into direct consumption (Yi) is equal to the total output of that sector
'Xi) minus the sum of the products

( aijXj)

formed by multiplying the input-output coef

output of the consuming sectors (Xj). Althoug
volved to this point, the algebra of computat

X - a12X2 - a13X3 - a14X4 alnX - - Y1

-a21X1 + X2 - a23X3 - a24X4 - a2nXn = Y2
(3)

--lXl - ar2X2- an3X3 - an4X4 *** + Xn = Yn

In the notation of matrix algebra, (3) can now be written in the manner
of (4).

all - a12 -- a13 - a14 . -aln- -Xi- "Y1-

- a21 a22 - a23 -a24 -a2n X2 Y2
(4)

- anl -an2 --a3 -- an4 * * * -an_ _ Xn_ LYn_

The figures in (4) are copied directly from (3), but are simply put in
different form. By the rules of matrix algebra, (4) has exactly the same
meaning as (3). Three matrices (arrays of numbers in a particular order
have been set up in (4). There is a matrix of the input-output coefficient
(with a1n, a22 etc., for all aij's where i = j, equal to 1 where intrasector
flows are not considered) in exactly the order of (3). Also there is a matri
or vector of total outputs, with the row order transposed into a column o
the same order, and a matrix or vector of the quantities used in direct
consumption, in exactly the order shown in (3). The algebraic system o
(4) says that if we multiply the matrix of input-output coefficients by the
matrix of total outputs, we obtain the amounts of each producing sector
moving directly ino final consumption. This is exactly what the arithmet
system in the equations of (3) states. In fact, (3) can be derived directly

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT MODELS 749

from (4), as well as vice versa. In matrix multiplication, we multiply the

element or number in the first row of the first matrix by the corresponding
element or number in the column of the second matrix, and add the
quantities to give the corresponding element or number in the third
(product) matrix. Using this rule, we obtain a,lX1-a12X2-al3X3-a14X4 ....
-ainXn = Yi, the first equation of (3), by multiplying the first row of the
first matrix by the column of the second matrix. Then by multiplying the
second row of the first matrix by the column of the second matrix, the
second equation of (3) results; etc. The only purpose of this explanation
is to show that (4) is the equivalent of (3), and, hence, to allow further
calculations to be based on the concepts of (4).
One of the goals of input-output analysis is to relate output of one
sector to final or direct consumption of the output of other sectors (i.e.,
to express output of producing sectors as a function of final demands).
The equations of (3) do not show this, but rather indicate the amount of
Yi remaining with particular outputs of the various producing sectors
(i.e., express the final bill of goods as a function of output in producing
sectors). The task is to use the data of (3), and hence (4), in expressing
the relation of production in one sector to the final demand for products
of the various sectors or industries. To facilitate the presentation, the
"abbreviations" in (5) are used for the quantities in (4). That is, we let A
represent the matrix of input-output coefficients; X the matrix of total
outputs; and Y the matrix of final demand quantities.4
(5) AX = Y
In simple arithmetic or
any change in Y by sim
inverse of A

i.e., or A-togiveX = - Y or X = A-1Y.

A A

However, s
each but fo
used. Matri
than divisio
(5), the elem
multiply eac
matrix mult
(6) A-1AX = A-1Y
(7) X = A-1Y

4 Each of these abbreviations re

in (4) and the product of the fir
vidual numbers or elements

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
750 EARL 0. HEADY AND JOHN A. SCHNITTKER

