The G"

 

This is to certify that the

thesis entitled

AN INFORMATION SYSTEMS ANALYSIS
OF USDA FARM INCOME DATA

presented by

Charles Henry Riemenschneider

has been accepted towards fulfillment
of the requirements for

Ph-D- degree in Agnimliuxal Economics

 

Mid, / (4%44/44/

Major professor

Date OCtOber 30, 1978

0-7 639

OVERDUE FINES ARE 25¢ PER DAY _

PER ITEM
w Return to book drop ‘to remove
a this chemfrom your record.
MIU
. $6223 "5

 

 

 

 

 

AN INFORMATION SYSTEMS ANALYSIS
OF USDA FARM INCOME DATA

By

Charles Henry Riemenschneider

AN ABSTRACT OF A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Agricultural Economics

1978

 

 

ABSTRACT

AN INFORMATION SYSTEMS ANALYSIS OF USDA
FARM INCOME DATA

By

Charles Henry Riemenschneider

Problems relating to the data base in agriculture have been of
concern in the agricultural economics profession in recent years. U.S.
Department of Agriculture aggregate farm income data have been the sub-
ject of a number of studies during this time. The failure of these
studies to consider explicitly the ultimate data users as key variables
in the analysis of farm income data has often led to an incorrect defi-
nition of the nature of data system problems and has made it difficult
to establish meaningful priorities among the recommendations of these
earlier studies.

This research is based on an information systems paradigm which
emphasizes the use of information in decision making. A mail survey and
personal interviews with farm income data users were the main research
methods used. These yielded results which provided a description of the
farm income information system and helped to define the nature of the
problems in the system.

The theoretical basis for the research was further expanded by
developing the economic implications of the information systems paradigm.
Emphasis in the theoretical area was on economic structure and the dis-

tributions of information and income in determining the appropriate

sir-ix

 

 

Charles Henry Riemenschneider
government role in supplying information.

The farm income information system was found to have four major
components, a primary data subsystem, a formatting and communication sub-
system, an analysis subsystem and a decision making subsystem. Public
policy uses of the farm income data dominated in the system. Major pri-
vate sector uses were in the areas of estimating the demand for farm in-
puts and for credit decisions relating to agriculture. The lack of use
by many of those receiving the data or the low weight often attached to
the farm income data in policy decisions was also a significant finding.

The descriptive results pinpointed a number of problems in the
farm income information system. Two major problem areas were identified
by users. First, conceptual obsolescence is a major problem in the sys-
tem. Through time the issues in agricultural policy have changed but
the concepts of farm income have not. The current system fails to pro-
vide adequate information on the distribution of farm income, especially
by commodity and by legal organization, which are needed to address cur-
rent policy issues. Conceptual obsolescence of a different type is also
apparent because the national family farm data concept currently used is
not a true representation of the reality of the farm sector. This lat-
ter type of conceptual obsolescence does not appear as serious as the
first since aggregate farm income data are used more as social indica-
tors and thus do not require a one to one relationship with reality.
Credibility is a second major problem area. Data revisions through time
have tended to create a credibility problem for the USDA, especially
with regard to analysis of farm income related issues in the policy pro-
cess.

Other minor issues were also addressed. These included the

Charles Henry Riemenschneider

  
  
  
 
 
 
 
 
 
  
 

i'ttjon of USDA farm income data into the national income and product
éll"qnnts, the usefulness of the information systems paradigm as a re-
3.:stlrch methodology, the political sensitivity of data and the ability to
hi Aiake changes in ongoing public data series.

J 0
..{p The major recommendations for improving the farm income informa-

fv .. '
'ijtton system were to improve data on the distribution of income, to make

;;iore use of directly reported data on production expenditures, to give

. flower priorities to earlier suggestions that farm income data be pre-

n.~
' “l-sented in national income and product accounting formats, and to expand
57' 1the farm sector performance measures emphasized to improve the credibili-

    

.1
.,‘
I

:.' ty of the existing data and analysis done by the USDA.

 

.51". - “'4
I. -- _‘~I ~ I. J we
' . -. ‘.
’ ,1. HJ' LI} .—
t- * .7
x . _. x
..
I_- .n ‘

 

{big rda few if. I; ,_ . ' . :
’ tog this n , r . . y. 14‘»...
“m. Chairman cf » , , . _ l ,. _.. ., Kunm'
mil. Adlai. .. . ‘- 1 122m: we

“I ECODIJ‘I. l .;1.1' ‘r .i‘i“:"u€;

,lor.Deput,viavI~r-Iw _ - I. , 5‘ ..

_. for. inrov‘ma‘m; . ' > ., ~ r . . '-'.I. mm. «on-

t ”In the furic; IA; ‘. . , ‘ , ’ ' w ". . - I “‘3‘; ’i’x‘ir'r'.

”Sututions don‘- ire" .; fl . .- . - .. mt.“ .- I.” .r "."v€5'.-
.1» To my parents

“ow '
‘ , . . ‘4.’ 5') lfif!‘ .kaJa

. .n .1 , Unfortunately, ~ ,
“‘fl not as car. * :~ super. -» ~ I . s W ’I II' ‘I i 3'
t”. to cWietely raid. -' ‘ .. v u‘ _;,_.»..-3,.,:, . . .' in". only
I . ~ . It.» rr~:; than: imi‘ia‘h

:1“. ”wt. ‘0' lil‘i: c’n'l in?" . , .3 .. _
' ‘ . CuI if- T‘ 2 - .' . X ._ _ . .V--'.3".""f'. am en—
“I: who lid! n) i , ‘

    
    
 
 
  
 
 
  
 
 
 
 
 
 
  

‘oo ”It
. 5.5.:
To Dr Jamar I. mm» . -- .- A~ chairman
' 7.9-... “1m and their 71". " , . l_ »_ <I a. cc; 0121;)!
1 mt. His (or. *rfv am... .. ....- 1-,: s an economist
dese'ves .‘ . in»; u r < m. ,1 in‘is‘ ‘: .39 reg”-
‘ ~n’"’!‘_. 3n musL

credit fir -m ‘~ .. .. . , I: ,
II] z to his. Ponce-r, oi i at: U! 2‘ "w. s. ».,-.mtn and
mrtino with ht? .9 1r!»s (fie-2.“.-
fifth mention 5 .s. I cite-res L: im- '89:- e *1. arm-'1 0? my th-
WW I I‘D patiently strugg ed rm r. a VMI r n1 craft of this
~ Their cements greatly errant“ the 9.4- it, :4 the Hui
, U. L'bb] «1:7'ngiy 9AM '\ 1 {tr-ii: H, ._ W of
' clad nonacademic. .trwgmm a; qf‘k’i‘ié caviar. no
- ' I: blissful during the final stages 2' ”.e ete- amen of g
. . ’ .. I I“: observations on the put. c. new: ago. ,5 a: W
_ .‘ .. ‘tion oystus further «picnic-a: 2m W‘”§?=Ff' er'
1 1 ,, (ti; Or. Lester r. Wa 131843 Ire. the ash» ,
I ‘” W :9 the “$730 0‘ '8 fl’i‘ fittest: V“ f". =’
' with the submit Wimp n 5.; and“. ;-
to an“: in this cm W gen 7: 51- 2.». 1.

 

"I

 

ACKNOWLEDGMENTS

Needless to say, I have incurred a great many debts during my
stay at Michigan State University. Those of a monetary nature were made
easier to repay through the financial assistance of the Michigan Agricul-
tural Experiment Station and the Economics, Statistics and Cooperatives
Service (ESCS) of the United States Department of Agriculture (USDA). In
this regard a few individuals deserve special mention for their roles in
supporting this research. I am especially grateful to Dr. Harold M.
Riley, Chairman of the Department of Agricultural Economics, Dr. Kenneth
R. Farrell, Administrator of ESCS, Dr. John E. Lee, Director of the
National Economic Analysis Division of ESCS, and finally, Mr. William E.
Kibler, Deputy Administrator for Statistics in ESCS, whose advocacy for
research on information systems was instrumental in initiating and con-
tinuing the funding for this study. It is my sincere desire that these
institutions and individuals realize an adequate return on their invest-
ment.

Unfortunately, most of the debts which remain are intellectual
and are not as easily settled as monetary liabilities. It is never pos-
sible to completely repay one's intellectual indebtedness, so I can only
hope that a few written words will begin to compensate the many indivi-
duals who made my four years of graduate school both rewarding and en-
joyable.

To Dr. James T. Bonnen I owe my deepest gratitude. As chairman
of my guidance and thesis committees, he influenced all aspects of my
graduate work. His contribution toward my development as an economist
and as a person deserves far more in thanks than I could possible repay.
Much of the credit for the ideas expressed in this dissertation must
ultimately go to him. Moreover, on a personal side, his warmth and
openness made working with him a true pleasure.

My appreciation also extends to the remaining members of my the-
sis committee who patiently struggled through a rough first draft of this
dissertation. Their comments greatly enhanced the quality of the final
product. Dr. Lawrence W. Libby willingly gave his advice in a number of
areas, both academic and nonacademic, throughout my graduate career. He
was particularly helpful during the final stages of the completion of
this research. His observations on the public choice aspects of improve-
ments in information systems further augmented the theoretical arguments
in this dissertation. Dr. Lester V. Manderscheid notably provided assis-
tance in matters related to the design of the mail questionnaire and in-
terview schedule along with the subsequent interpretation of the results.
His careful attention to details in this area added greatly to the

ii

 

 

iii

usefulness of the empirical findings. Dr. Lindon J. Robison used his
understanding of the information systems of the USDA to provide insights
during all phases of the development of this research. These insights
both strengthened and confirmed the overall results. Dr. Byron w. Brown
reviewed and earlier draft from the perspective of a non-agriculturalist.
The incorporation of his comments added clarity to the final presenta-
tion.

My fellow graduate students also deserve my sincerest thanks.
Many of these individuals gladly shared their knowledge and insights on
a variety of subjects which added immeasurably to my experience at Michi-
gan State University. Even more importantly, they provided the friend-
ship necessary to have made the experience enjoyable.

Special thanks are in order for those who provided clerical as-
sistance for this research. My gratitude goes to all of the secretaries
in the Department of Agricultural Economics who typed earlier drafts of
this dissertation along with many pages of interview notes and much
associated correspondence. I would like to especially thank Cathy Cooke,
who cheerfully coordinated or performed most of the clerical tasks that
were part of this research project. Linda Brown also merits a word of
thanks for her careful typing of the final thesis draft.

Finally, I must sincerely thank my family and especially my
parents. Their encouragement and support throughout my life made the
tasks of completing this degree significantly easier. For their part~
they deserve to share equally in the credit.

TABLE OF CONTENTS

LIST OF TABLES ........................... vi
LIST OF FIGURES .......................... vii

1. INTRODUCTION

1.1 Problem Definition ..................... 1
1.2 History and Current Situation ................ 4
l 1.3 Objectives ......................... 9
. 1.4 Research Design ....................... 10
2. CONCEPTUAL FRAMEWORK: THE THEORY AND ECONOMICS OF

INFORMATION SYSTEMS
2.1 Introduction ........................ 14
2.2 Information and Inquiry ................... 14
2.3 An Information Systems Paradigm ............... 19
2.3.1 Data and Data Systems ................. 19
2. 3. 2 The Nature of Information ............... 21
2. 3. 3 Application to Operating Information Systems ..... 24
2.4 Economic Implications .................... 26
2.4.1 Characteristics of Information ............ 27
2.4.2 Supplying Information for Private Use ......... 32
2. 4. 3 Information for Public Use .............. 38
2.4.4 The Demand for Information .............. 40
2.4.5 Income Distribution and the Distribution of Information 44
2.5 Summary ........................... 49

3. REVIEW OF LITERATURE 0N AGGREGATE FARM INCOME DATA
3.1 Introduction ........................ 51
3.2 Analyses of Farm Income Data ................ 52
3.2.1 Grove ......................... 52
3. 2. 2 Task Force on Farm Income and Capital Accounting 1972. 55
3. 2. 3 Task Force on Farm Income Estimates 1975 ....... 62
3. 2. 4 Gross National Product Data Improvement Project . . . . 70
3.2.5 Guebert ........................ 71
3. 2. 6 Agricultural and Rural Data Workshop ......... 74
3.3 Alternative Conceptualizations ............... 75
3.3.1 National Family Farm Concept ............. 76
3. 3.2 Establishment Concept ................. 77
3. 3. 3 Product Concept .................... 78
3. 3. 4 Value Added Concepts ................. 79
3.3.5 Taxfilers Income Concept ............... 83
iv

 

 

 

1‘.
-:T
‘r-‘ '

v
3.3.6 Combining Income and Health .............. 84
3.3.7 Parity Returns Concept ................ 85
3.4 Summary ........................... 87
4. DESCRIPTIVE RESULTS: THE FARM INCOME INFORMATION SYSTEM
4.1 Introduction ........................ 89
4.2 Overview .......................... 90
4.3 Users and Uses ....................... 93
4.3.1 Limitations of Mail Survey Population ......... 93
4.3.2 Selection of Sample .................. 95
4.3.3 Mail Survey Results .................. 97
4.3.4 Nonrespondents .................... 121
4.3.5 Indentification of Decision Makers to be Interviewed . 124
4.3.6 Relevant Interview Results .............. 128
4.4 The Farm Income Information System ............. 141
4.4.1 The Primary Data Subsystem .............. 141
4.4.2 The Formatting and Communications Subsystem ...... 145
4.4.3 The Analysis Subsystem ................ 147
4. 4. 4 The Decision Making Subsystem ............. 149
4.5 Summary ........................... 152
5. EVALUATION OF THE SYSTEM AND IMPLICATIONS
5.1 Introduction ........................ 153
5.2 Design for Whom ....................... 154
5.3 Conceptual Obsolescence ................... 161
5.3.1 Policy Agenda Changes ................. 162
5.3.2 Changes in Reality .................. 168
5.4 Credibility ......................... 174
5.4.1 Reliability of USDA Data ............... 174
5.4.2 Accuracy ....................... 177
5 4.3 Other Aspects of Credibility ............. 180
5 5 National Income Accounting and Farm Income ......... 183
5.6 Farm Income Statistics in a Developmental Mode ....... 186
5 6.1 Political Sensitivity ................. 187
5.6.2 Future Policy Issues ................. 188
5 7 Analysis of Methodology ................... 190
5 8 Summary ........................... I91
6. CONCLUSIONS AND RECOMMENDATIONS
6 1 General Conclusions ..................... 194
6 2 Recommendations ....................... 199
6.2.1 Distributional Data .................. 199
6. 2. 2 Farm Production Expenses ............... 203
6. 2. 3 National Income Accounting .............. 206
6. L 4 Improved Credibility ................. 209
6.2.5 Government Data Provision: Rationale and Priorities . . 212
6. 3 Additional Remarks ..................... 215
6. 4 Further Research ...................... 217
APPENDIX A ............................. 220
BIBLIOGRAPHY ............................ 224

 

4-1
4-2
4-3
4-4
4-5
4-6
4—7
4-8

4-9
4-10
4-11
4-12
4-13
5-2
5-3
5-4

5-5
5-6

LIST OF TABLES

Mail Survey Characteristics .................. 97
Respondents and Users by User Category ............ 99
Primary Use by Category .................... 101
Secondary and Tertiary Uses by Category ............ 106
All Uses by Category ..................... 108
Crosstabulation of Users by Primary Use ........ ‘. . . . 109
Crosstabulation of Users by All Uses ............. 112
Relative Usefulness of Selected Formats and Canponents

of USDA Farm Income Data .................. 114
Percent of Respondents Using Forecasts of Farm Income by Source 117
Source of Forecasts by Primary Use .............. 119
Source of Forecasts by User Category ............. 120
Nonrespondent Survey Results ................. 123
Nonrespondents by User Category ................ 123
Willingness to Pay for Farm Income Data by User ........ 158
Willingness to Pay for Farm Income Data by Primary Use . . . . 159
Mean Priorities Assigned to Design Criteria .......... 170
User Perceptions of Accuracy of Annual Realized Net Farm

Income ........................... 179
Revisions and Accuracy of Recent USDA Farm Income Estimates . . 181

Priorities Between Cash and Accrual Accounting by Uses . . . . 186

vi

 

LIST OF FIGURES

2-1 An Agricultural Information System .............. 23
4-1 Comparison of An Idealized and Operating Information System . 91
4-2 The Farm Income Information System .............. 142

vii

 

CHAPTER 1
INTRODUCTION

1.1 PROBLEM DEFINITION

The Economic Statistics Committee of the American Agricultural
Economics Association characterized the major problem of the U.S. agri-
cultural data systems as one of conceptual obsolescence. Changes in the
economic structure and organization of agriculture and in the important
policy questions involving the agricultural sector have not been matched
by corresponding changes in the concepts which are expected to capture
that reality and which are measured by the agricultural data system. This
change in the policy agenda and the resulting conceptual obsolence is
particularly disturbing when new problems become all too apparent.

In most instances, governments fail to act before some means is
found to measure a problem. The manner in which new problems are identi-
fied in a social policy setting often leads to this failure of government
action. For the most part, problems are identified through the use of
some combination of a specific model of reality and empirical measurements
relating to the model (de Neufville, p. 7). Thus, when the existing
models or the empirical measurement no longer mesh with reality, it is
nearly impossible to identify relevant problems. The lack of awareness
of problems arising from conceptual obsolescence can explain the failure
of policy makers to act on problems.

These principles might also explain the findings of Judith Innes
de Neufville (p. 7) that government policy makers generally concentrate

1

 

2
on problems where progress can be easily measured and demonstrated. It
would seem reasonable to assume that the ability to substantiate progress
in solving a problem would be related to the ability to identify the pro-
blem at the outset.

In some cases, it goes even beyond identifying a problem or
demonstrating progress in its solution. Often it appears that government
decision makers use methods which allow performance to be measured, i.e.,
available data may even dictate or suggest the solution to the problem.
Inaccurate or obsolete data may lead to a poor solution or to no solution
if the problem cannot be clearly defined. Thus, difficulties arising
from conceptual obsolescence in the data system can have important ramifi-
cations for problem solving decision making.

Conceptual obsolescence dOes not extend to the entire data system
in agriculture nor is it the only type of data problem evident. There
are many instances where the accuracy of the data has improved through
time, particularly when the concepts are based on biological or physical
processes which have changed little, if any, and where measurement,
collection and processing techniques have been improved (Bonnen 1975a,

p. 754). However, while the conceptual accuracy of some basic economic

statistics, such as prices, seems to haVe remained valid, changes in

‘ market structure create problems in operationalizing and measuring some

of these concepts and consequently have made these statistics less reli-
able (Riemenschneider, p. 27-36). Thus the problems of agricultural data
systems are not restricted to conceptual difficulties alone.

The farm income data system appears to be a prime candidate for
evaluation. In recent years two separate task forces have studied the

United States Department of Agriculture (USDA) farm income data and have

 

 

3

pointed out many actual and assumed deficiencies in the system. For exam-
ple, the current farm income data system often fails to distinguish be-
tween some current expenses and capital expenses and nets out many impor-
tant income flows in the farming sector, which along with other related
problems leads to a conceptual labyrinth so complex that one often does
not know quite what is being measured (Weeks 1971). In addition, it has
been suggested that the farm income data are not always compatible with
components of the national income and product accounts (Hildreth, gt. al.;
Heeks, gt, 31.). There appears to be ample justification to study the
farm income data system simply to learn more about these problems. How-
ever these other studies have left other important questions unanswered.
These studies assumed a set of uses for the farm income data and based
recommendations for improving the data on this assumed set of uses with-
out ever determining if these were the most important uses for current
decisions. More explicitly, these reports seemed overly concerned with
insuring that the farm income accounts were easily comparable with the
national income and product accounts without ever asking the question of
whether comparability between these accounts serves a useful function in
policy decisions. Hence these earlier studies failed to justify adequately
their recommendations and neglected to ascertain the improvements which
would be of greatest net value in the most important current uses of the
data. Furthermore, consideration was not given in these studies to the
question of the acceptability of the implied changes by the relevant data
users.

As important, these apparent conflicts and inconsistencies suggest
that agricultural policy decision makers may not be receiving adequate or

accurate data to use in reaching decisions. As long as farm income data

__ _ -.,-A

___.....-.—~

 

remain the only integrated measure of changing prices, sales, production,
and production costs they will most likely play a role in the decision
process and hence efforts must be taken to insure the accuracy in concept,

definition, measurement, and use of farm income data.

1.2 HISTORY AND CURRENT SITUATION

The beginnings of the farm income series can be traced to the
1909 Census of Agriculture. The report of the Census was released in
1913 and provided the first estimates of farm income. Over the next
decade work continued on estimates of farm income culminating around 1924
when estimates of farm income were started on a calendar year basis along
with a series on farm production expenses and national net farm income.
It is important to note that the current system used to estimate farm
income was developed prior to the time when the national income accounts
were established, thus explaining, in part, the differences in format
between the farm income accounts and the national income accounts. The
farm income series were originally set up to measure the economic welfare
of farmers, who at the time of the development of the series made up a
significant proportion of the population. This segment of the population
was later able to use this measure of welfare to justify price and income
supports through Federal legislation in the 19305.

The relationship between farm income measurement and the agricul-
tural policy of the 19205 through the 19505 is more direct than is appa-
rent on the surface. The pioneering work of the National Bureau of Eco-
nomic Research (NBER) in the area of national income measurement set the

tone for the important agricultural policy debates of this era. The

- _.-.—. *-‘_--‘4

 

5
National Bureau estimated agriculture's share of national income and the
purchasing power of farmers relative to the nonfarm population for the
period 1909 to 1920. These NBER studies portrayed the national income
share and purchasing power of agriculture in a poor light relative to
other sectors. John 0. Black in 1927 noted that the NBER results were
widely circulated and along with other analyses which relied on the NBER
findings had an important influente on the farm policy legislation of
this period.

.Numerous other statements during this period confirm the influence
of the NBER results on the government policy toward agriculture. J. I.
Falconer, writing during the time of the development of the USDA farm
income accounts, pointed out that one of the most important areas of
research on farm income was in the area of comparing the purchasing power
and well being of farmers with that of the urban population.. H. R. Tolley
is even more explicit in expressing this relationship between aggregate
farm income data and farm policy. In presenting the objectives of agri-
cultural policy, Tolley sets a farm income goal as the highest priority
on his list of ten objectives.‘

"First, a fair share of the national income for agriculture.

Undoubtedly there is disagreement upon its precise measure-

ment. Still, the idea of securing to the average farmer as

much purchasing power relative to that of the average non-

farmer as obtained in a more normal period is a definite and

tenable objective (Tolley, p. 24).

The notion of income parity developed in response to the perceived
plight of farmers relative to non-farmers. Income parity is defined in
the farm legislation of the 19305 and 405 in terms of an historical ratio
between per capita income of the farm and non-farm population. While the

ultimate concern of the farm legislation was with income parity, the

 

6

policy to achieve income parity revolved around the use of commodity
price supports. Thus, parity prices became the principal data used in
implementing farm policy in the 19305 and are used even today in some
con'lnodities. Karl T. Wright confirms the important relationship between
price and income parity by noting that one of the major assumptions of
the 1933 agricultural legislation, which provided the basis for most of
the subsequent major farm legislation, was ". . . that price established
by the parity formula would provide parity income to farmers." (p. 294).

Farm income measurement seems to have played a significant role
in the policy process during the period between 1920 and 1960 because of
the importance of the income questions raised in the farm debate. At this
time the major equity questions were concerned with comparing the welfare
of the farm and non-farm sectors in justifying government action to im-
prove farm welfare. Thus, the public policy users of farm income data
prior to the 19605 seem to have been concentrated on this objective.

'By 1940 the farm income estimates had reached a stage of develop-
ment which led to the publishing of the first Farm Income Situation. This
report was published regularly until 1975 when its functions were replaced
by parts of the Agricultural Outlook publication and the statistical bul-
letin Farm Income Statistics. In the early 19505 state estimates of gross
and net farm income were developed and published. Two significant changes
or additions occurred in the early 19605. At this time one of the first
measures of the distribution of farm income was made. The distribution

of income and production expenses of farm operators by value of sales

-class are available starting with 1960. At this same time the calculation

of the income of the farm population was changed from a national income

basis to a personal income basis which allowed more direct comparisons

A

 

7
between the fann and non-farm populations (Upchurch 1977, P. 325).

Currently there are two major concepts of farm income used by the
U.S. Department of Agriculture. The first of these is “income from pro-
duction." In general this concept is concerned with productive activities
of businesses or government. Farmers' net income is the current measure
of income from farm production. This concept treats farming as an indus-
try and views farming in the United States as a single national family
farm. Gross farm income and farm production expenses are estimated and
net farm income is calculated as a residual. Income from production was
the primary concern of the earlier studies on farm income and as such will
remain the principal focus of this research.

The second major concept used by USDA is the personal income of
the farm population. Generally personal and disposable income are used
to measure the purchasing power or economic welfare of individuals or
families (Hildreth, gt. El-: p. 3). The USDA concept accounts for income
from both farm and non-farm sources. In addition to farm operators this
concept includes resident farm workers and other farm residents as well.
Conceptually at least, it is a measure of the total income available to
operate the farm business and to maintain the standard of living of the
farm population (U.S. Department of Agriculture 1969, p. 1).

These two concepts lead to two specific data series: 1) Farmers'
Total or Realized Net Income and 2) Personal Income of the Farm Popula-
tion from All Sources. The former series is on a calendar year basis
covering the U.S., each state, and six regions while the latter is also

on a calendar year basis but only is available for the United States as a

whole.

The total net farm income series has six major components:

. '_._. _.-_.—_-- __

 

L;

8
1) cash receipts from farm marketings, 2) government payments to farmers,
3) non-money income which includes the imputed value of farm dwellings
and farm products consumed directly in farm households, 4) other farm
jpgpmg which is made up of recreational, machine hire and custom work
income, 5) fann production expenses, and 6) net change in farm inventories.
The inclusion of this last component distinguishes total net income from
realized net income. The personal income series measures the sum of all
income from farm and non-farm sources received by the farm population,
i.e., those actually living on farms.

Except for the government payments these components are not mea-
sured directly. Instead, primary data such as prices and quantities are
first measured. Then through various aggregations and accounting rules
estimates are developed for each of the components and for the residual,
farm income. Many sources of primary data are used but the principle

suppliers of data are the Statistics branch of the Economics, Statistics,

.and Cooperatives Service (ESCS) in USDAl/and the Bureau of Census. The

Farm Income Unit, which is part of the Economics branch of ESCS§/, then
uses this data to construct estimates of the components of farm income.
For any given year, a preliminary and three revised estimates of
farm income and each of the major components are made. The preliminary
estimate is made in January following the year in question. The second

estimate is made six months later in July when more complete data is

I] This branch was formerly known as the Statistical Reporting Service,
prior to January 1978, and hereafter will be referred to as ESCS: Stat-
stics.

g/ This branch was formerly known as the Economic Research Service,
pripr to January 1978, and hereafter will be referred to as ESCS: Eco.
nom cs.

 

9
available. The third and fourth estimates are made in July of the two
following years, incorporating even more complete and revised primary

data.

1.3 OBJECTIVES

This research has multiple objectives. An important part or ob-
jective of this research will be descriptive. An attempt will be made to
outline the relationships among data design, data collection, analysis,
interpretation, and decision making for the farm income information sys-
tem. This descriptive evidence should shed some light on any problems in
the system and lead to a better understanding of information system de-
sign.

A second concern is to develop further the economic and social
theory with regard to information and information systems. Since there
has only been a limited amount of theoretical development in this area, a
major task will be to provide a more coherent theoretical basis for eval-
uating the specification and economics of information systems.

The information systems paradigm, which provides the foundation
for this research, has not been applied extensively to specific informa-
tion systems. Thus a third objective will be to test the usefulness of
this approach or paradigm as a research methodology. That is, does this
paradigm provide a means for identifying researchable areas and for eval—
uating a specific information system or is it only useful for understand-
ing the more generic system?

A final objective is to suggest changes and improvements in the
current farm income information system. This will include the establish-

ment of some priorities among improvements already recommended in the

 

10

literature as well as the recommendation of changes based on this research.

1.4 RESEARCH DESIGN .

A preliminary examination of the U.S. Department of Agriculture's
farm income information system suggests that there are two basic types of
evaluative research that might be conducted in the context of this system.
First, the operation of the current data system could be analyzed for the
purpose of reducing revisions and improving the statistical reliability
of the existing accounts. A systematic examination of each account would
be in order in this case to reduce the variance between the January and
the subsequent July estimates. This type of research would stress the
methods and data sources used in the current farm income data system.
From a practical standpoint this type of research is probably best done
in the USDA by those who have a close working knowledge of the methods
and data sources used in estimating the various accounts. Therefore, we
will not focus on an effort to refine the current system.

Instead a second type of evaluative research was undertaken. It
is more closely akin to what the National Academy of Sciences has called
'research on "setting statistical priorities.“ While the Panel on Method-
ology for Statistical Priorities was concerned with setting priOrities
among many different data systems their suggestions are equally applicable
to establishing priorities within a given data system. Furthermore, they
state that setting priorities might even be easier within specific data
systems when compared to the difficulties encountered in establishing
priorities among very different types of data.

The type of research suggested by the National Academy panel and
that done in this study goes beyond what might conventionally be thought

 

 

11
of in setting priorities. More than simply choosing between data sources
is involved; this research goes further and attempts to answer some of
the important questions concerning the design and redesign of information
systems. It is basically an attempt to apply some of the findings of the
National Academy panel to farm income data. Their report summarized a
methodology in,a general way as a set of questions to be asked concerning
the use of data. When applied to the farm income information system this
methodology can be paraphrased as follows--the relevant question is not
simply, who uses farm income statistics, but, instead, a series of ques-
tions--what are the important decisions to be made concerning farm income,
who makes those decisions, and what are the data needed to make those
decisions effectively? (Panel on Methodology for Statistical Priorities,
p. 6).

Thomas A. Miller studied methods for valuing information and

‘ summarizes the general methodology for this type of research as a "prag-

matic user-oriented approach" to valuing information. Thus, the basis
for this research seems to be well grounded in the recent literature.
This research also builds on earlier studies of farm income data to the
extent that these suggested alternative conceptualizations of farm income
and recommended changes in the system for which some priorities can be
established using the findings of this research.

The research had three steps. The first stage included an exami-
nation of the literature, personal interviews with the data producers in
USDA, and a mail survey to determine the principal uses and users of the
farm income data. The identification of the current uses of the data
provided a list of important decision makers in the system. These indi-

viduals were then interviewed to ascertain the important issues facing

 

 

12

them with regard to farm income. This second step of personal interviews
with decision makers was also used to obtain a more detailed description
of the system and to assess the potential for changing the system. The
third step was the analysis of the current data system with regard to its
ability to meet the needs of decision makers. From this analysis came a
better understanding of the operation of the farm income information sys-
tem which provided a basis for setting priorities and making choices with
respect to improvements in the system.

The remainder of this study is divided into five chapters. Chapter
2 develops the conceptual framework. Included are the develOpment of an
information systems paradigm and its economic implications which provide
the basis for the evaluation of the current farm income data system.

Any changes in an information system normally imply different
distributional impacts from the original. The question then becomes how
to gauge whether or not the change is "better" since in nearly all cases
some individuals benefit and some are disadvantaged by any improvement in
the infonnation system. The criteria developed for analyzing and setting
priorities among improvements in the existing farm income data are de-
rived from the information systems paradigm and its economic implications.
In this sense the research has a normative focus since the results follow
from a particular view of the problem. Hence, the criteria used for
judging the system are somewhat unique to the approach used in the re-

search and follow from the aims and purposes of this study. However, it

.must be emphasized that in a different context with dissimilar goals ,

other, perhaps equally justifiable criteria, could exiSt for evaluating
information systems.

Chapter 3 is a summary review of the relevant literature which

A

 

 

13
concentrates on previous analyses of the USDA farm income data as well as
summarizing some of the alternative conceptualizations of farm income
suggested in the literature.

Chapter 4 presents the descriptive results of the study, by using
the mail survey results and the personal interviews to present a picture
of the current farm income information system. The Operating system is
contrasted with the idealized system developed in chapter 2 to provide a
means for classifying the components of the existing system.

Chapter 5 presents the evaluation of the system as it now exists.
It is concerned with defining the nature of the problems found in the
system and drawing some of the implications from these problems.

Chapter 6 outlines the recommendations for improvements and for
setting priorities in the system and reiterates the major conclusions in

summary form.

 

CHAPTER 2
CONCEPTUAL FRAMEWORK: THE THEORY AND ECONOMICS OF INFORMATION SYSTEMS

2.1 INTRODUCTION

There has been only a limited amount of theoretical development
in the area of agricultural information systems. Therefore, a principle
task of this chapter will be to provide a summary of some of the relevant
theory in this area. More importantly, this theory will be extended to
provide a more coherent basis for evaluating information systems. Heavy
emphasis will be placed in three areas. First, the philosophical basis
of an information systems paradigm will be developed. Following this,
the paradigm itself, along with some of the difficulties arising in apply-
ing it to operating information systems, will be presented. Finally, the
economic implications of this paradigm will be laid out. The theoretical
aspects in this area will concentrate on the economic characteristics of
information, the role of market structure in the supply of and demand for
information, and the effects of the distribution of information on the

distribution of income.

2.2 INFORMATION AND INQUIRY

In order to assess and evaluate a functioning information system,
it is helpful to have a framework within which to make comparisons and to
act as a guide in determining the relevant points of concern. Economics
has numerous theories which provide a framework to direct researchers to
the germane aspects of various economic issues and problems. In this

14

 

 

15
same vein one would expect that a theory of information would provide use-
ful direction in examining an information system. However, there is no
single theory of information. Engineers, computer scientists, business
management specialists, and economists, among others, all have their own
theories of information which focus on their unique disciplinary concerns.
So a principle concern must be with developing a framework for understand-
ing information systems.

Information systems in agriculture are designed to help either

* public or private decision makers or, in some cases, both, make decisions

to solve problems that arise at the farm, farm industry or the national
economy level. This would suggest that any theory of information should
focus, at least in part, on problem solving. Since problem solving deci-
sion making normally requires information from a variety of disciplines,
any theory of information must have a multidisciplinary perspective to be
useful in understanding information systems. Thus, to a55ess an operating
information system we are not concerned with developing a separate theory
of information for any particular discipline but rather with a way to
view and understand the process of inquiry and decision making.

To avoid any confusion, the ideas which follow in the next section
are best referred to as an information systems paradigm and not a theory
of information. Before developing the theoretical aspects further a
statement of the phi1050phical basis of this paradigm will be presented
to help clarify the importance of viewing information in a systems con-
text.

In the broadest and most general sense information is the product
of some basic process of inquiry. C. West Churchman (1971) emphasizes

this point and goes further to add that to a model a problem, (using

 

 

16
model in the generic sense) is to conduct an inquiry about the problem,
which is in turn an attempt to produce information on the very nature of
the problem. Thus, any knowledge about a specific problem is dependent
on the system of inquiry used in obtaining that knowledge. In this sense
information must be considered as a function of epistemology (Mitroff and
Sagasti, p. 119).

How one models or conceptualizes any information system can be
traced back to some philosophically based inquiry system. To attempt to
represent reality, as any information system must, while maintaining
strict adherence to a narrow philosophical mode of inquiry will limit one's
ability to capture reality adequately. The limitations of a single re-
search philosophy are particularly evident when the problems one is at-
tempting to solve are ill-structured, i.e., where the greatest difficulty
lies in defining the nature of the problem itself. Well-structured pro-
blems, on the other hand, are subject to more precise analytic methods of
attack (Mitroff and Sagasti, p. 121).

A Some general examples should make this clearer. Well-structured
problems in decision theory are of the type where a known deterministic
or probabilistic relationship exists between the choice of an act by a
decision maker and the occurence of a specific outcome. Further, these
acts must in some sense optimize the value to or utility of the decision
maker given a known set of states of nature (Mitroff and Sagasti, p. 120-
121). Thus, these types of problems are subject to more precise analyti-
cal methods of attack in that there are unambiguous rules for deciding on
an optimal course of action as well as known relationships between actions
and outcomes.

The problem of how much fertilizer to apply to an acre of corn,

 

 

17

given other inputs, is an example of a well-structured problem. In this
case, a decision rule of profit maximization is appropriate. Once the
physical relationships between-fertilizer and corn production are known,
along with the values of the inputs and outputs, it is then possible to
decide on a level of fertilizer application that is consistent with the
decision rule. The mode of inquiry is quite straightforward in this case
and the problems of information system design are relatively simple.

Ill-structured problems in decision theory are such that one's
knowledge about a given problem is limited by the fact that one or more
sets of acts, outcomes, utilities of the decision maker, or states of
nature are not known with confidence. With these types of problems deci-
sion makers are frequently concerned with the achievement of multiple
desired outcomes under the conditions of imperfect knowledge. The uncer-
tainty inherent in ill-structured problems often means that the greatest
difficulty in the solution lies in actually defining the problem.

Churchman (1971), Mitroff and others demonstrate this relationship
between epistemology and information by pointing out the differences be-
tween the pure philosophical systems of inquiry of Leibnitz, Locke, Kant
and Hegel. Each of the pure philosophical systems of inquiry is useful
in gaining insights into various aspects of problems. However, each has
its shortcomings in dealing with the whole of certain types of problems.
The solution of ill-structured problems requires a more general philoso-
phical system of inquiry that goes beyond those of Locke, Leibnitz, Kant
or Hegel. Singer-Churchman inquiry systems integrate these other inquiry
systems into a multidisciplinary approach to problem solving that stresses
the interaction between the different modes of inquiry (Mitroff and Pondy,

p. 476). This type of inquiry system is similar to what can be thought of

 

as the systems approach.

It is this systems approach to problem solving that provides the
foundation for the information systems paradigm which follows. The con-
cern of information systems and the systems approach in general are with
solving practical problems and consequently with the decisions addressed
to those problems. Hence an information system is teleological. The
goals and values of the decision maker must impact on the design of the
information system since these goals and values provide insights into the
nature of the problem on which the system focuses. As was noted earlier
the nature of the problem dictates the mode of inquiry used in arriving
at a solution, so the decision maker must be part of the information
system. As long as the values of the decision maker impact on the way in
which that same person defines a given problem, then the decision maker
must be considered as a part of the information system since the product
of the system, i.e., the information, is determined by the mode of inquiry
used in defining the nature of the problem. Thus, an information system
designed for farm policy decisions by government policy makers would
generally be different than an information system for use by a farmer in
the day to day operation of a farm business because of the difference in
decision makers. An important attribute of the information systems para—
digm developed here is its generality. This should permit its use as a
guide in assessing the quality of both government agricultural statistical
information systems as well as the informal information systems of individ-
ual farmers.

A further concern of the systems approach is with workability.

The roots of Singer-Churchman Inquiry Systems arise directly out of Amer-

ican pragmatism (Mitroff and Pondy, p. 477), so the implementation phase

 

19

of problem solving takes on a greater importance than with many other
modes of inquiry. This also follows from the holistic concern of the
systems approach which would imply that all important aspects of the pro-
blem should be studied including the implementation of possible solutions.

Many persons equate the systems approach with computerized model-
ing or simulation. This is far from a correct appraisal of either the
systems approach or the teChnique of simulation. Not all systems need to
be modeled mathematically in order to assist in solving problems. Often
a simple conceptual model of the system which identifies the various com-
ponents and interrelationships of the system will be all that is necessary

or desirable in developing solutions to problems.

2.3 AN INFORMATION SYSTEMS PARADIGMl/

Agricultural economists as well as other social scientists often
erroneously equate data to information. As Edgar S. Dunn (p. 20) notes,
“There is a pervasive tendency to assume that information is an intrinsic
property of symbolic data." But he follows this statement with the dis-
claimer that it is clearly untrue. The following paradigm is an attempt
to make clear the difference between data and information and to relate
the data collection process to the analytical process or system of inquiry

used in solving practical problems.
2.3.1 DATA AND DATA SYSTEMS
A data system is fundamentally an attempt to represent reality

empirically.‘ Since reality is infinitely complex and is not readily

1] Sections 2.3.1 and 2.3.2 are based on Bonnen 1975a.

 

20
grasped in total by the human mind, it is necessary to first break down
these experiential phenomena into a set of categories or classes that can
be counted or measured. This counting or measuring is usually thought of
in quantified terms but our arguments apply equally to numerical or non—
numerical data. Subjective impressions and simple relative comparisons
such as good or bad and high or low can be treated in similar fashion.
However, for ease of presentation, the remainder of this section will be
discussed in terms of numerical data, remembering that the ideas remain
valid for both quantitive and qualitative data.

Data collection is usually thought of in terms of measuring or
counting using sampling or complete enumeration of a certain pepulation.
But problems of sampling or measurement only arise after the prior ques-
tion of "what phenomena is to be counted or measured?" is answered. Given
the phiTOSOphical basis of our approach presented earlier the most rea-
sonable answer to this question is that it depends on the ultimate deci—
sions to be made. To maintain logical coherence and to represent reality
adequately these quantified phenomena must be related to each other and
to reality in a meaningful manner. Thus, there must be a concept of real-
ity to be measured and to be meaningful this concept must be capable of
accurately systematizing and categorizing reality so it can be understood
by those using the data. This categorization must also be such that the
concepts are relevant to the ultimate decisions. Improvements in sampling
procedures or other measurement techniques will be of little value with-
out this solid conceptual base.

While any data colleCtion must be preceded by some conceptualiza-
tion of reality, a concept is an abstract idea and it is not possible to

measure a concept as such. Instead it is necessary to operationalize or

 

21

define these concepts so the definitions (categories of empirical varia-
bles) are as nearly representative as possible of the chosen concept.

Therefore, data collection is made up of more than the simple
step that one might initially perceive, it is really three distinct steps:
1) conceptualization, 2) operationalization of the concept, and 3) mea-
surement. These are the essential components of a data system. With this
in mind, statistical reliability takes on three meanings. First, is relia-
bility of concept, i.e., is the concept representative of reality and are
the concepts pertinent to the decisions being made. Second, the data
accuracy is affected by the reliability in operationalizing or defining
the concepts, i.e., the categories of empirical variables should be as
highly correlated as possible with the reality of the desired concept.
Third, there is measurement reliability which follows from the statisti-

cians' usual definition of the term. 2!

