r 7/ ,
Staff Papers Series
STAFF PAPER P78-10
INSTITUTIONAL FACTORS AFFECTING THE ADOPTION OF AGRICULTURAL
SECTOR ANALYSIS METHODOLOGY IN LDC'S
H. Klein and T. Roe
D, ,. i ., i U :r
*''"i ~ ~O' r j 1,;tt^r'^'st*^< E" r1" -l*\''y W\\ T^\ \ 631 }i' 01'^ m'V' r- 11' 4.i' J, I"
University of Minnesota
Institute of Agriculture, Forestry and Home Economics
St. Paul, Minnesota 55108
/ ^/n/ /^/,4elL
INSTITUTIONAL FACTORS AFFECTING THE ADOPTION OF AGRICULTURAL
SECTOR ANALYSIS METHODOLOGY IN LDC's
H. Klein and T. Roe*
Major efforts have been undertaken to both develop and apply large
scale agricultural sector analysis models (ASA) in a number of LDC's.
Well over a dozen LDC's have ASA models either in advanced or final
stages of development including Brazil, Columbia, Egypt, India, Ivory
Coast, Korea, Mexico, Nigeria, Pakistan, Thailand and Tunisia to name
some of these (13). All of these efforts have been funded for develop-
ment and directed by foreign donors with heavy participation by FAO and
the U.S. Agency for International Development. These ASA models have
been developed under a wide variety of circumstances; some have been
devised almost wholely by foreign technicians while others have been
joint efforts between foreign technical advisers and host country
Most of the literature in this area has focused on the theoretical
and analytical problems that have to be confronted in model development.
However, there is very little data concerning the institutionalization
of ASA modelling techniques within governmental decision making and planning
processes (3, p. 35). What little data there is does not look encouraging.
A 1972 USAID study evaluated agricultural sector studies to that date and
concluded that the utilization rates of these ASA efforts was quite low
and, with few exceptions, the resources expended in these efforts were
Staff papers are published without formal review within the Department of
Agricultural and Applied Economics
not justified by their usage. Interestingly, the study concluded that
there did not appear to be any relationship between the utilization rate
and the quality of the ASA effort.(10).
The general purpose of this paper is to consider some issues and
constraints relating to the process of using ASA as an operational
piece in the on-going policy decision making process of a developing
countries host institution. We shall attempt to elaborate from our
experience what some of the constraints on the institutionalization of
ASA techniques appear to be and to introduce some notions on how ASA
efforts might be modified to alleviate or to better deal with these
constraints. These issues are considered in light of (a) the generally
competing need to maintain a sufficient degree of model or technique
isomorphism so as to yield insights into the fundamental directional and
magnitudional impacts of policy instruments on variables (activities)
endogenous to the private sector, (b) the data, resource and human capital
endowments upon which an ASA effort is based and constrained, and (c) the
organizational structure of the institution hosting the effort and/or
the institution in which the effort is based. For our purposes here, we
accept the premise that sector models comprising at least some compre-
hensiveness and detail of the agricultural sector are useful policy
formulation and evaluation tools, among other uses, and that they offer
competitive alternatives to the sole use of other methods in many LDCs.
The plan of the paper is to first briefly, and non analytically,
describe the agricultural sector analysis framework developed in Tunisia
and to identify the principle participants involved in its construction.
The intent of this section is to provide a glimpse of the product in
order to focus the discussion on the constraints and issues considered in
its development. Next, a review of the literature is presented for
purposes of highlighting the concern others share for actually using
these frameworks as a comprehensive policy-decision making tool in LDC's.
Then a discussion of the issues and constraints which appear to effect
the institutionalization of ASA methods in Tunisia is presented.
II. DESCRIPTION OF THE TUNISIAN ASA MODEL
The sector analysis model was developed in Tunisia. The model is
in the tradition of the Mexican ASA model (5) in the sense that a linear
programming algorithm is used to solve it and in the sense that the
procedures for incorporating nonlinear relationships into a linear
programming format (6, 12) are also included. A statement of the
initial version of the model appears in (11) and a later version in (4).