gives a product "equivalent" to I (i.e., an identity matrix). Thus (7) is de-

rived from (6): stating that the values of X (the output of particular sectors)
can be derived by multiplying the quantities of final demand by the in-
verse of the matrix of input-output coefficients. Hence, once the quantities
of (4) are established, purely clerical or arithmetic step, equation (7) can be
established to relate the Xi values (the output of particular sectors) to
knowledge of input-output coefficients and the final bill of goods, or final
demand. However, since (7) is in "abbreviated" or symbol form, we use its
equivalent form in (8). In this case, we have substituted a matrix with ele-
ments (cij) to represent A-1. If we multiply these elements by those repre-
senting final demand quantities (Yi), we can "predict" the Xi values (out-
put of particular industries or sectors). Just as the two matrices in (4) can be
multiplied to give the set of equations in (3), the two matrices in (8) can be
multiplied to give the set of equations in (9). Hence, if we know the ele-
ments of the inverse matrix of the input-output coefficients, we can multiply
them by the quantity moving into direct consumption from each sector
(Yi) and predict total output of each sector. We multiply the first row of
cij values (i.e., cij) times the Yi values, the sum of these quantities being
equal to X1. The same procedure is used for X2, except the Yi column
(vector) is multiplied by the second row of cij values (i.e., c2j), etc.

"X1- Cnl c12 C13 c14 * Cln- -Y1-

X2 21 c22 c23 c24 * . C2n Y2
(8) . . .. .

-Xn- -Cnl Cn2 Cn3 Cn4 * Cnn- Yn_

cllYl + C12Y2 + C13Y3 + C14Y4 * * cmnYn = X1

C21Y1 + C22Y2 + C23Y3 + C24Y4 * * C2nYn = X2
(9) .... .

CnlYl +

The cij v
analysis
sociated
The inte
tem, rel
demand
not only
for dire
indirect
mand for

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT MODELS 751

might express the effect of final demand for automobiles (Yn) on the out-
put of hides for leather (X1) and soybeans for plastics (X2) by agricultural
sectors. But, given the appropriate model, indirect demand also would be
reflected back to the agricultural sector through other sectors. Output
of autos requires output of machine tools which might come from an in-
dustry using soybeans for plastics in some of its equipment.
Thus the main mathematical task is computing the interdependence
coefficients. It is not necessary to set down equations and symbols in the
manner of (1), (3), (4), (5), (6), (7), (8), and (9). We did so only for ex-
planatory reasons. One simply computes the flows (xij and Yi values),
and the input-output coefficients (aij values). The latter then are arranged
in tabular or matrix form, in the order presented in the left-hand brack-
eted member of (4). (The aij values along the diagonal, for i = j are 1.)
From these data we compute the inverse directly and obtain the interde-
pendence coefficients. The interdependence coefficients are the "un-
knowns" of our analysis and are solved from the known quantities, the
input-output coefficients.
A simple example of inverting a matrix and obtaining the input-output
coefficients follows. Suppose that we have a two-sector economy where
X = 40, X2 = 80, x12 = 32, 21 = 20, Y1 = 8, and Y2 = 60. By simple
arithmetic an =- 1, a12 = -.4, a,, = -.5, and a22 = 1. (The input-output
coefficient of a "sector on itself," a where i = j, is always 1 and
al1 = a22 = 1.) Thus, we have the input-output matrix of known values in
(11) and wish to compute the unknowns of its inverse in (12). We now

(11) (12) c11 12

--.5 _1 c21 C22J

eCll C~q[ i- -.2

C2(1 C22 -[-.5 1] LO 1i
use the procedure of (13) and set the product of the invers
output matrices to equal the unit matrix in (13). The inv
matrix is one which multiplied by the given matrix will give
By carrying out the multiplication as explained above (th
cij matrix multiplied by the columns of the one with input
cients), we obtain the four equations below. From these w
values of c11, c21, c12, and c22 by simple elimination rul
Hence, we have 1.6 c11 = 2 and c11 = 1.25, etc. The sam
followed and all cij values are computed for the invers
using the procedure outlined for (8) and (9) we can predic
Ci - .5c12 = 1 Ci - .522 = 0 c1n = 1.25 c12 = .5
or
-.4cll+ C12 = 0 -.4C21 + C22 = 1 C21 = .625 C22 = 1.25

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
752 EARL 0. HEADY AND JOHN A. SCHNITTKER