2.3.2. THE NATURE OF INFORMATION

The data system outlined above produces data not information. To
become information, data require analysis and interpretation to place it
in a decision making context. Raw data or even semiprocessed data are
rarely used directly by decision makers. Instead decision makers require
analysis to impart meaning to data so they can be used to solve the pro-
blems of concern in the decision making process. In this sense an infor-
mation system is a process which imposes form and gives meaning to data.

Economic analysis can be a part of this process but it is not necessarily

g/ Bonnen attributes this expanded notion of reliability to
L. V. Manderscheid.

 

 

22
the only part. Practical problems require knowledge from many disciplines
so that the information system must go beyond economic analysis for solu-
tions. Given this understanding of an information system three major
components are obvious: I) a data system, 2) the analytical capability
necessary to transform data into information and 3) the decision maker.
This is depicted in Figure 2-1.

Among analysts trained in the social sciences there seems to exist
a common conceptual basis for analyzing and solving problems. This usually
starts with a received body of theoretical concepts which are a perception
of reality. Concepts are then defined through some form of model which is
subsequently tested against empirical evidence and conclusions are drawn.
Hence, data systems must share common ground with epistemological systems
of inquiry.

The representation of-a data system (left side of Figure 2-1) and
an inquiry system (right side of Figure 2-1) points to the necessary over—
lap between theoretical concepts and the operationalization of these con-
cepts in both data and inquiry systems. Without this common conceptual
ground any attempt to use theory and empirical analysis together would be
fruitless and the fit between the deductive and inductive processes of
inquiry could not exist.

Information systems are teoleological because they are subsystems
within social systems which are in turn designed to solve social problems
(Bonnen 1976, p. 6). This is an important observation in that if data
collection and analysis are purposive in nature, then it is only possible
to evaluate information in a social system or decision making context.
Thus, the concepts which underlie the data system can only be derived

when the context of the decision is known. Furthermore, the system itself

23

0'00 .

2m~m>m zo_p<=mouz_ 4<z=p4=u_zw< z< .Hum “mam—L

xu*_cwz

 

 

I1
acetone...
cannons-nu.

muawocou Lo :owuu~__m:o_uccvno

        

 

 

 

 

 

 

 

 

 

agozwsagu
_cu_ua_m=c Lo
mcpummh v=m .
:o_umu_m_uwam , usauzo moan

 

 

 

 

maqsxs KJInbUI

 

n canoe-cocoa.
- u o... ca. 0 o a o a a o o a o o enactment-0.0.0.0...unto-Iona

w_m>_m:< ecu co.umumgngmuc_

 

 

 

mexmz :o_mmowc Lou :o_ueELoL:_

 

  

u:_xmz :owm.uun

v

mansxs eaeg

 

  

 

weasxs UOIQEWJO;UI

 

I‘llIIVIlil.l|l'|[\l‘ll. ‘i‘llir

 

24
must adapt as societal goals change while at the same time providing the

information that leads to changes in society.

2.3.3 APPLICATION TO OPERATING INFORMATION SYSTEMS

What insights does this abstraction of an idealized information
system provide when applied to a specific type of operating information
system? To begin, there seem to be some minor deficiencies in separating
the various components of the system. For example, in many ways economics
revolves around a common basis of value. At first glance this might lead
one to conclude that all economic statistics have a common conceptual
basis. This erroneous conclusion ignores an important point of the para-
digm. The needs of decision makers must feed back in the system to pro-
vide the grounding for the concepts used in the data system. So the pur-
pose of the systems determines the level of generality or specificity used
in forming the concepts of the data system. Even for a concept as seem-
ingly simple as a price, it appears that for different types of decisions
different concepts are needed. Long run or strategic decisions by firms
often can use data based on a concept of price that is more of an average
or unit value concept, while more of a detailed specification price con-
cept seems to be of greater utility for the short run or tactical deci-
sions within the same firm (Riemenschneider, p. 34-35).

While in an abstract sense this paradigm distinguishes between
data and information, a difficulty is encountered in attempting to deter-
mine the difference between data and information within the context of a
specific information system. Very few decision makers or even analysts
can use raw data, so almost all data are presented within a chosen format.

The choice of a fonnat implies some level of interpretation of the data.

 

25

So formatting clouds the distinction between data and information. More
importantly this low level of interpretation or formatting points to the
critical nature of the communication function in any operating information
system. Except in the very limited cases where a single decision maker
undertakes the design and collection of data and their subsequent analysis,
the usefulness of data is tied to their interpersonal transmissability.
Communication theorists are concerned with the process of encoding mes-
sages prior to the transmission of signals in a communication channel.
All data undergo initial formatting so that they can be understood by ana-
lysts, this formatting seems to be analogous to the encoding of messages
before transmission.

The understandability or interpersonal transmissability of data
is often directly correlated with the chosen format of the data. For
instance, a general politician with only a passing interest in agriculture
might find a statistic labelled “hog farrowings" completely useless for a
policy decision if he or she does not know the meaning of the word "farrow-
ings." This same politician might be able to perform appropriate analyses
and reach a decision concerning an aspect of farm policy relevant to hog
farmers if this statistic were called an index of future pork production.
Labelling of data is not the only aspect of formatting relevant here.
Choosing the appropriate level of aggregation and the ability of users to
access data as well as other related aspects are critical. Thus, when-
ever the data collection process is organizationally separate from the
analysis and decision making, this formatting and communication process
becomes an integral part of the information system.

Conceptual obsolescence is a problem for most operating infonna-

tion systems and can occur in two ways. First, concepts can become obso-

A

 

 

26
lete when reality changes in such a manner that the concepts are no longer
representative of reality. A second type of conceptual obsolence is often
more critical; this occurs when the agenda for decisions changes so the
concepts of the data system are no longer pertinent to the decisions that
are being made. The American Agricultural Economics Association's Eco—
nomic Statistics Committee cited conceptual obsolescence as a major source
of data problems in agriculture. The majority of agricultural statistics
in the United States are collected around a concept of the "family farm"
which has not changed in over 50 years.§/ The structure of agriculture
has changed greatly during this time but the concept remains the same.
Conceptual obsolescence in agricultural statistics has also come about
because the policy issues facing agriculture have changed dramatically in
recent years. Issues which relate to energy, the environment, consumers,
and the world food situation have all had a substantial impact on agri-
culture yet the data system for the sector is not designed to answer many
of the questions which have arisen in response to these issues (Bonnen

1975a).

2.4 ECONOMIC IMPLICATIONS

The major implication of the information systems paradigm pre-
sented here is that information only becomes an economically valuable
commodity in the context of decision making. For those theoretical econ-
omists living in a world of perfect knowledge the disclaimer must be
added that information first becomes valuable under conditions of uncer-

tainty. But in most practical applications consideration of uncertainty

3/ While it is true that the concept has not changed, the manner in
which this concept has been operationalized has changed periodically.

 

27
is a fact of life.

When viewed in an information systems context, information can be
treated as a commodity. Kenneth J. Arrow (1962) provides a link between
information as a commodity and information systems. His concern was with
"inventive activity" which he equates to the production of information.
Arrow's notion of inventive activity seems to be analogous to an informa-'
tion system since both processes yield an output of information. In this
sense, inventive activity would seem to include data design and collection
in addition to the analysis of data to produce information for decision
makers. ‘

Since information only acquires value in a decision making frame-
work the value of decisions is a primary determinant of the value of in-
formation. Therefore, the value of an information system depends on the
types of decisions for which it is used and consequently on those who make
the decisions.

Infonnation has many characteristics which provide insights into
questions concerning information system design. Once the characteristics
of information as a commodity are understood it is possible to look at
general used of information to get an indication of the determinants of
the supply of and demand for information. While not explicitly providing
a measure of the value of information, the determinants of infonnation
supply and demand should illuminate some of the difficulties in deter-
mining the value of information and the appropriate role of government in

the provision of information.

2.4.1 CHARACTERISTICS OF INFORMATION

The characteristics of information as a commodity affect its

 

28

allocation in the economy. Information possesses some of the attributes
of public goods which lead to allocational inefficiencies when compared
to purely private goods in a competitive market. The attributes of un-
certainty, indivisibility, and nonappropriability all violate the classi-
cal properties of purely private goods. The existence of uncertainty is
inherent in our definition of information. Information is also by defi-
nition indivisible.

As Kenneth Boulding points out, the absence of any unit of infor-
mation makes the pricing of information difficult and hence even makes it
difficult to think of information as a commodity. The electrical engi-
neers and data processors break information down into "bits" and this con-
cept is basic to their theory of information processing.

"The bit, however, abstracts completely from the content of

either information or knowledge, and while it is enormously

useful for telephone engineers, who have no interest in what

is being said over their telephones, for the purposes of the

. social system theorist we need a measure which takes account

of significance and which would weight, for instance, the

gossip of a teenager rather low and the communication over

the hot line between Moscow and Washington rather high."

(Boulding, p. 3)

Geoffrey Newman (p. 486) notes two other related problems in defining in-
formation in term of bits. First “bits“ may vary with the problems of
the decision makers, and second even if simple factual propositions could
be broken down into bits, how can theories based on deduction be broken
down into bits since theories do not necessarily have a basis in fact.

For our purposes the problem of nonappropriability as a property
of information is particularly important because of the implications it
has for market structure. Producers cannot nonnally charge for further

uses of information once it is disseminated so the returns to the supplier

of infonmation are not fully appropriable. As Boulding answers, only

 

29

things clearly appropriable can become property and be exchanged; if
something cannot be property, it cannot be a commodity. The problems of
appropriability of information make it a peculiar kind of property which
affect its supply and demand.

The question of appropriability cannot be separated from the
issue of property rights for information. Copyright and patent laws make
the appropriability of returns to information easier for certain types
and certain uses of information but the costs of enforcement make this a
reasonable alternative only in selected cases. For instance, if one pos-
sesses information about a commodity that is traded in a market, one must
trade in the market to get a return on the information. However, by com-
pleting a transaction in the market at least the nature of the information
that one possesses is released to others in the market. Thus no copyright
or patent laws could prevent others from using this information. Many
cases still remain, though, where the tradeoff exists between changing
the mechanisms for supplying information and changing the property rights
to information in order to get a more optimal allocation of resources for
the production of information. Changing the supply mechanisms is for the
most part easier than changing the property rights and hence our later
analysis assumes that the structure of the property rights for information
is relatively constant. V

The incomplete appropriability of information suggests that in
cases where data or information are sold to individuals that it should be
presented in a manner that is somewhat ephemeral so that only those who
originally pay the information supplier are likely to receive the infor-
mation. If the information supplier can present it in such a way that

the original purchaser of the information does not have a relatively

 

30

permanent document containing the information then it is more difficult
for the original purchaser to pass the information along to others. For
instance, Maynes, pp. 31., (p. 27) suggest the use of cable television as
a means to present consumer price information which would reduce the po—
tential for unauthorized resale of the information. This also suggests
that any user fees charged for the information should be low enough so
those who buy the information have little incentive to resell it.

Reselling of information is related to the characteristic of
increasing net returns in the use of information. This phenomenon stems
from the indivisibility of information taken in conjunction with the high
fixed costs usually associated with acquiring infonnation relative to the
costs of transmitting the same information once it is acquired. The ini-
tial purchasers or users of information are able to pass along the infor-
mation at a cost lower than the original supplier. Increasing returns to
the use of information arise as long as the value of the information is
relatively constant for each subsequent use. The incomplete appropriabil-
ity attribute only exacerbates the difficulties brought on by increasing
returns in use, since it prevents the original supplier of information
from charging for the subsequent uses of the information once it is dis-
seminated. Thus, the high fixed costs of acquisition cannot be spread
over all users.

The fact that information only acquires value in a decision making
situation gives rise to a fundamental paradox. A decision maker or pur-

chaser of information does not know the exact value of the information

"until it is acquired and used, but to determine precisely its value prior

to buying it the purchaser must in effect obtain the information without

cost. The problem caused by this paradox would be alleviated if the

 

31
seller retained the property rights to the information, but as was pre-
viously mentioned the lack of complete appropriability is a basic charac-
teristic of information. The importance of credibility and reliability
of sources of data and information is stressed by this paradox. When the
purchaser of the information is forced to estimate its value prior to
receipt, the value is often determined from previous experience with the
same supplier.

Another common way to judge reliability, especially of statistical
data, is through the methodology used in the data collection. This accen-
tuates the critical role of documentation of statistical procedures in
operating information systems. When there is no other way to estimate the
value of certain kinds of data except through an assessment of the data
gathering procedures, this documentation of the procedures is important.
In cases where new suppliers or new users of the data arise documentation
is even more essential since prior experience cannot be used to place a
value on the data. Given the three types of statistical reliability im-
plied by the information systems paradigm, this documentation should in-
clude a statement of the concepts underlying the data, how these concepts
are operationalized, as well as some notion of the statistical sampling
methods used.

There are further characteristics which affect both the supply
and demand for information as a commodity. The production of infonnation
is a risky process. The output of the production process, i.e., the in-
formation, cannot be predicted perfectly from the inputs so the process
has uncertainty associated with it (Arrow 1962, p. 616). For an informa-
tion system this problem arises because of the nature of the decisions

for which the information is to be used. The same data can be analyzed

 

32
to produce information that is different depending on the problematic
situation.

The characteristics outlined above, i.e., the riskiness of infor-
mation production, the indivisibility of information, its nonappropriabil-
ity, increasing returns in use, all cause the competitive model to lead to
a sub-optimal allocation of resources from society's point of view for the
production of information. Arrow (1962) shows that these attributes cause
an underinvestment in and an underutilization of information in the free
enterprise economy. The same conclusion is reached if one considers that
information has many of the attributes of public goods and thus will be

underproduced relative to a purely private good in a competitive system.

2.4.2 SUPPLYING INFORMATION FOR PRIVATE USE

The characteristics of information outlined above create some
difficulties in determining a suitable means of organization for the pro-
vision of agricultural information for private sector decision making.
There seem to be three basic organizational arrangements for supplying
information on a given industry. Each firm could purchase information
from a specialist firm, all firms could work together in data gathering,
using an industry or trade association to provide the information, finally
governments could gather and provide information to all of the firms.

The social returns to information are not estimated in most in-
stances, perhaps, because of the inherent difficulties in valuing informa-
tion. However it does seem clear that the social returns to information
often exceed the sum of individuals' private returns, particularly in a
decentralized economic system where infonnation is needed to coordinate

economic activity among firms. Without information the prospects of

 

33

realizing the full potential for increases in productivity from technical
change in a sector would be greatly diminished. The initial work in this
area by Yujiro Hayami and Wilis Peterson tends to confirm these hypotheses.
Their results show that, at least at the margin, improvements in the mea-
surement accuracy of data can cause increases in social welfare in a market
situation, as measured by losses of consumer surplus, beyond the benefits
estimated by merely summing the individual private benefits.

While in an aggregate sense data improvements might be in the over-
all interest of society, it is clear that any improvements in an informa-
tion system generally benefit some groups more than others or benefit
certain groups at the expense of others. Thus, it is likely that improve-
ments or changes in data design will be non-Pareto better. This stresses
the nonnative tone associated with any decisions in this area. As in most
public choice situations tradeoffs between different groups must be weighed
to determine the sagacity of any improvements in a public information sys-
tem.

It should also be noted that arguments presented in this section
begin with the implicit assumption that social benefits to information do
exceed private benefits, i.e., positive externalities exist. From this
starting point it is then easier to discuss economic considerations in
evaluating different information systems. While related to questions con-
cerning the economics of information, the following discussion is perhaps
best described as dealing with the economics of information systems. This
subtle distinction is necessary to maintain the generality of the results
and to avoid the overwhelming difficulties associated with determining the
social value of information for each existing or pr0posed information sys-

tem.

 

34

Of particular interest is the effect of an industry's market
structure on organization to supply information. Since there are gener-
ally high fixed costs in information production relative to the variable
dissemination costs, one might expect a firm to exploit these decreasing
average costs by monopolizing the collection and dissemination of informa-
tion for an industry. The incanplete appropriability of the returns to
information production is one factor which decreases the likelihood of the
development of information specialist firms. Also, as Oliver Williamson
notes, the opportunistic behavior of firms reduces the probability of
existence of these information provision specialists. There is a risk
that any firm specializing in information provision will selectively dis-
tort the information it sells. Since the information is not easily veri-
fied, usually only by collecting original data again, exchange between
firms in an industry and the specialist firm will fail. Thus, the impor-
tance of credibility, pointed to by the fundamental paradox of information,
suggests that as long as firms are opportunistic in their bahavior, it is
unlikely that firms will purchase information from a profit seeking firm
specializing in infonnation.

Recent hearings by the United States House of Representatives
Small Business Committee concerning the manipulation of meat prices by the
National Provisioner Daily Market Service, a private meat price data col-
: lection firm, suggest that very little incentive is needed by specialist
firms to distort the information it provides. Testimony by the committee
staff investigator stated that the National Provisioner Daily Market Ser-
vice reported prices where only a limited number of trades or no trades
‘ at all took place. The implied incentive in this case was to maintain

sales of data to the specialist firm's customers.

 

35

As the number of transactions in the meat industry declined it
became more difficult to report prices for all types, weights, and grades
of meat. Instead of admitting this, the National Provisioner appears to
have continued reporting prices based on a small number of transactions
simply to give the appearance that it was doing its job. These actions
appear to have been undertaken to maintain customers. While these allega-
tions have not been proven it is suggestive of problems arising from the '
reliance on specialist firms for information (United States House of Re-
presentatives, pp. 245-292).

The argument against specialist finns hinges on the notion that
these firms will be opportunistic in their behavior, which Williamson de-
fines as seeking self-interest with guile. If opportunistic behavior is
not assumed, then the risk of strategic misrepresentation disappears and
specialization in the production of information is possible. This stresses
both the need for an unbiased, nonopportunistic firm or organization to
collect market information as well as the importance of reliability and
accuracy in data collection.
information production also makes the possibility for individual firm
production of market information less likely, except in the case of mono-
poly. Since there is only one firm in the industry in a monopoly situa-
tion, the benefits of any investment in market information for that indus-
try accrue directly to the mon0polist. Hence, it can justify its expense
and can expect to reap the benefits of any investment in information to
manage the industry.

Advertising by farm input firms is another example of some of the

problems of information specialization. The advertising function is

‘l‘ .l‘
IE

 

36
usually a case of specialization of information provision within part of
a larger firm. Here the obvious incentive is for the advertising branch
strategically to misrepresent or distort the information it provides to
farmers in order to increase the profits of the firm. Thus, the informa-
tion provided by advertising is of limited use to the farmer. For the'
most part, the useful portion of the information is only that which can be
easily verified such as price, product availability, and those quality
characteristics identifiable by in5pection. While truth in advertising
laws can be used to alleviate problems of gross misrepresentation by op-
portunistic firms it is unlikely that these will eliminate all biases.
When the advertising concerns product characteristics that can only be
determined through experience, such as the durability of a piece of farm
machinery, rather than by simple inspection, the opportunities for mis-
representation multiply.

The public good attributes of information suggest that some form
of collective action in information production should lead to an increase
in social welfare. .However, no indication is given from the characteris-
tics of information, per se, as to whether a voluntarily organized private
effort is possible rather than government intervention. In making this
choice the theory of groups can provide some insights.

If industry is viewed as a group of finns and information as a
public good, then Mancur Olson's theory of groups can be used to show the
effect of economic structure on information supply. Olson shows that the
ability to organize a group depends on whether the individual member is
able to obtain benefits in excess of costs. Olson has shown that some
small groups can organize to provide a public good without any benefits

other than those provided by the good itself. In cases where groups are

37

very small, i.e., where each member gets a major proportion of the total
benefits of the public good simply because the members of the group are
few in number, this public good can often be provided by the voluntary
action of the individuals in the group purely on the basis of the self-
interest of the group members. This suggests that as industry structure
moves toward oligopoly that market information is more likely to be pro-
vided by an industry association and that government collection of data
for private use in the industry is probably not necessary.

As groups get larger some other incentive such as government sub-
sidy or selective benefits in the form of private goods might be necessary
to organize a group to provide public goods. Since a public good is such
that use by one individual does not preclude the consumption by another
individual, even small groups will fail to provide an amount of public
good near the amount provided in the case where information possessed the
characteristics of a purely private good. The divergence between these
two amounts increases as group size increases (Olson). The conclusion
can be drawn from this that government intervention might be appropriate
to achieve a desired level of information production as an industry be-
comes more atomistic. This would also force some of the "free riders"
in an industry, who would use information provided by a trade association
but would not support the group financially, to pay for the use of the
information through taxes. It must also be kept in mind that government
information provision subsidized by tax revenues would force some taxpayers
who do not benefit from the information to pay for it anyway.

To summarize the arguments about the relationship of economic
structure to the supply of information for private use: First the problems

of indivisibility and nonappropriability make private data collection and

38
analysis unlikely under the conditions of a competitive market structure.
As an industry becomes more and more concentrated, it will be increasingly
in the self-interest of the firms in the industry to supply information
for their own use and hence government provision of information for pri-
vate use is less and less necessary. However, as the industry structure
moves toward more atomistic competition, then the argument for government
provision of data collection and analysis can be made on the grounds of
improved efficiency in the allocation of resources because the industry
is net likely to provide an amount any where near Optimum, assuming that
there are net social returns to the information in addition to the sum of

the private returns.

2-4. 3. INFORMATION FOR PUBLIC use
The previous discussion centered on providing data and information

For private sector decision making. For public sector decision making,
da ta collection and analysis is by definition a government activity.
GoVernment information provision is usually relied on to assure accurate
and credible information for public policy decisions and to avoid the
s‘:"‘ategic misrepresentation of information supplied by the private sector
‘PQF public decisions. This is not to say that government data and infor-
Wag tion is always beyond reproach. When data are used as performance mea-
S“ has for public policies it is easy for the data to become politicized
and lose credibility. Even the private use of data can cause it to be
“()1 ‘lticized. Private sector lobbies have politicized some public data
used in critical private sector decisions, e.g. the use of the Consumer

1:
hi ce Index in collective bargaining has tended to enhance the political

““essure on this data preventing timely modifications to maintain or

39
Somewhat similar

improve the reliability of the Consumer Price Index
pr1vate sector pressures have developed around the USDA measure of aggre-

gate farm income.
Pol1tic1z1ng data rarely involves changing the numbers, per se

Usually more subtle methods are found such as speeding up publicat1on of
favorable data or delaying unfavorable data to mesh more closely with
polf cy pr0posals, controlling the format of the data to 1nsure that tech-

nical interpretations fit with policy pronouncements or the fa1lure to make

changes in existing data concepts or definitions to fit reality since

these changes might make a situation look better or worse and thus make

the current government look bad.
Comments on personnel and other changes in the Federal Stat1st1cal

System of the United States in the early 19705 by Phil1p M Hauser show
that the integrity and credibility of data can be affected without actu-
Some of these events, such as appointing political

a 1 13' changing numbers. .
cronies to head statistical agencies and the cancellation of regular sur-

veys while yielding no evidence of direct political man1pulat1on of data,

least suggested possible impropriety and thus reduced the cred1b1l1ty

at
In agriculture the political difficulties encountered in

O
f the data.
I'"anging the definition of a farm gives an indication of why 1t 15 often

I.)
Q 1 itically difficult to make improvements in data once they have become

‘F‘i "mly institutionalized (Hildreth and Worden).
One should not paint too bleak a picture for government prov1ded

data for public decisions since the alternative is often much worse Many

ther types of difficulties are encountered when the government relies on
The current debate

phi vate sector information to make policy dec151ons
P“ the Un1ted States over the level of natural gas reserves points to the

40

Problems of strategic misrepresentation of data supplied by private firms

for public use (James N. Miller). Therefore, in designing a government

information system various efforts to guard against politicization of data
should be considered. Some of the important means of avoiding politiciza-
tion of data include the complete documentation Of methodologies used in
obtaining data, the encouragement of ties between government statisticians,
economists, and other analysts and academic and other nongovernment pro- '
fessionals in these fields, the selection of leaders Of statistical agen-
cies on merit not political acceptability, and an attempt to maintain

appropriate distance between policy formulation and evaluation and the

collection of statistics. In this last case we are not necessarily argu-

ing for a separate statistical agency, only that the functions of data

conection and policy formulation be quite distinct within an agency to

a\Hlficl even the appearance of imprOpriety.

2 - 4. 4. THE DEMAND FOR INFORMATION
The value of information is discovered only in a decision making

Qt)“text, so the demand for information is determined by its value in the
dQQision process of firms or government. The value of information, then,
‘T 3 not known with certainty until it is Obtained and used, so problems

a“? se in estimating the demand for information. Firms that are risk ad-
VQ "se will tend to demand less information because of the uncertainty of

t
heir returns, a priori, to investments in information.

As industrialization occurs in a country, production processes

b‘acome specialized so more informtion is required for firm and market

Q'Oordination. Thus, for the same investment in market information, returns

‘10 the firm should increase as an industry becomes more specialized. As

41
industry structure becomes concentrated, the returns to information invest—

ment for private use can be captured by the small number of firms in the

This ability to capture returns to information investment also

industry.
At the other end

affect the demand for information by an individual firm.
of the industrial structure spectrum where there are many independent firms,

as in agriculture, the amount of private sector investment in producing

information will probably approach zero since only a small portion of the
returns can be captured by any one firm. In this case the public returns

to information in the form of better coordination would probably exceed

the private returns. On this same continuum, the public use demand for
i nfor'mation will decline as one moves away from atomistic markets then

increase as information to monitor and regulate monopolistic industries

is necessary. This is expecially true where regulation on monOpolies

throngh antitrust laws is viewed as a socially desirable goal.
The need for data on monopolistic industries raises the additional

q“'esdzion of public access to data collected by highly concentrated indus-
1:
hi es and monopolies. There is a public interest in this type Of infor-

m
ati on which should temper any discussion of confidentiality and disclo-

Sub .
‘3. Data on these types Of firms are Often sensit1ve because Of the

Q
thentrated nature of the industry and the immediate effects of any
‘F
i i"hrs actions on the market. However, this makes this same data ex—
th '
Ql‘nely critical for public decisions. Thus the benefits to society of

t
hQ preservation of privacy, particularly among corporations, must be

Q i shed against the information needs of public policy decision makers

(BQHnen 1975b. Pp. 101-102).
The ability to capture returns is related to the ability to use

‘“formation. A farmer who has sold all of the farm's grain for a year

f '
)\¢

ES

42
will find further information on grain prices of little immediate value,
indeed, it has been known to lead to high blood pressure and other aggra-

The ability to use market information is also related to the
Therefore,

vat'ions.
accessability of sellers to different buyers of the commodity.

the market area covered (in terms of number of buyers) by an information

system should be related to the demand for the market information. A

farmer in California is not likely‘to be as concerned with spot market

prices in New York as those prices in California. This question of acces-

sabi ‘l ity to markets must be answered to define the area of coverage for

certain kinds Of market statistics. It should not be implied from this

that markets are not related or prices from other areas cannot enter into

the decisions of farmers in a given area. However it seems reasonable to

aSsume that sellers would be most interested in the prices from the spe-

(:1 fi c markets where their sales occur.
Up to this point we have dealt in terms Of the effect of indus-

tr»: al structure on information 'in a reasonably Obvious fashion. There is
a More subtle effect which runs in the Opposite direction, i.e., the
e"T'irect of information on industrial structure. Earlier some of the eco-

hQ'hic characteristics of information were mentioned. These can affect

15‘ _
i "‘111 size and industrial structure. The riskiness of information produc-

t‘i On is Often such that outside insurance cannot be purchased to Offset
that risk. Self-insurance in the form Of diversification is Often used
‘9 deal with such risk. This suggests that in order for a firm to be
ab‘le to produce information through data collection and analysis, it must
b5 large enough to internalize the risk Of losses in information gather-
‘ “9. Thus, information production is usually done by large firms and

‘ arge firm size is generally related tO industrial concentration.

43
The indivisibility of information also can affect industrial con-

centration, in that it leads to increasing returns in the use of informa4

Roy Radner notes that "the acquisition Of information Often involves

tion.
a 'set-up cost'; i.e., the resources needed to Obtain the information may

be independent of the scale of the production process in which the infor-
Robert Wilson calls this "informational econo-

mation is used" (p. 457).
mies of scale," and notes that this phenomenon is self-reinforcing in that

a higher scale Of operation justifies better information acquisition and
increased information acquisition will justify a higher scale of Operation.

Hence . economies in the acquisition Of information can increase firm size
Theoretically this will occur as long as infor-

t0 the point Of monOpOly.
While Radner and Wilson seem

mati on is acquired in an Optimal fashion.
to have horizontal firm structure in mind, as Williamson (p. 86) notes,

the same argument can be made for vertical integration in many cases.

Thus . indivisibilities in information and the resulting increasing re-

turns in use of information can also affect the vertical structure of a

s . . . .
QtI‘lzor by providing an 1ncentive for vert1cal integration solely to re-

d
u Ce uncertainty.
The arguments presented in this section demonstrate that it is

not a political or bureaucratic accident that government collects more
QTiailed statistics and does more analysis for private use on highly com-
QtZitive, atomistic industries, such as agriculture, than it does in more
QQ"boentrated industries such as steel or autos. Publicly collected data
fab private management decision making have played a substantial role in
the great increases in agricultural productivity in the United States
“Var the past century. Society has captured the returns to improved re-

SOUrce use in agriculture through lower food costs and the availability

44

(H much of the former farm labor force for nonagricultural production.

The greater returns to society through improvement in resource allocation

from better public information on competitive industries when compared to
concentrated industries provide the primary basis for allocating public
monies for statistical systems to support private decisions. Hence, the
logical allocation of public resources not just for public decision needs.
but for private uses follows from the nature of the industrial structure

itself (Bonnen 1976, pp. 14-15).

2.4. 5 - INCOME DISTRIBUTION AND THE DISTRIBUTION OF INFORMATION
Modern economic theory also has been particularly deficient in
deal ing with distributional issues while concentrating on problems of
in"Ocative efficiency. Just as uncertainty is usually assumed away in
models of resource allocation, the distribution of income is Often assum-
‘3‘: lto be Optimal to begin with and hence is not treated. The connection
betiween the distribution Of information and the distribution of income
Seems to be a key but Often-overlooked notion.
As Lester C. Thurow argues,
"The factors that cause changes in the distribution of
income are themselves distributions. The distribution Of
education and training affects the distribution of income.
Thus, to adequately study the American distribution Of
income, it is necessary to develop methods of explaining
the distribution Of income in terms of the distribution
of causal factors which influence it.“ (p. 261)
“‘Formation is clearly one Of those causal factors to which Thurow refers
and thus the study Of income distribution requires an understanding Of
the distribution of information.

The distribution of income can be discussed at two different levels

a“cl the types and impacts of information will differ at each of these

45

First, one can consider the overall or size distribution of in-

levels.
Secondly, there is the question Of income

come in the entire society.
distribution among given individuals or groups Of individuals within
society. This distinction may be viewed as a macro-micro delineation of

the problem.
Keeping in mind that we are considering information in a systems

context, the distinction between data and information has important im-
plications for income distribution. Data require analysis and interpre-
tati on in the context of a specific decision to become information (Bonnen

In general, a more equal distribution Of data among mem-

1975a, p. 758).
bers of society is likely to have quite different effects on income than

an equal distribution of information because of the disparity in analyti-

Mpabilities Of those receiving the data. It is this analytical
capability that Thurow and others seem to have in mind when discussing

the relationship between the distribution Of education and training and
the distribution of income. Insofar as education provides superior data
h terpretati on capability among members Of society, one would expect that

I:
he distribution Of education and hence information would in turn be re-

1 g . .
ted to the distribution of income. This expectation is supported by

‘t:
he literature (Thurow).
This relationship between education and income has been directed

Dh‘i marily at the most general level, i.e., its effect on the size distri-
Eution of income in society. As Donald M. Lamberton (p. 462) notes, the
SQ heral expectation of improved information is to reduce inequality in

DQ‘Wer, wealth, and income. HOwever, it is at this most general societal
~‘e‘lel that improved information is Often least likely to cause the desired

reduction Of inequality because of the different capacities of firms and

46
individuals to use or acLOLthg information that they receive, even given
the same capability to analyze and interpret data. When new information
becomes available to both concentrated buyers and dispersed sellers, the
buyers are at a great advantage. Not only dO the buyers have greater
analytical capabilities and capacity to use information but they also have
a greater capacity to take counteraction. For example, published prices
can also make price fixing agreements between buyers easier to maintain.
While there is no evidence Of this actually'happening, it is conceivable
that an agricultural information system could encourage price fixing
among Oligopsonistic buyers Of agricultural conmodities. Furthermore, in
a recent unpublished survey of nonrespondents to U.S. Department of Agri-
culture surveys, the two comments most frequently given by farmers for
their refusal to respond were that 1) the surveys benefit others more than
farmers and 2) the surveys hurt farmers. Whether or not these statements
are completely true is debatable but at least it suggests the possibility
0f data being used against the more dispersed traders in a market.

These insights have important implications concerning the relation-
5hlp of information to income distribution. First, since the value Of in-
1"ovulation arises only in its use, the capacity to use or act on informa-
tion will have a significant effect on income distribution. Second, eco-
nomic structure again influences the distributional consequences Of infor-
mation. Insofar as the size Of firms are related to market structure, e.g. ,
flmS in OligOpOlistic industries are assumed large enough so their deci-
lenS will influence the market, large firms probably have both superior

a"i”.Y’Cic capability and a greater capacity to use or act on information
than do small firms. These larger firms can be expected to use their

superior information to influence in its favor transactions with smaller

‘7»
Hi

'{n
'1‘

47
less informed firms, so the subsequent income distribution will favor
larger more concentrated firms. The distribution of income is primarily
determined by the outcome Of the market in the private sector (Weisbrod,
p. 2178). SO as information affects market structure and behavior, it
also affects income distribution.

This reliance on market transactions to determine income distri-
buti on highlights the importance Of the distribution Of information be-
tween individuals in an exchange situation. The problems here are at a
more micro-level than those discussed earlier and the effects of the dis-
tribution of information are on the distribution Of income between the
individuals involved. This is similar to the market failure brought on
by what is called information impactedness by Williamson. He argues that
this phenomenon,

". . . is attributable to the pairing Of uncertainty with

Opportunism. It (information impactedness) exists in

circumstances in which one of the parties to an exchange

is much better informed than is the other regarding the

underlying conditions germane to the trade, and the second

party cannot achieve parity except at great cost--because

he cannot rely on the first party to disclose the informa-

tion in a fully candid manner." (p. 14)

When trading occurs under the circumstances of asymmetrical information,
one can only expect a redistribution of income in favor of those who pos-
5855 the supperior information when compared to the case when information
Impactedness does not exist.

In the previous section, it was argued that the governmental col-
1ECtion of data and its production Of information for agriculture could

be Justified in terms Of improved resource allocation. Burton Weisbrod
(p. 179) makes the point that income redistribution can be undertaken in

a number of ways including the use Of redistributional "side effects“ Of

P°HCles that are usually considered to have efficient resource allocation

48
as their goal and not income redistribution. To the extent that govern-
ment wishes to redistribute income in favor of agriculture, one can argue
that many of the programs to improve the information system for agricul-
ture are achieving this Objective, even though many of the programs are
aimed at the resource allocation problems caused by uncertainty. However,
even those programs which tend to equalize the access of information in
trades, such as price and production estimates, might not have desirable
income distribution effects because of the market structure in agriculture.
The predominance Of atomistic producers and concentrated buyers in this
sector may prevent any major redistributions Of income between buyers and
sellers because Of the superior analytical capability and ability to use
government produced information possessed by the larger firms in the agri-
cultural sector. However, tO the extent that government research and data
collection tend to equalize the information of individuals involved in ex-
changes Of agricultural commodi ties, there will be a change in the distri-
buti on of income toward greater equality. Many programs have tended to
achieve this desired income distributional change. For example, the
land grant college system and extension education programs probably have
increased the analytical capability of farmers relative to those with
“Men they deal, and the establishment and regulation Of futures markets
91 Ve farmers a greater capacity to use or act on information that did not
previously exist for farmers.
The major arguments Of this chapter suggest that there exists a

Chronic or absolute underinvestment in information production because Of

the economic characteristics of information. Perhaps of greater impor-

tance are the implications Of these theoretical findings for the relative

distribution Of investments in data collection and analysis. In those

H
*H
Li

"'1
u.

f:'
"l

is.

rx:
4

49
cases where changes in the information system benefit some groups more
than others or at the expense Of different groups, a basis is required
for determining which changes represent the best use Of government re-
sources.

The criteria for making improved government investments in data
collection and analysis follow from the market structure of the industry
which benefits from the information and from the effects Of the informa-
tion on income distribution. Using these criteria it is possible to es-
tablish relative priorities for public investment in information. Wise
public investment in data collection and analysis should help to equalize
the information Of individuals involved in transactions and thus will

lead to greater equity in the income distribution.

2.5 SUMMARY

In order to evaluate an Operating information system it is parti-
cularly useful to have a framework to serve as a guide in structuring the
evaluation. For the purposes Of this study an information systems para-
digm provides the framework for the subsequent analysis. The main focus
of this paradigm is with problem solving and to the extent that problem
solving decision making requires information from a variety Of disciplines
this framework is multidisciplinary.

An information system as presented here has three major parts:
1) a data system which includes data concepts, the operationalization or
definition of these concepts and the measurement Of these defined concepts
to produce data; 2) an analysis or interpretation step to transform data
into information for decisions; and 3) the decision maker. When applied

to Operating information systems this paradigm has some minor deficiencies

'b
12.5

, ,
i'nh
1' w

I c:
95..

ME

CECE

IAI-

*le

50
with respect to the formatting and communication of data and information.
Problems of conceptual Obsolescence become apparent in Operating data
systems when the concepts used in the data system are no longer represen-
tative of reality or when the agenda for decisions changes so the concepts
are not pertinent for the types Of decisions that must be made using the
data.

The primary economic implication Of this paradigm is that infor-
mation only becomes valuable in the context of decision making. The pub-
lic good characteristics Of information coupled with the implied condition
that information can be treated as a commodity within a decision making
framework has major consequences for understanding the role of government
in producing information.

With these characteristics and conditions in mind, two variables
or relationships become critical in the design or redesign of an informa-
tion system. The first Of these is the configuration Of the relevant eco-
nomic sectors. Economic structure affects the supply and demand for infor—
mation in both the public and private sectors, has consequences on the
distribution of information and is critical in understanding the appro-
priate role Of government in providing information.

The second key theoretical relationship considered was that be-
tween the distribution Of information and income distribution. Economics
Often neglects the effects of the distribution Of information on the dis-
tribution Of income but, in many instances, equity concerns lie behind

the reason for allocating public funds for data collection and analysis.

CHAPTER 3
REVIEW OF LITERATURE ON AGGREGATE FARM INCOME DATA

3.1 INTRODUCTION

A review Of the relevant literature for this study is in order.
A comprehensive treatment of the literature on the economics Of informa-
tion or information systems in general is not intended. These were
covered in part in the development of the conceptual framework in the
preceding chapter. Instead, this literature review will focus on some of
the recent literature on aggregate farm income accounting. The intention
is to concentrate on the important works related to this study and to
point out the relevant aspects of these for this analysis. These earlier
studies provide a set Of recommendations for farm income data improvements
but their approach and intent often differ from the type of analysis un-
dertaken in this study. After examining the USDA farm income data in an
information systems context, it should be possible to set priorities
among the improvements recommended in the earlier studies.

Within the area Of aggregate farm income accounting two aspects
will be emphasized. First, some Of the recent reports and evaluations Of
USDA farm income data will be summarized. Second, some alternative con-

ceptualizations Of farm income or economic well being Of farmers will be

examined.

51

 

H' 1
I'd-

'_1
(t)

L“..
«'12

h

lE‘.

are
ll“;

is

52
3.2 ANALYSES OF FARM INCOME DATA
3.2.1. GROVE

Ernest W. Grove was one Of the first agricultural economists to
question the relevance of a concept Of farm income to the current needs of
data users. Grove's focus was rather narrow and concerned the apprOpri-
ateness Of the realized net farm income concept as Opposed to the total
net farm income concept. Data on both are estimated by the U.S. Depart-
ment of Agriculture.

The concept of realized farm income does not consider increases
or decreases in farm output inventories as part of income, while the total
farm income concept does account for the value of crap and livestock in-
ventory change. Grove argues that there are six major deficiencies in the
realized farm income concept that make it less desirable than the total
fanm income concept. First, since total farm income is needed for the
Department of Commerce's estimates Of farm product, the publication of
both realized and total farm income can cause confusion among unSOphisti-
cated users of the data. This dilemma suggests that understandability Of
data is an important consideration in data system design.

Second, he argues that the realized farm income concept is con-
trary to generally accepted practice and the theory Of income measurement.
This stems from the usual notion in economics that increases in inventories
are actually a form of savings from current income and real income can only
arise out Of current production. In part the distinction between realized
and total farm income is somewhat analogous to accounting on a cash or
accrual basis. However, the analogy is not exact since producer owned
purchased input inventories are not included in total net farm income. 50

at least by analogy it seems that Grove is arguing that the accepted theory

53
and practice in income measurement is to account for income on more of an
accrual basis. While accrual accounting dominates other sectors of the
economy because Of Ferderal income tax rules, it is clear that cash ac-
:counting is much more prevalent in agriculture. Harrison notes that 98%
of all 1969 farm tax returns used cash accounting methods. But Grove is
correct in saying, in effect, that the accrual method is the most accepted,
method in the sense that the National Income Accounts Of the Department
of Commerce are on an accrual basis. Grove seems to imply this when he
traces the develOpment Of the total net income series. This series arose
out of a need by the Commerce Department to have a measure of farm in-
come more closely in line with its accounts.

The third Objection that Grove has to realized farm income is that
it provides an undesirable choice Of statistics even among sophisticated
users. Grove Observed that the realized income figures were used by de-
cision makers in USDA except in those cases where the total income data
better proved the point that these decision makers were trying to make;
so that having a choice Of two series might tend to confuse some issues.
This argument is certainly not unique to farm income data and applies to
almost any data sets which measure similar phenomena. SO the usefulness
of this argument alone is somewhat suspect.