The present version of the model represents substantial inputs and
efforts by FAO and by our Tunisian counterparts who now have full
responsibility for and control of the model.
The model is static and small by linear programming standards
having about 400 rows and 1800 columns. It is specified so that activities
identify four regions of the country, where traditional modern and government
farms are identified in each region. Within each region, production
activities are specified for the major enterprises of wheat, feed grains,
sheep, cattle, irrigated crops produced in and out of season and tree crops.
Resource transfer activities are also specified and include seasonal labor,
machinery, credit, fertilizer and forage and for traditional farms,
The constraints are specified to capture the regional seasonality of
irrigated land, water, labor, and machinery use. Also included are com-
modity balance equations, and zero balance constraints reflecting livestock
feed activities. A set of convexity constraints also appear and correspond
to the demand functions for wheat, feed grains, legumes, various irrigated
crops, mutton, beef and various fruits. The most recent addition to the
model has been the incorporation of Cobb-Douglas production functions for
four different varieties of wheat.
Other sectors of the economy are included exogenously in terms of the
capital, credit, construction materials and other resources that they
have historically supplied to the agricultural sector.
The points of view of others concerning the use of these techniques
in a policy-decision making environment are presented below.
III. THE PERSPECTIVE OF OTHERS ON THE ADOPTION OF ASA METHODS
The literature suggests that over the last five years some attention
has been given to the problem of using ASA frameworks as an operational
tool in the planning process. In an earlier paper by Atkin and Rossmiller
(1), they pointed out that sector analysis and system simulation is a
"... very useful tool in the decision makers toolbox..." but concluded
that it is not a panacea to any of the policy decision makers problem.
In his 1974 paper "General Systems Stimulation Models for Sector Analysis",
Glenn Johnson (8) reflected his concern for the problem of using analytical
frameworks in the policy-decision making processes when he pointed out the
distinction between problem-solving research and disciplinary research. He
pointed out that while the former is designed and focused to find solutions
to problems faced by real world decision makers it is often a difficult
task which does not seem to be well suited to a strict interpretation of
the scientific method. Similarly, Thorebecke (13) concluded "... that a
large gap exists between what the models can deliver and what the users
need and desire for policy formulation purposes." More recently, Bassoco
and Norton (2) demonstrated how the results might be formulated from a
single sectoral model in a manner and form to shed some light on a rather
wide variety of issues of concern to agricultural policy makers. Yet,
difficulty exists in assimilating these techniques into an ongoing policy-
decision making process.
In our opinion, Kornai (9) very incisively lists nine lessons drawn
from his experience in institutionalizing mathematical models developed
for planning purposes. While our summation of this article does not do it
justice, he essentially emphasizes the need for allocating resources to
maintain a continuous dialogue between model builder and planner even at
the expense of decreasing the "mathematical completeness" of a model. In
a similar vane of thought Sherbini (7) was perhaps on the right track when
he suggested that at least part of the problem of institutionalizing sector
analysis efforts into the decision making process is related to behavioral
organization problems such as motivational, structural and communication
constraints. He suggests that using sector analysis to demonstrate how it
can be used to improve policy rather than change it tends to engender
support on the part of central decision making authorities. He points out
that organizations tend to be compartmentalized and self contained, thereby
reducing coordination and that "... feedback communication is invariably
weak and provides poor channels for sector analysis."
These views suggest that the resources, structure and behavior of host
institutions) in developing countries play a critical role in the adoption
and effective use of ASA methodology. Consequently, the extent to which
structural and behavioral factors are considered and incorporated into
the design and use of ASA methods, the more effective these methods may
become in the policy decision making-implementing process.