X1 and X2, the net outputs, as in (14) below. The predicted values of
X1 and X2, are identical with those specified at the outset. Thus the em-
pirical part of the input-output

1) Xi = (1.25)(Y1) + (.5)(Y2) or Xi = (1.25)(8) + (.5)(60) = 40

1 X2 = (.625)(Y) + (1.25)(Y2) or X2 = (.625)(8) + (1.25)(60) = 80
analysis is complete and the computational work is not beyond the
average clerk and arithmetician.
The only obstacle is in the amount of arithmetic involved. Although the
"hand arithmetic" outlined above could handle larger problems, the task
of eliminating variables to determine values of the cij coefficients would
become long and tedious with more than 3 sectors. Hence, the least-cost
method of operations is to follow the clerical routine through computa-
tion of the input-output table or matrix, then have the inverse matrix (i.e.,
calculation of the cij values) computed by a large-capacity calculator.
Costs for this step are much less than the costs of clerical operations for
most problems.

An Application to Agriculture

We now present the results of an input-output analysis, with emphasis

on agriculture, based on 1949 data. The small model is inadequate for
any detailed analysis of interindustry relationships. It should be looked
upon as illustrative in nature.
The model includes 12 agricultural sectors; a crop or primary sector
(sectors 1 to 6) and a livestock or secondary sector (sectors 7 to 12) for
each of the six geographic regions shown in Table 1. It includes 8 indus-
trial or nonfarm sectors, with aggregation to emphasize sectors that
furnish inputs to or process products from agriculture.
The objectives of the study include describing relationships between

TABLE 1. COMPOSITION OF AGRICULTURAL REGIONS BY STATES

Region 1 Region 2 Region 3 Region 4 Region 5 Region 6

Me. Ohio Va. N. D. Mont. Ariz.
N. H. Ind. W. Va. S. D. Ida. Wash.
Vt. Ill. N. C. Nebr. Wyo. Ore.
Mass. Wis. S. C. Kan. Colo. Calif.
R. I. Mich. Ga. Okla. N. M.
Conn. Minn. Fla. Tex. Utah
N. Y. Iowa Tenn. Nev.
N. J. Mo. Ala.
Pa. Ky. Miss.
Del. Ark.
Md. La.

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT MODELS 753

agriculture and certain components of the nonagricultural economy. In-

dustry aggregation is based mostly on a detailed study of the United
States economy made by the Bureau of Labor Statistics for 1947.5 The
industry sectors are:
Sector 13. Industries processing the products of primary agriculture, chiefly
for food use, but including livestock feeds as by-products.
Sector 14. Industries processing the products of primary agriculture, chiefly
for nonfood use.
Sector 15. Industries processing the food products of secondary agriculture.
Sector 16. Industries providing machinery, machine services, fuel and oil
to all sectors of the economy.
Sector 17. Industries furnishing fertilizers, seeds and other supplies to agri-
culture, as well as many products to other sectors.
Sector 18. All other industries, including most services and transportation
and merchandising.
Sector 19. Foreign trade.
Sector 20. Government.

Since agriculture is subdivided both geographically and by process,

industry sectors might also be handled in the same way. A regional model
developed by Isard included industrial product detail within regions.6
He cautioned, however, that for such a model ". . . appropriate data are
not currently available." Moses developed a similar model empirically,
using relationships between national industries, and associated data on
regional trade to estimate regional and interregional trade coefficients.7
However, given the data problems of regional consideration of industry,
and the objectives of this study, to stress agricultural production, it
appeared less important to consider industries regionally even at a high
level of aggregation.
Data for 1949 were of very uneven quality, with agriculture data gen-
erally more accessible than industry data. Even in agriculture, data short-
ages required either conceptual changes or estimates with low reliability.
As in the 1947 Bureau of Labor Statistics study, "Judgment, informed
guesses, and just plain hunches were required at many points to make the
detailed allocations."8 Nearly all agricultural data were available in their
earliest form in Agricultural Statistics, with partial revision in later pub-
lications. Industry, trade, and government data came from many sources,