The fourth Objection arises indirectly from measurement difficul-
ties in estimating the value of inventory change. Grove claims that
since physical inventories are the residual Of production less marketings
and that production is measured with more relative accuracy than market-
ings, then total income is more accurate than realized income. In the

case of total income, any errors in estimating marketings are Offset by

54
changes in inventories since inventories must equal production plus be-
ginning inventories less marketings and home consumption. C. Kyle Randall,
in his comment on Grove's article, points out that it is not always as
easy to measure total income since production is not measured for meat
animals. SO inventories here tend to be more difficult to estimate.
Related to this, and shown in Breimyer's comment to the Grove paper, is
the difficulty in valuing inventories. The inelasticity Of demand for
agricultural products means that using average yearly prices to value in-
ventories tends to overvalue large year end inventories and undervalue
small inventories.

These phenomena also, as Harold F. Breimyer suggests, tend to
exaggerate year tO year swings in the income actually received by farmers
when compared to total net farm income published by USDA. This tends to
refute another of Grove's Objections to the realized income concept, that
being that it has resulted in an unrealistic smoothing out of farm income
estimates by eliminating increases and decreases in output that are added
to or taken from inventory. The smoothing occurs because inventory change
is not considered in the realized farm income concept, hence this form
Of forced saving from current production is not counted. Breimyer's point
that large inventories are overvalued using the current methods also re-
duces the significance of Grove's other Objection, that over decades when
aggregate inventories are increasing an omission Of income occurs by using
the realized farm income concept.

For the purposes of this study the measurement problems associated
with realized farm income are not as critical as those involved in the
first three Of Grove's.objections which suggest that there might be diffi-
culties with the concept used by USDA in making policy decisions. While

55
his case is far from convincing, Grove at least raises the question Of
conceptual relevance which is a first step in understanding the problems
that should be considered in any prOposed redesigning of the farm income

information system.

3.2.2 TASK FORCE ON FARM INCOME AND CAPITAL ACCOUNTING 1972

‘ In 1972 a Task Force Of U.S. Department Of Agriculture economists
and statisticians headed by Eldon Weeks completed the first comprehensive
evaluation Of the department's farm income and capital accounting data.
In addition to the final Task Force Report (Weeks, g5, p1,, 1972), numer-
ous other publications resulted from this study, e.g. Weeks 1971, Weeks
1972, Carlin and Handy, Carlin and Smith. These will all be considered
as a single develOpment in this literature review, except for those which
summarize the Task Force's recommended alternative accounting system
which will be dealt with in the following section on alternative concep-
tualizations. Unless otherwise noted the following discussion Of this
development will be based on the Final Task Force Report (Weeks, 35, 31.,
1972).

The Task Force report covered the basic accounting for aggregate
farm inputs, outputs, capital, farm Operator inputs and to a more limited
extent the distribution Of these aggregates by farm size, type and loca-
tion. For the purposes of this study the capital accounting aspects will
not be considered except as these relate to farm income. More important
are the areas not covered by the Task Force. Given the information sys-
tem paradigm that provides the conceptual framework for this study, a
serious shortcoming of the 1972 Task Force was their explicit failure to

consider farm income accounting designed specifically to aid decision

56
making.

"Finally, possibilities of accounting for institutional

mechanisms designed to facilitate or exercise decision

making and control were considered out Of scope for the

pgrposes Of this Task Force." (Weeks, 31,131., 1972, p.

While it may be true that the 1972 Task Force was directed to stop its
analysis at the data output stage of the information system, it does not
relieve them of the necessity to consider how the data are ultimately
used in order to make apprOpriate recommendations concerning the concepts
to be used in redesigning the data system.

The first step in understanding the 1972 Task Force report is to
look at some Of their assumptions used in evaluating the current account-
ing system. This is also significant since much of the analysis in the
later chapters of this study will at least implicitly test the validity
of some of these assumptions.

The Task Force assumes three major purposes for an aggregate farm
sector accounting system (Weeks, 33, 31,, 1972, p. 5). First, the system
should describe the major economic features of the farming sector when
presented in aggregate terms. Implied in this is that the concepts of
data system.should bear a close correspondence to the reality of the
farming sector in order to be useful. The second assumption is that the
system should yield aggregate and individual series which are measures Of
performance for the farm sector for both public and private uses. The
final general assumption is that the system should provide data which is
easily compared with other sectors of the economy. This would seem to
imply that these types Of comparisons are useful in public and private

decisions.

From these three general assumptions six more specific assumptions

57
were made concerning improvements in the current farm income and capital
accounting system. First, the current system should not be abandoned,
i.e., any recommended system should not be substituted for the current
series. This at least implicitly assumes an almost unlimited budget for
farm income data provision in the short run. Second, any recommended
system should be readily reconcilable with the Department Of Commerce's
National Income and Product Accounts. Third, the system should be for-
matted in such a manner that the data can be used in national input out-
put models. Fourth, the basic farm income and capital accounts should
provide the basis for describing the performance Of the farm sector
through time. Fifth, the data system should have a conceptual basis that
allows for easy cross sector and within sector comparisons. Sixth, the
system should distinguish between farm businesses and farm households and
between the long and short run while reflecting the unique characteristics
of the farming sector.

The major difficulty for the purposes Of this study with the as-
sumptions outlined in the Task Force Report is that it is never stated
why these Specific assumptions are made or on what ground the assumptions
are based. SO before one can establish priorities for improving the
current farm income data system using the Task Force recommendations, it
is necessary to assess the relevance Of these assumptions in an informa-
tion systems context. This topic will be addressed later.

After presenting a recommended accounting framework, the Task
Force appraised the existing USDA accounting formats for farm income, us-
ing the recommended framework as a norm. The farm income data series were
evaluated along three separate lines: 1) overall format, 2) the concep-

tual basis of the most important series and 3) the linkages between

‘ 58
series. In their appraisal Of the overall format, the Task Force con-
cluded that the farm income accounting format does not resemble the formats
used in other sectors or countries and does not yield measures Of perfor-
mance similar to those found in other sectors. This arises in part be-
cause much Of the capital formation in agriculture takes place on the
farm with the expenses being measured as current production expenses.
Thus, the input and output sides of the current accounts are inconsistent.
and the distinction between the production and capital accounts are un-
clear.

Certain conceptual problems arise within specific components of
the farm income data that cause difficulties in using the farm income
series. The 1972 Task Force cited five Of these problem components that
relate to farm income. Government payments can be treated in two basic
ways in the accounting system. Direct government payments can be thought
of as compensation to farm businesses for the production of public goods
or as income transfers tO farm families. Since the costs of compliance
to the programs are currently considered as production expenses, an in-
consistency arises when government payments are treated as income trans-
fers. 50 depending on the concept Of production expenses used, the
treatment of government payments might be different.

Farm housing is another component which causes some conceptual
difficulties. The maintenance and operation costs of farm housing are
treated as production expenses, so it is necessary to consider the gross
rental value of farm housing as farm output in order to be consistent.
Under alternative accounting systems this treatment of farm housing might
be different depending on whether one includes housing as part Of the

Output of the farm sector. For instance, under the national income and

59

and product accounting formats, one might wish to consider farm housing
as part of the real estate sector. If this were implemented, then it
would be necessary to change the maintenance and Operation costs of farm
housing to the real estate sector rather than the farm sector.

Depreciation was the third problem identified by the Task Force.
In addition to the problems associated with using book value or replace-
ment value in measuring capital consumption a further difficulty arises
because Of the own account capital formation on farms. As noted earlier,
own account capital formation in the farming sector is generally not ac-
counted for in the current data system. SO depreciation using either
book value or replacement value is understated in the existing series.

The conceptual basis of the inventory measurement in farm income
also poses a dilemma when the current system is considered. Questions
arise because of the content of some of the inventories. For example,
beef breeding herds, dairy herds and laying flocks are all considered as
part of the current livestock inventories when these are actually capital
items in most uses. SO the current accounts might reflect changes in
capital items when inventory change is measured. Under alternative con-
ceptualizations one might also want inventories Of work-in-progress in
addition to finished goods, both Of which are reported for other sectors.
The Task Force recommended that work-in-progress, such as cattle on feed
and crops in pre4harvest stages of develOpment, should be included in the
linventory change measured for the farm income accounts.

A final quandary pointed out by the Task Force was in the area of
measuring unsold output. The current farm income accounts do not provide
a measure Of total output of the farm sector. Without a measure of this,

increases in cash receipts or purchased inputs could be interpreted as

60
changes in industry size and/or structure. But in reality, these might
only reflect an increase in specialization. For instance, feed grain
farmers raise fewer livestock and sell the grain they had previously fed
to livestock. Output in this case is sold and shows up in the accounts as
cash receipts for the feed grain farmer and a production expense for the
livestock feeder but the total production Of feed grains did not increase.
Thus the conclusion that a better accounting of intermediate products is
necessary formeasuring aspects such as the productivity and size of the
farming sector.

Definitional and conceptual difficulties arise when the relation-
ships between the various components and series of the farm income data
system are examined. The linkage between returns to resources in farming,
as measured by realized gross and net farm income, total net farm incOme,
and personal income of the farm population highlights one of the major
problems in the current system. In tying these series together it is
assumed that for every farm there is only one farm Operator. This seems
to be a rather tenuous assumption and makes it quite difficult to assess
the impact Of the changing tenure structure in agriculture.

In the formal accounting sense there are very few linkages between
the current accounts in the farm income series and the capital accounts in
the balance sheet series. The problems associated with inventory and de-
preciation concepts and definition weaken the linkages between current
land capital accounts”

Gross farm income is derived mainly from ESCS estimates of sales
of commodities, government payments, and imputed nonmoney income, while
production expenses are estimated on a national basis primarily from

census benchmarks, ESCS prices paid, and industry data. Thus, the

61
empirical linkages between these two aspects are also tenuous. The fact
that some production expenses for one type Of farmer can be cash receipts
_for another only complicates these linkages. SO this system of indepen-
dent estimation Of income and expenses may cause the linkages between the
series to be less direct.

In considering the various disaggregations and distributions Of
farm income the linkages in the system vary from quite direct to nonexis-
tent. The disaggregation Of cash receipts by state is very direct since
prices and sales are estimated at the state level. However, on the ex-
pense side disaggregations are based on much more limited data, since ex-
penses are first estimated at the national level. The distribution of
income and expenses by sale class is based on Census Of Agriculture bench-
marks, hence this series has a number of consequent measurement problems
in intercensal years. However, this series for the most part provides
useful information as judged by the Task Force. The disaggregation of
these sales class distributions by state or region is not even attempted
by USDA, suggesting rather weak linkages between these aspects.

The Task Force also notes that no attempt is made on a regular
basis to distribute farm income and expenses by type of farm. A lack of
appropriate data is blamed but they do not conjecture why these data do
not exist. Perhaps the conceptual problems associated with determining
Vpreciselv what constitutes a certain type of farm leads to this data gap.
Given the probelms USDA encountered in redefining a "farm" (Hildreth and
Worden), one would expect similar problems in defining a "cash-grain farm"
or most other types Of farms.

All in all the 1972 Task Force made a significant contribution in

suggesting improvements in the USDA farm income extimates. However,

62

various reorganizations within ESCS: Economics and revisions Of the cur-
rent farm income data series has precluded the implementation Of very
many of the Task Force recommendations (Guebert).

For the purposes of this study, two major criticisms Of the 1972
Task Force Report are apparent. First, in making recommendations for
improvements, no basis was provided for establishing priorities within
their suggested accounting framework to guide those who would wish to im-
plement these recommendations in an incremental fashion rather than as a
whole package. While the task force did recommend that as much as possible
be done in implementing the alternative accounting system over a two year
period and that any residual work be prioritized at the end of two years,
a means for setting priorities was not presented. Second, and even more
significant, was the fact that the Task Force assumed certain purposes for
a farm income accounting system without determining if these purposes

suited the users Of the data.

3.2.3 TASK FORCE ON FARM INCOME ESTIMATES 1975

In 1975 a second Task Force comprised primarily of non-U.S. De-
partment of Agriculture employees was established to evaluate the Farm
Income estimates. It did not examine the overall income and capital
accounts, as did the earlier Task Force. The "Report Of Task Force on
Farm Income Estimates" (Hildreth, 31, 31,) will be summarized in the fol-
lowing paragraphs. The exceptions tO this will be some Of the recommen-
dations of the Task Force which refer to alternative conceptualizations
Of farm income which will be summarized in the next section.

The 1975 Task Force was primarily interested in reviewing the

methods and techniques for estimating farm income. Hence most of their

63
recommendations centered around these type of improvements. The formation
Of this Task Force was prompted by the large revisions required between
the January and July, 1974 estimates Of 1973 farm income, which helps to
explain their focus on measurement problems. During the period 1960-1971
the changes in farm income were gradual and the revisions of the data pub-
lished by USDA were relatively small. From 1972 to 1973 farm income in-
creased nearly 78% and the subsequent July 1974 revisions of total net
farm income and realized net farm income showed increases of 35% and 23%
respectively. The revisions focused attention on the procedures used to
estimate farm income. Many of the problems on which the Task Force fo-
cused were caused by USDA's reliance on historical marketing patterns to
allocate crop and livestock sales during the year.

One Of the principal areas addressed by the 1975 Task Force was
the integration of farm income data with the national income and product
accounts. GNP and other summary measures provided by the national income
and product accounts force a certain consistency in accounting among sec-
tors and make farm income accounting potentially more difficult than would
be the case if only agricultural sector uses were considered. The Task
Force Report cites a U.S. Department of Commerce, Bureau Of Economic
Analysis study which showed that the quarterly estimates Of farm income
had the largest revisions Of any income type in the national income and
product accounts over the prior ten year period. The Task Force suggests
lthat this may occur because unrealistic measures are used. An additional
explanation not considered by the Task Force is that, since more data are
available for the agricultural sector relative to other sectors, farm in-
come might be subject to greater revisions than income in other sectors.

Further, the greater inherent variability in farm income relative tO other

64
sectors would lead to greater expected revisions for the farm sector even
if the measurement procedures were comparable among sectors.

Farm inventory estimates also caused a problem with revisions
even during the 19605. The importance of inventory estimates is magnified
somewhat since the USDA estimate Of net change in farm inventories enters
into the change in business inventories component Of GNP in the national
income and product accounts on a quarterly basis. In order to be consis-
tent with other sectors the farm sector also needs to account for inven-
tories of purchased inputs and "work-in-progress." The Task Force then
suggests that this latter component is probably not feasible to measure
at this time, since weather and other factors can significantly affect
growing crops. Thus, treating growing crops as work-in-progress may
cause some difficulties.

The separation of corporate income of farms is also difficult
under the current system but is required in order to arrive at an accurate
measure Of national income. The income of farm establishments owned by
corporations in non-farm industries can be double counted in the current
national income and product accounts. This also raises a question Of the
need for disaggregated or distributional information on the claimants to
income in the farm sector. Increases in corporate ownership are not well
documented nor are other organizational or ownership pattern changes in
the farm sector. Thus, little data exists on the distribution of income
by the legal organization of farms. While suggesting that the proposed
Census classification of primary, part-time and business associated farms
would improve the data on farm income by giving estimates by organizational
form, the Task Force avoids many Of the distributional questions by argu-

ing that these are not in the scope of their study.

65

The recommendations of the 1975 Task Force were in four broad
areas: accounting rules and definitions, baSic data, timing and revisions,
and improved techniques for data use. Under accounting rules they suggest
that the farm should be treated as a business establishment rather than a
family or household and thus the term "net income of farms“ should be sub-
stituted for "net income of farm operators." In addition to this they re-
commend that economic activity in the farm sector be measured and empha- ‘
sized primarily as gross value addedl/ rather than net income of farms.
This former concept is a more comprehensive one and includes the latter
as a principal component and can be reasonably estimated with existing
data. They also recommended that farm income be measured on an establish-
ment basis in preference to a product basis,§/ but at the same time recog-
nized the difficulty in develOping a precise definition of an establish-
ment consistent with other sectors. A further examination of the estab-
lishment concepts used in other sectors shows that these concepts are
quite varied and even incompatible in aggregation. The differences arise
primarily in the operationalization of the establishment concept, however,
the vehicles used for data collection in other sectors also tend to aggra-
vate these differences. Thus, the difficulty in precisely defining an
establishment concept, noted by the Task Force, might have been due to
the existing inconsistencies in the operationalized concepts used in other
sectors.

The Task Force also perceived a need for better data on industry

I] See next section of this chapter for a definition of this concept.

g/ Ibid.

66

size and specialization and hence argued for better measures of interfarm
transactions with offsetting entries in cash receipts and production ex-
penses to assure accurate net income estimates. ApprOpriate changes to
separate capital formation on farms, paricularly own account capital for-
mation, from current production were recommended. Specifically, removal
of increases in beef breeding and dairy herds and laying flocks from in-
ventory change was recommended along with the inclusion of depreciation
on own account capital in the production expenses. The Task Force also
recommended that the treatment of CCC loans be changed to include crops
under loan as farmer owned inventories rather than sales unless the far-
mer forfeits the collateral for the CCC loan. The current practice treats
CCC loans as sales at the time the loan is made and appropriate adjust-
ments are made when CCC loans are redeemed. The current practice was
probably more apprOpriate during the period of the 19305 through the 19605
when chronic surpluses dominated in the farm sector. The establishment of
farmer held grain reserves under the Food and Agriculture Act of 1977 also
gives added significance to the Task Force recommendation.

Consistent with the stated earlier recommendations of this Task
Force, they also recommend that inventories of purchased inputs be mea-
sured so that current and suggested income measures more accurately re-
flect income from current production. They further suggest that the
gross rental value of farm dwellings and associated production expenses
'be included in the real estate industry rather than farming industry.
The net rental income of nonoperator landlords is excluded from farm in-
come but to be consistent with the gross value added on farms concepts,
this should be included along with appropriate depreciation, indirect

taxes and interest payments of nonoperater landlords. The final

67
recommendation concerning accounting rules and definitions follows Grove's
recommendation that "realized net farm income" be dropped as a separate
series.

In the area of basic data the Task Force recommended that better
quarterly data be collected on: 1) crop movements, 2) expenditures for
feed and livestock, 3) inventories of cattle and calves, and 4) inventories
of purchased inputs. The probability surveys of grain buyers now used by I
ESCS: Statistics is a reSponse to this first request since it now provides
.data on the movements of major grains within a month after the sales. At
the time of the 1975 Task Force Report, data on crop movements often was
not available until 18 months after the end of the calendar year. However,
this does not help to measure interfarm sales nor does the probability
survey cover all cr0ps, so the basic data in the area of crop movements
still does not meet all of the recommendations. 0n the livestock side
the Task Force noted the inadequacy of data on interstate movement of
feeder and stocker cattle for which better data is needed for accurate
farm income estimates. In line with their earlier recommendations on in-
terfarm transactions, the Task Force recommended that investigations be
made as to the feasibility of collecting such data for livestock.

In the area of production expenses the recommendations centered
around making better use of and increasing the sample size of the ESCS:
Statistics Farm Production Expenditure Survey to obtain expense data by
region and size and type of farm." Among particular expense items the need
for quarterly data on feed, livestock and fertilizer was deemed most cri-
tical because of the inherent variability in the purchases of these items
by farmers during the course of a year.

Since changes in inventories displayed the most need for revisions

68
in the past, this area was also cited as needing more and better primary
data. Cattle and calf inventories, especially cause difficulties because
of the failure to separate livestock on feed, which are really work-in-
progress, and beef breeding and dairy herds which are actually capital
items. Thus improved data of sufficient detail to separate out these
aspects are needed. The absence of data on purchased inputs also prevents
accurate income estimation on an accrual basis and this type of data is
necessary to have accounts consistent with the national income and pro-
duct accounts.

The price data used in estimating farm income was judged adequate
for the most part by the Task Force. Problems do occur when only season
average prices are available for certain craps, especially when prices
and sales are changing during the year. Contract prices where the terms
of trade are unknown also cause some problems when these prices are used
in estimating income. For income estimation the Task Force was particu-
larly concerned with matching prices with quantities sold rather than the
conceptual problems in discovering prices. This is in keeping with over-
all concern of the Task Force with the estimation accuracy of revisions
of the farm income data.

The final two areas of recommendations concern the relationship
of farm income work within the U.S. Department of Agriculture and between
USDA and the Bureau of Economic Analysis (BEA) in the Department of Com-
merce. The constraints placed on USDA by BEA can lead to problems in
the timing of revisions. For the purposes of this study the recommenda-
tion for a higher priority for farm income work within the USDA particu-
larly within ESCS: Economics is important. The implementation of many of

the recommendations of the Task Force is impossible without increased

69

staff and budget resources. As in the earlier Task Force study the ques-
tion of priorities among the recommendations is not well developed so
that it is difficult to make improvements within an incremental budget
process. Thus a study of the uses and users of the data seems necessary
to a better understanding of the benefits of various improvements, so
that some priorities can be set on implementation of the suggested changes.

A point raised by Emanuel Melichar, concerning the recommendations
of the Task Force to separate out the inventory change of beef breeding
and dairy herds from inventories of current production, suggests that the
overall impact of these data improvements might not be very significant
in terms of the effect on total farm income. However he is quick to add
that such an action is conceptually sound. This does suggest that in de-
termining priorities among different types of changes in estimating pro-
cedures or accounting rules, those which have the greatest effect on the
aggregate income figures might be given priority. However one must also
give consideration to the importance of individual components of the
aggregate which often have separate uses. In the case of farm inventories
it might be very important to have measures which are conceptually sound
since these data also affect business inventories in the National Income
and Product Accounts which have uses distinct from the aggregate farm in-
come data.

In general the motivation and recommendations of the 1975 Task
'Force on Farm Income are quite different from the 1972 Task Force. The
more recent group was concerned with improving the preliminary farm income
estimates each January and with the current quarterly farm income estimates.
So in general the recommendations were not centered on conceptual issues

but rather on the provision of key bits of data on a more timely basis in

70

order to improve accuracy and reduce revisions.

3.2.4. GROSS NATIONAL PRODUCT DATA IMPROVEMENT PROJECT

The Gross National Product Data Improvement Project was undertaken
in 1973 by the Statistical Policy Division of the Office of Management and
Budget (now the Office of Federal Statistical Policy and Standards in the
U.S. Department of Commerce) to examine the accuracy and timeliness of
the underlying data base of the National Accounts. This project ended in
1976 with their report following later. Chapter six of their final report
is entitled “Improving Non-Benchmark Estimates: Farm Sector." They justi-
fied this separate study on farming sector income even though gross farm
product has traditionally been only about 4% of GNP. They cited four
reasons for examing farm income data. First the volatility of farm out-
put in the short run can greatly affect quarterly changes in overall na-
tional output. Second, farming still is a major industry in the economy.
Third, the importance of U.S. Farming in the world food economy has in-
creased in recent years. Fourth, the current agricultural statistics
show more clearly than other sector statistics the problems of obsoles-
cence in the Federal statistical system arising from structural changes
in an industry. 6

The concern of the GNP Data Improvement Project was almost exclu-
sively on the current quarterly estimates of national income. In the farm
'sector accounts this concern was manifested by examining the data base for
the current quarterly estimates of income originating in the farm sector
and to a slightly lesser extent the January estimates of farm income. The
quarterly farm inventory change data also received heavy emphasis. Given

that the current methods used by USDA in estimating quarterly farm income

71
rely heavily on extrapolations of annual data and other historically
based procedures, the GNP Data Improvement Project concentrated their
recommendations on a shift away from the indirect data sources presently
used to more directly reported figures. For the most part these recom-
mendations paralleled those of the 1976 Task Force (this latter group in
their report noted that their ideas were supplemented greatly through the
help of the GNP Data Improvement Project). The 1975 Task Force placed a
higher emphasis on the data problems associated with the January prelimi-
nary farm income estimates as well as the subsequent July revisions but
their recommendations were still reasonably congruent with the GNP project.
The GNP project's recommendations also focused somewhat more on the Bureau
of Economic Analysis data on gross farm product and income originating in
farming than on the USDA's farm income estimates. This is not as serious
a difference as it appears, since most of their recommendations also in-
fluence USDA in that USDA is the main data provider for BEA regarding
quarterly farm income.

One of the spinoff benefits of the GNP project's work is that they
documented the procedures currently used by USDA in making the quarterly
estimates of farm income. Most of the methodology that USDA uses in esti-
mating farm income is only published for the annual July estimates with
little information on how quarterly or preliminary estimates are made.
Hence this study of the GNP Data Improvement Project is an important con—

‘ tribution in this area.

3.2.5. GUEBERT
In a paper presented at the WorkshOp on Farm Sector Financial

Accounts in April 1977, Steven Guebert reviewed some of the data and

72

concepts used in the farm income accounts as well as identifying some of
the major users and uses of the farm income data. Guebert notes that the
farm income data enter into the deliberations on national level farm poli-
cy and he considers this one of the major uses of the data. There is also
a program implementation use for the data in that the state estimates of
farm income enter into the revenue sharing allocation mechanisms. At the
state level he suggests that government budget planning is an important
use. Guebert also noted that private sector uses of the aggregate farm
income data lie in three major areas. First, U.S. farm income data are
used in advertising and investment decisions in what can be called the
farm input sector, e.g., machinery manufacturers, fertilizer supplying
firms as well as those firms which supply inputs to these firms. Second,
the local and national political interest groups concerned with agricul-
ture use the data. Third, the banking and financing industry is a user
of farm income data. Guebert then makes the statement that the primary
interest of the private sector appears to be more with the overall eco-
nomic activity or total dollar turnover for specific commodities or ex-
pense area rather than net income per se. For the most part the percep-
tions of the author were not substantiated by any study of users and uses.

In evaluating the data base used in estimating farm income,
Guebert considers the measurement of quantities of inputs as the major
data constraint to better farm income estimates. The second data base
problem is in the area of timeliness of the marketing data for cash re-
ceipts estimates. A third area of difficulty is in the determination of
farmer owned inventories and inventory change. In this case the problem
centers on substituting the value of the change in on-farm stocks for the

value of the change in farmer owned stocks. The former neglects farmer

73
owned inventories held in elevators or other facilities off the farm. In
addition to the conceptual problem concerning the inclusion of breeding
livestock as part of current inventory Guebert notes that a further data
problem arises from measuring inventory change as the value of the change
in the number of head of livestock. For current production this method
neglects the change in weight of the national livestock herd. Guebert
appears to overstate his case in this area since it is the value of the
inventory change that is used in the national accounts not the physical
change. So as long as the weight of the animals is discounted in the
pricing system, measuring value on a per head basis would not seem to be
a significant problem. Also as long as beef breeding and dairy livestock
are included in the inventory account, valuing inventory change on a per
head basis is probably not that serious a problem.

Guebert also raises some questions as to the apparent difficulty
of reconciling the USDA farm income accounts to national income and pro-
duct accounts of the Department of Commerce. This was a major concern of
the two Task Forces on farm income and the GNP Data Improvement Project.
Guebert argues that the problem is not one of estimation but that dif-
ferences only arise in the manner of presentation. He goes further to
state that the concepts used by USDA in the farm income accounts fits
more closely the data needs of the Department of Commerce than the data
from any other sector.

Guebert's analysis while not as detailed as some of the earlier
projects and Task Force reports raises a number of important issues. By
identifying user and uses the seeds are sown for a better understanding
of the needs and design of the system. Given the nature of his study,

Guebert cannot really be blamed for not making the explicit connection

74

between uses and redesign of farm income data system. His approach at
least reflects a concern for these issues. Guebert's findings also raise
a question as to the significance of the alleged problems associated with
bringing the farm income accounts more in line with the national income
and product accounts. If the USDA data fit the needs of the Department
of Commerce as well as any other sector then improving the data for the
national income and product accounts should probably be given a lower

priority than was suggested in some of the earlier studies.

3.2.6 AGRICULTURAL AND RURAL DATA WORKSHOP

Lee Bawden, £3, 31,, presented a paper at the AAEA-USDA Agricul-
tural and Rural Data WorkshOp which analyzed the USDA farm income data
from a slightly different perspective than some of the earlier studies.
Their main concern was with measuring the well being of farm operator
families so they concentrated on the areas of personal income and wealth
of farm people. Their findings were that the existing farm incOme and
wealth data were inadequate to measure the well being of farm pe0ple.
The differences here come about primarily because of problems in defining
the farm papulation and in the treatment of capital gains. The concern
of Bawden, 33, 21,, with farm families as opposed to farm businesses is
somewhat different than the studies summarized earlier. They state that
the agricultural community seems to prefer a farm operator family concept
as the appropriate unit of observation but it is unclear on what grounds
this statement is based.

One aspect that they point out but do not develop to any extent
is the need for distributional data on farm income and wealth in order to

understand the well being of farmers and to measure the impact of

75
government policy. Some types of distributions suggested include those
by size and type of farm, time spent farming, tenure, age, net income,
education and geographic region. They note that the current series on
the personal income of the farm papulation does not allow one to evaluate
the impact of government farm policies because the "farm population" in-
cludes farm residents whose only ties to the sector are as a place to
live and excludes farm operators who happen to live off the farm. Bawden,
g1, 31,, suggest that distributional data which allow one to examine the
impact of policies are needed to understand how the change in income
brought on by policy changes affects agriculture.

As with the earlier studies, this group seems implicitly to assume
an unlimited budget in that they accept the recommendations of the earlier
Task Forces and recommend that their findings be implemented in addition
to the earlier recommendations. They fail to identify the users and uses
of this data so that some notion of the priorities for these improvements
can be established nor do they give any other criteria for determing the

importance of their suggested changes.

3.3 ALTERNATIVE CONCEPTUALIZATIONS
Many of the studies summarized in the previous section suggested

alternative accounting systems for farm income. This section will pre-
sent the concepts of the farming sector currently used by USDA and some
alternative conceptualizations of income and economic well being put forth
in the literature. These alternative conceptualizations provide the basis
for some of the recommendations of the final chapter of this study in that
the results help to establish priorities in choosing between these con-

cepts.

76
3.3.1 NATIONAL FAMILY FARM CONCEPT

The current farm sector accounting system used by the U.S. Depart-
ment of Agriculture treats the sector as if it were a single national
family farm. Thus, the measure of income derived using this concept in-
cludes the value of farm products consumed in farm households, government
payments or transfers to farmers, and the imputed value of farm dwellings
in addition to the cash receipts from the sales of farm products. This
concept also implies that net returns to nonoperator landlords be excluded
from farm income so this is included on the production eXpense side. The
other expenses include depreciation, taxes, and interest on farm mortgages
in addition to the current operating expenses. So in effect, this con-
cept of the farming sector produces an accounting system similar to that
used by a typical farm operator for Federal income tax purposes (Carlin
and Smith, p. 2).

This tax accounting analogy can be carried somewhat further in
that the farm income accounts under this concept are on a cash accounting
basis as are most individual farm tax returns rather than an accrual
basis. The inclusion of inventory change in the total net farm income
series is a departure from this cash accounting basis, but even this de-
parture is not complete since inventories of purchased inputs are not
deducted from the measured operating expenses.

In estimating income using this national family farm concept, as
with any concept of the sector, it is critical that the production ex-
penses and income sides be comparable. Expenses must only be associated
with the income generating activities included on the gross income side,
e.g., if government payments to farmers are included on the income side

then costs to farmers associated with program compliance should be

77
included on the expense side or if own account production of capital
items is excluded from the income side then the associated production

expenses should be excluded (Weeks, 1972).

3.2.2 ESTABLISHMENT CONCEPT

This characterization of the farm sector views it as one made
up of farm establishments. An industry, such as farming, is made up of
those establishments for which more than half of their production is of
the commodities characteristic to the industry. Establishments for
which the production of agricultural commodities is their major activi-
ty would be considered farm establishments.

Thomas Carlin and Allen Smith provide a more general definition
of an establishment as it is used in the national accounts and other
Federal statistics.

"An establishment is defined as an economic unit, generally

at a single physical location, where business is conducted

or where services or industrial operations are performed.

Nhere distinct and separate economic activities are performed

at a single physical location, each activity should be

-treated as a separate establishment wherever: 1) no one in-

dustry description in the classification includes such com-

bined activities, 2) the emplo ent in each such economic
activity is significant, and 3 reports can be prepared on

the number of employees, their wages and salaries, sales or

receipts, and other establishment type data. Establishments

are the basic unit of account. Firms would be composed of

one or more establishments." (p. 4)

In many ways this establishment view of the farm sector is si-
milar to the national family farm concept, since most family farms are
probably each a separate establishment. However, this specific defini-
tion of an establishment does not coincide on a one to one basis with
the definitions used in arriving at the existing farm income accounts.

The establishment view of the farm sector makes it clear that a fanm

78
can use its resources to produce both characteristic farm products,
(e.g., cattle, corn) and ancillary products which are nonagricultural
(e.g., trucking). Thus, in measuring income using this concept one
would wish to determine all outputs of farms including those minor out-
puts not normally considered as farm output in addition to the associ-
ated production expenses for both the agricultural and nonagricultural
products. This view of the farming sector would be more useful in an-
swering questions of decision makers concerning the performance and eco-
nomic behavior of farms, farm business-household relationships, the
structure and control of the industry, and would make comparisons with
other industries easier because data for these are also estimated on an
establishment basis (Weeks, gt,.31., 1972). As was noted earlier these
comparisons may be difficult because of the heterogeneity in the Opera-

tionalization of the establishment concept in other industries.

3.2.3 PRODUCT CONCEPT

Another characterization of the farm sector suggested by the 1972
Farm Income and Capital Accounting Task Force is a product concept. This
concept implicitly recognizes that agricultural commodities are in some
cases produced by establishments which do not produce characteristic farm
products as a major activity. Thus, if one is interested in the output
of all agricultural commodities then a product concept is needed to in-
sure the consideration of farm products produced by both farm and non-
farm establishments.

On the output side of the production account only the output of
characteristic agricultural products would be measured. But at the

same time only those inputs associated with the production of these

79
characteristic products would be deducted in arriving at a net income
estimate. This concept then would be particularly useful in assessing.
the relationships between the total output of food and fiber commodities
with the inputs used to produce them. Information concerning the pro-
ductivity of the sector can also be derived more readily from data
based on a product concept when compared to data with alternative con-
ceptualizations. Other questions relating to aggregate supply of farm
products are in most instances better approached with data organized on
a product basis than on an establishment basis.

The existing farm income accounts rely quite heavily on data
which are closely related to this product concept, especially on the
output side. Cash receipts for farm marketings are for all farm products
and do not exclude production from non-farm establishments. However, on
the input side the existing methodology uses data more closely akin to
data collected on an establishment basis since it is impossible to sepa-
‘rate out those production expenses used in the production of ancillary
products on farms. Thus, the empirical problem associated with using
these alternative conceptualizations would be difficult to overcome un-

less the existing data base were greatly modified.

3.3.4 VALUE ADDED CONCEPTS

The production or value added approach to measuring income and
product by industry is perhaps the most basic approach to national ac-
counting in that it considers aggregate production as the sum of the pro-
duction statements of the producing units in the economy (Kendrick, p.
39). Conceptually, value added can be estimated in two ways for a par-

ticular sector. Using what is called a production approach, value added

80

for a sector is determined by taking the value of total production in
the sector and deducting from this the cost of the intermediate products
used in the production of the sector's current output. The second ap-
proach is called the income approach. Value added for a sector, using
the income approach, is estimated by summing the factor and nonfactor
costs used in production. Factor costs include such items as labor com-
pensation, net interest, and corporate profits while nonfactor charges
might include depreciation and indirect business taxes.

Value added for a sector should be identical no matter which ap-
proach is used. The production approach is preferred when one wishes to
deflate industry output to obtain a measure of real output. This ap-
proach also provides a beginning point for the development of the input-
output matrix for the economy. The income approach provides a statisti-
cal check and allows for the analysis of factor shares by industry in
addition to making it possible to deflate to obtain real factor cost
(Kendrick, p. 40).

For agriculture the value added concept is the desired concept
'of the Department of Commerce which they use in their estimates of farm

output, gross product, and income published in the Survey of Current

 

Business. However, in practice the value added concepts are modified
somewhat by the Commerce Department in estimating the farm income and
product accounts.

Theoretically, at least, one should begin estimating the value
added in farming using the production approach by estimating the value of
production. Assuming that all production is sold, the value of produc-
tion would equal the value of sales plus inventory change. But in agri-
culture many farm products are not sold, instead these are used as in-
termediate products on the same farm where produced. For example, feed

grains are often fed to cattle on the same farm where produced. The

81

estimation of value of production as the value of sales plus inventory
change is the basis for the existing U.S. Department of Commerce esti-
mates of farm product. This underestimates the value of production by
the amount of unsold or own account production in the sector. Simunek
‘notes that the value of sales is apprOpriate for cash income analysis,
but also points out that a measure of value of production which includes
own account production is necessary to avoid distortions in productivity
analysis, input—output studies, size and type of farm classifications,
and estimates of total capital formation. Thus, it appears Simunek is
arguing that while the value added concept might be appropriate for cer-
tain uses, the current definition of farm output is an inadequate opera-
tionalization of this concept because it neglects own account uses.

Gross farm product is used by the Bureau of Economic Analysis
(BEA) in the U.S. Department of Commerce as the estimate of value added
in farming. The current measure using the production approach starts
with farm output. As noted above, farm output is not really the total
value of production in farming; rather it is the sum of cash receipts
from farm marketings and CCC loans, other farm income, farm products con-
sumed on farms, change in farm inventories, and gross rental value of
farm dwellings. From farm output the value of intermediate goods and
services consumed is deducted. This latter item includes net rent paid
to nonoperator landlords as part of the intermediate goods and service.
Gross farm product is the measure of value added in farming since in
theory at least it is an attempt to measure total value of production
less intermediate purchases. It should be noted that BEA also adjusts
the gross farm product figure by adding adjustments for such items as
wage supplements, social security contributions from wages and salaries,

interest received and Federal fines.

82

Using the income approach, gross farm product is the sum of the
factor and nonfactor costs, i.e., net interest, corporate profits, pro-
prietor's income, employee compensation, indirect business taxes, cap-
ital consumption less direct government subsidies to operator landlords.
This differs from the USDA definition of net farm income which includes
proprietor's net income, corporate profits and direct government sub-
sidies (GNP Data Improvement Project).

National income originating in farming is also derived in the
BEA estimates. For the economy as a whole, national income is derived
in theory by subtracting depreciation and indirect business taxes from
gross national product (Schultze, p. 45). National income originating
in farming is derived in a similar manner. Gross farm product less
capital consumption allowances and indirect business taxes plus subsi-
dies to operator landlords equals national income originating in farm-
ing using the existing BEA accounts. National income originating in
farming can also be estimated by summing net interest, corporate pro-
fits, proprietors' income and employee compensation.

The 1975 Task Force on Farm Income recommended that USDA use the
value added approach in estimating and presenting data on farm income.
They noted that the gross farm product and income originating in farm-
ing could be approximatedby merely changing the accounting rules now
used in estimating farm income so that a change in the accounting sys-
tem is not necessary. Since they also recommend an accounting for all
output, their recommendations would require collecting new data on own
account uses of production and on some interfarm sales which are now
excluded. Therefore, changing the accounting rules would in effect

require more primary data than is now available.

83
3.3.5 TAXFILERS INCOME CONCEPT

Periodically the Internal Revenue Service (IRS) publishes data on
income obtained from samples of individual and business income tax returns.
The concept of income used by IRS is sufficiently different from the USDA
income concept so that the two concepts are not readily comparable.

Edward Reinsel has noted several differences between the farm in-
come concept of IRS and USDA. On the gross receipts side he found that
the IRS and USDA estimates were reconciled fairly easily. The IRS figures
include intrastate livestock sales to other farmers which are netted out
of the USDA estimates. The IRS estimates exclude some sales of breeding
livestock which are treated as capital assets, however USDA includes these
as receipts for livestock. In the IRS data a significant amount of farm
income is not measured because crop share tenants report only their own
share of the farm receipts while the landlord might report income from
the farm operation as rent rather than farm receipts. Since the prices
(used to estimate farm income by USDA include normal marketing charges, the
gross receipts data of USDA would then be higher than the IRS estimates

of gross receipts,'ceteris paribus, in that receipts are reported to IRS

 

without these marketing charges. Since the USDA subtracts out these
charges on the production expense side, the net income estimates should
not differ dramatically because of this. IRS data also exclude some
receipts from corporations that are primarily non-farm businesses. How-
ever, IRS data also include some corporate receipts from foreign countries.
The sum total of these conceptual and definitional differences on the
gross receipts side appear offsetting since gross receipts on tax returns
and from USDA estimates were about the same in 1962 as noted by Reinsel.

Data for 1974 suggests that this same relationship holds.

84
When comparing USDA net farm income with IRS net farm profits

there are large differences in the reported data. Since the gross income
estimates are basically the same, this suggests that the major conceptual
differences exist in expense accounting between the two data suppliers.
For instance, depreciation in the USDA accounts is estimated at replace-
ment cost while IRS data can be based on the book value of the assets and
IRS considers gross rent as an expense while USDA only uses net rent to
nonoperator landlords as an expense because of USDA's desire to measure

income based on a national family farm concept.

3.3.6 COMBINING INCOME AND WEALTH

Income is often used by economists and public decision makers as
a proxy measure of well being in society. Burton Weisbrod and Lee Hansen
have developed an approach that, while not a perfect measure of current
welfare, attempts to go beyond traditional income measures as indicators
of well being. Their suggested measure is a combination of current in-
come plus current net worth. Net worth is converted to an annuity to
overcome the difficulties of combining a stock concept like net worth with
a flow concept such as income. By adding the annual lifetime annuity va-
lue of current net worth to current income Weisbrod and Hansen develop a
measure of welfare that can be used to assess the economic position of
various sectors of society.

Thomas Carlin and Edward Reinsel used the methodology developed
by Weisbrod and Hansen to compare the economic position of farmers with
other sectors and the economic well being among different units within
the farm sector. Their results show that when wealth is added to income,

the distribution of well being in the farm sector became more equal than

85

when income alone was considered. Weisbrod and Hansen's findings for the
entire U.S. economy showed well being to be more unequally distributed
when wealth is added. 50 the apparent disparity in welfare between the
farm and non-farm sectors when income alone is considered is less pro-
nounced when wealth is also considered.

This points out some of the difficulties in using farm income as
a measure of well being for the farming sector. Not only does it neglect
the distribution of income within the sector but important wealth effects

fail to receive proper attention.

3.3.7 PARITY RETURNS CONCEPT

The concept of parity returns was first put forth in a 1967 USDA
report to the U.S. Congress concerning parity income, in reSponse to a
mandate for this type of study set out in the Food and Agriculture Act of
1965. The concept of parity returns was used in place Of the parity in-
come concept laid out in legislation in the 19305 and 40s. Parity income
is defined in the 1936 and 1938 agriculture legislation and is in terms
of the historical ratio, using a 1910-1914 base, between per capita in-
come of the farm and non-farm population. The 1948 definition of parity
income eliminates the fixed base period and attempts to establish an equi-
valent standard of living between the farm and non-farm sectors using a
moving average of the 10 preceding years. These parity income concepts
have some obvious limitations such as reliance on a fixed base, the ex-
clusion of off farm income of farm operators, and the reliance on averages
of all farmers which neglects important distributional effects by size
and type of farm.

To overcome some of the disadvantages of the parity income

86
concept the USDA report developed an income concept called parity returns.
Parity returns are defined as "income required to make the current rate of
return to the labor, capital and management employed in farm production
equal to the current rate of return to comparable resources employed in
other sectors of the economy." (USDA, 1967, p. 9) In operationalizing
this concept both the income and changes in net worth of farmers are com-
pared to the opportunity cost in nonagriculture uses of the resources
currently used by farmers. The approach also took into account distribu-
tional differences by size and type of farm and did not rely solely on
averages as did earlier parity income measures.

Parity returns standards were developed for capital and labor in
farming which in theory reflects the returns to farm capital and labor in
comparable nonagricultural uses. So in effect these returns exclude off
farm income of farmers. Another limitation of the concept is that the in-
clusion of capital gains might not be appropriate for short run compari-
sons when farm capital would tend to be illiquid and thus spending from
net worth to increase well being might not be an available option.

Two aspects of the parity returns study are important in the con-
text of this analysis. First, the idea of separating out returns to vari-
ous resources used in farming provides useful information for agricultural
policy decisions, e.g., in choosing between programs for improving agri-
cultural credit and programs aimed at giving tax relief to farmers.
Secondly, this study showed the importance of considering the income and
returns position of farmers by size and type of farm. The finding of the
USDA on parity returns showed that farmers with gross sales over $20,000
were earning parity returns or more in 1966 under each alternative method

of calculation used while farmers under $20,000 gross sales earned less

87
than parity returns under all methods. The smallest farmers (under $5,000
gross sales) earned only one third to two fifths of parity returns in 1966
(USDA 1976, p. IV). 50 the average income figures even for a relatively
good year like 1966 fail to show the important differences in income with-

in the sector.

3.4 SUMMARY

Six specific analyses of farm income data were summarized in the
first major section of this chapter. Each of these point out some of the
important shortcomings of the existing farm income data system but all
differ in approach from this study. These earlier evaluations often be-
gin to question the conceptual relevance of the USDA farm income data and
all make recommendations for improvements in the data. However in all
but one case they fail to identify the users and uses of the farm income
data. For the most part assumptions are implicitly or explicitly made
about the uses of the data without any attempt to substantiate the con-
gruence of these assumptions with reality. This failure to consider the
uses of the data in decision making makes it difficult to place any
weights on their recommendations or to establish priorities for improve-
ments. The consideration of users and uses is not the only means of set-
ting priorities but_the literature summarized provided few other criteria
for determining the significance of the recommended improvements.

These studies of farm income as well as other research provide a
set of alternative conceptualizations of farm income or well being which
are also summarized in this chapter. These include: 1) the national
family farm concept, 2) an establishment concept of the farming sector,

3) a product concept of farming, 4) value added concepts of sector income,

88
5) taxfilers income, 6) a concept combining income and wealth, 7) a con-
cept of parity returns. The results presented in later chapters will be
used to show the strengths and weaknesses of these alternative concepts
as bases for the information to be used in decision making concerning

farm income.

CHAPTER 4
DESCRIPTIVE RESULTS: THE FARM INCOME
INFORMATION SYSTEM

4.1 INTRODUCTION

This chapter presents a description of the existing farm income
information system. A complete description of this information system
does not now exist anywhere in the literature. The information systems
paradigm developed in Chapter 2 provides a convenient framework for under-
standing and analyzing the farm income information system and as such will
be the basis for organizing the overview and summary presentations in this
chapter. A descriptive examination of the users and uses of the USDA will
also be presented. This latter section is based on the results of a mail
questionnaire sent to a random sample drawn from the mailing list of the

USDA statistical bulletin Farm Income Statistics with foreign addresses

 

and libraries eliminated. In addition some of the more detailed observa-
tions with regard to data uses by some public and private users are de-
veloped from personal interviews. Thirty-five of these interviews were
with peOple in different roles in all the organizations that normally par-
ticipate in public policy decisions for agriculture. Ten others were

follow up interviews with individuals on the Farm Income Statistics mail-

 

ing list. Finally the relevant parts of the primary data used in the
USDA farm income estimates will also be described based on written docu-
mentation and personal interviews with the providers of the data.

89

90

4.2 OVERVIEW

It is difficult to separate a discussion of the uses of farm in-
come data from a description of the farm income information system. In
describing the components of the system it is necessary to start by con-
sidering the ultimate decisions that must be made and work back to the
data sources used to estimate farm income. The farm income information
system as perceived in this research has four basic components: a pri-
mary data subsystem, a formatting and communication subsystem, an analy-
sis subsystem and finally a decision making subsystem. Figure 4-1 sche-
matically outlines these components. The distinctions between these parts
are not as clear as the titles would imply but this categorization does
allow one to focus on the most important interralationships in the system.

When comparing the existing farm income information system with
the idealized system outlined in Chapter 2, one finds many similarities.
This paradigm pinpoints five basic steps or actions that make up an infor-
mation system, beginning with theoretical concepts and ending in decision
making. The farm income information system begins with a theoretical con-
cept of farm income. The basis of this concept is developed in section
3.3.1 and revolves around what is called the national family farm concept.
The operationalization of this concept of farm income involves defining a
set of accounting rules or relationships which relate various price and
quantity data, yielding farm income as a residual. Thus, the Operation-
alization phase actually involves the conceptualization, operationaliza-
tion and measurement of the different sets of primary data needed. The
primary data underlying the farm income series are provided by a number
of government and private sources, such as the Statistics Branch of the

Economics, Statistics, and Cooperatives Service (ESCS) in USDA, the

91

AN IDEALIZED AGRICULTURAL COMPONENTS OF THE USDA FARM
INFORMATION SYSTEM INCOME INFORMATION SYSTEM

Decision Making Decision Making

Subsystem

   

 

 

I

Information for Decision Makers Farm Income Information for
Decision Makers

 

I

 

 

 

[ Interpretation and Analysis I l Analysis Subsystem ]

i % i
W /A W/ %

 

 

 

 

 

 

 

 

 

 

 

MeQSUrement I Comunicatl' on and
Formatting Subsystem
.L
Measurement
7 Operationalization Of PrImary Data
of the Operationalization
Operationalization of Concepts Farm Income of Primary Data
Concept Concepts

 

Primary Data Concepts
Primary Data Subsystem

 

 

 

 

 

 

 

 

A
[ Theoretical Concepts I I Concept of Farm Income I

 

Reality

 

FIGURE 4-1 COMPARISON OF AN IDEALIZED AND OPERATING INFORMATION SYSTEM

92
Bureau of Census in the U.S. Department of Commerce, the Farm Equipment
Institute, and the Market News Branch of the Agricultural Marketing Ser-
vice in USDA. These, among other, government and private data sources
make up what can be called the primary data subsystem.

Since'farm income is calculated as a residual, the measurement
phase of the idealized information system becomes more of a formatting
process when applied to the farm income data system. In cases where de-
cision making is not tied organizationally to the data system the communi-
cation between those who produce the data and the ultimate decision makers
is important. With respect to the farm income infonnation system these
formatting and communication functions are performed primarily by the
Farm Income Unit in ESCS and to a lesser extent by the Bureau of Economic
Analysis (BEA) in the U.S. Department of Commerce. In addition to these
government agencies, a communication function is also performed by edu-
cational institutions, private publishers of agricultural and trade maga-
zines and other government agencies. These organizations constitute what
can be called the formatting and communication subsystem of the farm in-
come information system.

The analysis and decision making phases of the farm income informa-
tion system are at least conceptually the same as in the idealized system.
However, the diverse nature and number of decision makers and analysts
dictates that these phases be viewed as subsystems rather than merely
steps in a process. The analysts include government agencies, private
consultants, educational institutions and the staffs of the primary deci-
sion makers intereSted in farm income. Decision makers are in both the
public and private sectors and make decisions with regard to public poli-

cy, demand estimation, credit needs and other areas.

93

4.3 USERS AND USES

Government data and information systems often begin with one type
of ultimate use or decision as a primary focus. However, even in those
cases where data are collected strictly for administrative purposes, mul-
tiple uses for government provided data usually develOp despite the ori-
ginal intent and design of the data. Farm income data are no exception
to this. Through time a number of different uses of this data have de-
veloped both in the public and private sectors. This section will attempt
'to identify the various user groups associated with the farm income data

and to categorize the various uses of the data.

4.3.1 LIMITATIONS 0F MAIL SURVEY POPULATION

Before proceeding into a discussion of the results of the ques-
tionnaire a few statements on the survey design would seem appropriate._
This should reveal some of the limitations of the results with regard to
identification of the relevant p0pulation.

Farm income information like other types of information possesses
attributes of high fixed costs of collection relative to the costs of
transmission of the data to additional users. Therefore, one would ex-
pect many individuals to obtain the farm income data from sources other
than the original supplier, in this case from sources other than the bul-
letin Farm IncOme StatiStics each year from the U.S. Department of Agri-
culture. Thus the population by definition excludes those users of USDA
farm income data who receive it from other sources. Even within the De-
partment of Agriculture these data are supplied to other individuals
through the Agricultural Outlook publication, various press releases, and

other sources such as the annual volume, Agricultural Statistics, and

 

94
extension service reports. In the private sector sources such as maga-
zines, newspapers, trade publications, radio, television, libraries, and
private consultants report USDA farm income statistics.

The existence of multiple sources of USDA farm income data creates
some difficulties in using the results of a sample drawn from the mailing
list provided by USDA. Very little can be said about the overall magni-
tude of the use of farm income based on this survey since many users are
probably excluded from the population. However, as long as those who are
on the mailing list are fairly representative of the entire population of
users, one can use the survey to gain an understanding of the various
types of users and uses of the farm income data. If the assumption is
made that the total number of different types of users is directly propor-
tional to the number of the types of users found on the mailing list then
one can begin to obtain an indication of the relative importance between
different users and uses when measured in terms of absolute frequency of
use.

A comparison of numbers of users and uses based on the mail sam-
ple does not take into account the importance in terms of value or social
welfare, etc., of any one type of use but instead treats all users and/or
uses as of the same importance. This is an inherent difficulty in using
random sampling; one must assume that those who are not selected are of
equal importance to those selected in some respects.

The relationship among economic structure, income distribution
and the supply and demand for information developed in the conceptual
framework for this study suggests that in some cases different types of
users and uses should receive higher priority than others in the design

of a_government information system. Others can undoubtably develoo

95

different criteria for determining the importance of various users and
uses. Implied by this is that frequency of use is not necessarily a good
proxy for the importance of a given use. So a mail survey also has some
shortcomings in this respect.

These problems in defining the p0pulation and in sampling begin
to point out some of the limitations of this methodology for evaluating
the farm income information system. However, it is not unreasonable to
assume that those individuals who are the principal users of fanm income
data are likely to be on the mailing list, especially those users outside
of the Federal government, since the most detailed data on farm income
are not readily available from the other sources. This assumption would
not appear to hold in all cases within the Federal government where inter-
nal lines of communication would probably transmit much of this informa-
tion and direct distribution rather than mailing is more likely. Further-
more, users of detailed data are more likely to understand the concepts
and definitions and to be able to respond intelligently to alternatives.

Casual users tend not to know the limits of the data.

4.3.2 SELECTION OF SAMPLE

The total mailing list contained 4224 names and addresses, of
these 268 were foreign addresses, and 290 were libraries or duplicates
leaving a pOpulation of 3666 to be sampled. Libraries were removed from
the population because the ultimate concern of this study was to obtain
information about the use of the data in decision making. It was felt
that libraries were not ultimate users of the data and therefore could
be left out of the survey. Foreign organizations or individuals were also

excluded since it was felt that input from these users was not relevant

96
for questions concerning the apprOpriate design of a United States farm
income information system funded at least, in part through U.S. tax dol-
lars.

In selecting the apprOpriate sample size for the mail survey no
clear cut method could be used without making some assumptions because of
the multi-purpose nature of the questionnaire. As Moser and Kalton (p.
149) note there is "no perfect solution" to this difficulty. Thus, for
this study it was assumed that the determination of the users and uses
of the farm income data was the most important objective of the survey.
From this an arbitrary decision concerning significant users and/or levels
of use was made so a sample size could be ascertained. A priori, fourteen
different user groups were tentatively identified. If all groups were
equally represented in the pOpulation then 1/14 of the population would
be in each group. This arbitrary proportion of 1/14 or .07 was used to
provide a basis for selecting the sample size.’ A 1 to 2 percent error in
this was deemed acceptable. Using standard methodsl/ this decision then
led to a selection of a sample size of 268. A 50 percent response rate
was also assumed so the actual sample size was doubled and then rounded
up to 545. These 545 names and addresses were selected at random from

the USDA mailing list.

1/ Where n is sample size, p is proportion of p0pulation in a user group
and S.E. is the standard error, then without the finite p0pulation cor-

rection
’ _ Egl-p).
" ' S.E.
For this study p=.O7 and S.E.=.015, so n=290. Using the finite population

correction of n' = _B_.where N is the population size, 3666, n' is 268.
1+9

97
4.3.3 MAIL SURVEY RESULTS
Of the 545 samples, some form of response was obtained from 307
individuals. These responses are characterized in part in Table 4-1.
A total of 270 questionnaires had responses to the questions concerning
the use of farm income data. These provide the primary basis for the

descriptive results which follow.

TABLE 4-1 MAIL SURVEY CHARACTERISTICS

 

Total USDA Mailing List 4224

Foreign Addresses 268
Libraries and Duplicates 290

Population for Survey 3666
Sample Size 645
Total Responses 308
Problem Responses:

Bad Address 2
Library

Foreign Institution

Refused to Answer

\OWOON

Usable Responses 270
Usable Sample* 517
Response Rate} 52.22%

 

*Sample size less bad addresses, libraries, and foreign institutions.

The information systems paradigm used in this research provides a
a guide for analyzing the farm income information system. The impor-
tance of decision making in this paradigm stresses the need to identi-
fy users and uses of information in order to evaluate the system.

Table 4-2 presents data on the frequencies of different types of

users obtained from the mail survey. The fourteen different user groups

98
identified prior to sampling seemed to be an adequate categorization of
the users since less than 2 percent of the respondents and users were
placed in the miscellaneous category. Perhaps the most significant fact
to be gleaned from Table 4-2 is the comparison of respondents to users.
While there were 270 usable reSponses to the mail survey, only 205 of
these respondents actually made use of the farm income data provided by
USDA. In three categories the rate of nonuse of the respondents seems
worth noting. Only 53 percent of the farmers or ranchers who responded
to the survey actually used the data. For commodity trading firms and
farm product processing firms the rates of use among respondents were 50
and 43 percent, respectively. On the other side of the coin, all 23 farm
input supply firms reported using the data. The rate of data use of re-
spondents from most other major user catergories varied between 70 and 90
percent.

In terms of the number of individuals in a user group, four major
users of USDA farm income data are apparent. In order of number of users
reported from the sample, educational institutions, the Federal government,
farm input supplying firms, and banking and financing firms are the most
prevalent users, each with greater than 10 percent of the sample. Educa-
tional institutions made up 26.3 percent of the reported users of the
farm income data. The relative importance of educational institutions as
users is somewhat surprising but included in this total are state and
county extension service personnel which accounted for 35 percent of the
uSers within the educational institution category. If these extension
service personnel were included as part of state and local government
then this category would contain over 15 percent of the users. However,

it was felt that the extension personnel, especially those at the state

99

 

 

 

 

 

 

 

 

 

~.eN mo o.ooH mON o.oo~ oNN 4<pop
o.ooH o o o N.N c umucoamc aaogm com: oz
o.o¢ N m.H m m.~ m maomcoppoum_z

o.om m ¢.e m N.c NH as?»
mmmcmxoca so m:_umca xuwvoeeou
m.mm H o.~ N H.” m Neeeeeu sue__b=
m.m~ NH m.mN em ~.¢N mm coruauwpmcw pmcowumuscm

m.¢~ N a.m NN N.m «a :ovumumpnag
segue co sawums mzmz
m.mN m m.m efi o.N mH at?» mcwupzmcou mpm>wca
H.3H a m.m m m.m m asocm umdedue_ Fecsspsuwca<
o o o.N e m.~ w .m>wpmcmaooo chaapzuwcm<
e.aN m m.m NH m.o NH newscgm>om Page, Lo mpeum
H.~N m m.e~ om H.eH mm newsccm>om poemuwu
o.NH m N.o~ NN m.m mN me_ueeepc Le mewxeam

~.Nm e m.~ m o.N N meP._em Le
\ucm mcpmmmuoca auscoca scam

o o N.HH mN m.m mN manger
scc$ merapqq=m\mcwcauumw=:mz
H.N¢ m ¢.¢ m m.m NH coumcmao gucmc co Scum

mpcmucogmmm Aucmugmav Mncmugmav
yo ucmucmm zucmacmcm zucmaamcu aucmaamgu ocmzamgm accmacocu
a m< muzpomn< m>wuapmm auspomn< m>pumpmm muapomn<
xgommumu
cam:
mgwmacoz mgmmz mucmvcoammm

 

>moamk<u «mm: >m maum: oz< mhzmozommum

~i¢.m4m<h

100
level, were really part of the educational system rather than the state
government policy making process. It should also be noted that the per-
centage of users to respondents in these four categories was higher than
the average for all categories which should also give some indication of
the usefulness of the farm income data to these types of users.

The information systems paradigm that undergirds this research
stresses the importance of the use of data and information in the design
of an information system. Thus, while information on the users of farm
income data is important for improving these data, an understanding of the
uses of these data is of equal or even greater importance.

Table 4-3 summarizes the response to the question, "What is your
primary use of USDA farm income data?" This was the first question on
the questionnaire.g/ A priori, all of the uses of the data were not known
so it was felt that an open ended question would best identify the uses of
the data. An additional question was asked about the types of decisions
on which farm income data have the most impact in order to clarify the
answers to the questions concerning the use of the data. From these an-
swers a list of uses was develOped and the responses were categorized
accordingly.

Seven major categories of uses of the data were determined from
the mail survey. Within four of these major use categories, subclasses
of uses were disaggregated where appropriate. The first major category
was that of general information or no specific use. In this case the re-

spondent could not identify any specific types of decisions that the

g1 See Appendix A for a copy of the questionnaire.

101

TABLE 4-3 PRIMARY USE BY CATEGORY

 

 

 

Respondents
Use Category
Absolute Relative Adjusted
Frequency Frequency Relative
Frequency
(Percent) (Percent)

General Information--

No specific use 41 15.2 20.0
Pass On to Others 40 14.8 19.5
Public Policy Uses

General policy uses 30 11.1 14.6

Program evaluation 10 3.7 4.9

Tax-revenue planning 3 .1 1.5

Allocation of

reasearch funds 7 2.6 3.4

Demand Estimation Uses

Production planning 6 2.2 2.9

Marketing-advertising 26 9.6 12.7

Within firm investments 3 1.1 1.5
Credit and Financing Uses

Future loan volume 6 2.2 2.9

Repayment ability 4 1.5 2.0

Investment in

agricultural firms 3 1.1 1.5

Land Valuation 10 3.7 4.9
Miscellaneous Uses

Futures market trading 9 3.3 4.4

Other 7 2.6 3.4
Total Users 205 75.9 100.0
Do Not Use 65 24.1
Total Respondents 270 100.0 100.0

 

 

 

 

102
farm income affected but the data were used by the receiver for maintain-
ing a general knowledge of the sector. Thus, the data in this case are
not likely to be used directly in specific decisions but instead to pro-
vide background information for any number of decisions. This category
contained 20 percent of the users of the data and 15.2 percent of the
respondents. As was noted earlier, 24.1 percent of the respondents did
not use the USDA fann income data so for each category of use the percen-
tage of respondents will be less than the percentage of users.

The second type of use is one of passing the data along to others.
This is not really a use of the data in the sense that it impacts on the
decisions of those who initially receive the data. However, it is impor-
tant to understand the communication function of the system. This cate-
gory contains 19.5 percent of the users who responded to the survey, again
stressing one of the limitations of this methodology in identifying all of
the ultimate decisions based on farm income information.

One would expect that an important use of aggregate farm income
data would be in public policy decision making; in the sample pOpulation
this seems to hold with 24.4 percent of the primary uses falling into the
area of public policy decisions. Within this category most respondents
identified only general policy uses, such as in policy research, but some
expressed uses of a more specific nature. Almost 5 percent of the total
uses were in the area of program evaluation; examples of these were pro-
jecting the farm income effects of various pesticide regulations and water
development projects. Among state government users who responded, 25 per-
cent reported the utilization of farm income data in estimating taxes and
revenues within the state. Only 1.5 percent of the total uses reported

were in this group but it appears to be important since one would expect

103
that the total population of users of this type to be at most 50. A final
specific policy use of the farm income data was as a guide in allocating
research funds. This seems to indicate that a goal of agricultural re-
search is to enhance farmer income. Government and university research
administrators were those most likely to use the data in this way but in
total only 3.4 percent of the primary uses of the farm income data fell
into this category.

Prior to the survey, use of the data by farm input supplying firms
was thought to be a major use. An implication of this is that farm income
is an important variable in estimating the demand for inputs and other
goods purchased by farmers. Three different types of decisions appear to
be dominant within the category of demand estimation. Farm income data
are important in marketing and advertising decisions of firms. For in-
stance one pesticide manufacturer used data on crop receipts by state to
decide if enough of the cr0p is grown in a state to justify expenditures
to register the pesticide in a state. A farm equipment manufacturer used
state cash receipt data and annual net income figures to allocate market-
ing resources between regions. Other firms use similar data in the de-
ployment of sales staff between regions. One firm in a reasonably oligop-
olistic industry even mentioned that it used farm income data in deciding
on the price of the farm inputs that it sold. These marketing and adver-
tising type decisions comprise almost 13 percent of the total uses re-
ported.

A second type of demand estimation use is in production planning;
this made up about 3 percent of the reported uses. These uses are of
course closely related to the marketing uses but some respondents speci-

fically mentioned such uses as production scheduling, new product

104

development, or inventory management. A final use related to these is in
the long run planning of firms supplying the farm sector. Capital expen-
ditures within the firm and long range planning were cited as decisions
for which the data were used. Thus, it would look as if for a small num-
ber of reported uses, about 1.5 percent, farm income data have an impact
on estimates of the long run demand for farm inputs. However, the econo-
mic impact of this use would tend to be understated by considering only
the frequency of the use. The value of the information might be much
higher because the decision involves relatively significant expenditures.

Credit and financing uses were a fifth major category of use of
farm income data. Within this category the data were used in estimating
future loan volume, in determining the repayment ability of farmers on
loans or inputs sold on credit, and as a guide in evaluating the future
profitability of investments in agriculturally-related firms. In total
these credit and financing uses accounted for 6.4 percent of the total
primary use of those sampled. Again the disclaimer must be added that
the frequency of use is not a good proxy for the economic impact of the
decision and hence, the value of the information.

Almost 5 percent of the primary uses of this data was in valuing
farm land. One firm used the data to compare farm incomes in various
parts of the U.S. to farm land costs in order to ascertain where the best
agricultural investment values are located. Another was attempting to
use the data in an econometric model to forecast farm land values. In
government, a senior appraiser for the Army Corps of Engineers noted that
the farm income data were used as check data to compare with locally col-
lected data on income used in determining land values.

The final major category of use was a miscellaneous category.

105
Within this class the most readily identifiable subgroup was the use of
farm income data in commodity futures trading. In this case it seems that
income and receipts data are used to get an indication of planting inten-
tions for certain crops which in turn cause fluctuations in the price of
futures contracts. Over half of the miscellaneous uses reported were in
the area of commodity trading while the remaining miscellaneous uses were
in such areas as bargaining in cooperative contract negotiations, location
of commodity processing plants, and in making wage adjustments for farm
employees.

When the term multiple use data is mentioned it is usually thought
of in terms of multiple users of the data. However, there are also multi-
ple uses by the same user. Nearly one third of the users of farm income
data in the mail survey reported a secondary use of the data in addition
to a primary use. Furthermore, almost 5 percent of the total users re-
ported a third use of the data as well. Table 4-4 summarizes these secon-
dary and tertiary uses of farm income data by the same categories devel-
oped for primary uses. The most important of the major secondary and
tertiary uses was in passing the farm income data on to others, again
this only points out the problems in defining the p0pulation of users.
Over 31 percent of the users (64 out of 205) perform a communication func-
tion in the system as either a primary, secondary or tertiary use of the
farm income data. The decentralized social and economic organization in
agriculture would lead to the expectation that a high proportion of users
would pass the data on to others. The results tend to confirm this hy-
pothesis. The value of information in coordinating and managing the farm-
ing sector is also stressed by the fact that the data are shared with

others by many of those who receive them.

106

TABLE 4-4 SECONDARY AND TERTIARY USES BY CATEGORY

 

 

 

 

 

Uses
Use Category
Absolute Relative
Frequency Frequenc
(Percent)

General Information 11 14.3
Pass On to Others 24 31.2
Public Policy Uses

General policy uses 5 6.5

Program evaluation 0 O

Tax-revenue planning 0 0

Allocation of research funds 1 1.3
Demand Estimation Uses

Production planning 9 11.

Marketing-advertising 10 13.0

Within firm investments 0 0
Credit and Financing Uses

Future loan volume 5 6.5

Repayment ability 6 7.8

Investments in agricultural firms 0 0
Land Valuation 5 6.5
Miscellaneous

Futures market trading 1 1.3

Other 0 0
Total Secondary and Tertiary Uses 77 100.0

 

107

Public policy and miscellaneous uses make up a much lower per-
centage of the secondary and tertiary uses than of the total primary uses,
while relatively more respondents make additional uses of the data in
credit and financing, demand estimation, and land valuation uses as com-
pared to primary uses. The relative importance of these latter three
categories in secondary and tertiary uses suggests that the overall im-
portance of these three categories might be higher than the results com-
paring the primary uses alone indicates. Table 4-5 adds together both
the primary, secondary and tertiary uses to paint a slightly different
picture of the uses of the data.

An important consideration is the relationship among different
users and uses of the data. Target groups of users can be identified and
recommendations made to improve the data to meet the needs of these groups.
Given the importance of group interests in the political decision process
and the need for government funds to make improvements in data, it is
often crucial to develop support from interested data users in order to
obtain the budgetary support needed for improvements. The ability to
translate user support into political support and consequently budget
support is a strength of the decentralized Federal statistical system in
the U.S. Without this ability to foster user political support, improve-
ments in government provided data might be even more difficult. Thus, if
improvements in farm income data can be offered which resoond to the uses
of the data, then identifying the relevant user groups associated with
these uses is an important step toward realizing action on the recommended
improvements.

The results of a cross tabulation of major users and primary data

use for farm income is presented in Table 4-6. For clarity, only the

108

TABLE 4-5. ALL USES BY CATEGORY

 

 

 

Uses
Use Category
Absolute Relative
Frequency Frequency
(percent)

General Information 52 18.4
Pass On to Others 64 22.7
Public Policy Uses

General policy uses 35 l2.4

Program evaluation lO 3.5

Tax-revenue planning 3 l.l

Allocation of research funds 8 2.8
Demand Estimation Uses

Production planning l5 5.3

Marketing-advertising 36 12.8

Within firm investments ' 3 l.l
Credit and Financing Uses

Future loan volume ll 3.9

Repayment ability lO 3.5

Investments in agricultural firms 3 l.l
Land Valuation l5 5.3
Miscellaneous Uses

Futures market trading l0 3.5

Other 7 2.5
Total Uses 282 lO0.0

 

 

 

1(39

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

No.o¢_ NN.N NN.. NN.¢ N_.N_ N...N um.m_ No.cN .esON

NON e. e. N_ mm an 84 .e ee=_eN
NG.N_ N¢.m ac.N Nm.c Nm._ N4.N no._ Nm.e

NN __ c _ N N N o. aseeea__edm_z
Ne.m c c c a “N.N N¢.N e seeseeesoe

N. c c o a N e o .ados to abeam
NN.m o o o NN.N c NO.N o e_eez

N. c o o N a e o azez
NN.e Nm.. ac.p Nm.o no.N Nm.¢ o Nm._ mueaupsmeeu

N. N N _ N _ c N dues.ca
NN.o_ e um._ Nm.e am.c c e NN.N me.dea=.u

NN a N a. _ o c N me_xeaa
NN... o o o NN.N e a Nc.N N_ea=m

NN o o c a. o o 4 Sage. seem
an... o um.e am.e c NN.N Ne.N N¢.N seeeeLdseN

on o _ _ o N. m m .aceeda
NN.eN ue._ o o c NN.N “N.N_ Ne.m ee_3=3_eme_

on N o o c e. mN .p paeetuaeseu

accumuau
pouch «mm: co.uoapn> og—uwuu_u co.uua.umu xuppoa mcoguo a» :o—aaeLou=_ can:
so: «zoocoppoum.x can; as season uppaaa cc mung pocmcoo om:
Hut U Ne—gwkm
um: >¢<:_¢N NN maum: No zc_h<s=m<hmmoau .e-¢ NSN<N

110
major use categories and major users (those with more than 5 percent of
the total users) are shown.

For almost all categories of users there appears to be one or
occasionally two dominant uses. Perhaps the two most surprising findings
shown by the crosstabulation are in the user categories of educational
institutions and the news media. The domimant use by educational insti-
tutions is in passing the data on to others, accounting for 46 percent
of the primary uses in this category. This communication function appears
to be performed primarily through extension and classroom teaching pro-
grams. It should also be noted that public policy uses, usually in the
area of policy research, make up nearly 30 percent of the primary uses
reported by educational institutions.

The dominant primary use of farm income data among news media
users is also somewhat surprising. Two-thirds of the media users reported
that their primary uses were in the area of demand estimation. A priori,
one might have expected the news media, for the most part, to provide a
communication role in the farm income information system by simply re-
printing the USDA data. However, it appears that many publications, es-
pecially those aimed at a specific group of farmers, use the data to de-
termine which groups their publications should cater to in terms of adver-
tising and editorial content. For example, one publisher for a group of
regional farm magazines reported that they use the data in presentations
to advertisers defining the size, scope and character of their market.
Another noted that media buyers refer to farm income data for advertising
decisions. The USDA data tends to have special value in these uses because
it is provided by an objective third party. It should be noted that half

of those in the news media who reported that advertising decisions were a

111

primary use, also reported that a secondary use was in passing the data
on to readers in various magazine and newspaper articles. Thus, when all
media uses of the data are considered, the same number of uses were re-
ported for demand estimation and for passing the data on to others.

Table 4-7 shows this relationship as well as the distribution of total
uses among users.

Within the Federal government and state and local governments the
primary use of the data appears to be in public policy uses, as one might
expect. In each of these cases a secondary and somewhat less significant
use in terms of number of uses reported is in transmitting the data to
others, again not an unexpected use in these categories.

Among farm input suppliers, the clearly dominant and anticipated
use was reported in the area of estimating demand. Over 80 percent of
these firms reported this as a primary use. Almost 50 percent of the
banking and financing firms cited the various credit uses as a primary
use. The other major use within this banking and financing category was
for general information. However, given the types of decisions typically
made by these firms one would presuppose that the farm income uses in this
case would be similar to those users within this same category who cited
credit and financing uses as a primary use.

Private consultants showed no dominant primary use of the data. Since
these firms typically perform analysis of the data for other firms one
would expect that the primary uses of the data would cover a wide range,
depending on the needs of the clients of these consultants. The miscel-
laneous users also reported uses in all of the primary use categories.
The fact that over one quarter of these miscellaneous users reported that

their primary use was for general information or no specific use gives

I'l2

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

uc.oo_ no.m um.m am.m u_.a_ am.m_ uN.NN Ne.m_ Neuop
NmN N— m. «N em am we Nm ce=Nou
no.9. am.m um._ se.: uN._ um.N um.N Nm.¢
me _. m N m N N N. maoocc__oom_z
ac.c o o c o uN.N av.— o Newsccw>ou
m. o c c o o v o Nouoa so oumum
NN.m o o c am.N o um.N o NNuo:
o. o c c N c N c mzmz
xv.o u_.— uN.o uN.c um._ No.o e um._ mucaupzmcou
a. m N N m N o m ouo>_cm
am.—N o um.. um.m uN.c o uN.o RN.N m:_o:N=_N
Nm o m m. N a N N m:_xcam
“4N. o o :5 2.2 o c N: 32.3.
mm o o N aN o o N u=a=_ scam
“N: o :6 "Nd a... 3N nmN um; 22.538
cc o p N m —N m m potato;
u—.oN u_.— nN.c «N.c se.. u¢.m u¢.N— u¢.o cawuau_um:_
Nm N N N e N. mm mp NagoNNNust
someone
.ouoN mum: co.un=_o> mcpucacpm :oNNNENNNN ao._om meoguo o» :oNNascoN:_ cum:
soc maoocappoumvz tau; NNuacu teases u.Nn=m cc mmam —oco=ou um:
mum: 44¢ >N mzuma mo zo_p<4=N<Nmmo¢u .N-e NNN<N

113
support to the miscellaneous character of the uses among this category.

When the crosstabulation of users by all uses (Table 4-7) is
examined the dominant uses do not change and in fact even become more
dominant in some cases. The exception to this, noted earlier, was the
news media uses.

Question 5 on the mail survey identified 24 different components
of the USDA farm income data or formats in which these data are pre-
sented. Users were asked to indicate the usefulness of these components
or formats by checking whether these were very useful, moderately use-
ful, rarely useful or not used. Table 4-8 presents the frequency dis-
tributions of the responses from this question.

For the purposes of priority setting it would seem relevant to
identify those components which are most useful so one might insure that
maintaining the accuracy of these series takes precedence over other
series. For different formats of the data this information on useful-
ness gives an indication of how the data are used since each different
format implies a certain level of analysis of the data.

The frequency distributions in Table 4-8 shows the greatest
variability among formats and components in the "very useful" category,
with only 3.7 percent of the respondents finding the series on value of
home consumption very useful while 37.8 percent found the data series
on farm production expenses very useful. It is interesting to note
that those series found very useful by more than 30 percent of the re-
spondents were divided equally among the summary income measures, dis-
tributional data by commodity and state, and the production expense
data. The fact that data on production expenses are found very useful

by nearly 40 percent of the respondents is somewhat surprising since

H4

.oc.n=:o.um~:a mg» 50 NN coNNQNm emu

-o_aeou a—co can cast «sauc. use» on: No: u_u has» «ecu coauum as: mucoecoamoc omega Lou m. Ngomouau mpg»:

 

c.nN
o.n~
o.m~

c.m~
c.m~
o.nN
c.n~
o.m~

o.mN

o.m~
c.mm
o.m~
o.m~
o.n~
c.m~
c.mw

c.mm
o.m~
o.MN
c.n~
o.m~
o.n~
o.n~
o.nN
o.n~

N.c
v.v

O
Q

'MNQOCNOV QCON'OO Q QQOVO
QMMQVQMQV GVQMQQQ Q Q'QQ?‘

~.e~
n.m~
m.m~
N.m—
c.v~
o.mN
c.o~
o.N~
m.n~
m.N~

$0
0-.

o o o I
“MI-n90“
"I‘N

O O I 0 O
NN

NQQQéONOQ

o
d

ommaomcn—c «shaman

«.mg
m.m~
c.m~
v.NH
v.e~
°.N~
o.-
—.-
~.e~

@Q
pow-o

o O I I I O
H—O—O

one...
fifiNN

”OMNMNQ¢D mmmcmmm
NROOMOGQQ “NOV”

0
0-.

«.mN
N.nN
m.n~

N.mN
m.-
o.nN
a.e~
o.m~

~.m~

n.m~
N.nN
m.e~
¢.¢N
m.~N
«.mN
m.¢N

o.NN
o.m_
~.-
N.m~
~.~N
—.NN
N.-
o.-
c.n~

o.cN
c.N_
~.¢~
~.m~
m.-
c.o~
m.v~
m.-
c.NN
".m—

o
N

O
‘lflflI-OLDMODMN “509°
o-GMMMNM

c-HOIDOHDOMNQ moon

ammucoucom mancm><

mun—u mm_om No mapu> an acouaconc

ELNN so mamcoaxu :o_au:vcca use «sous. scum
mac—u mopam No mapo> an ace oucaom

cones an x—NENN gauacoao Ego» can esou=N

mucosxaa acoscco>cc

«Noun an mmmcmaxo :oNNuauoga Esau

mmapu mopam No cape) N: macaw No coasaz

oc_asnm seem mcouacoqo scam No oeouc. gmnu uwz
cc.uae:mcou

_aa_aau gonna use copuo—uocaov scam
acosumm>=p

an: new mmcau.u=aaxm poo—qnu «mega scam

menu. pea—auu ELNN No :oNNNcmao ecu aspen»:

some. Esau vac—g so» mmmcoaxw

nonconxo mcpuucoao scam Newscau

mmmcoaxo :o_uu:eocq scam

mance. ELNN Leguo use am:o5-:oz

co—uopaaoa Esau Na venue. uu_aau can
:opuo_:aoa grow we «saucy

—a:cmcoa «panacea—u ago «save. Nucomcom

acoN can usage. umuco>o wee mace» No guess:

capuassmcou was: we mapa>

gazes an maa.cuog gmuu

NJNeossou an munvuuoc zmuu

«Noam an mun—mums smug

«Noam an wanes? Esau

osoocw scam no: page» Noacc<

venue. Esau an: ego mmocm eNNNNNuc Noacc<

mcougoac an usage. scam am: as. mmoco

 

Nucmugoav Nucousamv Nucuucomv Nucougumv

cause
usou:_ ago;
on: we: on
cazmc< oz

sgzuc<

«02 on

«m: Nahum:

Nagaucomv Nucoucomv

pace»:

apogoc »_uuacouoz

Nauoma
Ngo>

Nassau so acagcnsou

 

<h<a uzouzu x¢<u (an: ma mhzwzcmzcu
az< mpczaom ampuuANm no mmmzaauum: u>~h<4u¢ mic m4m<h

115
those who produce the data suggest that in terms of accuracy the produc-
tion expense data are one of the weaker component series.

The usefulness of different types of distributional data is also
suggested by these results. The current users of the farm income data
would seem to find income data distributed by commodity to be slightly
more useful than income data by state and each of these somewhat more
useful than income distributed by size of farm as measured by value of
sales class. This would tend to imply that some measure of net returns
by commodity or type of farm might be a useful addition to the current
series since the only data of this type currently available is gross
cash receipts distributed by commodity.

While it might be expected that the two summary formats for
presenting the USDA data, i.e., realized and total income, would be
more useful than some of the component series to those who responded,
it is interesting to note that the realized series was found to be
slightly more useful than the total net income series. In the face of
the criticisms of the realized income format in the literature, this
finding is somewhat surprising. One possible explanation for this
might be that the realized net income series has been emphasized by the
USDA in the past and there is a carryover effect that causes some users
to place more emphasis on the realized income statistics.

The two least used series or components of the farm income data
are the value of home consumption and nonmoney and other farm income.
This latter category actually includes the former along with gross ren-
tal value of farm dwellings and income from custom work, machine hire
and recreation on farms. These first two components are based on im-

puted values estimated by USDA rather than collected data and as such

116
might explain some of the lack of use. Furthermore, this suggests more
Of a concern for data on farm business rather than farm families.
A priori, it was conjectured that because of the list of de-
tailed components presented in this question that it would be useful
for respondents to refer to the USDA statistical bulletin Farm Income

Statistics in answering the questionnaire. In the cover letter sent

 

with each questionnaire, it was recommended that users refer to the bul-
letin in answering the questions. So an additional question was asked
to determine whether individual respondents referred to this bulletin

in answering the questionnaire. It was also felt that the question on
the usefulness of components and formats would be the one most likely
affected by whether or not respondents consulted this bulletin. So a
crosstabulation of answers for each component or format was made with

the answer to the question on the use of the Farm Income Statistics

 

bulletin.. A chi square test was used to determine if there were dif-
ferences in the usefulness of the component series or formats depending
on whether or not this bulletin was used. Differences which were sig-
nificant at the 10 percent level were found for only 4 of the 24 compo-
nents: value of home consumption, number of farms and average income
per farm, farm production expenses, and repairs and Operation of farm
capital items. In general, though, this would seem to suggest that
those who referred to the bulletin in answering the questionnaire did
not answer this question in a manner significantly different from those
who did not consult this publication. For 24 questions taken at random

one would expect 2 or 3 to appear significant at the 10 percent level.