5See U. S. Dept. of Labor, Industry Classification Manual for the 1947 Inter-
industry Relations Study, Washington, 1953, (Mimeo.); and Interindustry Flow
of Goods and Services by Industry of Origin and Destination, 1947. Washington, 1952.
'W. Isard, "Interregional and Regional Input-Output Analysis," Review of Eco-
nomics and Statistics, Vol. 33, pp. 318-328.
7L. N. Moses, "Interregional Input-Output Analyses," American Economic Re-
view, Vol. 45, pp. 803-832.
National Bureau of Economic Research, Input-Output Analysis; An Appraisal,
Princeton: Princeton University Press, 1955.

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 754 EARL 0. HEADY AND JOHN A. SCHNi-rrAriH

but the transactions table of the 1947 interindustry study was the basic
reference for the latter sectors.9 No attempt is made here to give a
detailed description of the methods and sources used to estimate net out-
puts and their distribution. Such a description is available from another
source.'0

Interdependence Coefficients
The interdependence coefficients computed for the model outlined
above are included in Table 2. These quantities are equivalent to the
cij values shown in the matrix of (8) and the equations of (9). If put in
matrix form and multiplied by the 1949 direct consumption of each of the
20 sectors, the product provides the output of the producing sectors. In
the conventions of input-output literature, these coefficients might be
interpreted as indicating the change in output of one producing sector
associated with a dollars worth of change in final demand (direct con-
sumption) for the output of any other sector. However, we prefer to
interpret the quantities shown as the average amount of product in a
particular sector associated, in 1949, with each dollars worth of product
consumed directly from each other producing sectors. In this vein, we do
not suppose that the "fixed mix" representing output of one sector will
be projected into the future as national income increases. Neither do we
suppose that the technical coefficients will remain constant as demand
for the product of any one sector increases. Although the fixed-mix restric-
tion is not a serious limitation when emphasis is on industry,l1 the prob-
lem is more difficult for agricultural sectors that specialize in products
with definite interregional differences in income elasticities of demand.
In the simple model, intersector flows for agricultural regions were
computed only for primary crops. Data on feeder stock were inadequate;
all secondary livestock products were treated as if they went directly to
sector 15. Although there were blanks in the table of input-output co-
efficients, there are none in the table of interdependence coefficients, how-
ever, because interrelationships are expressed both directly to a sector,
and indirectly back through other sector.
Given the model employed, the important elements affecting farm
sectors are magnitudes of final demand for the products of industries
processing the products of agricultural sectors. The figures presented
represent interindustry relationships for a given point in time, namely,

9 U. S. Department of Labor, Interindustry Flow of Goods and Services by In-

dustry of Origin and Destination, Washington, 1952.
10 John S. Schnittker and Earl 0. Heady, Application of Input-Output Analysis to a
Regional Model Emphasizing Agriculture, Iowa Agr. Exp. Sta. Report (unpub-
lished).
1 J. M. Ryan, "The Leontief System," Southern Economic Journal, Vol. 19, pp. 481-
489.

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 TABLE 2. INTERDEPENDENCE COEFFICIENTS BETWEEN THE FINAL BILL OF GOODS AND N

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Primary (crop) agriculture sectors Sec
numbSetor Crop- Nonfood Live- Machin- M
number Region Region Region Region Region Region Region Regio
1 2 3 4 5 6 1 2 3 4 5 6 prod- prod- prod- and sup
ucts ucts