117

TABLE 4-9. PERCENT OF RESPONDENTS USING FORECASTS
OF FARM INCOME BY SOURCE

 

 

Absolute Percent Percent Percent of
Frequency of Total of Those Total
Respondents Using Sources
Forecasts
Total Respondents 199 100.0 100.0
00 Not Use Forecasts 24 12.1
Use Forecasts 175 87.9 100.0
Source of Forecasts
U.S. Department of
Agriculture 151 75.9 86.3 53.5
In-house or internal
estimates 42 21.1 24.0 14.9
Radio, TV, Newspaper 23 11.6 13.1 8.2
Trade Publications 26 13.1 14.9 9.2
Private Consultants 22 11.1 12.6 7.8
Other 18 9.0 10.3 6.4
Total Sources of
Forecasts 282 100.0

 

 

 

 

As a proxy for identifying major sources of outside analysis of
the farm income data, a question was included on the mail questionnaire
to ascertain the major sources of forecasts of farm income used by the
respondents. Simple forecasting of future farm income was thought to
be one of the most basic forms of analysis of the historical data. Of
the 199 users of the farm income data who responded to this question,
88 percent reported using forecasts of farm income from at least one
source while 12 percent did not use forecasts. Table 4-9 shows the
sources of farm income forecasts as a percentage of the 282 responses,
as a percent of those who use forecasts and as a percent of total re-
spondents. Since it was possible for an individual to give more than

one source of farm income forecasts these latter two percentages sum to

118

a total of more than 100 percent. For example, 86.3 percent of those
using forecasts reported receiving them from the USDA, while 24 percent
reported making their own forecasts. Thus some individuals must have
used USDA forecasts in addition to their own. From this table it is
clear that the U.S. Department of Agriculture is a dominant analyst in
the system when one considers forecasting of farm income. Not only do
86.3 percent of those who use forecasts report that they use USDA fore-
casts but also 13.1 percent report using forecasts obtained from radio,
television or newspapers and 14.6 report receiving forecasts from trade
publications. Table 4-11 shows that 100 percent of the news media re-
spondents used USDA forecasts of farm income so it is likely that those
individuals who reported using radio, television, newspapers or trade
publications also received USDA forecasts. Of the 13 percent who re-
ported using private consultants many named one of the econometric
modelling companies which have agricultural sector models such as Data
Resources Inc. or Chase Econometric Associates. In the category of
other sources of forecasts state universities were mentioned by many as
a provider of farm income forecasts.

Perhaps of more interest are the results presented in Tables
4-10 and 4-11 which show the sources of farm income forecasts by use
and user categories respectively. These show a relationship among
users and the source of forecasts used in private sector decision mak-
ing. Those who use the data for demand estimation and credit or financ-
ing decisions seem to rely on their own or private consultant forecasts
to a much larger extent than those who make other uses of the data.
Also, those who use the farm income data only for general information

tend to use USDA forecasts much less frequently relative to other use

I'I9

.uuo .N :E:.ou N N ca:.ou .N :a=.ou N . :E:.ou Nucoacocu ..u.. .oogaom ca.=u.ucoa N N=.m: Neououau om: Na mucoucoamoc No szogoge

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

eucoucoe N... N.N N... N.N. N... e..N N.NN .Nuo.
Nueaaawcu u.=.oma< NN. NN N. NN NN NN Ne .N. ce=.ou
«acougmm N.N c c N.N. N.N N.N. N.NN
Nucuaaocu oN=.oNn< N. . o o N . N N. Naooce..oum.z
eueoucoe N.N. o N.NN N.NN N.NN N.NN N.NN =o..a=.e>
35:8... 3282 o. . c N N N N N 2.3
eucougoa N.N N.N. N.NN N.N N.N. N.NN N.NN N:.u:N=.N
.3533... 328.... N. . N e . N o N. .35. 38.5
aucmucog N.N N.N N.NN N.v. N... ..eo N.NN co.uoe.umu
.3838... 33.32 ..N N N N . N N N. NN ness...
eucmucme ..N ..N N.N N.N. ..N N.N. N.NN New:
Noemaoocu o.:.cma< No N N e N N N .e Nu..oN u..a:N
success; .... N.N. N.N N.N N.N. .... ..NN «ensue oN
Nucmaaocu mu:.omn< am e N . . m e .N co mama
cueoocoe ..NN N.N. ..N N.N. N.N. N.N. N.NN co.NNscoN=.
3:38.... 33.3.... N N N N N N N NN .828
macoucoqNoN mumauocom muceu.:mcou m:¢.uau..g=g cocoamzmz .occaac. co oc:N.:u.NN< am:
3.... M2. .55.. 32...... 3.... N. .23.. 38..-... . a N... . .22....
ass: N 2 on a on N a amouocou
N N N m e N N . No magnom
mm: >¢<z.¢a >N mhm<um¢cN no uuuaom .o.i¢ NNN<N

.uum .N :e:.ou N N =E:.ou .N cs:.ou

. ca:.ou Nucmzcugm ..u.. .oucaoN

ca.=u.uc~a

N N:.N: Neoamuuu goN: Na mucoucoach Na Nemucaae

 

 

 

 

lZC)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

..eeeee. N... N.N .... ..N. N... ...N N.NN .NNN.
Nagasaec. N.N.NNNN NN. NN N. NN NN NN NN .N. ess.eN
..eeecee N.N. N.N N.N. N.N. N.N. N.N. ...N
Nucmacocu mu:.oNn< NN N . N N N N NN Naomcu..mum.z
«Needeee o N.NN ..N o ..N N.NN N..N Nemseeesou
Nucmacmcm ou:.cNN< .. o N . c . N N .Nuo. go «Noam
..eeecee o N.NN N.N. N.N. N.NN N.NN N.NN. N.NN:
Necesaee. N.N.eNNN N. N N . . N N o. Nzez
..eeetee N.N. ..N o N..N o N.N. N..N N.NN..:NeoN
Neceseece N.N.eNNN N. N . c N o N a. N.N>..N
..eeueee ..N N.N. N.N. N..N N.NN N.NN N.NN Ne.eeee..
Neeaseoe. N.N.NNNN NN N N N N N N N. Ne.NeeN
..eaucee N.N. N.N N.NN N.N. N.N N.NN N.NN N.NNNN
Nucmaamcu mu:.oNa< NN c . N N N N N. Naac. scum
..eeueee ..N o o N.N o N.N. N.NN Neeeeceseu
Necesaee. N.N.NNNN NN N c o . o N NN .NLNNNN
.NeedeeN ..N. N... N.N N.N N.N. N.N. N.NN ee..=..NNe.
Nucoacocm cu=.oNa< NN N N N N N N cc .NcoNNNuacN
N cm :oamm N Nauugo
a v56 N N am: N cog» NN=NN.:N=ou Nee—uuu..aaa .maoqmzmz .occouc. so aesuhwu.gm< xgwmwcm
cease: No: on N Nuns... tem.. .N. .e.NNN eN=e=-e. ..NNN .N .N
uNaumcoN
N N N N N N N . Ne eueseN
.NNNNNNN NNNN .N N.NNNNNN. Na NuaaoN ...-N N.NN.

121
categories. The low relative level of use of USDA forecasts found in
this category seems to be attributable to the low level of use of fore-
casts in general by this group since they have over twice the percen-
tage of nonusers of forecasts compared to the other groups.

When analyzing user categories and sources of forecasts the
results tend to confirm the findings mentioned previously. In this
case, the two major private sector farm income data users, farm input
suppliers and banking and financing institutions, show the lowest level-
of direct use of USDA forecasts and among the highest levels of use of
both internal and private consultant forecasts. Two possible explana-
tions for these phenomena are apparent. First, the importance of the
decisions in these firms require multiple sources for validation and
the reduction of uncertainty. Second, the ability of these firms to
capture the returns to improved information causes these firms to be

willing to make expenditures for more information.

4.3.4 NONRESPONDENTS

In part, the applicability of the mail survey results to the
entire population of farm income data users, or that part of the popu-
lation on the USDA mailing list, depends upon whether and how the non-
respondents to the survey differ from the respondents. Given the level
of nonuse of the farm income data found among those who responded to
the questionnaire, one reasonable hypothesis would be that those who did
not respond to the questionnaire, for the most part, did not use the
USDA farm income data. A counter hypothesis is that those who responded
made up an inordinately large portion of the nonusers of the farm income

data. This might be expected since it was more difficult to explain how

122
the data was used than to simply write that the data was not used.

The responses to a telephone survey of the nonrespondents pre-
sented in Table 4-12 seem to refute this latter hypothesis and support
the hypothesis that those who did not respond also did not use the data.
0f the nonrespondents to the mail survey who were contacted, 56.7 per-
cent said outright that they did not use the data while in another 13.3
percent of the cases, the individual to whom the mail survey was origi-
nally sent was no longer with the firm or agency and the person con- °
tacted on the telephone could not refer the call to anyone who used the
data. So almost 70 percent of those contacted did not use the data.
This proportion is significantly different at the .01 level from the
proportion of nonuse found among the respondents. Even the 56.7 percent
level of nonuse is significantly different at the .01 level when com-
pared to the 24.1 percent of nonuse found among the respondents.g/
Thus, if this percentage of nonuse is applied to 46.85 percent of the
sample that did not respond and added to the 24.1 percent of the respon-
dents who did not use the data then over 45 percent of the sample did
not use the USDA farm income data. The fact that nearly half of those
surveyed do not use the data is even more significant when one considers
that the sample was drawn from a population that receive the data on a
regular basis.

Table 4-13 summarizes the breakdown of the nonrespondents by
user category. While the sample size is relatively small, the nonre—

spondents identified seem to be represented in about the same

2] The procedure for testing the difference between sample prOportions
is presented in Freund, p. 285-287.

123
TABLE 4-12 NONRESPONDENT

SURVEY RESULTS

 

 

 

 

 

 

 

 

Adjusted
Absolute Relative Relative
Frequency Frequency Frequenc
(Percent) (Percent{
Contacted ‘ 30 83.3 100.0
Use for general information 5 13.8 16.7
Use specifically 4 11.1 13.3
Do not use 17 47.2 56.7
No longer with firm or agency
and could not refer to a user 4 11.1 13.3
Not Contacted 6 16.7
No telephone listing 4 11.1
Unpublished telephone 1 2.7
Unable to contact 1 2.7
Total Sample 36 100.0
Total Nonrespondents 237
TABLE 4-13 NONRESPDNDENTS BY USER CATEGORY
Relative
User Category Absolute Frequenc
Frequency (Percent{
Educational Institution 4 13.3
Federal Government 6 20.0
Farm Input Supply 2 6.7
Banking or Financing 3 10.0
Private Consultants 2 6.7
News Media 1 3.3
State or Local Government 1 3.3
Miscellaneous 6 20.0
No User Group Given 5 16.7
Total Contacted 30 100.0
Unable to Contact 6

 

 

 

124

proportions as the respondents. Using a 2 test to test the differences
between prOportions and the null hypothesis that the proportions of the
contacted nonrespondents in each category are equal to the proportion

of respondents in each category, at the 10 percent level the null hypo-
thesis cannot be rejected for any category of user.§/ Thus, the nonre-
spondents seem to be similar to the respondents in most ways except that
the nonrespondents have a much lower level of use of the data than do

the respondents.

4.3.5 IDENTIFICATION OF DECISION MAKERS TO BE INTERVIENED

The main thrust of the personal interview portion of the re-
search was aimed at the public policy uses of farm income data. A some-
what more limited number of interviews with private sector data users
were made to gain a better understanding of the uses of the farm income
data in the farm input industry and the banking and financing industry.
However the majority of those interviewed were in the agricultural poli-
cy decision process at the national level.

As was stated earlier, the value of information stems from the
value of the decisions in which the information is used. Accepting the
fact that some decisions are more significant than others forces one to
have misgivings about relying on a random sample of users to set prior-
ities on the redesign of a data system. The use of publically collected
farm income data in public policy decisions was felt to be a major use,

the importance and details of which would not be brought out in a

_3/ Ibid.

125
strictly random sample of users.

Given the papulation chosen for the mail survey, i.e., subscri-
bers on a USDA mailing list, it appeared that public policy users might
be underrepresented since more direct channels for supplying this in-
formation exist for public policy decision makers when compared to the
use of a mailed printed bulletin. Thus, the population was expanded to
include more of these users. In a sense the total population of farm
income data users was stratified with the major public policy decision
makers being sampled at nearly a 100 percent rate while other users
were sampled at a lower rate. The basis for selecting the public poli-
cy users is outlined below.

Through time the decision process in agricultural policy has ex-
panded to include more actors than the traditional "Farm Bloc" of the
19205. The increased specialization of production in farming has in-
creased the fragmentation in the policy decision process (Bonnen 1977).
Hence the number of important actors who use farm income data in policy
making has also increased since the time the data system was first de-
signed.

Within the Federal government there are two main branches which
are concerned with policies affecting farm income. The executive branch
has remained a principal locus of decision making in agricultural poli-
cy. However, through time the primary nodes of decision activity have
moved to higher levels in the executive branch hierarchy when compared
to the 1920-1940 period when a high proportion of the decision making
in the executive branch was focused at the bureau chief level. Decision
making at this earlier time was characterized by what Randall Ripley

and Grace Franklin call the "subgovernment phenomena" where interest

126

groups, Congressional committees and agency head or bureau chief level
members of the exesutive branch dominate the decision process. When
subgovernments are strong there is generally good agreement as to the
appropriate direction of policy in a given area. In agriculture the
fragmentation of interests brought on by increased specialization has
led to greater conflict in agricultural policy and thus has moved the
decision making up to higher levels in the process. This has weakened
the subgovernment in agriculture so that important decisions on agri- _
cultural policy in the executive branch are now being made at the
Assistant Secretary and Secretary level in USDA, in the Executive Office
of the President particularly at the Office of Management and Budget and
the Council of Economic Advisors, and even in the White House itself
(Bonnen 1977; Cochrane and Ryan; Stucker, Penn and Knutson). In recent
years, many other actors in the executive branch, such as the Secretar-
ies of State, Commerce, and Treasury, the Special Trade Representative,
the Special Assistant for National Security, etc., have also played an
important part in the policy process (Bonnen 1977; Stucker, Penn, and
Knutson). A priori, it was felt that these other executive branch ac-
tors would be less concerned with farm income and more concerned with
issues of direct interest to their specific positions than the actors
mentioned earlier. The interviews with individuals in these tradition-
ally nonagricultural_agencies and others seemed to confirm this hy-
Pothesis. Farm income was not as crucial to the types of decisions
with which these individuals normally dealt.

In the legislative branch the important actors concerned with
farm income fall into three basic areas: the Senators, Congressmen and

their staffs on 1) the agricultural committees, 2) the House and Senate

127
Budget Committees, and 3) two of the research arms of Congress, the Con-
gressional Research Service in the Library of Congress and the Congres-
sional Budget Office. The appropriations subcommittees concerned with
agriculture in both the House and Senate were at first thought to have
a major interest in farm income data. However, since most of their ap-
propriations work concerns the operating budget of the U.S. Department
of Agriculture and only indirectly the income support programs, the in-
terviews revealed these subcommittees to be less concerned with farm
income data relative to these other Congressional users. Again there
are other committees such as the Interior Committees which affect vari-
ous aspects of food and fiber policy but their concern with farm income
issues are not as critical as the groups mentioned earlier.

Various private interest groups are also influential in the
agricultural policy process. Thus, interviews with the general farm
organizations such as the National Farmers Union, the American Farm Bu-
reau Federation, the National Farmers Organization and the National
Grange or commodity groups like the Milk Industries Foundation might
also give insights into the private use of government data in public
decision making. Also some other interest groups representing consu-
mers and other interests were chosen for interviews to see how some of
these groups outside of the "agricultural establishment" might use the
farm income data.

Thus, 35 interviews with the various actors in the public deci-
sion process in agriculture were undertaken to obtain a more in-depth
understanding of the use of farm income data in the process. Those in-
terviewed were selected to represent a crossection of the roles and or-

ganizations that currently participate directly in public policy

128
decisions for agriculture. Those selected for interviews included pre-
sent and former members of the Department of Agriculture such as the
Assistant Secretary responsible for commodity programs, the Director of
Agricultural Economics, and various USDA staff and advisors dealing
with price and income policy; agricultural policy experts in the Office
of Management and Budget, on the staff of the Council of Economic Advi-
sors, and on the White House staff; the relevant staff peOple of the
Congressional committees and their research organizations; and the lob-
.byists and economists of many pertinent interest groups. Interviews
with 35 of these individuals were conducted during the fall of 1977 to
determine the uses and needs of these policy makers with regard to farm
incane. All interviews were made under conditions of confidentiality of

individual respondent indentity.

4.3.6 RELEVENT INTERVIEW RESULTS

The common expectation concerning the use of data in policy de-
cision making would seem to indicate that data are important in the for-
mulation of policy. Data should provide a source that along with analy-
sis can produce information about problems that require government ac-
tion. Insofar as farm income is a measure of the combined effects of
changes in prices, sales, production and costs it should be a useful
guide in policy formulation. Thus, farm income data might be used to
monitor developments in the fann sector and to provide an impetus for
policy action.

However, none of the policy makers interviewed could cite any
instances where the data on farm income alone led directly to any spe-

cific policy action. Aggregate farm income data are used in the policy

129
process primarily to support positions taken by decision makers after
these stands are taken. As one Congressional staff person put it,
"knowledge fortifies bias.“ Most uses of aggregate farm income data
seem to be in justifying positions that were already held and not in
initial policy formulation, per se. Farm income data do not seem use-
ful in placing issues on the policy agenda, rather, in most cases, the
data are used merely as a tool in the debate.

The nature of the political process itself may force farm in-
come to a lesser role in policy formulation. Farm income tends to be
used in reinforcing policy positions rather than in forming policy pre-
scriptions because most policy formulation arises from constituent
pressure or pressure from other decision makers in the process which
occur before the income data are available to reflect the source of the
pressure. One former Executive Branch decision maker noted that the
timeliness of the farm income data, since they are for the most part,
published on a yearly basis, causes many of the uses of these data to
be ex poste, that is the data are used to justify positions rather than
formulate policies. Another Congressional staff person felt that many
politicians like to use the aggregate farm income figures because changes
in the aggregate are much more dramatic in their impact. Since the main
uses are primarily in speechmaking and the ex poste justification of
policy proposals, the dramatic nature of changes in farm income are
much easier to show when dealing with aggregate figures rather than
averages.

Another reason that farm income data are not used in the policy
formulation stages of decision making on agricultural policy is due to

the way the legislation is generally written. Agricultural policy has

130
tended to be commodity specific, aggregate farm income data are not, so
in most cases farm income data must be used in conjunction with price
data or cash receipts data by commodity to influence commodity policy
decisions.

If farm income data are not used to a great extent in the for-
mulation of alternative policies the next logical place to examine
their usefulness is in evaluating alternative policies after these are
formulated. Farm income data do seem to have some influence in the de-
cision process with regard to certain issues. In terms of Congression-
al decision making a few interviewees felt that USDA farm income data
are more likely to influence those Congressmen and women on the fringe
of agricultural issues, i.e., those with nonagricultural constituencies
or who are undecided on a particular issue. The major agricultural
leaders in Congress tend to have their own informal information sources
and thus rely less on published sources of data on farm income. This
use by nonagriculturalists stresses the importance of understandability
as an attribute of data on agriculture. It is doubtful whether any per-
son unfamiliar with agricultural data would be able to make the distinc-
tion between realized net farm income and total net farm income, so
some confusion might arise because of this. The recent turnover in mem-
bers of Congress and their staffs only exacerbates this problem. The
apparent preference of Congress to use the notion of "cost of produc-
tion“ as a basis for setting target prices as opposed to "parity" seems
to be a reflection of the understandability of the data. Most urban
members of Congress at least implicitly feel that they know the meaning
of costs of production but for the concept of parity this understanding

is not as prevalent, even though both of these in a sense are variations

131
of the same general concept.

While no one suggested that farm income data were disregarded
in decisions concerning various policy alternatives, many suggested
that aggregate farm income as a statistic was given a rather low weight
in the process. The analysis concerning the 1978 set-aside program for
wheat provides an example. The Agricultural Stabilization and Conser-
vation Service of USDA did a study that showed lower aggregate farm in-
come under the various proposed set-aside programs. One high level Ex-
ecutive Branch decision maker stated that this aspect of the analysis
was given a very low weight in the process because a large error in the
farm income data was assumed which would most likely overshadow the ef-
fects of the set-aside on farm income, Second, the more important
question of whose income is reduced was not answered, so the distribu-
tional aspects of the effect of the set-aside might offset the aggre-
gate effects. Another decision maker noted that aggregate farm income
data were not very useful in evaluating different program alternatives,
because the data are not sensitive to changes among policy alternatives.
He felt that this arose primarily because of the way the data were cal-
culated., The reliance on trends around census benchmarks and historical
patterns in estimating the data tends to smooth out the income estimates
and thus reduces the sensitivity of the data to specific dimensions of
proposed policy changes. Other policy makers suggested that cash re-
ceipts data are often more useful for comparing policy alternatives be-
cause these tend to be more sensitive to the policy Options.

These problems in using farm income data in program evaluation
seem to have reduced the effectiveness of farm income data as a prime

mover in decisions on agricultural policy. MoSt of the uses of the

132
data in policy were indirect or ad hoc, so other variables were often
more important in the decisions. Variables such as budget costs and
consumer prices and, in 1977, cost of production often become dominant
in the policy deliberations. Another reason that farm income data have
not been as critical as other types of data in agricultural policy de-
cisions is the failure to debate the question of the need for government
intervention in agriculture. Aggregate farm income is the primary sec-
tor performance measure available for farming. Without serious ques-
tioning of the need for special treatment for the farming sector, aggre-
gate sector performance data which can be used to compare agriculture
with other sectors are not as critical. Instead, the farm policy de-
bates center on the distribution of the benefits and costs of the pro-
grams. This creates quite different data needs which will be discussed
in the next chapter. The principal focus of the debate on farm policy
also effects the type of analysis needed. For instance, input-output
models might be useful in program evaluation but as long as the farm
policy debate does not explicitly consider the impact of farming on
other sectors then the importance of this type of analysis is reduced.
However, an observation is in order. A case can be made that an impor-
tant consideration for agricultural policy is the increasing impact on
agriculture of changes in other sectors. An obvious example in recent
years was the concern among policy makers over the impact of changes in
the energy sector or in environmental policy on farming. Thus, input-
output studies and similar types of analyses might provide needed infor-
mation for decisions in agricultural policy in some cases. Conversely,
the greatly increased importance of agricultural exports in the balance

'of payments and the potential impact of farm policy changes and farm

133
sector performance on inflation reputedly now cause concerns among U.S.
policy makers outside agriculture because of the farm sector's impact
on various aspects of the U.S. economy. Nevertheless, no evidence of
any of these policy concerns showed up in the interviews with policy
makers primarily concerned with agricultural policy decisions.

Prior to studying the system, a further policy use of farm in—
come data that was considered was in program implementation. For exam-
ple, many government programs or grants are made available to certain
localities based on measured levels of unemployment as published by the
Bureau of Labor Statistics. There seem to be no specific examples of
program benefits or other aspects of programs tied directly to USDA farm
income data. Guebert noted that farm income data enter into the revenue
sharing mechanism but this use is only indirect in that the farm income
data enter into the overall income measures of a geographic area com-
puted by the U.S. Department of Commerce. These are then used to allo-
cate the revenue sharing dollars. This laCk of uses in implementation
might also explain the seemingly low level of direct use of this farm
income data in the policy process. It also suggests that changing the
concept or definition of farm income might run into less political road-
blocks. since few program benefits are directly affected by changing
the concept or definition. However, the implicit ties of farm income
to price parity might create difficulties since parity prices are pre-
sently still used in implementing a few government programs.

The sources and types of analysis found in the public decision
process are also important in understanding the operation of the farm
income information system. Until recently the primary government ana-

lysts for public policy decisions in agriculture were in the U.S.

134
Department of Agriculture. The subgovernment phenomena and its domi-
nance in earlier years of the decision process gave analysts at the
lower levels of USDA influence in the process. However, the lack of
conflict in the process also tended to reduce the need for analysis of
alternatives. Recent changes, particularly in Congress, have shifted
the relative importance of different analysts. Congress has increased
its ability to do analysis in the agricultural policy area. The budget
process in Congress has increased the demand for both analysis and more
detailed data in Congressional decision making. The conflict between
the Executive Branch and Congress that dominated the later Nixon Admin-
istration years has also had a carryover effect which has caused Con-
gress to rely less on USDA analysis of farm policy issues. Increased
staff on Congressional committees, increased reliance on the major re-
search arms of Congress, i.e., the Congressional Research Service of the
Library of Congress, the Congressional Budget Office, and the General
Accounting Office, and the increased use of paid private consultants
have all reduced the relative importance of USDA analysis in farm policy
decisions. Analysis is also provided by the various interest groups in
i the agricultural policy process. While recognizing some of the problems
mentioned in Chapter 2 concerning the use of privately supplied informa—
tion in public decision making, the continuity of the decision process
through time reduces the incentive of private interest groups to distort
selectively the information that they provide. The use of private in-
terest group analysis seems to be most important in Congressional deci-
sion making and among those decision makers in the Executive Office of
the President. Private interest groups are also more likely to be used

for analysis on commodity specific issues since many of these groups

135
have a commodity focus and considerable access to information on their
subsector of agriculture.

Private consultants, especially those offering large econome-
tric simulation models of the agricultural sector such as Data Resources
Inc. and Chase Econometric Associates Inc. have become more important in
decisions on agricultural policy. The Congressional Budget Committees
and the agricultural committees all subscribe to and use these services
as do most of the Executive Branch decision makers outside of USDA. One
reason given for the rise of these private analysts relative to USDA ana-
lysts are that the private consultants are willing to take more risks
and project farm income estimates much further into the future than USDA.
When dealing with a four- or five-year farm bill this can be significant,
although it must be added that the USDA does produce unofficial fore-
casts for these types of uses. This willingness to take additional
risks is perhaps more important among the private sector users of fore-
casts where USDA forecasts of sufficient length into the future might
not be available.

University researchers also provide a source of policy analysis
in the public decision process. The type of analysis provided by these
researchers is usually more of a long run, in-depth nature rather than
evaluations of selected policy alternatives under deliberation at a
given time in the decision process. As one staff person noted, it is
difficult to get university researchers to reSpond fast enough to be of
use in the policy process. It is apparent that timeliness in analysis
is often as important as timeliness in the provision of the data. Lags
in either of these steps hinders the use of information in the decision

process. University researchers also influence government analysis

136
through professional ties with government analysts. Nearly all of the
government decision makers reported using, to some extent, university
sponsored research but the usefulness of it varied depending on the
types of decisions to be made. The remarks of a Congressional staff
person as to the relationship between research and policy making are
somewhat enlightening in this regard. It seems that researchers tend
to have quite different aims than policy makers. For the most part re-
searchers want a product that is entirely defensible in the sense that
it is comprehensive and correct, while policy makers must define a pro-
duct that will be acceptable to their constituents. The policy maker
does not necessarily care if he or she is right for the right reasons,
so a policy prescription does not have to be entirely defensible. Thus,
the data requirements for research and policy making are somewhat dif-
ferent. When a researcher encounters insufficient data on a given as-
pect of a problem, he or she is able to shift the focus of the research
and thus can shift the risk to others by not dealing with this aspect of
the problem. A policy decision maker is not often able to shift the
risks brought on by insufficient data and must deal with the issue at
hand with or without the relevant data. So insufficient data or gaps
in the data on a given issue create problems which are often more criti-
cal to the decision maker than the researcher.

The use of various components and formats of farm income data
seem to be somewhat different among those interviewed and the mail ques-
tionnaire respondents. The data on farm income and number of farms by
value of sales class seemed to be much more important than suggested
by the mail survey. A reason for this, which was cited by many of those

interviewed, was that for most of the decisions relating to income

137
distribution this was the only data available, so it received a signifi-
cant amount of use. Comparisons of cash receipts data among various
cr0ps and between crops and livestock also appears to be very important,
this finding concurs with the mail survey. It was also suggested that
cash receipts were used primarily because net income data by commodity
are not available. Most policy makers recognized the problems associ-
ated with the data on average income per farm and hence these were not
used often in the decision process.

Decision makers in agricultural policy at the national level
also seem to indicate a much lower use of the production expense data
than those who responded to the mail survey. Some interviewees sug-
gested that the data was too crude for any major types of analysis.
This coupled with the relative insensitivity of these data to changes
in the short run also reduced its usefulness. A further consideration
of the mail survey results would tend to confirm this. Only 18.9 per-
cent of the Federal government respondents to the mail questionnaire
found production expense data very useful while 40.8 percent of the re-
maining respondents found production expense data very useful. While
it is not statistically correct to compare these prOportions taken from
an ordinal scale ranking, these mail survey results at least suggest an
explanation for the interview findings.

Another and much more limited set of interviews were used to
develop information on the use of farm income data among two of the ma-
jor user categories in the private sector. Six personal and telephone
interviews with representatives of farm input manufacturing and banking
firms were used to gather more in-depth information on their uses of

farm income data. Individuals were selected from the Farm Income

 

138

Statistics mailing lists and an attempt was made to choose firms repre-

 

sentative of these two user categories in the judgment of the researcher.
All individuals were assured confidentiality in responding to the ques-
tions. The schedule of questions was similar to that used for the pub—
lic policy user interviews but concentrated primarily on how data and
analysis was used and on deficiencies in the existing data.

In the farm input supply industry, farm income data are used to
gain an understanding of changing economic conditions in agriculture
which then can be related to the demand for farm inputs. Cash receipts
also tend to be more important relative to other farm income data in
any econometric analyses of demand done by these firms. In part this is
because of the way farm income data are estimated. Cash receipts ele—
ments seem to induce most of the year to year changes in farm income.
These changes are what effect the forecasts and demand estimates of
these firms, so cash receipts become key data.

As agriculture has become more specialized, inputs have become
more commodity specific. The fact that cash receipts data are availa-
ble by commodity increases its usefulness for input suppliers. Even in
the case of a general input such as a tractor, commodity specific data
are important. In general, the size of tractors used on various types
of farms is different. For instance, knowing that dairy farms and to-
bacco farms use more small tractors makes cash receipts data for these
commodities more useful in estimating demand for these types of equip-
ment. Since data on income and number of farms by value of sales
class can be used as a proxy for income distributed by farm size, this
type of data is useful in estimating demand for inputs which are used

on different sized farms. The experience of a farm equipment

139
manufacturer also explains the relative importance of cash receipts as
compared to realized or total net farm income in estimating demand for
farm equipment. This firm found that most farmers pay for purchases of
farm equipment out of the gross farm receipts and in reality the farmer
does not generally distinguish between current and capital expenses. A
further finding by the firm was that farm operators tend not to pay for
farm inputs with off-farm income. So data on off-farm income are not
as important in estimating demand.

Three main types of farm input firm decisions are affected by
farm income estimates: 1) production scheduling and inventory manage-
ment, 2) marketing, advertising and sales strategies, and 3) financial
planning for the firm. Production scheduling is often influenced the
most by changes in estimated demand brought on income changes. If the
firm already has an idea of its actual market share and desired market
share and its actual and desired inventory levels, in the short run,
most changes in demand will first affect production scheduling. Over a
slightly longer period, changes in income may lead to changes in market-
ing strategies which can be planned to account for these income induced
changes in demand. Finally and to a much smaller extent, farm income
data might influence investment decisions with regard to expansion of
plant and equipment for manufacturing farm inputs.

In the banking industry the uses of farm income are at the same
time both more and less direct than in the farm input supply industry.
Indirectly these data are used as an indicator of the economic health
of the farm sector and as such are useful in decisions regarding the
allocation of funds of the bank between agridultural and other types of

loans. In this same manner the ability of farmers to repay outstanding

140
loans plus some notion of the future credit needs of farmers can be an-
ticipated by monitoring farm income estimates. However, since most
banks do not have truly national markets, aggregate farm income data
are only useful in giving general indications. At the national level
institutions interested in farm credit policy, such as the Federal Re-
serve Board and the Farm Credit Administration, make similar uses of
the data. The aggregate national level data are also moreimportant
for these agencies than any specific bank for obvious reasons. Thus,
for estimating the debt financing capabilities of the agricultural sec-
tor or future loan volumes, aggregate farm income data have more of a
direct use for these national level users.

Larger banks or the "money center" banks often make a more di-
rect use of the farm income data. The contribution of farm income to
total gross national product and farm inventories to business invento-
ries are of concern to these banks. This arises because of the need to
anticipate reactions in bond markets. The bond markets respond very
directly to changes in key economic indicators. Movements in personal
income tend to be one of the critical variables and farm income is a
major component of the personal income data series. Using monthly cash
receipts estimates, a projection of the contribution of farm income to
the monthly personal income estimates of the U.S. Department of Commerce
can be made. Using these along with other data, an estimate of personal
income can be made which allows the bank to anticipate the reactions of
other banks in the bond market and thus make better decisions about the
purchase and sale of bonds. It is also interesting to note that very
little difficulty is encountered in integrating the fann income data

into the national income accounting format used in the personal income

141
accounts. Uses similar to those described in estimating the contribu-
tion of farm income to personal income arise because changes in farm
income and farm inventories can often mask changes in the non-farm com-
ponent of GNP and in business inventories. These latter two data ser-
ies are important indicators in analyzing the business cycle which con-

sequently impact on many economic decisions.

4.4 THE FARM INCOME INFORMATION SYSTEM

Figure 4-2 is a schematic representation of the major components
of the aggregate fann income information system. The following sections
will describe each of the subsystems which make up the overall system.
While these will describe the important individuals and organizations
involved in the system, it must be kept in mind that the importance of
farm income data in decisions varies to a large extent. The high level
of nonuse of the data among those receiving the data through the mail
from USDA coupled with the low weight often applied to these data in

public policy decisions should be considered in evaluating this system.

4.4.1 THE PRIMARY DATA SUBSYSTEM

The primary data subsystem of the farm income information arises
because estimates of farm income, per se, are not obtained by sampling.
Instead these estimates of aggregate farm income are built up from pri-
mary data consisting, for the most part, of price and quantity data for
farm inputs and outputs.

The Statistics Branch of the Economics, Statistics, and Cooper-
atives Service (ESCS) in USDA is the principal source of primary data

for estimating farm income. For estimating livestock receipts the

142

 

 

 

5.53 22253“...— ES... 52.. N... -.. $52..

 

 

 

 

 

 

 

 

      

 

   
      

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

33.37.? No TSPGNN

 

     
 

 

 

 

 

 

 

 

 

132:2; 3N:—

 

 

 

 

_ ..ufio a mumm—

 

 

nzoé NNNLNNE .N-.3.=u...N< F

i

 

 

_

 

3.8.532..-

N.... .3353
...N—3.60....-
mN—NN N..N 9.29... N...-

 

co..Ne..m. NcasoN

 

 

NE... .233 3...: EC

 

 

22.23.; 9.5.5. Nay—N.NN: _

xuhmamam N335?

— 55593 gar—325:8”. 93 9::580...

.3 ..N 9.22.3.8
9.3.... .2328.
€2.93 5...... $95
«:33 En“.

 

3:9... 322:3
«.33.... 3959.3

 

 

8m:
m .m» — u..<
u .535 No 323

 

_

 

3...“. No.3 .0
03!. .3 cogs—.539

5322.2. 3.8 9...
No 8.3... .23.:

09.2.”. .9352: .02

32.2.3
8.339... 5.3...

333?. ..NN”.

033...
EN». «0.. .33.

93o... .63.. No:
N..N N33 N322...

 

.553 _ 38.25
c2332. v.3.- NGNLN 2.2.5”. .I.
3.5.532..- .cuuom 83.... .5: No 3?..5
£33.. :3.- _
N..N...»an- 9.3.3:: N..N 9.255 .833 3N3... 2.2.3.3....
mum... ..NN... .225 SN:
. 2:33.
N..N 3.5.6 .283
E N N moo—to 3N2... Neon-.33.. ...N—..N _
N .N ..NN .
. 285352..
33:55”. 32.5 .035 3.3—:33
N .NgoENoLNcou
5...: II.
$3.92. $3.25 ...N; _ goo—«.5
3.3—«3.559 39.9.8 .0 ENE: 3.3—3...:
1 3.333... I
.:3... 2.3... E:
3:... N32. 3:...
.23 ...u ...N< 2853. 3....
<mu _ , gm:
N.N...
2.3.3....
w... N0
81.5 2.2336 N NE...
...N—3.3“.
Ne..Nx
.3333 _ 39......
33.533... .58 .N< No .33.... _
N
37.39.92... .32 ”may“

 

:2:
gut. 52... mumu

 

 

II. A

 

3 ..o.umm..w...89
N...

 

 

 

.53. ..NN-8 .8...
i=8 3.53 N.N—.5:
.53 Efzuuéagaz

no» Sofas.
..N... a: .33 ...N.
55......

 

 

 

2.8.... 5.2.8:
momma—u 3.3

:o—uaawguma
N35 .6 .3333...

flow-=5 mamcwu

 

 

 

3.523.. «3.3...
..NN.N a.N. .o N=.N.
3mm - 3.5.53

 

 

 

«N.N»NNN N:..u .Nz
wogau.c=w xm
co.uu=vo.a NENN
co.ou=ua.a
mw_.o»=w>=.
mwo_.m

mum“ - 333:;

 

 

 

 

      

S.E.-h... acme-:33
33...”... ..N... 8N

 

3.3 233 .83..
3:32.
a .3: .3533

“go: .N..Nz - NNN

 

 

 

 

£3333 :3 55:5.

143
Statistics Branch provides production, disposition, and income reports
as well as monthly marketing percentages to be used in distributing
cash receipts over the year. The Statistics Branch provides somewhat
more limited data on slaughter and feeder cattle sales. In the area of
crop receipts the data from the Statistics Branch are clearly preemi-
nent. Monthly prices for major crops by states and season average
prices for minor crops and minor states are provided. State estimates
of monthly marketing percentages and quantity marketed are provided for
major field cr0ps.

0n the expense side, data are provided by the Statistics Branch
on monthly prices paid for production items, interest, taxes and wages.
The annual Farm Production Expenditure Survey data are important, parti-
cularly for minor items in the expense accounts. Data on fertilizer
consumption and purchases of livestock from inshipment estimates are
also provided. Inventory quantity estimates for both livestock and
crops are also obtained primarily from the Statistics Branch.

The Census of Agriculture provides important benchmark data for
estimating production expenses. 0f the approximately 50 separate ex-
pense accounts estimated, most are extrapolations of benchmarks provided
by the Census Bureau. Major accounts such as purchased feed and seed,
fertilizer, fuel and oil, custom work and cash wages all are based on
Census data. Benchmark distributions of farms by value of sales class
are obtained from the Census of Agriculture and then adjusted each year
depending on changes in prices and quantities of sales. The data on
value of home consumption and non-farm income are extrapolations from
Census data also.

The Agricultural Stabilization and Conservation Service (ASCS)

144
provides monthly data on direct government payments which enter, with-
out major changes, into the published government payment series. Com-
modity Credit Corporation (CCC) loans and redemptions are also reported
by ASCS and these are added in or deducted from the cash receipts esti-
mates for those crops with CCC loan programs.

The Economics Branch of ESCS also provides some data on the ren-
tal value of farm dwellings and the value of farm buildings. This lat-
ter item is important for adjusting estimates of depreciation and re-
pairs of capital items. 0n the expense side the Economics Branch also
provides estimates of taxes and interest.

The Market News Branch of the Agricultural Marketing Service
(AMS) in USDA provides data that are used primarily in estimating live-
stock cash receipts. Data on slaughter numbers, the number of head
shipped to public stockyards, placement numbers and weights from seven
major feeder states are developed from AMS data.

Private trade associations such as the Farm Implement Equipment
Institute, the Limestone Institute and the American Feed Manufacturers
Association are sources of data on capital expenditures on farm machin-
ery, limestone expenses, and feed purchased, among other items. Private
data also provide check data for the estimates developed in the ESCS
Farm Income Unit.

The Internal Revenue Service (IRS) produces data on compensation
to corporate officers and corporate profits which are used by the Bureau
of Economic Analysis (BEA). This agency uses the IRS data to adjust
USDA farm income estimates to get farm pr0prietors' income.

Much of the data collected in the primary data subsystem are not

designed explicitly or solely for the income estimates. Thus, the

145
concepts used in the design of these data can cause problems which re-
verberate through the system. One of the less apparent reasons that
cash receipts data are used more often than many other components might
be that the price concept used to estimate the cash receipts is expli-
citly designed to yield aggregate income estimates. The concept used in
ESCS: Statistics estimates of prices received by farmers is "That of a
price which, if multiplied by the total quantity of the commodity sold,
would give the total amount received by all farmers for that commodity"
(USDA 1975).

Many of the difficulties in the timing of the release of primary
data used in farm income estimates alluded to in earlier studies (Hil-
dreth, gt, 31,, Weeks, £3, 31,) can be viewed as measurement problems
in the primary data subsystem. Since estimates of farm income are built
up from primary data on prices and quantities rather than direct mea-
surement of income, difficulties in measuring the primary data cause
problems in the current system analogous to those which would arise in
attempting to operationalize a concept of farm income under a system
where income is measured directly. Thus, problems associated with the
concepts, operationalization of concepts and the measurement of primary
data then will tend to reduce the accuracy of the farm income data in
the same manner as problems in the operationalization of a concept of

farm income that is measured more directly.

4.4.2 THE FORMATTING AND COMMUNICATIONS SUBSYSTEM
The formatting of this primary data into a form that has meaning
and can be used as an input into the analysis subsystem is done primar-

ily through the Farm Income Unit in ESCS. The various accounting rules

146
and relationships among the primary data are somewhat analogous to sta-
tistical sampling procedures in other data systems. In this latter
case statistical procedures provide rules for gathering individual
cases into a single summary statistic. Accounting rules and relation-
ships also provide a way for summarizing the primary data into a single
or limited number of income measures.

The Bureau of Economic Analysis uses the farm income data sup-
plied by USDA but has a slightly different set of accounting rules and
uses some additional primary data to arrive at a slightly different
measure of farm income. Even though the underlying concepts of farm
income used by USDA and BEA are different, the resulting income esti-
mates are not essentially different. The reason for the similarity in
measures arises in the Operationalization of these different income con-
cepts. Since both agencies basically rely on the same primary data,
differences in concepts cannot be easily operationalized. For instance,
BEA attempts to measure a value added concept. Under this concept data
are needed on total value of production to get fann output. Since these
data do not exist the value of sales or receipts is substituted for
value of production. This in turn forces the measurement of income or
actually the formatting to be similar for the different concepts of farm
income.