1 1.0239 0.0016 0.0017 0.0012 0.0013 0.0015 0.3013 0.0025 0.0056 0

2 0.0099 1.0365 0.0068 0.0055 0.0067 0.0068 0.0336 0.4014 0.0174 0.
3 0.0076 0.0063 1.0202 0.0051 0.0060 0.0059 0.0184 0.0085 0.4239 0.
4 0.0079 0.0064 0.0058 1.0321 0.0070 0.0064 0.0184 0.0086 0.0139 0.
5 0.0022 0.0016 0.0016 0.0012 1.0376 0.0015 0.0061 0.0025 0.0044 0.
6 0.0041 0.0028 0.0028 0.0022 0.0025 1.0128 0.0176 0.0064 0.0125 0.
7 0.0760 0.0012 0.0011 0.0010 0.0011 0.0011 1.0240 0.0013 0.0020 0.
8 0.0046 0.0814 0.0035 0.0033 0.0039 0.0037 0.0090 1.0338 0.0064 0.
9 0.0012 0.0010 0.0366 0.0008 0.0010 0.0009 0.0021 0.0011 1.0161 0.
10 0.0019 0.0017 0.0015 0.0598 0.0017 0.0016 0.0035 0.0019 0.0028 1.0
11 0.0007 0.0006 0.0005 0.0005 0.0822 0.0006 0.0013 0.0007 0.0010 0.0
12 0.0008 0.0007 0.0006 0.0006 0.0007 0.0346 0.0017 0.0008 0.0013 0.0
13 0.0309 0.0161 0.0178 0.0140 0.0149 0.0158 0.2557 0.0823 0.1770 0.0
14 0.0267 0.0249 0.0233 0.0210 0.0248 0.0228 0.0386 0.0245 0.0325 0.0
15 0.0121 0.0108 0.0095 0.0089 0.0106 0.0099 0.0202 0.0114 0.0161 0.0
16 0.1640 0.1696 0.1357 0.1463 0.1845 0.1608 0.1127 0.1104 0.1094 0.1
17 0.1448 0.0866 0.1056 0.0475 0.0521 0.0817 0.0615 0.0437 0.0584 0
18 0.4674 0.4659 0.4076 0.3991 0.4768 0.4243 0.4624 0.3928 0.4310 0
19 0.0281 0.0252 0.0213 0.0206 0.0248 0.0234 0.0445 0.0256 0.0352 0
20 0.1653 0.1554 0.1140 0.1247 0.1540 0.1457 0.1351 0.1143 0.1132 0

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
756 EARL 0. HEADY AND JOHN A. SCHNIrrL'KR

1949. They represent a fixed mix in the product reflected for each produc-
ing sector; or between producing sectors for any particular Yi quantity
or change. However, these relationships express some interesting condi-
tions for agriculture in 1949. The coefficients in the table show the amount
of output in the row sector per dollar's worth of final demand for the
products of the colum sector. (We can think of the table as the matrix of
interdependence coefficients such as the matrix of cij's in (8); there, the
column headings indicate the Yi elements in the Y matrix and the row-
stub headings for represent the Xi values of the X matrix.) Thus a dollar's
worth of final demand for crop-food products, sector 13, is associated with
an output of only $.0211 in sector 1 (the Northeast), and $.0655 in sector
6 (the Pacific Coast), where a large proportion of fruits and vegetables
move into sector 13 processing, and then to final demand. The interde-
pendence coefficient of sector 13 with sector 5 (crop production in the
Mountain states) is only $.0207. The bulk of crop production there con-
sists of forage crops, which move to livestock in the same region.
The sum of the first six rows in column 13 is only .291. This magnitude
points up, in numerical terms, the existing situation in respect to the
farmer's share of the consumer's dollar spent for crop-food products. Each
dollar of final demand or household consumption of products in sector 13
requires only a 29-cent output by all agricultural crop sectors. The large
interdependence coefficient, 45 cents (column 13, row 18), indicates that
each dollar's worth of consumption of products in sector 13 is associated
with a large output by sector 18. Sector 18 includes mainly transportation
and merchandising services.
A dollar of final demand for sector 15 or livestock products is associated
with a total mix value of 72 cents (i.e., the sum of rows 7-12 in column
15), for the six secondary agricultural sectors. The fact that this figure is
much greater for livestock than for crops expresses the fact that a much
larger proportion of the consumer's dollar reaches the farmer for livestock
products. More than 33 cents of the 72-cent total is drawn from the Corn-
belt where the main farm product is livestock and represents the major
part of the pork, beef and milk consumed by the nation. The second
largest interdependence coefficient is for sector 10 which includes the
western portion of the hog-raising and beef-feeding area; and which pro-
vides a considerable amount of beef processed directly from the range.
Although livestock is the important product of sector 11, most of this
product is range beef and sheep, which moves to the feed lots of sectors
8 and 10, rather than directly to processing in sector 15.
The magnitude of the interdependence coefficients of sector 15 on
regional crop-producing sectors also is of interest. The largest coefficient
again is for the Cornbelt (sector 8). In 1949, a dollar's worth of final