Through government publications, particularly in the USDA's

Farm Income StatiStics and Agricultural Outlodk, the data on farm income

 

are then transmitted to analysts or directly to decision makers who can
use the data. Some nongovernment groups also provide a communication
role in the system. The news media, primarily newspapers and trade pub-

lications receive the data and pass them along to other potential users.

147
Educational institutions also have a communication function. Teaching
and extension education programs often are used to pass the data along
to others as well as to provide some additional analysis on the farm in-
come data.

The BEA publication Survey of Current Business presents their

 

estimates of farm output and gross farm product. However these data are
not used to any great extent by agriculturally oriented decision makers.

Most of these individuals seem to rely on USDA data.

4.4.3 THE ANALYSIS SUBSYSTEM

The analysis function in the farm income information system is
performed by many different firms and organizations. The Economics
Branch of ESCS remains a major analyst of data on farm income which are
used primarily in the public policy process. Indications are that the
USDA analysis with regard to farm income for policy decisions is not as
prominent as it might have been earlier. The increased ability of Con-
gress to do policy analysis has perhaps reduced the role of USDA analy-
sis somewhat. Professionalization and increases in the size of Con-
gressional staffs have led to a greater capacity within Congress to do
analysis on farm income issues. The Congressional budget process has
also put new demands on data and the addition of the Congressional Bud-
get Office has provided another source of analysis.

Private consultants and interest group analysts also provide
important sources of data interpretation concerning farm income. While
the provision of information has traditionally been a function of inter-
est group lobbyists in the decision process, the use of private consul—

tants seems to be a more recent phenomena. Firms with econometric and

148
simulation models of the agricultural sector and the entire economy seem
to be able to provide information on farm income that is both timely and
pertinent to agricultural policy decisions. In the policy process these
consulting firms as well as the interest group analyses are used the
most in Congressional decision making and by those in the Executive Of-
fice of the President as opposed to USDA.

Educational institutions also provide analysis that is used in
the public decision process but their analysis is in some ways different
from other analysts. Academic research often suffers from a timeliness
problem. It is difficult to use university analysis to a great extent
in actual policy deliberations because policy makers are not often aware
of ongoing research or when they are aware of this it is difficult to
get a rapid response which is often critical. On line computer records
of research currently underway in land grant universities as provided
by the Current Research Information System (CRIS) might possibly help
cape with this problem. University research is perhaps most useful in
suggesting broader alternatives or on more longer run analysis of future
problems and programs or in analyzing the impacts of existing programs.
The maintenance of professional ties among government and academic ana-
lysts is critical in the communication of the results of analysis and
helps to foster the use of academic research in policy decisions.

In private sector decisions, analysis by private consultants and
more internal or in-house analyses are prevalent, although most finns
seem to rely in part on USDA analysis. Banking firms might rely on some
aggregate analysis done by the Federal Reserve Board or Farm Credit Ad-
ministration but insofar as certain kinds of information have returns

to uses that are almost fully appropriable by the firm, these firms are

I49

willing to do their own analysis in these areas. In most of the pri-
vate sector uses of data it is more difficult to separate the analysis
and decision making functions because they are both undertaken within
the firm. Thus, separate organizations do not exist for identification.

The goals of firms in the private sector tend to be more easily
identified than goals in many public policy areas so the analytical
methods for dealing with farm income issues are probably more precise
in private uses than in public uses of farm income data. For example,
the uses of cash receipts or income data in econometric modeling seemed
to be much more prevalent in those firms estimating the demand fer a
product than in the public policy uses of the data. Thus, the types of
analysis would appear to be more easily identified in the private sec-

tor.

4.4.4 THE DECISION MAKING SUBSYSTEM

The types of decisions on which farm income data impact can be
classified into a rather small number of categories. Farm income data
impact on public policy in agriculture but these are rarely key or cri-
tical data because of the way policy is formulated. As long as the
critical questions in agricultural policy do not center on the desira-
bility of government intervention in agriculture then the crucial income
data needs will be for more disaggregated measures, particularly distri-
butional data, rather than for aggregate measures.

The central actors in the public policy decision process in
agriculture are in three main areas. First, in the Executive Branch,
the Secretary of Agriculture and others in the Department are major

actors and users of farm income data in this process. The important

150
economic advisors in the Executive Office of the President also have in-
put into Presidential decision making. Agricultural economists in the
Office of Management and Budget, on the staff of the Council of Economic
Advisors, and on the White House staff itself are the most important
farm income data users and policy decision makers in this part of the
system.

The second area of users and decision makers in the agricultural
policy process are in Congress. In terms of their use of farm income
data, four committees are most important. The agricultural comnittees
in the House and Senate are users of the farm income data but more in
the justification of given policy alternatives rather than as indicators
for anticipating problems and formulating policies. The budget commit-
tees in both houses of Congress are also important and their uses are
somewhat similar to the agricultural committees. The apprOpriations
subcommittees for agriculture do not appear to be major farm income data
users. This arises because these committees only appropriate funds for
the commodity price and income policies of the Commodity Credit Corpora-
tion two years after the losses occur. These committees have no alter-
natives in their ex poste decisions, thus their influence over current
farm legislation is not as important as some other committees.

Agricultural and other interest groups also influence farm poli-
cy. The uses of farm income data by these groups varies somewhat, with
the traditional agricultural interests concerned for the most part with
commodity specific data. The way in which agricultural legislation is
written leads to the concern for commodity oriented data. Even those
interest groups outside of the traditional agricultural establishment

are concerned more with the distribution of farm income and program

151
benefits than aggregate income. One might have thought that these groups
would be questioning the appropriateness of government intervention in
agriculture rather than expressing concern over which farmers benefit
from the farm programs. Their attitudes might be explained by the na-
ture of the public decision process, i.e., decisions tend to be incre-
mental except in times of major crisis.

At the state level the public policy uses were not as easily
identified. Many of those receiving data merely act as a data source
for others. However, one significant use of the data appears to be in
estimating tax receipts for the state income and sales taxes. Some
states also use econometric models in their forecasting of general tax
fund receipts.

Two major private sector end uses of the data were found. First,
many farm input supply firms use cash receipts and farm income data to
estimate the demand for farm inputs. In the short run these demand es-
timates affect production scheduling within the firm and then such areas
as marketing and advertising strategies. To a lesser extent decisions
on plant expansion or other investment within the firm might be influ-
enced by farm income.

Among banking and financing firms, fann income data in a more
general way affect the firms' estimates of the repayment ability of far-
mers and thus affect decisibns on the allocation of funds between agri-
cultural loans and other loans. Banks with a high percentage of farm
loans might also use the data to obtain an indication of future loan
volume. Trends in farm income also seem to be correlated to some extent
with farm land values which also enter into financing decisions on farm

mortgages. Investments by banks are also affected by farm income data.

152
First, the farm income projections give some idea as to the profitabil-
ity of investments in farm input firms. Second, the effects of aggre-
gate economic statistics on bond markets makes fanm income data impor-

tant in that these are a component of these other aggregates.

4.5 SUMMARY

The farm income information system has four major components.
First, the primary data subsystem is composed of those public and pri-
vate data gathering activities which provide the price and quantity
data used to estimate the various farm income data series. Second,
the formatting and communication subsystem uses a set of accounting
rules and economic relationships to create income statistics from the
primary data which then can be passed on to analysts and decision
makers. Third, the analysis subsystem is comprised of the public and
private organizations and firms which transform the farm income data
into information for the ultimate users. Finally, the decision making
subsystem includes the decision makers and decisions that rely on farm
income information. Public policy uses dominate in this subsvstem but
significant private decisions on the demand for farm inputs and the
credit needs of farmers also are apparent. However, any discussion of
the importance or value of farm income information should be tempered
by the survey findings which indicate a high level of nonuse of the
existing data by those who receive it. Possible explanations for this
nonuse and its implications will be the subject offa major portion of

the subsequent chapters.

CHAPTER 5
EVALUATION OF THE SYSTEM AND IMPLICATIONS

5.1 INTRODUCTION

Information systems, as part of social systems, tend to be
characterized by problems which are ill-structured. The farm income
information system is not an exception. A basic dilemma in dealing with
any ill-structured problem is the imperfect knowledge which character—
izes the problem itself. So solving the problems of the system are pre-
cluded by the absence of knowledge about the exact nature of the pro-
blems. Out of necessity, a major portion of the time and effort in
evaluating the farm income data system will be spent in trying to de-
fine the nature of problems in the system.

In this way, the subject of this research differs from earlier
studies in that by examining the nature of problems, rather than assum-
ing a set of problems or purposes, it is possible to establish priori-
ties for data improvements to meet the needs of decision makers. This
chapter will concentrate on defining the nature of the problems in the
farm income information system in order to obtain insights into the ap-
propriate redesign of the system and will be divided into seven sec-
tions. First, the question of priorities among users is addressed.
This is followed by sections on the problems associated with conceptual
obsolescence, credibility, national income accounting, and the feasibi-
lity of changing the system. Finally, an assessment of the usefulness

153

154

of the methodology used in evaluating this system is presented.

5.2 DESIGN FOR WHOM

The importance of farm income data in public policy uses as
noted by the number of those surveyed who cited this as a primary use
points to the role of government in supplying data on farm income. As
was noted earlier, data collection and analysis for public policy uses
is by definition an appropriate government activity. Since there are
also some private sector uses of the farm income data, questions arise
concerning the tradeoffs between designing the system to meet public
versus private sector needs.

In atomistic markets the social returns to improved data for
private use often appear to justify assigning a high priority to pri-
vate sector needs in designing an information system. Much of the pro-
duction data and some of the price data supplied by USDA seem to have
taken into account the private sector needs in designing these data
systems. The fact that farmers are perceived to be the main benefici-
aries and users of these data would also attest to the appropriateness
of designing production and price data for private users since the mar-
ket structure in agriculture would probably prevent farmers from or-
ganizing to supply the data themselves.

For farm income data, the survey results seem to pinpoint a
different set of private sector users distinct from actual farmers or
ranchers. Farm input suppliers are a principal private sector user of
farm income data rather than the farmers themselves. This has impor-
tant implications for the design of a farm income data system. The

market structure among the major farm input supply industries is quite

155
different from the market structure of farming itself.

0f the 23 firms that were identified by the mail survey as farm
input suppliers, almost 57 percent were manufacturers or suppliers of
farm machinery and equipment. In a USDA report on the structure of the
farm input industries, Strickler noted that seven firms were responsi-
ble for almost two-thirds of all sales of farm machinery. The ferti-
lizer and pesticide industries accounted for about 30 percent of the
respondents. These industries seem to be best considered together since
the majority of the farm input firms reporting use of farm income data
in the pesticide industry were also in the fertilizer manufacturing in-
dustry. This finding is consistent with the findings of Duane A. Paul,
gt, al, (p. 13), who reported that 76 percent of the anhydrous ammonia
producers or their parent firms produce chemicals and allied products
in addition to fertilizer. The diversity in the fertilizer industry
prevents one from estimating exact levels of concentration in this in-
dustry. However, Paul, gt, 31,, did summarize their findings by sug-
gesting that the fertilizer industry is relatively concentrated and
this concentration is increasing due to horizontal and vertical inte-
gration and certain barriers to entry. Both of the cited studies taken
together would seem to indicate that overall, the farm input manufac-
turing industries which use farm income data are relatively concen-
trated.

Given the theoretical relationships between market structure
and government provision of information develOped in Chapter 2, the
appropriate recommendation concerning the design of a farm income sys-
tem would be to emphasize the needs of public policy decision makers

relative to those private sector decision makers in the farm input

156
industries. Excluding the communication function performed by educa-
tional institutions, their main use of farm income data was also in the
area of public policy research. Thus, the needs of academic users of
farm income data would seem to coincide generally with government and
other public policy users.

The other major private sector users of the farm income data
were in the banking and financing industry. Of the 22 firms or organi-
zations in this industry which reported using farm income data, 41 per-
cent were institutions under the direction of the Farm Credit Adminis-
tration. Thus, insofar as government credit policies toward agricul-
ture are carried out through this agency the usefulness of the data in
banking and financing industries is related to the usability of the
data in public policy making in general. Based on the mail survey,
local banks make up less than 30 percent of the users of farm income
data in this industry, while large money center banks and investment
firms account for nearly 25 percent of this category.

All in all it would seem that the highest priority for the re-
design of the farm income information system should be given to meet-
ing the needs of public policy users of this data with private sector
users receiving somewhat lower priority. This is not as important
a distinction as it appears since many of the data needs of public po-
licy makers were echoed by the private sector users of the data. These
areas of overlap will be brought out, where appropriate in the follow-
ing sections of this chapter.

In order to approximate an intensity of use of farm income

data, a question was asked on the mail questionnaire to assess the re-

157

spondent's willingness to pay for USDA provided farm income data.l/
The results from this question can only be used to make comparisons
between the private sector users of the data since a substantial por-
tion of the government users were unable to place a value on the farm
income data or felt that because of their association with the govern-
ment that it was inappropriate for them to pay for the data. Table
5-1 summarizes the results of this question categorized by user group.

Based on willingness to pay for farm income data, the farm in-
Put supply users seem to be more intense users of the data than the
banking and financing users. The proportion of banking and financing
users that are willing to pay nothing for the data is significantly
greater at the 10 percent level than the prOportion of farm input sup-
pliers who would pay nothing. At the same time the proportion of the
farm input suppliers willing to pay more than $75 per year for the data
was significantly greater at the 5 percent level than that prOportion
willing to pay the same in the banking and financing user group. A
possible conclusion from this is that among private sector users farm
input suppliers should be given a higher priority since their demand
for the data as measured by willingness to pay might be greater.

These results do not hold when a comparison of the primary uses
of data are made with the willingness of the user to pay for the data.
Table 5-2 shows these results of this crosstabulation. The same limi-
tations mentioned earlier also hold in this case. However, when com-

paring the willingness to pay between those who use the data to estimate

1/ See Question 3 Appendix A.

1553

 

 

 

 

 

 

 

 

cm _ c _ o p N m N. gazm=< oz
3 a m N m m c m : 33 Ammewomcmm
“caucus sea—cu o.oo_ o.cop c.co— o.oo— o.oo_ o.°o_ o.oop o.coF c.oc_
Hemogma so: o.oo— ..cw m.e ..e v.m a... N... o.e_ m.m~ .euop
mecwaaacm as. wm m _— m. pm —~ mw we csapou
ou:_omn«
accuse; ce=_ou ..mw ~.cp o.cm ¢.om ~.m~ c.ap c.a_ o.om m.m sumo go
“cause; 3cm o.oo— m.m_ m.m m.m o.m 0.x o.m m.mm m.m m=~o> a mum—a
35:3: .3 m e e a w e 3 e 3 use...
muapoma<
unease; ce:_ou m.m m.m o.m~ ..m o m.m m.mn o.e e.~
acoogom so: o.cc— e.g. m.—_ a.m o m... ~.pq ¢.m a.m mm»
zucmacogm s- m w _ o N n p _ cash smuomcw
mu=_omn<
acougua :e:_ou «.mm n.mm o.m~ m.- «.mn ..mm m.mm o.c— «.me
“caucus 30¢ c.cop m.o~ v.m ..m m.« o.mp m._— m.c m.om mka
xocozaocu an N. N m m m s e m— cash mmoa
ous—omo< .1
acoucam ca:_ou «.mm ~._o a m.- o.ce «.mm m.e_ o.o~ ~.mo
acougoa 35¢ o.co— m.m~ o m.m m.o— _.~_ ~.m o.m “.mm cemN
35:3: an m. o m c A m m. 2
ous—omn<
. so; cu
page» maoocup acoEecm>ou u_uwz u:ou.:m=ou mc_u:e:.u apaaam u:u§:cm>oc copuau.um=~ mmwcm:——..:
2e: -.oum.x pauea mxmz muu>.cm agfixzem usa:_ pocoumu .ocepuauaum
so macaw Esau com:
mum: >m <h<a uzouz_ z¢<m com ><m op mmuzuz_aa_3 .-m mam<~

l59

 

 

 

 

 

 

 

on N c o N m cp N cmzm=< oz
acoucoa gas—cu c.cc_ c.co— c.oo_ c.cc— o.oo— o.co— c.oo— o.oo—
acoucom 3oz o.oo_ c.m N.m e.N o.mp ¢.mN ..N. e.m_ pouch
aucmacocm mN— e. o— m. an _e an em ass—cu
ous—cma<
«smegma ass—cu N.mN e._N c.o— N.NN «.mm c.cm N.c— c.oN can: go w:_a>
ucwucma 3oz c.oo_ N.c N.N N.o ¢.¢N N.NN —.__ m.m_ a mun—a
accesses» me n — m pp mp m N cu m—acca
waspomn<
accuse; :E:_ou N.o c o N.N_ N.¢N N.N. o m.m
“gouge; 36: o.cc. o c m._— ..Nc v.aN o N.—_ mN»
senescent N. o o N m m o N cogu gwueogm
manpoma<
acaucmm cszpou N.NN m.N¢ c.om m.em N.¢N N.NN c.cm m.mN
accuse; soc c.oo— N.o_ m.m N.o_ c.m— o.m_ v.mN o.m_ mN»
aucoscmgm am e m o a _~ m— N can» mmmd
manpOmn<
“cause; gas—cu m.cm N.mm o.ov N.N— N.N— e.¢N m.mn o.om
acougma 3oz c.oo— m.m o.N N.m ..pp m.N— m.N— e._m oEoN
aucoacmcm on m e N m o— o— N—
ous—cma<
xeg ca
~ou¢h manage. co.uu:pa> uNuogu was :ONuaewumm xuwpom mgmnuo ca :orua5comc~ mmocmc——__:
30¢ -.mum_z use; o:.ucoc.u tease: u—paam go «we; pucucww
mm: >¢<z_¢m >m <h<c mzouz_ :¢<m can ><¢ oh mmmz¢z_44~z N-m Ngm<h

160
demand and those who use it in making credit decisions these uses do
not seem to be significantly different. As one might expect, the demand
for the data between those that use the data for general information or
pass it on to others seems to be less than those who use it for demand
estimation or credit decisions; comparing both the proportions of those
willing to pay zero and those willing to pay more than $75 between these
two groups shows the above relationship to be significant at the 1 per-
cent level.

Even with the apparent difficulty that public policy users had
in placing a value on government provided data, a comparison of those
who use data for general information or pass it on to others with the
public policy users suggests that these latter users have a somewhat
higher demand for the data. Comparing the combined proportion of those
in the general information category and those who provide a communica-
tion function with the proportion of public policy users the former
group has a significantly greater proportion at the 5 percent level
who are willing to pay zero for the data and the latter group has a
significantly greater prOportion at the 5 percent level willing to pay
more than $75 per year for the data. As expected, at least when com-
paring the major uses of the data in terms of number of users, those
who use the data in specific types of decisions seem to have a higher
demand for the data when measured by willingness to pay than those who
use it for general information or pass it along to others. Thus, if the
demand for data by specific users can be utilized to establish priori-
ties in information system design, then those who make a specific use of
the data should presumably be given a higher priority than the more gen-

eral or nonspecific users of the data.

161

Among public policy uses this willingness to pay criteria for
determining the level of demand for data offers an interesting rela-
tionship among state government users. Of the three users in this cate-
gory reporting a use of farm income data in tax and revenue planning,
two were willing to pay over $500 per year for the data and one did
not answer the question. While this is a relatively small sample it
does appear suggestive of the importance of this use relative to some

others in the system.

5.3 CONCEPTUAL OBSOLESCENCE

Conceptual obsolescence is a problem with many Operating infor-
mation systems, the farm income information system is not an exception.
Understanding the nature of this problem of conceptual obsolescence is
critical for making improvements in the farm income information system.
Conceptual obsolescence can be of two types. First, a data concept is
obsolete when it fails to represent adequately the reality of the sit-
uation for which it is designed. A data concept is also obsolete when
it fails to meet the needs of decision makers as the issues facing the
decision makers change. Hence, data concepts can become obsolete when
either reality changes (actually our perceptions of reality) or when
the agenda facing decision makers changes.

These two types of conceptual obsolescence are both evident in
the farm income information system. This section will identify the root
causes of this problem and will also attempt to assess the severity of
this difficulty as it relates to the functioning of the information sys-

tem.

162

5.3.1 POLICY AGENDA CHANGES

Many of the agricultural policy decision makers interviewed
stated that aggregate farm income was "meaningless." However, in most
of these cases, it appeared that "meaningless" was more of a code word
for "not usable in the type of decisions that I make." This arises for
a number of reasons, predominant among these is the fact that most of
the decisions which concern policy makers are in reality equity issues.
Aggregate farm income data are not very useful in addressing distribu-
tional issues. As one Congressional aide argued, aggregate data are
not that useful because it is difficult to tell which farmers you are
discussing and hence the use of aggregate data could tend to distort
the policy. For instance, it might be true that farm income is adequate
for cotton producers but not for corn producers. The amount of consen-
sus on this issue was almost unanimous. Given the more parochial views
often found in Congress, one might expect a certain desire by them for
data that provide clear ties to an identifiable group of farmers. How-
ever, this concern for better distributional data was expressed with
nearly the same force by decision makers in the Executive Branch and
among the leaders of various interest groups. Even the limited number
of private decision makers interviewed listed better distributional
data as an important data need. For instance, banking firms found com-
modity oriented data very critical since this tended to give the best
indication of the repayment ability of certain types of farmers. Input
suppliers noted the importance of distributions of income or receipts
by size of farm and commodity.

Perhaps the most important reason for the need for better dis-

tributional and disaggregated data occurs because of the types of

163
questions raised in agricultural policy. The interview process sug-
gested that there exists an overriding ethic of farm politics, that
being the responsibility of the U.S. government to maintain farm income.
As was mentioned earlier, when the aggregate farm income accounting sys-
tem was originally designed in the 19205, the data were used to point
out the need for and to support the role of government intervention in
agriculture because of the plight of farmers relative to other sectors.
Rarely is the question posed at the present time as to why government
should maintain fann income. Only the question of how is raised. Thus,
the need for aggregate sector level performance measures, like farm in-
come, are not needed to justify farm programs.

The fact that the data is not used to justify farm programs has
important implications concerning the recommendations of the earlier
task forces on farm income. These earlier studies stressed the need
for comparability among different sectors of the economy. The 1972
Task Force even listed comparability across sectors as one of the prin-
cipal purposes of a farm income accounting system. According to deci-
sion makers interviewed, comparability between sectors is not as impor-
tant a consideration as comparability between groups within the farm
sector. Since there is very little debate on the justification of
farm programs in general, only the costs of programs receive serious
debate, comparisons between the farming and other sectors are not as
important to current decision makers. Instead the distribution of
costs and benefits within the sector become the important decision
criteria. This gives rise to an inordinate amount of interest for data
by size and type of farm. Since the agricultural legislation is writ-

ten on a commodity by commodity basis, commodity specific data such as

164

relative profitability or returns to resources by commodity were often
suggested as gaps in the current data.

Comparability between sectors is one of the principal arguments
in favor of national income accounting formats, however, the ability to
make comparisons among sectors is not limited to the case where the na-
tional income accounts data format exists. For most types of analysis
the need to have a greater ability to combine microdata files from
different sectors is more critical. For instance, the importance of
the inflation issue in economic policy implies that price data from
different sectors need to be comparable in order to obtain information
on aggregate inflation and determine the impact of price increases in
one sector on the remainder of the economy. The growing internal in-
terdependence of economic sectors of the U.S. economy, indeed of the
world economy, generates greater need for intersectoral comparability
of data.

A more serious problem suggested by this type of conceptual
obsolescence has been pointed out in the interview process. The current
fann income data are really misleading in that they do not account for
many distributional aspects of the situation. Since the farm income
data do not take into account these distributional issues, many pro-
blems brought on by skewed income distribution, such as concentration
of land ownership, are not identified and dealt with in the farm legis-
lation. The picture of the agricultural sector held by most non-agri-
culturists active in the decision process is shaped by how the statis-
tical system is set up. Since the distribution aspects are neglected
in the farm income data system, many data users do not obtain the true

picture of the sector and hence important policy areas are overlooked.

165

Even those distributional aspects currently reported by USDA, i.e.,
distribution by value of sales class, are relegated to the back pages
of the annual summary. A

The statistics themselves may also lead to a policy which im-
plicitly assumes away distributional issues. A few decision makers
suggested that the reliance on aggregate data may lead to a policy
which assumes homogeneity between different farms. One noted that it
is rare that the difference between the livestock and cr0p subsectors
are dealt with in current policies. For instance, the loan rate for
corn might not be considered in conjunction with its effects on live-
stock production and the income of livestock producers. Another deci-
sion maker in the executive branch argued that the effects of homogene-
ous data and a policy which assumes homogeneous farms are most likely
reinforcing. 50 changes in the data will most likely be met by changes
in policy and vice versa.

When the early farm income data system was designed in the mid
19205 there was little call for distributional data. The farming sec-
tor was relatively homogeneous and the major farm programs similar to
current programs were not implemented until the 19305. Perhaps the na-
ture of those who designed the original system might explain some of
the difficulties now encountered. The farm income information system
was designed by economists and agricultural economists who by virtue of
their training tend to deal with questions of efficiency. So the sys-
tem appears to have been designed more readily to answer efficiency
questions. Whether this was an explicit consideration is unknown, but
treating the U.S. farming sector as if it were a single national family

farm would certainly seem to make more traditional types of microeconomic

166
analysis easier to perform on the aggregate level. The conceptual obso-
lescence problem emerges because of the nature of the political deci-
sion process. If one accepts that the value of information is discov-
ered only in a decision making context and that farm income informa-
tion should be used in public policy decision making, then data on the
distribution of income are essential because the policy questions which
face public decision makers today revolve around equity issues. Thus,
while economists are often trained to deal primarily with issues of
allocative efficiency, many political issues involve equity considera-
tions. Problem solving decision making in this area requires informa-
tion on both equity and efficiency, to concentrate information produc-
tion in only one of these area will yield inadequate solutions to poli-
cy problems.

After recognizing the need for better distributional data on
farm income, consideration must be given to what types of distributions
are most useful. From the interviews it became apparent that data dis-
tributed by commodities or type of farm were important because of the
nature of agricultural legislation, which supports income through
price programs for specific commodities. The responses to the mail
questionnaire seem to bear this out also.

Table 5-3 presents the results to a question in the mail survey
which attempted to assess the priorities among different criteria for
designing a farm income information system. While the technique used
in asking this question produces an ordinal scale variable, mean scores
are calculated to provide a summary measure for each design question.
The calculation of means is not a statistically correct procedure for

ordinal variables, but is used for more conciseness in the presentation

167
of the results. Some comparisons between the means are made and the
nonparametric Mann-Whitney U test (Clark and Schkade, p. 390) is used
to estimate the statistical significance of these differences. However,
this makes the implicit assumption that these scales are the same for
all questions for each respondent. Consequently, these comparisons
must not be considered as analytically precise but rather as a general
indication of the results.

In comparing the mean scores of different criterion concerning
distributional data it is apparent that income distributed by commodity
is given a higher priority than income distributed by state or region.
Using the Mann-Whitney U test to compare these mean scores, commodity
or type of farm income data was given a significantly higher priority
by respondents at the 5 percent level. Furthermore, each of these types
of distributional data was accorded a significantly higher priority by
respondents at the 1 percent level when compared to data on the distri-
bution of income among individuals in the farming sector.

Congressional policy makers' interview responses perhaps gave
One reason for the lower priority given to data on the distribution of
income among individuals. The role of group interests in the American
political process is well documented. Congress tends to respond to
readily identifiable groups, among those suggested were by commodity,
by geographical region, or by size of farm. It is not surprising that
the distribution of income among individuals received less enthusiastic
support, since members of income groups are not usually a politically
distinct group. Also, the movement of farmers among income groups is
much more likely than movement of farmers from one geOgraphical region

to another or among different types of farms. Hence, these latter two

168
groups are more easily identified as a political group.

In summary, the policy agenda in agriculture has shifted to in-
clude many equity issues which are not easily addressed with current
data. This implies that the current data concepts are inadequate or
obsolete. Among the types of distributional data that currently appear
most important are the income differences among commodities or enter-

prises and disaggregations by states and regions and by size of farm.

5.3.2. CHANGES IN REALITY

Most decision makers also admitted that the aggregate farm in-
come concept was not an adequate representation of reality because it
implicitly assumes a homogeneous single national farm, i.e., current
farm income accounting does not paint a true picture of agriculture.
Thus, the type of conceptual obsolescence brought on by changes in re-
ality is also evident in the system. The two earlier Task Force Reports
stressed this type of conceptual obsolescence when discussing problems
in the existing system. However, the examination of how farm income
data are used suggests that this type of conceptual obsolescence may
not be as critical a problem when compared with the type alluded to in
the previous section. The way the data are used, in an ex poste justi-
fication of policies or in speechmaking, connotes a use of aggregate
farm income as more of an economic or social indicator rather than as an
explicit representation of the reality of the sector. There seems to be
a distinction between an economic or social indicator and a more micro-
economic statistic, such as a price, that is designed to represent re-
ality. While an indicator does have to be tied in general to reality,

ostensibly a more important consideration would be whether or not it

169

gave good indications of changes in the performance of the sector ra-
ther than measure the state of the sector at any one point in time. In
practical application this would imply that an indicator should measure
more closely changes in a variable from one period to the next instead
of the absolute level of that variable. The sensitivity of the meas-
ure to changes in the sector or to policies which affect the sector
would also seem to be an important aspect for designing an indicator.

Almost without fail those who were interviewed stated that the
change in aggregate farm income from one period to the next is more im-
portant than the absolute level of farm income at any point in time.
Those who were concerned about the level of farm income tended to ex-
press this concern in reference to some historical level. Analysis of
the mail survey results on the appropriate criteria for designing an
aggregate farm income accounting system yields findings which support
this concern for changes in income over the level of income. Questions
12 and 13 in Table 5-3 were included to test the hypothesis that the
measurement of changes in aggregate farm income should be given higher
priority over the measurement of absolute levels of income. A compar—
ison of the mean scores of these design criteria using a Mann-Whitney
U test found that respondents placed a significantly higher priority
at the 10 percent level on designing the system to estimate changes in
farm income rather than the absolute level.

The importance of sensitivity in an indicator is also suggested
by the interview findings. Many noted that aggregate net farm income
was not sensitive to changes in policy variables affecting the sector.
However, many individuals further stated that cash receipts data often

provided a reasonable but not totally acceptable substitute when data

I70

ms.a==c,umo=c _.ee _e=uua cc» < xeucoaa< mama

 

 

Nom.N Nue_—.soe age» we mecca. EcaN-NNo co Seneca: as“ m. “as: .m.
moo.m Nm_mea mmmm s cc so» voucaouua mace» No usage. oueomgmmu mgu m. gag: .N.
e_N.n Nm_mon fimmmmmm co co co» umuczouua mace» No usouc. ouommcmma as» m. ans: .m—
meo.N Ncouumm m:_sgou as“ e. vows—mg muomma new «sage. men go: .m—
coN.n Nx—_Eeu sLoN as» ou noun—as mam:_m:a see» mgu up to: .e.
ocN.n Nosoucv aces oueamcmma Na —m>o_ one m. ans: .m—
cwm.N Nmuovcoa we.“ mso+>wca EocN ummeazu usage. Eco» mo _m>m_ as» we: gqu 30: .N.
vao.N NLOuumm o:_ageu as» No macaueoe e.se=ouo cones use age “as: ..N
nem.N Nmeuuum co\u=o mco.mmc cooxuoa moucmemeu_u capes. age» as» use «an: .op
...NNN 3335.8 .5 5.3% No 3.3 3 39:232.. 2.3.: m5 9.... gen: .o
pmN.N Nocaupau.cae ca away—ssou mmuczommc ecu cu message any use can: .m
SNN 232: 22:2 2: co 52:83.... «5 3 2.5 .N
mmm.N Ncouuum m=.scou we“ ewes—3 m.n=u_>.u=_ meose peace. we :o_u:a.gum.u on» m. “as: .c
Noo.N Nmu——.Eau see» no o=_on ppm: oesocouo «cu m. «as: .m
amc.N Nassau hemuocoe c. gouuom meweLeN «ea mo oNNm mg» m. «as: .e
EN Nan—Nam Ea ca 2.85 ...5 a. :23 .m
mmm.N Naommmepmaa see» ya quu:. as» m. use: .N
mem.N Nmm_cum=e:_ cosao saw: oeuqecu menu—:u_cmu e? magnum; mgu on so: .—
uccum gee: Emumxm zm vommmcvu< on Oh meo.umo:a

 

.Aumozo— m. m ecu a»*ce_ga

ummcm—g m. — oguczv m cu — we o—aum a co zepoa team—p nee—“mono ecu No oucmucoqe_ may use: .mgmguo
co>o meovammac :_naeou cu co>.a on Nuvco.ca ween aega og_=aog ma:_ucumcou mucaemoc .go>ezez .m:o_umw=a
mo mmaxu u:ocou».u News coxmca cu uo=m_mou as :au soumxm u=_u:=ouue «sous. ago» muamocmmu —a:owuoc <

auco.umm=c .m:uu<

 

<~¢uh_¢u zmumuo ch Gaza—mm< mm—h—«O—gg z<uz Mpm u4m<h

171
sensitive to changes in the sector were needed. Many individuals attri-
buted the lack of sensitivity of the net farm income figures to the way
the data are estimated, especially production expenses. The reliance on
census benchmarks make this portion of the data less sensitive to quar-
terly and yearly changes in variables affecting the sector. Cash re-
ceipts data, on the other hand, are relatively responsive to changes
within sector and often these changes can be detected monthly. Thus,
while the sensitivity of the net income data may not be as high as de-
sired, the availability of data in the form of cash receipts that are
responsive tends to alleviate some of the difficulty. However, a pos-
sible implication of this is that the data on production expenses need
to be more responsive to short run changes in the sector. So production
expense data collected at more frequent intervals, perhaps with less
detail, might provide improvements in the sensitivity of the net farm
income to intrasectorial changes.

The use of aggregate farm income data as more of an economic
indicator is also borne out by the way decision makers appear to form
norms with regard to aggregate farm income. Most seem to develop a
notion of the economic health of the sector on the basis of information
from constituents or others concerned with agriculture. After obtain-
ing a subjective impression of well being, they turn to the measure of
aggregate farm income to get a quantified indication of sector welfare
that can be easily used in speechmaking or to give a quantified justi-
fication for a particular policy alternative.

A quite different method can also be hypothesized for formula-
ting norms about farm income. In this caSe a person would simply quan-

tify an adequate level of farm income for an individual farm and then

172
aggregate it up to a national level to get a standard measure of income
adequacy for the welfare of the sector. In this example it would seem
more useful to have a measure of farm income that is more clearly re-
lated to reality. This latter type of norm does not seem to exist to
any great extent for farm income. The relatively low mean score in
Table 5-3 for the design question, "What are the major economic features
of the farming sector?" would also seem to suggest that concern for data
which showed a one to one relationship with reality might not be as im-
portant as some other concerns.

One area where the current data may be inadequate even as indi-
cator is in the area of productivity measurement. The mean score for
the design criterion concerning productivity measurement showed this
receiving the highest priority among all others. The current data sys-
tem often does not provide data which are relevant to productivity
measurement because of the focus of the current system on cash receipts
rather than value of production in measuring gross income. The large
number of intermediate products produced on the same farm where used
causes most of the discrepancy between gross receipts and value of pro-
duction. To avoid counting increases in marketings as increases in
production, a measure of value of production might be needed. The high
priority accorded this area from the mail questionnaire is somewhat
surprising since this concern over productivity measurement was not ex-
pressed to any great extent in the personal interviews.

When asked whether the farm income accounting system should be
concerned with measuring the income of farm families or the income of
farm businesses, the respondents to the mail questionnaire seem to be

more concerned with the income of farm businesses. A comparison of

173
mean scores between criterion 2 and 3 in Table 5-3 shows that the indi-
viduals surveyed tend to give a higher priority to the income of farm
businesses. This difference is significant at the 5 percent level.
The fact that the measurement of nonfarm income of farm families was
assigned a relatively low priority by most respondents would also seem
to confirm this interest in farm businesses. Those interviewed eXpres-
sed similar concerns. It also suggests that the concept of a national
family farm might not be an adequate measure of reality and hence a
Problem of conceptual obsolescence could exist. But insofar as the
current concept mixes both concepts of the farm business and family at
least some information about farm businesses can be ascertained. A
further implication concerns the appropriate level of detail in the
current system, it might be important to insure that these farm busi-
ness and family aspects can be separated out. For instance, it might
be useful to provide estimates of net nonmoney income of farm operators
rather than gross nonmoney income as is now presented.

All in all it appears that the type of conceptual obsolescence
brought on by changes in the reality of the food and fiber industry as
compared to changes in the policy agenda might not be as serious in the
case of farm income because of the predominant use of farm income data
as an indicator. The obvious implication is that those aspects of
changing reality which affect the data's usefulness as an indicator
should be given a higher priority in making improvements in the system.
In many ways it is not possible to completely separate these two types
of conceptual obsolescence. Changes in the structure of agriculture
toward increased specialization and heterogeneity of farms has made the

question of equity within agriculture more of an issue. Hence, the

174

policy agenda has changed because the reality of the sector has changed.

5.4 CREDIBILITY

A second area of problems in the current farm income informa-
tion system concerns the issue of credibility of the U.S. Department of
Agriculture. A decision maker cannot know the exact value of any infor-
mation until it is received and used. So the principal way to judge
the value of information, before receiving it, is by the previous reli-
ability of the data source. One important means of assessing data ac-
curacy or reliability is by evaluating the methodology used in collect-
ing the data. Statistical sampling theory can be used to assess the
accuracy of certain types of data.“ However, in cases where there is
little documentation or understanding of data estimation procedures to
which the producing agency is committed then the reliability of the

source remains a critical variable in predicting the value of the data.

5.4.1 RELIABILITY OF USDA DATA

There seems to be a basic credibility problem in the farm in-
come information system. The methodology used to estimate the data on
farm income is very complex and not particularly well documented or
publicized. Even when the methodology is modified, very little notice
is given to the users of the data. This does little to foster the cre-
dibility of the data. 0f the 203 individuals who responded to the mail
questionnaire only 19 percent were familiar enough with the methodology
used to estimate farm income to cite some of the basic procedures used
in making the estimates. Thus, the main option left for an individual

is to estimate the reliability of the data by considering the previous

175
accuracy of USDA farm income estimates.

Normally, one would estimate the reliability of data after it
is received by collecting new data and comparing this sample to the
original data. For aggregate farm income data this is not an economi-
cally feasible alternative. So a principal way in which the reliability
of USDA farm income data is determined by a user is through examining
the number and size of revisions the data undergo after the first esti-
mates are published. In the early 19705 the number and size of the
revisions of the farm income estimates increased (Hildreth, 31, 31,).
This appears to have undermined the credibility of these data. The
emphasis which the USDA has historically placed on aggregate farm in-
come as the primary indicator of well being for the farm sector has
made the credibility problem even worse. With this in mind it is in-
teresting to conjecture as to what course agricultural policy would
have taken in recent years had the U.S. Department of Agriculture cho-
Sen in the past to place a much greater emphasis on data regarding the
increase in the value of assets in agriculture as taken from the Ba-
lance Sheet of Agriculture rather than farm income.

Strangely enough, this credibility problem does not seem to
have been manifested by a lack of use of historical data, although this
might be a partial explanation for the apparent lack of use of the farm
income data among those receiving it in the statistical bulletin, E3rm_
1ncome Statistics. Most individuals interviewed noted that they still
relied on USDA historical estimates of farm income, though some sug-
gested that their uses were much less critical than earlier and that
the frequency of revisions caused the data to be disregarded as a key

input in econometric models. But for the most part USDA historical

176
estimates of farm income are not seriously affected by this credibility
problem because no other historical estimates of aggregate farm income
exists. So users consider these to be the best estimates available,
since the accuracy of the data cannot be tested without incurring the
cost of collecting the data again. A second reason that the use of
whistorical data is not affected by revisions is that timeliness is a
key variable in the decision process. Most historical data arrives
too late to have any impact in and of itself on decisions about future
courses of action. So even historical data with very few revisions are
not used extensively in decision making.

The revisions of farm income data seem to influence the credi-
bility of the Department of Agriculture's analysis rather than data
collection. While there is only one source of historical data on farm
income there are numerous sources of forecasts in the private sector.
Congressional decision makers cited this lack of trust in USDA fore-
casts as a major reason for the choice by the Congressional Agricul-
tural and Budget Committees to subscribe to the agricultural models of
Data Resources Inc. and Chase Econometric Associates. Another Con-
gressional staffer noted that around Capitol Hill the USDA is believed
to have the facts, i.e., their historical published data is accepted
as fact. But he then went on to add that this is not so with their
analysis and other information. When the USDA might have a stake in
defeating a bill their analysis is much more suspect, so Congress re-
lies on its own or private analysis. One area cited was in the fore-
casts of farm income. USDA normally makes and publishes forecasts for
aggregate farm income for only a few quarters in advance. Forecasts

which are not widely published by USDA are generally viewed with more

177
suspicion by the users in Congress. In analyzing a farm bill which is
designed to operate for four or five years, the forecasts done by USDA
beyond one year in advance are not likely to be given as much credence
since these forecasts are not widely published, instead private sources
seem to be relied on heavily.

It is also interesting to note that the private consultants do
not stop using the historical estimates of farm income because of the
frequent revisions. These historical data must be regarded as fact
because without them the private consultants would have no benchmark
from which to make forecasts. Without some standard of truth there is
no way to judge the accuracy of the forecasts. In the case of aggre-
gate farm income the USDA historical estimates are accepted as this
standard of truth.‘

The apparent finding that this credibility problem affects the
forecasts and outlook work of USDA but not the historical estimates is
something of an anomaly since good historical data are usually neces-
sary to make forecasts. Insofar as revisions in historical data af-
fect forecasts of farm income, revisions do affect future policy. The
repeated revisions of the estimates also raise doubts as to the reli-
ability of the USDA farm income forecasts, so these forecasts are more
heavily discounted in making decisions. This credibility problem ex-
tends beyond the Congress. Members of the Executive Branch outside of

USDA expressed similar concerns as did some private data users.