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
 APPLICATION OF INPUT-OUTPUT MODELS 757

demand for the product of livestock processing in sector 15 was associ-

ated with a 14-cent output of crops insector 8. This is true because most
of the crop product in the Cornbelt (sector 2) moves to livestock in the
same region (sector 8), and then into the livestock-processing industry
(sector 15). A dollar's worth of final demand for products in sector 15
(livestock at retail) required a crop output in the Cornbelt greater than
the livestock output in any other region.
Based on the model, final demand for the product of industrial sector
18 had little relationship to the output of agricultural sectors. The co-
efficients range from .0011 for secondary output in the Intermountain
states (sector 11) to .0160 for primary output in the Cornbelt. In contrast,
however, final demand for products of agricultural sectors required a
much greater output from sector 18. These quantities (row 18, columns
1 to 12) range from .3928 in the case of secondary products in the Corn-
belt to .4768 in the case of primary products in the Intermountain states.
Similarly, sector 16 (machinery and fuel) final demand bears only a
trivial relationship to output of agricultural sectors (column 16, rows 1 to
12). However, the opposite is not true. One dollar of final demand for
crop or primary agricultural products in sector 5 (the Intermountain
states) was associated with an 18-cent output in the machinery and fuel
sector. The figure was 14 cents for the Southeast (sector 3), where more
of the work is done by man and horse power, and machine inputs per
dollar of crop output are generally lower than for other regions (row 18,
columns 1 to 6).
Limitations

Other interesting notes relating to magnitudes of interdependence co-

efficients could be mentioned. However. this detail will not be included in
order that a summary of limitations of the model can be included in this
paper. Input-output analysis is, in a sense, descriptive. Although the par-
ticular model used permits description of interrelationships among sectors
at a point in time, it is less appropriate as a tool for exlaining these rela-
tionships, or in predicting the producing structures under long-term eco-
nomic growth. It has projective value mainly for small changes antici-
pated in the immediate future. This last statement holds true because of
the restrictions on the fixed input mix forced into the system of equations
by the linear coefficients. If total demand (the final bill of goods) were
to increase by large proportions, it is not likely (a) that the mix of agri-
cultural products forthcoming within regions, (b) that the relative mix
between agricultural regions, or (c) that the relative mix between agri-
culture and industry, would remain in the constant proportions reflected
by the input and interdependence coefficients. By accounting for change
in coefficients, some useful projections might be possible, however.

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms
758 EARL 0. HEADY AND JOHN A. SCHNrrrITE

As a "diagnostic tool," input-output analysis is generally as useful, and

perhaps more useful for particular purposes, than the great body of
arithmetic, descriptive analyses used so often in statements on agricultural
policy or about changes in the agricultural economy. It has many of the
same limitations as these types of analyses, but it is relatively more formal
and precise. In a statistical sense, there are no degrees of freedom since
one observation is used to predict each parameter and the model has no
stochastic properties. When computed from value quantities and used to
predict physical quantities, the coefficients give rise to problems in index
numbers. Judgment in correct aggregation procedure is always difficult.
However, these assumptions and restrictions are no more serious than
those inherent in arithmetic, descriptive analyses of agriculture based on
agricultural census data.

This content downloaded from

116.73.75.4 on Sun, 09 Feb 2025 08:31:33 UTC
All use subject to https://about.jstor.org/terms