5.4.2. ACCURACY
The mail survey attempted to obtain some estimates of the data

users' perceptions of the accuracy of the farm income data. Beforehand

178

it was hypothesized that the data users might attribute more accuracy
to the data than was possible to obtain, given the current methods of
estimation. The current methods calculate net farm income as a resi-
dual. This method of operationalizing the concept of farm income cre-
ates difficulties in making accurate estimates of net farm income. A
.simplified example perhaps will best illustrate this problem. Assume
that net income is only the difference between cash receipts and pro-
duction expenses. Between 1973 and 1976 cash receipts averaged about
90 billion dollars, while production expenses averaged about 75 billion,
so net farm income in this simple case would be about 15 billion. An
optimistic but not unreasonable assumption might be that cash receipts
and production expenses could be estimated to within 3 percent of the
true value. Assuming the estimates above are the "true“ figures, cash
receipts could be estimated to be 92.7 billion and production expenses
as 72.75 billion and still be within 3 percent of the true figures.
But in this case net farm income would be almost 20 billion dollars.
So while maintaining a 3 percent level of accuracy in cash receipts
and production expenses, the accuracy of net income is only within 33
percent, assuming that the errors in estimating the two components are
independent. Even assuming a one percent error in the components
could lead to an 11 percent error in the estimate of net income in the
above example. Thus, it would not seem at all unjustified to conjec-
ture that the USDA net farm income estimates are within 20-30 percent
of the "true" figure for net farm income.

The results in Table 5-4 show the mail survey respondents'
perceptions of the accuracy of the annual realized net farm income

estimates published by USDA. Of those who were able to assess a level

179

of accuracy over 75 percent felt that the data was almost always within
0 to 10 percent of the true value. The type of use or user did not
seem to affect these perceptions of accuracy. Perhaps those who use
the data have a somewhat unreasonable expectation with regard to the
accuracy of the farm income data. This might also explain the apparent
,negative reactions to large revisions in the data, in that the magni-
tude of the revision might cause the estimate of farm income to be out-

side the limits of error that the users typically incorporate into

their plans.

TABLE 5-4 USER PERCEPTIONS 0F ACCURACY OF
ANNUAL REALIZED NET FARM INCOME '

 

Relative Adjusted Rela-

 

Users Perceived Absolute Frequenc tive Frequency
Level of Accuracy _ Frequency (Percent (Percent)
0-5 Percent 27 . 10.0 21.6
6-10 Percent 69 25.6 55.2
11-20 Percent 23 8.5 18.4
21-30 Percent 3 1.1 2.4
Greater than 30 Percent 3 1.1 2.4

Subtotal 125 46.3 100.0
Don't know 55 20.4
No Answer 25 9.3
Do Not Use--No Answer 65 24.1

Total 270 100.0

 

180

It does appear that the perceptions of accuracy are based at
least in part on the magnitude of past revisions in the farm income
data. This was frequently cited as the basis for making a judgement on
accuracy, although most individuals did not give a reason for their es-
timate of inaccuracy. The perceptions of accuracy do seem to be in
.line with revisions of the data in the recent past. Table 5-5 shows
the absolute and relative magnitudes of recent revisions in the farm
income data. It shows that the average revisions of the annual re-
alized net farm income data average about 9 percent. This seems to be
quite close to the average perception of the accuracy noted in the mail

survey.

5.4.3 OTHER ASPECTS OF CREDIBILITY

While not a major factor in the apparent credibility problem of
USDA, the fact that multiple sources of data are published by USDA for
measuring similar phenomena.adds to this problem. Most often noted in
the interviews were the two sources of data for production expenses in
farming. Farm production expenditures are produced by the Statistics
Branch of ESCS and are based on a survey of farmers. These data are
indirectly used as a source of the expense accounts but are not perfect
substitutes. A few public decision makers noted that the presence of
multiple data sources from the same department tended to confuse some
issues and further reduced the credibility of USDA in doing analysis.

The fact that the Department of Commerce also publishes farm
income estimates does not seem to pose any major difficulties. Most
agricultural decision makers rely on the USDA estimates and do not use

those published by the Bureau of Economic Analysis. Also, most users

181

TABLE 5-5 REVISIONS AND ACCURACY OF RECENT USDA FARM INCOME ESTIMATES

 

Preliminary Most Recent Absolute Absolute

 

Year and Component Estimate* Estimate** Difference Percentage
(Billion $1) (Billion 51) Difference

1973

Cash Receipts from Fanning 83.4 87.1 3.7 4.4
Production Expenses 64.4 65.6 1.2 1.9
Farmers' Realized Net Income 26.1 ~29.9 3.8 14.6
Value of Inventory Change 0.8 3.4 2.6 325.0
Farmers' Total Net Income 26.9 33.3 6.4 23.8
1974

Cash Receipts from Farming 95.0 92.4 2.6 2.8
Production Expenses 74.8 72.2 2.6 3.6
Farmers' Realized Net Income 27.2 27.7 0.5 1.8
Value of Inventory Change 2.4 -1.6 4.0 166.7
Farmers' Total Net Income 29.6 26.1 3.5 13.4
1975

Cash Receipts from Farming 90.6 88.1 2.5 2.8
Production Expenses 75.5 75.9 0.4 0.5
Farmers' Realized Net Income 23.7 20.8 2.9 13.9
Value of Inventory Change 2.3 3.5 1.2 52.1
Farmers' Total Net Income 26.0 24.3 1.7 7.0
1976

Cash Receipts from Farming 94.8 94.3 0.5 0.5
Production Expenses 80.9 81.7 0.8 1.0
Farmers' Realized Net Income 23.3 21.9 1.4 6.4
Value of Inventory Change -1.3 -1.9 0.6 46.0
Farmers' Total Net Income 22.0 20.0 2.0 10.0

 

* Source: Farm Income Situation, USDA-ERS, PIS-223, February 1974, PIS-225,
February 197SYTAgricultural Outlook, USDA, April 1976, March 1977.

** Source: 53;? Income Statistics, USDA Statistical Bulletin No. 547, July

182
are aware that the BEA estimates of farm income are basically the same
as the USDA estimates from which they are derived.

Farm income data tend to be politically sensitive in that it is
somewhat institutionalized in the policy making process. Since these
data are often used by politicians to justify a certain policy stance
and because they are the main measure of sector wide performance empha-
sized by USDA, farm income data maintain a certain amount of political
sensitivity. This sensitivity will probably remain as long as farm
policy and the farm population are politically important. Thus, farm
income data are politically sensitive because they give an appearance
of objectivity to this political concern for the well being of farmers
and the data are accepted as a valid measure by those in the system.

Revisions in the data are politically sensitive also. These
revisions can embarass a politician, who may have used the data to jus-
tify a position, by causing a certain policy position to appear wrong
simply because revisions caused the data no longer to support that
position. The types of positions taken using farm income usually make
downward revisions in the data more politically sensitive than revisions
which increase aggregate farm income. If farm income data are used to
justify a particular level of price supports for farmers and then re-
visions show income actually to be lower, then farm constituents are
likely to be clamoring for more assistance. Revisions in the opposite
direction tend not to have similar effects because strong anti-farm
groups do not seem to exist.

However, the way farm legislation is written also makes farm
income data less politically sensitive than it might be otherwise.

Commodity prices tend to be more politically sensitive than farm

183
income. This point was made by most of those interviewed and arises
because most current farm program benefits are tied to farm prices.
The hypothesis that data are the most politically sensitive when tied
directly to program benefits is reinforced by this phenomenon.

One final note before leaving the area of credibility. The
credibility problems of USDA seem to be lessening somewhat at this
time. The credibility of USDA, especially with Congress, was weakened
in the past when many members of the Democratic controlled Congress
used the data and analysis problems of the USDA to attack the Nixon
and Ford administration. This is not as important in the Carter ad-
ministration, but it does appear that a carryover effect of this lack
of credibility still remains and is likely to affect the USDA for some

time to come.

5.5 NATIONAL INCOME ACCOUNTING AND FARM INCOME

Many of the earlier studies of USDA farm income data stressed
the importance of these data in the national income accounts. Argu-
ments were made in the earlier Task Force Reports as to the need for
farm income data that could be easily integrated into the National In-
come and Product Accounts. Consequently these reports stressed the
need for data which were comparable between sectors and for data which
closely met the needs of the Department of Commerce.

As was pointed out earlier, this need for data which are com-
parable among sectors does not appear to be that crucial to users, par-
ticularly in government. Guebert's analysis, cited in Chapter 3, be—
gins to put to rest the questions concerning the reconciliation of

USDA accounts with the BEA accounts on national income. Interviews

184
with individuals in the Department of Commerce suggest that the pro-
blems of reconciling the USDA data with the BEA concepts are blown out
of proportion by some of the earlier studies. Conversations with those
responsible for reconciling these accounts pointed out that the pro-
blems were more in measurement than in concept. Most of the adjust-
ments made to the USDA data to get these to fit the BEA concept are
relatively minor considering the overall magnitude of farm income.
For example, they rearrange sales in some commodities because of the
way USDA treats Commodity Credit Corporation loans. Another important
difference is that BEA separates farm income into corporate and non-
corporate shares and takes out corporate officers' salaries, this has
only been about $500 million in recent years. Other minor adjustments
were pointed out by Guebert. But for the most, these adjustments are
no worse than the problems encountered in reconciling data from other
sectors.

The measurement problem arises because of the need for quar-
terly data for the GNP accounts. USDA seems to consider yearly esti-
mates as most important and hence the farm income unit in USDA has
tended to concentrate on making certain that the quarterly levels of
farm income come out to an accurate yearly total. BEA would prefer an
accurate accounting for the quarterly change in farm income. However,
it was also mentioned that no other sectors have particularly good
quarterly data. The yearly data supplied by USDA was said to be of
sufficient accuracy in concept and measurement for most Department of
Commerce needs for nonquarterly data.

The seasonality of and time required in the production of agri-

cultural commodities along with the uncertainty of the level of output

185

arising from a set of inputs makes accurate quarterly income estimates
in agriculture more difficult and less meaningful than for some other
sectors. Such aspects as measuring growing craps as work-in-progress
are probably not feasible given the uniqueness of agriculture. Recent
improvements in primary data collection especially in measuring total
cash receipts for grains within a month after the sales should serve
to improve the quarterly income estimates. .So given the adequacy of
quarterly agricultural data relative to other sectors, major improve-
ments in the farm sector data on a quarterly basis may not justify the
recommended high priority which was given these improvements by some
earlier studies. Other farm income data needs for agricultural policy
making seem to deserve equal or higher priority.

One of the main differences between the current fanm income
accounts and the national income accounts is that the latter are
basically on an accrual basis and the former on a cash accounting basis.
The farm income accounts do attempt to measure inventories of outputs
so that the data are at least partially on an accrual basis.

The mail survey was used to assess the importance of these
different accounting methods for aggregate farm income data. This
should also give an indication of the usefulness to farm income users
of data presented in the national income and product accounting formats.
When all users were considered, it is apparent that a higher priority
is given to cash accounting, as shown in Table 5-6. When the major
categories of end uses are considered, a slightly different picture
emerges. While those who use the data in public policy and demand es-
timation seem to place a higher priority on cash accounting principles

in designing a farm income accounting system, the credit and financing

186
users seem to place a higher priority on accrual accounting. Thus, it
would seem that the largest users of that data might prefer data pre-
sented in the current format over the national income and product ac-
count format, with the possible exception of banking and financing
firms. This might also explain why many users of the data seem to find
- the realized net income data as useful or more useful than the total

net income data currently published by USDA.

TABLE 5-6 PRIORITIES BETWEEN CASH AND
ACCRUAL ACCOUNTING BY USES

 

Mean Scores

 

All Public Demand Credit and
Design Criteria Users* Policy Estimation Financing
Users** Users*** Users***
Cash Accounting 3.005 3.116 3.000 3.333
Accrual Accounting 3.214 3.419 3.548 2.833

 

Scale: 1 is highest priority and 5 is lowest.
*Mean scores significantly different at 7 percent level

**Mean scores significantly different at 12 percent level
***Mean scores significantly different at 5 percent level

5.6 FARM INCOME STATISTICS IN A DEVELOPMENTAL MODE

Edgar Dunn discusses the importance of having an information
system that is in a learning or develOpmental mode. By this is meant
that the system must be capable of supplying information on changes
within the areas addressed by the information system as well as adapt-

ing itself to these changes. Thus, the ability to make changes in a

187

data system is critical for it to remain in a developmental mode.

This raises two questions of importance for this research.
First, how difficult is it to make changes in the current data system
given its apparent political sensitivity? Second, what issues or
changes in the sector which affect the farm income information system
.will become important in the future? Given the policy focus of the
current farm income information system this latter question concerns

the important future issues in agricultural policy.

5.6.1 POLITICAL SENSITIVITY

The political sensitivity of farm income data will make chang—
ing the data system somewhat sifficult, especially if these changes
affect the definition and level of farm income. Hildreth and Worden
cited many of the difficulties encountered in trying to change the de-
finition of a farm. Some of these types of political roadblocks are
bound to appear if major changes in the farm income data system are
attempted. The timing of changes seems to be critical. When net farm
income is at a relatively low level, as it is at the present time, any
changes in the estimates of farm income are bound to be interpreted,
by some, as politically motivated. If new procedures are used which
raise the aggregate figure, those in favor of increased support for
farmers are likely to claim that the data were changed to make farmers
appear better off than their true situation. If changes reduced the
level of farm income, opponents of farm programs are likely to argue
that the data were changed to gain more support for farm programs.
Those interviewed tended to confirm the hypothesis that changing the

concepts or definitions used in the farm income data system would be

188
more difficult when the level of farm income was at an historically low
level.

Farm income data are also a measure of performance for the cur-
rent Administration's farm policy. Thus, changes in the data system
are more likely to be acceptable early in the tenure of a given admini-
stration, since they are not as likely to have developed a vested in-
terest in the data. While most of those who were asked about this con-
sideration agreed in principle, most felt that the timing of changes
with respect to the level of aggregate farm income was most important.

The fact that farm income data are not tied directly to the
benefits of specific programs should make changes in the data system
easier than some other types of data. This should prevent groups which
are organized directly around program benefits from becoming active
opponents to any suggested changes.

While some of those interviewed were somewhat pessimistic about
the possibilities for making changes in the existing data system, these
individuals were not necessarily in the majority. Most suggested that
changes in the current system could be made if presented well and
backed up with appropriate research. Some individuals suggested that
the problems encountered in the attempted change of the definition of
a farm were not envisioned beforehand so the appropriate analytical
support was not available. Hence, the implication that changes could

have been made if the analytical support were available.

5.6.2 FUTURE POLICY ISSUES
The interviews with public policy decision makers suggest that

three major types of issues will be important in the future for

189
agricultural policy. This classification of issues will be made ac-
cording to how these issues are related to farm income data needs.

First, many felt that the current farm program issues would
remain important. Thus, the data needs pointed out in the previous
analysis in this chapter should remain in the foreseeable future. The
.second area of future concern can be called structural issues in farm-
ing. These primarily deal with the areas of legal organization and
control of agriculture. Included in this are problems associated
with the ownership of agricultural land, vertical integration, and
part-time versus commercial farming. The third set of issues might be
characterized as productivity issues. Many individuals argued that
world food problems and the role of the United States in producing this
food would be important. A set of issues concerning energy, the envi-
ronment, and land use are also really productivity issues in that these
all have a primary effect on the productivity of farming.

The importance of structural issues again stresses the need
for data on the distributional aspects of farm income. One of the
principal concerns in this area is the relative importance of large
corporations and part-time farms in agriculture. One possible way to
distribute income data to help in answering these questions is sug-
‘gested in the chapter on "Agricultural Statistics" in the U.S. Depart-
ment of Commerce's "Framework for Planning U.S. Federal Statistics
1978-1989.” This report recommends an allocation of total farm income
between non0perator landlords, production contractors, business asso-
ciated farms and to primary and part-time farms. How often this type
of data needs to be collected is also an issue. Structural change is

not likely to occur at as rapid a rate as prices or income so Census

190
benchmark data might be all that is necessary for collecting appropriate
data for these categories.‘

For the productivity issues it seems that data on total value of
production might be appropriate. Currently data of crop value of pro-
duction are provided by the Statistics Branch of ESCS, but data on live-
. stock value of production is lacking in some cases. Even where there
are data on value of production it rarely receives much emphasis from
USDA, so the potential users of the data may not be aware of its exis-
tence. It should also be noted that in terms of accuracy, the measure-
ment of value of production, especially for cr0ps, is probably one of
the most accurately measured of any aggregate income related figures
because of the relative accuracy in measuring cr0p production as 0p-
posed to sales. The multiple frame surveys used in estimating crap
production appear to yield more accurate estimates of production than
sales because of problems in defining a transaction and in determining

own account uses of craps.

5.7 ANALYSIS OF METHODOLOGY

Perhaps the greatest deficiency in using the information systems
paradigm as a guide for research arises in defining the population of
users and uses. In the previous chapter some of the problems encoun-
tered in defining a population of farm income data users were identified.
The problem arises because of the general characteristic of information
that the fixed costs of gathering the information are much higher than
the costs of transmitting the information to additional users. So the
identification of the population of users most likely will be a problem

in the design of any study using this information system paradigm.

191

A priori, it was hypothesized that a further deficiency of this
approach would be caused by the fundamental paradox of information.
Since decision makers cannot necessarily place a value on data until
they are used it was thought that these decision makers would have dif-
ficulties in deciding on their potential data needs. This proved to
.be false; most decision makers were able to identify gaps in the exis-
ting data and could point out the types of data that they needed but
did not have available. Thus, this information system paradigm is
much more effective as a guide in doing research since it is possible
to specify problems and data gaps in the system by examining the users
and uses of the information.

There remain a series of difficulties in placing a value on
different uses of data and information particularly nonmarket uses
which are likely to hinder the utilization of this framework in decid-
ing on specific improvements to suit one user over another. At this
point it may be easier to rank the importance of different users within
the same information system by using some criteria such as those de-
veloped in Chapter 2. However, until a common unit of measurement
exists for information, it seems unlikely that meaningful comparisons
and rankings between different information systems will be possible,

except at the most general level.

5.8 SUMMARY

Given the users identified in the previous chapter and the mar-
ket structure of some of the private sector users of the aggregate farm
income data, it is apparent that any redesign or improvements in the

system should probably be made to increase the utility of the data in

192
public policy decisons. Among private sector uses a willingness to pay
criteria was suggested and applied as a measure of the demand for farm
income data. Those who make specific uses of the data in such areas as
demand estimation and credit financing decision showed the greatest de-
mand for the data.

Two major problem areas were identified by the users of the
data. These can be characterized as conceptual obsolescence and credi-
bility. Conceptual obsolescence is of two types in the system. First,
and most important, is the conceptual obsolescence brought on by
changes in the policy agenda in agriculture. The current system fails
to provide adequate distributional data on farm income, especially by
commodity group, which appears necessary in answering questions which
arise in the agricultural policy process. The second type of concep-
tual obsolescence arises because the current concept of farm income
centers on a national family farm concept which does not seem to be a
true representation of the reality of the farming sector. Ostensibly,
this second type of conceptual obsolescence is not as serious as the
first because of the way in which farm income data are used in policy
decisions. Since aggregate farm income data are used more as social
indicators, the need for data which bear a one to one relationship with
reality is not as critical.

Revisions in the data through time have tended to create a
credibility problem for USDA, especially with regard to the analysis
performed by the department for others in the policy process. The
change of administrations after the last election seems to have les-
sened, but not eliminated, this credibility problem.

The importance of better integration of the USDA farm income

193
accounts into the national income and product accounts does not appear
to be as serious a problem as the earlier Task Forces have suggested.
Data on farm income, while showing some problems for quarterly data,
are not much worse, if at all, when compared to data on other sectors.
Except for the credit and financing users of USDA farm income data,
~most users tend to prefer the cash accounting methods used by USDA in
reporting farm income rather than the accrual methods which undergird
the national income and product accounts.

Because of the political secsitivity of the data, the ease in
making changes in the concepts, definitions, or format of farm income
data depends primarily on the level of farm income at the time changes
are prOposed. The fact that farm income data are not tied directly to
program bevefits is likely to make it easier to redesign the data sys-
tem. Future policy issues, in addition to the current issues, which
the information system will apparently have to address primarily re-
volve around structural and productivity questions. This points out
the need for better distributional data on legal organization and con-
trol in farming and for better data on the value of production.

In analyzing the methodology used, the major problems in using
the information systems paradigm as a guide in research is in defining
the population of users. The problems in defining a unit of measure-
ment of information make it difficult to place a value on information
and hence the usefulness of this paradigm in setting priorities among
different types of data is more suspect. However, this paradigm is
helpful in setting priorities for improvements within the same data
and information system in that it provides a framework for defining

problems in the system.

CHAPTER 6
CONCLUSIONS AND RECOMMENDATIONS

6.1 GENERAL CONCLUSIONS

A The specialization of functions and activities that has occurred
in agriculture over the half century since the farm income accounts were
designed has caused serious problems with regard to the usefulness of
aggregate farm income data. The aggregate data, based on the concept of
a single national family farm, are not utilized to any great extent in
public decision making except as general indicators of the economic
health of the sector. Aggregate sector performance measures, such as
farm income, are not as useful in policy decisions as in the past be-
cause the specialization of agricultural production has forced policy
to focus on issues which are commodity specific or specific to other
groups within the farming sector. This shift in policy has made the
current concept of farm income obsolete, because the data concept has
not changed in response to the changes in the policy agenda. The low
level of use found among those receiving the data directly from USDA
also tends to confirm the significance of this type of conceptual ob-
solescence. Most seemed to imply that their lack of use of the exist-
ing farm income data arose because it was not of value in the types of
decisions which they normally made. It would seem, beforehand, that

most persons who requested the bulletin Farm Income Statistics would

 

have expected to find fann income data of some use. The fact that many

did not use the data after receiving it would tend to imply that the
194

195
current data are not useful in their decisions.

The increasing heterogeneity among farms in the sector has made
the national family farm concept used in the farm income data system
obsolete in the sense that it does not represent reality. However,
users of the data seem to have already incorporated this fact into
,their use of the data. Most do not appear to expect the data to bear a
one to one relationship to reality, instead these data are relied upon
for a measure of the changes in income from one time period to the
next. So as long as the current concept is adequate for measuring
change in income this type of conceptual obsolescence is not as criti-
cal as that arising from the changed policy agenda. However, the fur-
ther task remains of assessing the ability of the existing data to mea-
sure year to year changes.

Major changes in the farm income data system toward national
income and product accounting formats, as recommended by earlier studies
of the fann income data system, will have little impact on the concep-
tual obsolescence caused by changes in the focus of agricultural policy.
Unless explicit consideration is given to various disaggregations in
the income data, changes in the farm income accounting system to make
it compatible with the national income accounts will not produce data
which can begin to answer the important questions in agricultural pol-
icy.

The results of this study also suggest that problems of con-
ceptual obsolescence may be more critical for some types of statistics
than others. In designing social or economic indicators the need for
concepts which represent reality exactly is not necessary or perhaps

not eVen desirable. There might be tradeoffs in measuring the level

196
of detail for income and expense accounts against measuring those ac-
counts which cause the greatest change from one period to the next.
Some social or economic indicators can be operationalized in such a
way to have no meaning in and of themselves. For instance, to know
that the Consumers Price Index is 150 means very little unless the
level of the index is known for other time periods. Aggregate farm
income is similar in the sense that knowledge of any individual esti-
mate does not mean much unless there are estimates for other periods
which can be compared with the current estimate. For example when
farming in the United States was reasonably homogeneous, it was pos-
sible to use aggregate estimates of farm income and the number of farms
to obtain an average income per farm that was useful alone since it was
easily comparable with a person's experience. The lack of homogeneity
among farms has raised doubts as to whether any measure of average in-
come per farm could be useful even if the problems with defining a
farm were solved. With adequate data on the distribution of income
among farms the problems in defining a farm disappear, since any rea-
sonable definition can be Operationalized by combining the distribu-
tional data.

Decision makers in the public policy process outside the USDA
also raise some questions about the credibility of USDA and their abil-
ity to do unbiased analysis and forecasting with regard to farm income.
Revisions in the historical data on farm income seem to be a major
cause of this lack of credibility since the size of the revisions are
often outside the range of accuracy expected by the users, i.e., most
-users expected the farm incomedata to be within 10 percent of the

"true" value yet revisions of greater than 10 percent are not uncommon.

197
The emphasis placed on farm income data by USDA as the measure of sec-
tor performance only serves to intensify the credibility problem by
making the revisions more visible. It is also rare that the reasons
for the revisions are presented when these are made, so even the more
sophisticated users of the data begin to doubt its reliability, when
’actually the revisions are supposed to enhance the accuracy of the
data.

Earlier studies have stressed the need for better integration
of the farm income accounts into the national income and product ac-
counts but this does not appear to be an immediate necessity. The
questions raised in agricultural policy tend to center on who benefits
from and who pays for the farm programs rather than whether it is ap-
propriate to have a national agricultural policy. So data which allow
for comparisons between the agricultural sector and other sectors is
not as necessary for this purpose, hence the importance of having farm
income data in a national income accounting format is diminished.

The apparent ease with which the current farm income data are
used by the Bureau of Economic Analysis also belies the importance of
having data in a national income and products accounts framework. As
long as the farm income data are no worse than the data BEA uses for
other major sectors, it would seem that the integration of these two
accounts would be less critical than other problems in the system.

Comparisons betweenagriculture and other sectors do not seem
to be as important a use of the data at the present time relative to
the need for comparisons between groups within the sector. In the long
run, however, it seems more likely that farming will become similar to

other business sectors. In all probability this will lead to questions

198
concerning the apprOpriate role of government intervention in agricul-
ture vis-a-vis other sectors. However, very few of the policy decision
makers interviewed foresaw this within the next decade. Thus, while
gradual changes in the farm income data toward the national income and
product framework would be apprOpriate, the urgency of these changes
.is not evident. More gradual change is also less likely to cause poli-
tical objections to be raised concerning changes in the data, or to
raise questions about changes being motivated by political purposes.

Comparability between sectors is one of the principal benefits
of national income accounting formats. However, the ability to make
meaningful comparisons among sectors is not limited to the case where
the national income accounts formats are used. In most instances the
need to have comparability among more basic microdata files is even
more critical since these types of data are fundamental for understand-
ing the behavior and policy interactions among sectors.

In the future the need for better distributional data will
also remain important. In particular, the disaggregation of the income
data according to legal organization or other structural aspects of the
sector will be of increasing importance. At the same time, the current
fanm policy issues are likely to maintain their significance so data
distributed by commodity, states or regions, and by value of sales or
any other size class will remain useful.

The methodology used to evaluate the system seems to have been
an apprOpriate choice at this time. The information systems paradigm
provided a guide that was useful in understanding and categorizing the
problems in the system. The ill defined nature of problems usually

associated with information processing makes this type of framework or

199
paradigm very useful in that it is possible to gain an understanding
of the relationships between problems in the system. In a sense it
provides a means to judge the system by providing an ideal against
which the Operating system can be compared and pinpoints areas where
potential problems exist in an operating information system. The major
_ difficulty faced in using this methodology is brought on by the in-
creasing returns in use characteristic of information. This makes it
difficult to identify the p0pulation of data users and the nonusers who
might be potential users. Since a principal focus of the paradigm is
on the use of information in decision making, the inability to deter-
mine the users of data and information of a given type will reduce the
utility of the paradigm in doing research in this area. However, this
does not appear to be a serious deficiency as long as the users identi-
fied are relatively similar to those users not identified. Further-
more, this would also seem to imply that this type of research might be
more appropriate for setting priorities among users within a specific

system rather than between different systems.

6.2 RECOMMENDATIONS
6-2.1 ‘DISTRIBUTIONAL DATA

The need for more data on the distribution of income was one of
the most obvious findings of this research. Solutions to the problem
of conceptual obsolescence assOCiated with this seem to deserve a high
priority in the redesign of the farm income data system.

0f the many types of disaggregated data which were cited as de-
ficient or nonexistent, two types are most significant. First, data on

net returns by commodity or income by type of farm would appear

200
deficient. The existence of cash receipts data by commodity reduces
the urgency of this problem but does not eliminate it.

As a first step toward obtaining better net income or returns
data by commodity it would seem useful to have the data disaggregated
into two different categories, net returns from livestock and livestock
,products and net returns from crops. In this latter case some poten-
tial for integrating the data from the cost of production studies with
the current cash receipts data might provide a means to estimate net
returns for a major proportion of the crops. One relatively easy way
to make these estimates would be to take the difference between average
monthly price and cost of production and multiply this by the amount of
monthly marketings. This type of data would be more of an indicator or
index of profitability for the subsectors within the farming sector.

The second type of distributional data which seems critical
given the nature of future issues in agricultural policy is data which
deal with the ownership and control of productive resources in the
farming sector. As the questions raised in agricultural policy begin
to focus on areas such as the role of large corporations in agriculture
or on foreign investors in U.S. agricultural land, data on legal organ-
ization and other aspects of resource control will become important.
Data distributed by legal organization or other claimants on income
seem to be even more critical than that concerning the lack of data on
income distributed by commodities. In the case of commodity specific
data at least some proxy variables in the form of cash receipts by
commodity exist. There seems to be very little time series data on
some of the major structural issues facing agriculture. To achieve

this goal of improved data on structural issues in farming, it is

201
recommended that a high priority be given to the disaggregation of to-
tal farm income and its allocation to nonoperating landlords, produc-
tion contractors, businesses with only a minority interest in farming,
and self-employed farmers. This latter category should be further dis-
aggregated into part-time and primary farms. Primary farms are those
.where the majority of the self-employed operator's time is spent in
farming.' This recommendation is essentially the same as one of the
recommendations made by the Office of Federal Statistical Policy and
Standards in the Department of Commerce in their framework for planning
agricultural statistics (p. 34). It is important to reiterate this
recommendation since it should be given a high priority in making im-
provements in the system.

The same U.S. Department of Commerce report notes that USDA and
the Census Bureau are currently using the Operationalized concepts of
"business associated", "primary" and "part-time" farms in some of their
data collection activities. Their beginning efforts should be ap-
plauded but better data on nonoperator landlords and production con-
tractors may be necessary. An important question that has not been an-
swered adequately is how fast changes are taking place in the structure
of agriculture. If these occur at a relatively slow rate then it might
only be necessary to collect data on the number and size of farms in
. each of these structure categories during each Census of Agriculture.
These Census benchmarks could then be used to allocate income among the
categories during intercensal years using techniques similar to those
currently used in allocating income by value of sales class in the
existing series. It would appear that a pilot study of some type

might be in order to determine if these structural changes are occuring

202
at a rapid enough rate to warrant collection of the data more frequent-
ly than every five years.

The farm income data are currently disaggregated by state and
by value of sales class. Both of these types of distributional data
should be maintained. However, some minor improvements seem in order.
With regard to the state data, the insensitivity of the net income data
to year to year changes is somewhat of a problem. This arises mainly
because of the way production expense data are calculated for each
state, i.e., data are estimated at the national level and then allocated
to the states based, for the most part, on cash receipts estimates.
Therefore, many users rely more heavily on cash receipts by states ra-
ther than net income that would normally be of concern to them. Recom-
mended improvements in measuring farm production expenditures, outlined
in the next section, should alleviate some of these difficulties.

Inflation has tended to make data on the distribution of income
by value of sales class less useful in answering questions concerning
changes in farm size. Since the economic classes of farms are deline-
ated according to fixed dollar amounts, where lower classes are asso-
ciated with higher sales, inflation can cause some farms to shift to
lower classes without any increase in physical farm size or productivi-
ty. Thus, the current data would not be adequate for use in decisions
concerning such areas as economic concentration in the farm sector.

Two approachs can be used to ameliorate some of the difficulties brought
On by inflation. First, measuring the sales classes and value of sales
in real dollars might be used. This would involve using a price index,
such as the index of prices received by farmers to deflate the value of

sales reported and then maintaining the same endpoints in dollar tenns

203
for the sales classes. A second way would be to estimate the number of
farms or income in each percentile of value of sales. Thus as the num-
ber of farms in a certain percentile of value of sales increases it is
more likely to be caused by changes in the structure of farming rather

than merely by inflation.

6.2.2. FARM PRODUCTION EXPENSES

Since farm income data are used primarily as an indicator of
the economic health of the sector, it would be helpful to have a mea-
sure of net income that is sensitive to yearly and quarterly changes
within the farm sector. Given the methods currently used, the produc-
tion expense data are relatively insensitive to short run changes in
expenses since extrapolations from Census benchmarks are the principle
means of determining these estimates. Thus, as the patterns of re-
sources and inputs used in producing farm commodities change, it is
less likely that these changes will be noted in the expense estimates.

The Farm Production Expenditure (FPE) component of the Annual
Economic Survey of Agriculture appears to be a ready vehicle for gain-
ing more sensitivity in the estimates of production expenses. The
results of the FPE survey could be used directly in estimating the pro-
duction expense component of the farm income accounts, but now these
are used only indirectly, if at all, in making expense estimates. The
fact that the current Census of Agriculture is no more than a list
frame survey of farmers raises questions as to whether the Census data
are any more statistically accurate than the Annual Economic Survey of
Agriculture data which are collected using a multiple frame survey.

Thus, it is difficult to make the argument that the Census farm

204
expense data are "better" than the data collected by ESCS. In order to
obtain data which are more sensitive to changes in expenses on a yearly
basis it is recommended that the results of the FPE survey be used
directly as the estimate of production expenses in the farm income ac-
counts. This will require some minor modifications of the current sur-
vey and the existing methods for estimating two of the expense accounts
must be maintained.

The current FPE survey (as was done in 1975 and 1976) will need
to be modified to obtain data on intrastate sales of livestock and feed.
However, reasonable approximations should be obtained by asking those
surveyed the percentage of feeder and stocker livestock and feed pur-
chased from out-of-state firms. Data on intrastate sales are needed in
making state farm income estimates. The sample size for this survey
should also be increased to secure better estimates of state level ex-
penses. Currently a sample of about 5000 has been used, sources in the
Statistics Branch of ESCS suggest that a multiple frame sample of be-
tween 15,000 and 20,000 would be necessary to obtain reliable state
level expense data.

One expense account currently estimated by the Farm Income Unit
that probably could not be estimated from a survey would be depreciation
or capital consumption based on replacement costs. Estimates of this
account should therefore remain a function of the Farm Income Unit.

The current FPE survey obtains data on some capital expenditures such
as machinery purchases and land and building improvements. While there
is no harm in asking these questions, in fact they are useful in esti-
mating depreciation in the future, it does create some confusion when

the total farm expenditures estimated by the Statistics Branch of ESCS

205
are compared with the Farm Income Unit's estimates of production ex-
penses. Using the FPE survey data directly would not cause a large
change in the expense estimates from the current methods. When the
capital expenditure items in the FPE survey are removed and replaced
by the Farm Income Unit's estimates of depreciation and other capital
consumption and the FPE rent estimates are replaced by the Farm Income
Unit's estimates of net rent paid to nonoperator landlords, the FPE
estimates of production expenses are within 2.2 and 3.6 percent of the
Farm Income Unit's estimates of production expenses published for 1975
and 1976 respectively.

Many of the other reports on farm income have also suggested
that the production expense data are probably the weakest component of
the data system. So this recommendation for increased use of more di-
rectly obtained expense data follows closely and is supportive of this
aspect of the other studies.

A principal concern of the Bureau of Economic Analysis is in
measuring quarterly change in GNP. Thus, the sensitivity of the farm
income data to quarterly change is a significant consideration for their
uses. Again the lack of sensitivity of the current farm production ex-
penses make these data less useful for users of BEA data. To secure
better estimates of quarterly change in expenses three major components
need to be measured: 1) feeder cattle, 2) feed and 3) fertilizer.
These three expenses make up nearly 50 percent of the current operating
expenses and are the inputs most likely to show the greatest variabili-
ty between quarters, so data on these inputs should be collected on a

quarterly basis at the national level.

206
6.2.3 NATIONAL INCOME ACCOUNTING

The concern for commodity specific data and the desire to mea-
sure productivity in the sector as expressed by the mail survey and
interview results has important implications for changing the farm in-
come accounts to be more in line with the national income and product
,accounts. Given the current uses of farm income data, the recommenda-
tions of the 1972 and 1975 Task Forces that farm income be measured as
Value added and the farm income accounts be put in national income ac—
counting formats, do not seem to warrant a very high priority over the
next five to ten years. However, as agriculture becomes more industri-
alized and specialized, it will be more useful to have farm income ac-
counts in the same format as some of the other sectors. A gradual move
toward these formats can be achieved by placing a greater emphasis on
the farm business aspects rather than farm families and by obtaining
some of the key primary data needed to build up to these national in-
come and product type accounts.

The concern of the data users for better information on the
productivity of agriculture suggests an essential first step in getting
the type of data to make the change to national income type accounts.
Value of production is the starting point for the value added concepts
of income. In agriculture, estimates of the value of production are
made for most crops and some livestock products but these tend to be
hidden in obscure publications and a measure of aggregate value of pro-
duction is not presented. One way to begin making this transition to
a value added concept of farm income would be to publish data on value
of production in aggregate and for some of the major crops and live-

stock products in the same publications as the current farm income

207
data.

Since for most crops value of production data are not available
until after the marketing year is over, the preliminary estimates will
need to be made before prices are available for the entire marketing
year. This problem cannot be overcome unless the marketing year for
crops coincides with the calendar year. The unique nature of agricul-
tural production insures that this difficulty will not be overcome in
the near future. Forecasts of prices and marketings consequently must
play an important role in the January estimates of value of production
but the problems would not seem any worse than the current difficul-
ties which arise in making the preliminary farm income estimates.

The concern expressed for data on farm businesses as opposed
to farm families pimpoints another area where gradual changes toward a
more farm business oriented national income accounting framework can
be made. At the present time a major link between farm businesses and
fann families in the income from farm production accounts is the area
of nonmoney income, particularly the imputed rental value of farm
dwellings. Nonmoney income is presented as a gross income figure and
the associated expenses are included along with the remainder of the
production expenses for the sector. For those users who wish to ex-
clude these components to more closely approximate their desired farm
business concept, it is not now possible to remove both the income and
expenses associated with these components. Also, if it is deemed that
the rental value of farm dwellings is more appropriately included in
the real estate sector of the national income accounts then it will be

necessary to separate the expenses associated with farm dwellings from

the remainder of the expenses.

 

208

Some of the improvements in the primary expense data suggested
earlier should also provide some of the data on the intermediate pro-
ducts consumed, which are necessary to estimate the value added by
farming. This should also make the transition somewhat easier.

The lack of concern for farm income data accounted for on an
accrual basis, relative to a cash basis, would seem to imply that data
on inventory change do not merit a high priority in improving the data
system. However, the spinoff effects of improvements in such expense
items as quarterly fertilizer expenses should make the estimation of
input inventories easier. In terms of output inventories the Task
Forces both cited the example of the inclusion of beef breeding herds,
dairy herds, and poultry laying flocks as part of current inventories
as a major problem with the existing accounts. Melichar's (p. 83-84)
counterarguments that the magnitude of the errors brought on by this
are not great imply that these types of improvements might rate a lower
priority. The findings which suggest that farmers in general make many
capital purchases from current receipts also would seem to infer that
the concept of inventories which includes both current production and
short term capital items could quite apprOpriately be utilized in un-
derstanding some aspects of farm operator's behavior.

The earlier Task Forces also noted two different views of the
farming sector which could be used in designing farm income accounts
as part of the national income and product accounts. The 1975 Task
Force recommended that an establishment concept be used in forming the
farm income accounts because this fits more closely with the tradition-
al notion of a farm as a "place." When considering the apparent de-

mand for farm income data that are commodity specific and the concern

209
for measuring productivity in the sector, it seems that the product
concept deserves equal attention relative to the establishment concept
in the development of the farm income accounts. However, it should be
remembered that the concept of an establishment is more compatible
with the idea of a "family farm" which remains central to much of the
agricultural policy debate. The establishment concept is also more
easily reconciled with disaggregations of income by legal organization
and other structural aspects.

The importance of family farms was a key part of the debate on
the "Food and Agriculture Act of 1977." Section 102 of Public Law
95-113, the Food and Agriculture Act of 1977, is entitled "Family
Farms" and reaffirms Congressional support for the maintenance of family
farms. More importantly for the purposes of this study, the law directs
the Secretary of Agriculture to provide written information to Congress
every year on the current status of family farms in the United States.
If this request by Congress is backed up with sufficient appropriations,
then this might provide the support for better data collection on farm
income, especially on some of the structural issues mentioned earlier.
This may also tip the scales in favor of an establishment concept as
the basis for a future accounting system. But the issues mentioned

above also show the importance of a product concept.

6.2.4 IMPROVED CREDIBILITY

The absence of knowledge by those using the data about even the
rudiments of the methodology used in estimating fann income statistics
suggests that if the users were more aware of the methodology that it

might be easier for them to assess the reliability of the data. Given

210
the current estimating techniques, some revisions are inevitable, how-
ever, this point is rarely made to the users. So revisions tend to
reduce the reliability of the data in the eyes of the users when just
the opposite should be true. Since the reasons for revisions are not
’often published along with the data, the users cannot determine why the
revisions are made. So they must assume that the data are of poor
quality. Also, the methodology is not well understood by the users so
the inevitability of revisions is not understood.

To improve the credibility of the USDA as a data supplier, a
statement concerning the methodology used in making the farm income
estimates which stresses the need for revisions should be included in
each Farm Income Statistics bulletin. This statement could be similar
in style and length to the statement on concepts now published in this
bulletin. It will also make it easier to present a statement as to
why revisions in the data have been made, so that users can be made
aware of the improvements in the data.

Some might argue that this merely attacks the symptoms of the
problem rather than the problem itself, in this case the problem being
that of revisions. Of course, by eliminating revisions in the data
the credibility would most likely be improved, however, the timing of
the release of the estimates almost insures some revisions. Recent im-
provements in the availability of primary data should help to reduce
revisions. For instance, estimates of monthly marketings of major
grains are now made within a month for the preceding month, while in
the past this data was not available until the end of the crop year.
The suggested improvements in the expense data should reduce the number

of revisions of this component, especially those occurring after the

211
first July estimate, since this estimate will be based on more directly
reported data.

The fact that USDA emphasizes farm income as the singular mea-
sure of sector performance only intensifies the credibility problem,
since these data become more politically sensitive. Earlier recommen-
_dations should also alleviate some of the credibility problems. Chang-

ing the name of the bulletin Farm Income Statistics to Farm Sector

 

 

Performance Measures and including in it other recommended series such

 

as value of production and indices of profitability for various com-
modities should also serve to reduce the political sensitivity arising
from an exclusive focus on farm income.

The Department of Agriculture still has a great deal of influ-
ence on what becomes an issue in agricultural policy. The political
sensitivity of farm income data is in part a carryover from times when
this data received more emphasis than they do even today. This has
caused farm income data to become institutionalized to a certain extent
in the policy decision process. Thus changing the farm income data
will be more difficult as long as the focus of many decision makers re-
mains so exclusively on farm income. USDA can choose to emphasize
different performance measures and these will become the universe of
data which have the potential for being politically sensitive. Hence,
by emphasizing a larger set of performance indicators rather than farm
income alone, making changes in the design of farm income data might
become easier.

One other approach to reducing the revisions of farm income
data and thus enhancing the credibility of USDA is to reduce the number

of estimates of farm income made before the last primary data become

212
available. The 1975 Task Force was particularly concerned with revi-
sions between the January preliminary estimate and the first revised
estimate in the following July. One way to avoid this would be to
eliminate the January preliminary estimate and replace it with a £353:
3333_made in December. Users are more likely to expect a forecast to
be different than the actual estimate, 30 revisions will tend to be
less sensitive. Given the current estimation procedures, the January
estimate is not much more than a forecast anyway since many of the key
pieces of primary data are not available in their entirety. But call—
ing this January figure a preliminary estimate seems to set higher ex-
pectations about accuracy. A December forecast could then be used in
the same fashion as the January estimate, except that the changes made
in July would be treated and perceived as inaccuracies in the forecast
rather than as revisions in an estimate. Inaccuracies in a forecast
should not be as politically sensitive as revisions in estimates.

The premise that data users view inaccuracies in forecasts dif-
ferently and more favorably than revisions in data estimates is an un-
tested hypothesis. However, in Canada no estimates of farm income are
made until July and the December Canadian forecast seems to be used in
the same manner as the January preliminary estimate in the United

States without apparent problems.

6.2.5 GOVERNMENT DATA PROVISION: RATIONALE AND PRIORITIES

The question has been raised as to whether government should
provide data that are used in market transactions. Arguments against
government data collection tend to center on the inequitable conse-

quences which can occur in many situations where government data is

213
provided. The conditions for this have been described in Chapter 2.
Many individuals feel that the rich get richer and the poor poorer when
the government supplies data because of the great disparities in analy-
tical capabilities between large and small firms. Farmers often insist
that they are hurt by USDA data because the firms buying agricultural
_ commodities or selling farm inputs use the resulting information to the
detriment of farmers who must transact business with these firms. The
implication is that these firms are often able to perform more 50phis-
ticated analyses on the USDA data than are farmers. An initial reac-
tion to this difficulty, and a solution advocated by some individuals,
is to stop all public provision of market data.

‘This Luddite view of the problem and its solution is likely to
widen only further the income inequities between large and small firms.
If the government does not provide the data, the largest firms will
most likely produce proprietary data to meet their needs. Hence, ter-
mination of government data collection will only serve to broaden the
disparity in income among the largest and smallest trading partners.

Despite the availability of government data many of the lar-
gest firms produce their own data on market conditions and thus only
use government data as a source of validation. For instance, Continen-
tal Grain Company maintains observers in most of the major grain pro—
ducing areas of the world to monitor production. The USDA used agri-
cultural attaches for similar purposes and publishes information on
production in foreign countries based on these observations. The grain
company would clearly be in a more advantageous position relative to
the farmer if the USDA stopped publishing data on foreign grain pro-

duction since the grain company would be the only party with data in

214
addition to their inherent advantage in analytical capability. The de-
gree of industrial concentration and the income distribution that would
follow would be even more extreme than now prevails.

The more reasonable conclusion to be drawn after examining the
problems of unequal information in trades is that the government should
do a more thoughtful job in managing its investments in data collection
and analysis. The criteria developed in Chapter 2 relating to market
structure and income distribution provide some means for establishing
relative priorities for public investment in data and analysis. Ac-
counting for income distributional consequences and the effects of mar-
ket structure on data design could help insure that new data invest-
ments would tend to ameliorate the inequities among large and small
firms in trades. However, the government must make this a conscious
goal in order to have an impact.

Currently some data series serve the interests of large units.
For example, the agricultural census includes detailed questions about
machinery on farms. Some of the data arising from this are of public
value but the principal benefits go to the oligopolistic farm machinery
industry. The same could be said of data collection on other oligopo-
listic industry sectors. Given the appropriate role of government in
data collection this part of the census seems to be a candidate for
rethinking the use of government funds for data collection.

Another way to counteract the inherent analytical bias in favor
of large firms would be for the government to undertake more analysis
of data. Implicit in this recommendation is the need for the govern-
ment to communicate this analysis to the small units that can use the

information. The timing of the release of data and analysis can also

215

reduce the bias in favor of large firms. If the data and analysis are
released simultaneously then large firms will lose one of their advan-
tages in doing analysis, that being, the ability to analyze data more
rapidly than small units. Thus, the argument that the government
should not produce new data because this only serves to skew the income
distribution further is perhaps the least attractive of the alternatives
for dealing with the bias that follows from greater analytical capabili-

ties of large firms.

6.3 ADDITIONAL REMARKS

The ability to generate political support for data improvements
is an important consideration for implementing many of the recommenda-
tions suggested in the previous section. Improvements in primary data
are often costly and thus political support is necessary in the budget
and appropriations processes to fund these improvements. Data gather-
ing is not often a glamorous activity for an agency that also has.
other functions such as research and policy analysis. So political
support muct go beyond that for allocating funds to an agency. It is
also necessary to insure that support exists for allocating agency
funds to these activities. The 1975 Task Force recommended that farm
income related work be given a higher priority within the Economic Re-
search Service of USDA. It does not appear that this recommendation
was implemented, if anything recent budget reductions have probably hit
the farm income area harder than others.

The lack of use of the farm income data in the private sector
is not encouraging in terms of the ability to generate political sup-

port for data improvements. The problems of conceptual obsolescence

216
and credibility have also reduced the demand for the existing data in
the public decision process. Improvements in these areas should in-
crease the demand for the data and hence lead to more political and
budget support for these activities. However, potential users of the
improved data are unsure of their demand for new data until these are
available, so it is difficult to generate the support necessary to make
the improvements. This is a corollary of the fundamental paradox of
information. An obvious implication of this corollary is that any
failure to improve data, simply because there is no inmediate payoff
to the agency for achieving these advancements, might be a very myopic
choice by a data producing organization.

Given the current level of staffing and funding of farm income
work in USDA, it is not likely that improvements can be made without
additional support. To obtain this support, agency officials and others
associated with budget allocation must be made aware that the payoffs
for improved data are often of a longer run nature when compared to
other activities such as direct assistance to farmers which tend to
generate more political support for the department. Improved public
data and information in a decentralized sector are likely to result in
better private decisions for the coordination of the sector. The tre-
mendous increases in productivity in agriculture over the last half
century, the benefits of which accrued to all of society, can be attri-
buted, at least in part, to the existence of government supplied data.
Unfortunately, the benefits of improved data are not always obvious
before they are collected since the value of the data arises in its
use and all potential uses might not be predicted beforehand. Thus,

criteria for budget allocation in government agencies for data

217

collection must go beyond the shorter run interests of political ex-
pediency and instead recognize that the benefits of better information
often are spread over a longer time frame than other activities and
serve other users besides government decision makers.

The benefits of public data collection for private uses often
. transcend the normal gains arising from uses by individuals in market
decisions. In a democratic social order an informed electorate are
better participants in the policy process. Hence, private sector users
of data also provide the basis in part for collecting data for public
uses. Unfortunately, the private use of government data in both pri-
vate and public decisions are easily overlooked in establishing budget
priorities for data gathering by an agency yet without private support

the possibility for data improvements are greatly diminished.

6.4 FURTHER RESEARCH

As was noted in the introduction, one potential type of further
research would be in the area of making improvements in the statistical
measurement accuracy of specific accounts within the farm income data
system. This type of research is probably best done within USDA as
quality and accuracy of data should be an ongoing concern of any sta-
tistical agency.

The research presented here is far from a complete statement on
farm income data and analysis. Unanswered questions arise concerning
the tradeoffs between improved concepts and improved accuracy in mea-
suring existing concepts or improvements in analyzing existing measures.
Further work on the inherent differences in data design between social

indicators and more micro—level statistics is suggested by the manner

218

in which farm income data is used in deCision making. Of a more spe-
cific nature is the need for research which gives a better understand-
ing of the relationship among specific types of similar data, such as
cost_of production and farm production expenses. It would seem that
these concepts should be related and the potential for collecting pri-
, mary data which can be used to operationalize either concept needs to
be explored.

' Many conceptual and theoretical aspects also need further work.
Questions concerning the value of information are often asked but
currently the conceptual basis for guiding one in this area is weak.

As zero based budgeting comes into vogue in government, estimating the
value of government information production will be critical. However,
the lack of a solid conceptual and theoretical basis for estimating

the value of information will raise doubts as to the validity of any
results in this area. There seems to be a general conclusion that in-
creases in information are always welcome. However, problems of infor-
mation overload can obsfucate issues and make decisions more difficult
to reach. The economics of obsfucation or whether more information is
always better is an area that deserves further study.

This research was primarily a case study which addressed some
aspects relevant to valuing information systems. Case studies of other
information systems also seem justified in order to arrive at a set of
principles or guidelines which can be used in a general way to place a
value on information systems. From this could come methods for estab-
lishing priorities among different information systems in addition to
setting priorities within systems. Until a more universally acceptable

method for valuing information is found, much more research in this

219
area is necessary. The difficulties in determining a suitable unit of
measurement for information is a hindrance in measuring the value of
finformation which also tends to make the case study method currently

more appropriate and likely to yield more results of long term value.

APPENDIX

22C)

QUESTIONNAIRE DIRECTED TO USERS OF USDA
FARM INCOME DATA

INSTRUCTIONS: If you do not use the U.S. Department of Agriculture farm income data please refer this
questionnaire to a person in your organization that uses it. If the data is no longer used by anyone in
your organization please answer all of the questions in section II of this questionnaire and return.
Please read each question carefully before answering. Complete sentences are not necessary -- use phrases
or whatever structuring suits your purpose in answering the questions. All information will be considered
confidential. No identities of individuals, firms or organizations will be revealed in tne report derived
from this :urvey. The form does carry a control number so that we can send reminders to those who do
not respon .

I. USES AND PRIORITIES
l. a) What is your primary use of USDA farm income data?

 

 

 

 

b) List any other of your uses of USDA farm income data.

 

 

 

2. On what types of decisions do the farm income data that you use have the most impact?

 

 

 

 

 

3. Hypothetically, if you could not obtain data on aggregate fanm income from any other source, how much
would you be willing to pay on an annual basis to receive the data on farm income as it is now
published by the U.S. Department of Agriculture?

 

4. a) Are you familiar with the general methods used in making the annual USDA farm income estimates?
yes no

b) If yes, describe these methods briefly.

 

 

 

 

 

 

221

5. Farm income data are made up of many different components and are presented in different formats.

Indicate the usefulness for your purposes of these various components and formats by checking the
most appropriate response given in front of each.

VU - Very Useful MU - Moderately Useful RU - Rarely Useful DU - Do Not Use
VU MU RU DU

a) Gross and net farm income by quarters

b) Annual realized gross and net farm income

c) Annual total net farm income

d) Gross and net farm income by state

e) Cash receipts by state

f) Cash receipts by commodity
‘ 9) Cash receipts by month

h) Value of home consumption

1) Number of farms and average income per farm

1) Personal income and disposable personal income of the farm population

k) Per capita income of the fanm population

1) Nonmoney and other fann income furnished by farms

m) Farm production expenses

n) Current farm operating expenses

0) Expenses for hired farm labor

p) Repairs and operation of farm capital items

q) Farm gross capital expenditures and net investment

r) Farm depreciation and other capital consumption

s) Net cash income of farm operators from farming

t) Number of farms by value of sales class

u) Farm production expenses by state

v) Government payments

w) Income per farm operator family by major source and by value of sales class
___ ___ ___ ___ x) Farm income and production expenses of farm operators by value of sales class
List any other components or formats of fanm income that you find particularly useful:

 

 

 

On whom do you usually rely for forecasts of aggregate fann income for future time periods?
U.S. Department of Agriculture

In-house or other internal estimates

Radio, TV or newspapers
Trade publications - specify:

 

___ Private consultant - specify:

 

____0ther - specify:

 

____Do not use forecasts of farm income

What types of economic indicators or statistics do you feel would best suit your needs for monitoring
the performance of the farming sector?

 

 

 

222

8. A national aggregate farm income accounting system can be designed to answer many different types of
questions. However. resource constraints require that some priority be given to certain questions over
others. Rate the importance of the questions listed below on a scale of l to S ( where l is highest
priority and 5 is lowest) by circling the appropriate number based on your own farm income data needs.

d

d—ld—‘dd—O—J-‘ddd—I—J—ld-fi

N

NNNNNNNNNNNNNNNN

u

a) How do the returns in agriculture compare with other industries?

b) What is the income of farm businesses?

c) What is the income of farm families?

d) What is the size of the farming sector in monetary terms?

e) What is the economic well being of farm families?

f) What is the distribution of income among individuals within the farming sector?
9) What is the productivity of the fanning sector?

h) What are the returns to the resources comnitted to agriculture?

i) What are the income differences by type of farm or comnodity?

j) What are the fanm income differences between regions and/or states?

k) What are the major economic features of the fanning sector?

l) How much has the level of farm income changed from previous time periods?

m) What is the level of aggregate farm income? A
n) How is the farm business related to the farm family?

0) How are income and assets related in the fanning sector?

p) What is the aggregate income of fanms accounted for on an accrual basis?

q) What is the aggregate income of farms accounted for on a 3333.basis?

3 4 5 r) What is the nonfarm or off-farm income of farm families?

 

mmmmmmmmmmmmmmmmm

uuuuuuuwwuwuuuuu
b‘b&.hhbfififib&hhh#

List any other questions that you feel should be given high priority in designing a national aggregate
farm income accounting system.

 

 

9. a)’

b)

a)

How accurate is the aggregate realized net farm income data published oy the U.S. Department of
Agriculture? Almost always within:

.___ : 0-51 ___ : 6-lOZ .___ : ll-ZOZ .___ : 2l-30% .___ Greater than t 30% of the true value.

On what do you base your perceptions of the accuracy of the farm income data?

 

 

Have your perceptions of the accuracy changed through time?
yes no ___-inadequate information to make a judgment

If yes. in what direction and why?

 

 

 

l0. What changes or improvements in the current USDA farm income data would be most useful for your
purposes? Why?

 

 

 

 

223

ll. What other U.S. Department of Agriculture data, in addition to that in the USDA Statistical Bulletin

Farm IncomeLStatistics, do you use?

 

If you use other USDA data, how does the overall usefulness of this data compare to the USDA farm
income data? (For example, I find that USDA data on production of comedities is more useful in general
than USDA farm income data)

 

 

 

II. RESPONDENT DATA

1.

For statistical purposes only, please identify your principal business or professional affiliation.

_ l) Farm or ranch operator _ 9) Private consulting finn

____2) Manufacturing/supplying fanm inputs ___JO) News medium or other publication
___ 3) Farm product processing and/or selling ___Jl) Educational institution

‘____4) Banking or financing ___12) Utility company

__ 5) Federal government _l3) Commodity trading or brokerage firm
.___ 6) State or local government ___}4) Foreign government or institution
.___ 7) Agricultural cooperative ___]5) Other - specify:

 

__ 8) Agricultural interest group

 

Please state your personal job title and description.

 

 

 

Do you wish to continue receiving the U.S. Department of Agriculture Statistical Bulletin Farm Income
Statist cs? —

yes no

Did you refer to the USDA Statistical Bulletin Farm Income Statistics while completing this questionnaire?
yes no

To be completed by those respondents who do not currently use USDA farm income data.
Why do you receive the USDA Statistical Bulletin Farm Income Statistics?

 

 

 

 

Please complete and return to: Mr. Charles H. Riemenschneider
Department of Agricultural Economics
Michigan State University
East Lansing, Michigan 48824

THANK YOU FOR YOUR HELP

 

BIBLIOGRAPHY

 

BIBLIOGRAPHY

Advisory Committee on Gross National Product Data Improvement, "Improv-
ing Non-Benchmark Estimates, III: Farm Sector," unpublished draft of
chapter for Gross National Product Data Improvement Project Report,
U.S. Office of Management and Budget, July 1977.

American Agricultural Economics Association, Committee on Economic Sta-
tistics, "Our Obsolete Data Systems: New Directions and Opportuni-
ties," American Journal of Agricultural Economics, Vol. 54, No. 5,
December 1972, pp. 867-875.

Akerlof, George A., "The Market for 'Lemons': Quality Uncertainty and
the Market Mechanism," Quarterly Journal of Economics, 1970, p. 488.

Alchian, Armen A., "Information Costs, Pricing, and Resource Unemploy-
ment," Western Economic Journal, Vol. 7, 1969, PP. 109-128.

Arrow, Kenneth J., "Economic Welfare and the Allocation of Resources for
Invention,“ in The Rate and Direction of Inventive Activity: Eco-
nomic and Social Factors, National Bureau of Economic Research,
Princeton University Press, 1962, pp. 609-626.

 

Arrow, Kenneth J., "Limited Knowledge and Economic Analysis, " American
Economic Review, Vol. 64, No. 1, March 1974, pp. 1-10.

Arrow, Kenneth J., "Vertical Integration and Communication." Bell Journal
of Economics, Vol. 6, No. 1, Spring 1975, pp. 173-183.

Barnard, C.S., "Data in Agriculture: A Review with Special Reference to
Farm Management Research, Policy and Advice in Britain," The Journal
of Agricultural Egonomics, Vol. 26, No. 3, September 1975.

 

Bartlett, Randall, Economic Foundations of Political Power, New York, The
Free Press, 1973.

 

Bawden, Lee, Thomas Browning, Thomas Carlin, Peter Emerson, George Irwin,
and Luther Tweeten, “Income and Wealth Data as Indicators of Well-
Being for PeOple Engaged in Farming," Proceedings of Workshop on
Agricultural and Rural Data, Series B, U.S. Department of Agriculture,
Egongmics, Statistics andTCOOperatives Service, May 4-6, 1977, pp.

-1 ,

 

 

Black, John 0., "Agriculture Now?" Journal of Farm Economics, vol. 9,
no. 2, pp. 137-162, April 1927.

 

224

 

225

Bonnen, James T., "Improving Information on Agriculture and Rural Life,"
American Journal of Agricultural Economics, Vol. 57, No. 5, December
1975a. pp. 753-763.

Bonnen, James T., "Problems of Agricultural Information Systems of the
United States," Workinng3pers on Agriculture, Food,_and Nutrition
Information Systems, Congress of the United'States, Office of Tech-
nology Assessment, June 1975b. PP. 97-128.

 

Bonnen, James T., "An Information System Approach to National Agricultu-
ral Data Collection, Analysis and Decision," paper presented at the
ADC-IADS WorkshOp on Minimum Information Systems for Agricultural
Development in Low Income Countries, Oxford, England, December 9,
1976.

Bonnen, James T., "Observations on the Changing Nature of National Agri-
cultural Policy Decision Processes: 1946-1976," paper presented at
the Conference on "Farmers, Bureaucrats and Middlemen: Historical
Perspectives on American Agriculture" sponsored by the National
Archives and the Agricultural History Society, April 27-29, 1977.

Boulding, Kenneth, "Richard T. Ely Lecture: The Economics of Knowledge
and the Knowledge of Economics," American Economic Review, Vol. 56,
No. 2, May 1966. PP. 1-13.

 

Breimyer, Harold F., "'Realized' Farm Income: An Outmoded Concept?
Comment,“ American Journal of Agricultural Economics, Vol. 50, No. 2
May 1968. PP. 432-434.

 

Browning, Thomas L., and David R. Dyer, "Alternative Methods of Account-
ing for Livestock Capital Formation: An Application to Southern U.S.
Agriculture,“ Southern Journal of Agricultural Economics, December
1976, Vol. 8, No. 2, pp. 25-29.

 

Caplan, Nathan and Eugenia Barton, Social Indicators 1973: A Stugy of
the'Relationship Between the Power of Information and Utilization by
Federal Executives, Institute for Social Research, The University of
Michigan, September 1976.

 

Caplan, Nathan, Andrea Morrison, and Russell J. Stambaugh, The Use of
Social Science Knowledge in Policy Decisions at the National Level,
Ann Arbor, Institute for Social Research, University of Michigan,
1975.

 

Carlin, Thomas A. and Charles R. Handy, "Concepts of the Agricultural
Economy and Economic Accounting," American Journal of Agricultural
Economics, December 1974, Vol. 56, No. 5. PP. 964-975.

 

Carlin, Thomas A. and Edward I. Reinsel, "Combining Income and Wealth:
An Analysis of Farm Family 'Well Being'," American Journal of Agri-
cultural Economics, Vol. 55, No. 1, February 1973, pp. 38-44.

 

Carlin, Thomas A. and Allen G. Smith, "A New Approach in Accounting for

 

226

Our Nation's Farm Income," Agricultural Finance Review, Vol. 34, July
1973. pp. 1-6.

Cobb, Charles and Wilson Kaiser, "Methodology for Estimating Distribu-
tions of Farms and Farm Income, by Value of Sales Class," Statisti-
cal Reporter, March 1969, pp. 141-145.

 

Churchman, C. West, The Systems Approach, New York, Dell Publishing Co.,
1968.

 

Churchman, C. West, The Design of Inquiring Systems, New York, Basic
Books, 1971.

 

Clark, Charles T. and Lawrence L. Schkade, Statistical Analysis for Ad-
ministrative Decisions, second edition, Cincinnati, South-Western
Publishing Co., 1974.

 

 

Cochrane, Willard W. and Mary E. Ryan, American Farm Policy: 1948-1973,
Minneapolis, University of Minnesota Press, 1976.

 

Crano, William D. and Marilynn B. Brewer, Principles of Research in
Social Psychology, New York, McGraw Hill Book Company, 1973.

 

Davis, G.B., Management Information Sygtems: Conceptual Foundations,
Stucture and Development, New York, McGraw-Hill, 1974.

 

 

Debertin, D.L., G.A. Harrison, R.J. Rades and L.P. Bohl, "Estimating the
Returns to Information: A Gaming Approach," American Journal of
Agricultural Economics, Vol. 57, No. 2, May 1975, pp. 316-321.

 

 

Demesetz, H., "Information and Efficiency: Another Viewpoint," Journal
of Law and Economics, Vol. 12, April 1969, pp. 1-22.

DeNeufville, Judith Innes, Social Indicators and Public Policy, New York,
Elsevier Scientific Publishing Company, Inc., 1975.

 

Duncan, Joseph W., "Developing Plans and Setting Priorities in Statisti-
cal Systems," Statistical Reporter, August 1976. PP. 281-286.

 

Duncan, Joseph W., "Priority Setting in the Coming Decade," Statistical
Reporter, April 1977, pp. 221-228.

 

Dunn, Edgar 5., Social Information Processing_and Statistical Systems--
Change and Reform, New York, Wiley - Interscience, 1974.

 

Dyer, David R., "Livestock Capital Formation: A Preliminary Estimation,"
Western Agricultural'Economics Association Proceedings, 1975, pp.
226-229.

Eisgruber, Ludwig M., "Micro- and Macro-Analytical Potential of Agricul-
tural information Systems," Journal of Farm Economics, Vol. 49, 1967,
pp. 15 1-52.

227

Eisgruber, Ludwig M., "Managerial Information and Decision Systems in
the U.S.A.: Historical Developments, Current Status, and Major Is-
sues," American Journal of Agricultural Economics, Vol. 55, No. 5,
December 1973, pp. 930-937.

Falconer, J.I., "Research in Agricultural Income," Journal of Farm Eco-
nomics, Vol. 10, No. 1, ll. 71-79, January 1928.

Freund, John E., Modern Elementary Statistics, third edition, Englewood
Cliffs, New Jersey, Prentice-Hall, Inc., 1967.

Gardner, Bruce, "Strategies for Long-Run Investment in Rural, Social, and
and Economic Statistics," American Journal of Agricultural Economics,
Vol. 57, No. 5, December 1975, pp. 892-899.

Goldman, Stanford, Information Theory, New York, Prentice-Hall, Inc. 1953.

Garden, Raymond L., Interviewing: Strategy, Technigues and Tactics,
Homewood, Illinois, The Dorsey Press, 1969.

Grommet, Allen C., The Influence and Behavior of Interest Group Leaders
in the Public Decision Process: A Model with Four Rural Applications,
unpublishéd'Ph.D. Dissertation, Department of Agricultural Economics,
Michigan State University, 1975.

 

Grossman, Sanford J. and Joseph E. Stiglitz, "Information and Competitive
Price Systems," American Economic Review, Vol. 66, No. 2, May 1976,
pp. 246-253.

 

Grove, Ernest W., "Realized Farm Income: An Outmoded Concept," Journal
of Farm Economics, Vol. 49, No. 4, November 1967, pp. 795-805.

 

Guebert, Steve, "Farm Income: Data and Concepts in Review," paper pre-
sented at Workshop on Farm Sector Financial Accounts, Washington
D.C., April 14-15, 1977.

Hadwiger, Don F. and Richard Fraenkel, "The Agricultural Policy Process,"
in James E. Anderson, ed., Economic Regulatory Policies, D.C. Heath
and Co., Lexington, 1976, pp. 39-50.

 

Harp, Harry H., "Measuring Income Originating in Marketing Food Products,"
Marke;jngand Transportation Situation, USDA-ERS, HTS-193, May 1974,
pp. 1 -19.

Harrison, Virden L., Accounting Methods Allowed Farmers: Tax Incentives
and Consquences, Washington D.C., U.S. Department of Agriculture,
Economic Research Service, ERS 505, January 1973.

Hathaway, Dale E. and Kenneth L. Bachman, "National and International
Institutional Framework for Collecting, Analyzing, and Disseminating
Agricultural Data: Some Issues and Alternatives," paper presented
at the WorkshOp on Minimum Information Systems for Agricultural
Development in Low Income Countries, Oxford, England, December 8,

Hau

Ha‘

Ha

H.

228

1976.

Hauser, Philip M., "Statistics and Politics,“ The American Statistician,
Vol. 27, No. 2, 1973. pp. 68-71.

Havrilesky, Thomas M. , "Information and Economic Analysis," The Annals
of the American Academy of Political and Social Science, Vol. 412,
March 1974, pp. 64-79.

 

Hayami, Yujiro and Willis Peterson, "Social Returns to Public Information
Services: Statistical Reporting of U.S. Farm Commodities,“ American
Economic Review, Vol. 62, 1972, pp. 119-130.

 

Hayek, F., "The Use of Knowledge in Society," American Economic Review,
Vol. 35, No. 4, September 1945. PP. 519-530.

Hildreth, James, et. al. ,“Report of Task Force on Farm Income Estimates,"
unpublished report, U. S. Department of Agriculture, Economic Research
Service, January 1975.

Hildreth, R.J. and Gaylord Worden, "Revisions of Statistical Definitions:
What Is a Farm?" Proceedings of the American Statistical Association--
Social Statistic Section, 1976, pp. 42-46.

 

Hottel, J. Bruce and Robert D. Reinsel, Returns to Equity Capital by
Economic Class of Farm, U. S. Department of—Agriculture, Economic
Research Service, Agricultural Economic Report 347, August 1976.

 

 

Internal Revenue Service, 313tistics of Income--1974 Business Income
Tax Returns, Washington D.C., U.S. Government Printing Office, 1977.

 

Internal Revenue Service, Statistics of Income--1974 Corppration Income
Tax Returns, Washington D.C., U.S. Government Printing Office, 1978.

 

Johnson, Glenn L. and Lewis K. Zerby, What Economists Do About Values,
East Lansing, Michigan, Center for Rural Manpower and Public Affairs,
Michigan State University, 1973.

 

Kendrick, John W., Economic Accounts and Their Uses, New York, McGraw-
Hill, 1972.

 

Lamberton, D.M., "National Policy for Economic Information,“ Internation-
al Social Science Journal, Vol. 28, No. 3, 1976, pp. 449-465.

 

Lewis, Edgar L. and Clinton F. Wells, "A Forward Look at Cash Receipt
Estimates in ERS," contributed papers on Emerging Issues, ERS For-
ward Look, USDA-ERS, 1976/77.

Machlup, F., The Production and Distribution of Knowledge in the U.S.,
Princeton University Press, 1962.

Mack, Ruth, Planning on Uncertainty, New York, Wiley-Interscience, 1971.

229

Manetsch, Thomas J. and Gerald L. Park, System Analysis and Simulation
with Applications to Economic and Social Systems, Parts I and 11,
second préliminary edition, Department of Electrical Engineering and
System Science, Michigan State University, August 1974.

Marschak, Jacob, "Economics of Inquiring, Communicating, Deciding,"
American Economic Review, Vol. 58, May 1968, pp. 1-18.

Marshall, J. M., "Private Incentives and Public Information," American
Economic Review, Vol. 64, June 1974, p. 373.

 

Maynes, E. Scott, James N. Morgan, Weston Vivian and Greg J. Duncan, "The
Local Consumer Information System: An Institution-To-Be?" The Jour-
nal of Consumer Affairs, Vol. 11, No. 1, Summer 1977, pp. 17-33.

 

Mayer, Leo V. and J. Dawson Ahalt, "Public Policy Demands and Statistical
Measures of Agriculture," American Journal of Agricultural Economics,
Vol. 56, No. 5, December 1974, pp. 984-988.

 

McDonough, Adrian M., Information Economics and Management Systems, New
York, McGraw-Hill, 1963.

Melichar, Emanuel, "Overall Measures of Financial Performance," Agricul-
tural Finance Review, Vol. 37, February 1977, pp. 79-96.

 

Miller, James Nathan, "They're Giving Us Gas, All Right," .The New Repub-
113, Vol. 176, No. 7, February 12, 1977, pp. 15-17.

 

Miller, Thomas A., "Value of Information--A Project Prospectus," unpub-
lished report, Economic Research Service, U.S. Department of Agricul-
ture, September 1977.

Mitroff, Ian I., and Louis R. Pondy, "On the Organization of Inquiry: A
Comparison of Some Radically Different Approaches to Policy Analysis,"
Public Administration Review, September/October 1974.

Mitroff, Ian I. and Francisco Sagasti, "Epistemology as General Systems
Theory: An approach to the Conceptualization of Complex Decision-
Making Experiments," Philosophy of the Social Sciences, Vol. 3, 1973,
pp. 117-134.

Mitroff, Ian 1., and Murray Turoff, "The Whys Behind the Hows,“ IEEE Spec-
trum, March 1973.

Morgenstern, 0., On the Accuracy of Economic Observations, second edition,
Princeton University Press, 1963.

Moser, C.A. and G. Kalton, Survey Methods in Social Investigation, second
edition, New York, Basic Books, 1972.

Moulton, Kirby, A. Levinson and P. Thomas, The Feasibility of Measuring
Benefits of the California Federal-State Market News Service, Univer-
sity of Cilifornia Cooperative Extension Special Publication 8003,
Berkeley, California, September 1974.

 

230

Nathan, Richard P., Allen D. Manvel, and Susannah E. Calkins, Monitoring
Revenue Sharing, Washington D.C., The Brookings Institution, 1975.

 

Newman, Geoffrey, "An Institutional Perspective on Information," Inter-
national Social Science Journal, Vol. 28, No. 3, 1976. PP. 466-492.

Olson, Mancur, The Logic of Collective Action, Cambridge, Harvard Univer-
sity Press, 1971.

 

Panel on Methodology for Statistical Priorities, Committee on National
Statistics, National Academy of Sciences, Setting Statistical Prior-
ities, U.S. Department of Health, Education, and Welfare, 1976.

 

Parker, R. H. and G. C. Harcourt, eds., Readings in the Concept and Mea-
surement of Income, Cambridge University Press, 1969.

Paul, Duane A., Richard L. Kilmer, Marilyn A. Altobello, and David N.
Harrington, The Changing Fertilizer Industry, U.S. Department of Agri-
culture, Economic Research Service, Agricultural Economic Report No.
378, August 1977.

 

Public Law 95-113, The Food and Agriculture Act of 1921, Washington D.C.,
U.S. Government Printing Office, September 29, 1977.

 

Purcell, Wayne 0., An Appraisal of the Information System in Beef Market-
iflfls unpublished Ph.D. Thesis, Michigan State University, 1966.

 

Radner, R., "Problems in the Theory of Markets Under Uncertainty," Ameri-
can Economic Review, Vol. 60, No. 2, May 1970, pp. 454-460.

 

Randall, Alan, "Information, Power and Academic Responsibility," American
gournal of Agricultural Economics, Vol. 56, No. 2, May 1974, pp. 227-
34.

Randall, C. Kyle, "'Realized' Farm Income: An Outmoded Concept? Comment,"
Amerlggnggournal of Agricultural Economics, Vol. 56, No. 2, May 1968,
PP- - -

Reinsel, Edward 1., Farm and Off-Farm Income Repgrted on Federal Tax Re-
turns, USDA, ERS-383, August 1968.

Rich, Robert F., "The Selective Utilization of Social Science Related In-
formgggop4by Federal Policy-Makers," nguiry, Vol. 12, September 1975,
pp. - 5.

Riemenschneider, Charles H., "Economic Structure, Price Discovery Mecha-
nisms and the Informational Content and Nature of USDA Prices,"
Procegdings of Workshop_on Agricultural and Rural Data, Series A.,
U.S. Department of Agriculture, Statistical Reporting Service, May
4-6, 1977. PP. 20-46.

Ripley, Randell B. and Grace A. Franklin, Congress, The Bureaucracy, and
Public Policy, Homewood, Illinois, The Dorsey Press, 1976.

 

231

Rosenblatt, David, Ezra Glaser, and.Marshall K. Wood, "Principles of De-
sign and Appraisal of Statistical Information Systems," The American
Statistician, Vol. 24, No. 4, October 1970.

 

 

Ross, Arthur M., "The Data Game" in Inside the System, second edition,
edited by Charles Peters and John Rothchild, New York, Praeger Pub-
lishers, 1973, pp. 193-205.

 

Rossmiller, George E. and Charles H. Riemenschneider, "Information Theory,
Systems Simulation and Decision Making," in Proceedings: Workshop
on Capital Investment Decisions, Electric Power Research Institute,
Palo Alto, California, and Michigan State University, East Lansing,
Michigan, November 14-15, 1977, pp. 183-197.

 

 

Ruggles, Nancy and Richard Ruggles, The Design of Economic Accounts, New
York, Columbia University Press, 1970.

 

Salop, Steven, "Information and Monpolistic Competition, " American Eco-
nomic Review, Vol. 66, No. 2, May 1976. Pp. 240-245.

 

 

Savage, Richard, "Setting Statistical Priorities," Review of Report by
same name by the National Academy of Sciences, Statistical Reporter,
December 1976, pp. 77-81.

 

Schultze, Charles L., National Income Analysis, third edition, Englewood
Cliffs, New Jersey, Prentice-Hall, Inc., 1971.

 

Shiskin, Julius, "Objectives and Priorities in Statistical Programs,"
Statistical Rgporter, October 1973, pp. 57-66.

 

Simunek, Richard W., "Cr0p Production Accounting in the Farm Sector Using
National Economic Accounting Concepts," Western Agricultural Economics

 

Association Proceedings, July 1975, pp. 230-233.

Simunek, Richard W. and Carson D. Evans, "Capital Finance Accounting in
the Farming Sector," Agricultural Finance Review, Vol. 36, April 1976,
pp. 42-49.

Spence, A. Michael, "The Economics of Internal Organization: An Introduc-
tion," Bell Journal of Economics, Vol. 6, No. 1, Spring 1975, p. 163.

Starrett, David, "Social Institutions, Imperfect Information, and the Dis-
tribution of Income," Quarterlyidournal of Economics, Vol. 90, May
1976. Pp. 261-84.

Stigler, George J., "The Economics of Information," Journal of Political
Economy, Vol. 69, No. 3, 1961, pp. 213-225.

 

Stine, O.C., "Parity Prices," Journal of Farm Economics, Vol. 28, No. 1,
February 1946, p. 301-305.

 

Strickler, Paul E., "Note 2--Farm Machinery and Equipment," in Structure
of Six Farm Input Industries, U.S. Department of Agriculture, Econo-
mic Research Service, ERS-357, Washington D.C., U.S. Government

232
Printing Office, January 1968.

Stucker, T.A., J.B. Penn, and R.D. Knutson, “Agricultural-Food Policy
Making: Process and Participants," U.S. Department of Agriculture,
ERS-AFPR-l,(Agricultural-Food Policy Review, January 1977, pp. 1-11.

Sundquist, W.B., "Changing Structure of Agriculture and Resulting Statis-
tical Needs," American Journal of Agricultural Economics, Vol. 52,
No. 2, May 1970. pp. 315-320.

Tolley, H.R., "Objectives in National Agricultural Policy," Journal of
Farm Economics, Vol. 20, No. 1, February 1938, p. 24-36.

 

 

Thurow, Lester C., "Analyzing the American Income Distribution," Ameri-
can Economic Review, Vol. 60, No. 2, May 1970, pp. 261-269.

 

Upchurch, M.L., "Toward a Better System of Data for the Food and Fiber
Industry," American Journal of Agricultural Economics, Vol. 56, 1974,
pp. 635-37.

Upchurch, M.L., "DevelOpments in Agricultural Economics Data," in A Sur-
vey of Agricultural Economics Literature, Volume 2, Quantitative
Methods in Agricultural Economics, 19405 to 19705, George Judge, 33,
31,, editors, Minneapblis, University of Minnesota Press, 1977.

U.S. Department of Agriculture, Farm Income Situation, various issues.

U.S. Department of Agriculture, Farm Income Statistics, Statistical Bul-
letins NOS. 547, 557, 576, July 1975, 1976, 1977.

 

U.S. Department of Agriculture, "Gross and Net Farm Income," Major Sta-
tistical Series of the U.S. Department of Agriculture: How They Are
Constructed and Used, Agriculture Handbook 365, Vol. 3, September
1969.

 

 

U.S. Department of Agriculture, Parity Returns Position of Farmers, Report
to the Congress of the United States, 90th Congress, lst Session,
Senate Document No. 44, Washington D.C., U.S. Government Printing
Office, 1967.

 

U.S. Department of Agriculture, Statistical Reporting Service, Crop Values,
Washington D.C., January 29, 1976.

 

U.S. Department of Agriculture, Statistical Reporting Service, Farm Pro-
duction Egpenditures, various issues.

 

U.S. Department of Agriculture, Statistical Reporting Service, Miscella-
neous Publication No. 1308, chpe and Methods of the Statistical Re-
porting Service, Washington D.C., U.S. Government Printing Office,
July 1975.

 

 

U.S. Department of Agriculture, World Food and Agricultural Outlook and
Situations Board, Agricultural Outlook, various issues.

 

233

U.S. Department of Commerce, Bureau of Economic Analysis, Survey of Cur-
rent Business, various issues.

 

U.S. Department of Commerce, Office of Federal Statistical Policy and
Standards, “A Framework for Planning U.S. Federal Statistics, 1978-
1989: Agricultural Statistics," Statistical Reporter, November 1977,
pp. 28-41.

 

U.S. House of Representatives, Committee on Small Business, Small Busi-
ness Problems in the Marketing of Meat and other Commodities, 95th
Congress, 2nd Session, Washington D.C., U.S. Government Printing Of-
fice, 1978.

 

Weeks, Eldon E., Aggregate National Agricultural.Data--Status and Alter-
natives, A Current Progress Report for the Economic Research Service,
U.S. Department of Agriculture, March 1971.

Weeks, Eldon E., "The Concept and Measurement of Farm Income and General
Welfare in Agriculture," Western Agricultural Economics Association
Proceedings, July 1972, pp. 259-266.

 

Weeks, Eldon, et. 31,, "Farm Income and Capital Accounting-Findings and
RecommendaETons of a 1972 ERS Task Force," unpublished USDA-ERS, July
1972.

Weeks, Eldon E., Gerald E. Schluter, and Leland W. Southard, "Monitoring
the Agricultural Economy: Strains on the Data System," American
Journal of Agricultural Economics, Vol. 56, No. 5, December 1974, pp.
976-983.

 

Weisbrod, Burton A., "Collective Action and the Distribution of Income:
A Conceptual Approach," in The Analysis and Evaluation of Public Ex-
penditure: The PPB System, JOint Economic Committe, 9lst Congress,
Ist Session, 1969, pp.'177-197.

 

 

Weisbrod, Burton A. and W. Lee Hansen, "An Income-Net Worth Approach to
Measuring Economic Welfare," American Economic Review, Vol. 58, De-
cember 1968. PP. 1315-1329.

 

Williamson, Oliver E., Markets and Hierarchies: Analysis and Antitrust
Imincations, New York, The Free Press, 1975.

 

 

Wilson, Robert, "Informational Economies of Scale," The Bell Journal of
Economics, Vol. 6, No. 1, Spring 1975, pp. 184-195.

 

Wright, Karl T., "Basic Weaknesses of the Parity Price Formula for a
Period of Extensive Adjustments in Agriculture," Journal of Farm
Economics, Vol. 28, No. 1, February 1946, p. 294-300.

 

Wunderlich, Gene, "PrOperty Rights and Information,“ The Annals of the
American Academy of Political and Social Science, Vol. 412, March
1974, pp. 80-96.

 

234

Youngberg, Garth, "U.S. Agriculture in the 19705: Policy and Politics,"
in James E. Anderson, ed., Economic Regulatory Policies, Lexington,
Massachusetts, D.C. Heath and Co., 1976, pp. 51-68.

“1111111111111“