IV. EXPERIENCE AND INSIGHTS DRAWN FROM TUNISIAN EFFORTS
The background to our efforts in Tunisia that are important to highlight
the issues advance here are: the point in the hierarchical decision making
structure in which the ASA was present, the data, human and other resources
at the disposal of the ASA effort the portfolio of activities assigned to
the effort and briefly the nature of the planning process.
Briefly, taking the above topics in order, Figure 1 approximately
depicts the organagram of the ministry of agriculture. The directions
appearing on the left perform the functions of information supply and to
some extent policy analysis. The offices appearing on the right are
program implementing agencies. The division of regional services refers to
approximately 12 regional offices located throughout the country in
administrative regions referred to as gouvernorats. These regional
offices primarily supply extension services and perform various data
collection and monitoring functions.
The ASA efforts were located in the direction of planning, formally
referred to as the Direction of the Plan, Economic Analysis and Project
Evaluation. The organagram of this bureau appears in Figure 2. This
Direction is composed of the four divisions with a total of about 15
people training through USAID funding to the equivalent level of a U.S.
Masters of Science in Agricultural Economics plus a cadre of people with
various lower levels of college training. The available data base at
the initiation of the sector analysis effort was "meager".
The portfolio of activities ascribed to the ASA effort included the
design of an annual planning system and the development of an initial
version of an agricultural sector analysis model, both to be located in
the Division of Planning (Figure 2). The initial endowment of resources
to undertake this activity included two counterpart Tunisians trained to
the Masters of Science level in Agricultural Economics, a small cadre
of technicians and the right to request data and other special services
from the other Divisions in the bureau, in addition to the use of an IBM
While time and space do not allow the opportunity to discuss in
depth the nature of the planning process some of its important characteristics
can be highlighted. The formal planning process in the Ministry of
Agriculture includes the periodic development of a series of documents
listing publicly supported projects, data describing the recent history
of agricultural production by subsector and its structure, and some subsector
Ministry of Agriculture
on Agric. Credit
-Bureau of State Agencies
-N Bureau of State Agencies
U *H M *i
0 0 0 .
H P Q) r4 P W $ -4 a )
c i a4 $j ~0 P4 a
*0o 0 *3 >
O U E C4 3 P
H 10 P 3S 'H E C C-
S4J 0 P4 d a)
a ** H t
0. C. Centers (145)
Farm Supply Cooperatives
* = Staff planning responsibility
0 = Office
ed 4j Wo E r ) .
ScU c -pi -r i 0 M PU f 4j a) Ti
W 6 <) X! o :2 4- : $ 0 j ca r +:
a) 0 1) *4 H 4-4 (a ) 1 U a) l
W i a) 0 0 0 '> r- 4 s- -rH
Ha) 4- O r4 0 W 0 4. i4
W a) 4- 4 44 a) P 4-4 4-J
4-4 4j 0 0 40 0 44 -$4 .0 -4
P PA 0 0
C C1 0 *) P 4-1 IT0
4-- 0 0 4 0 0 0
or-q oi P 41 P
Direction of the Plan, of Economic Analysis and
of Project Evaluation
-Service of the
Division of Eco. Analysis
and Agric. Policy
-Services of Production
-Service of Markets and
-Service of Other Policy
Division of Statistics
-Service of Economic
Division of Project
Personnel located at CRDA's
input and crop projections over the period of the plan. Committees
representing the various Directions and offices appearing in Figure 1
partake in this effort which is organized and directed by the Direction
of the Plan (Figure 2).
Now, from this background several constraints to institutionalization
of a sector analyses methodology becomes apparent. First, the Direction
in which the effort is located is not at a pivotal point in the hierarchical
structure so that the composition of the central decision making authority
is not obvious. This increases the difficulty of:
(a) formulating and articulating policy objectives,
(b) identifying policy constraints and the mechanism that
establishes these constraints,
(c) identifying policy instruments (and their dimensions) over
which central authority has discretion and the degree of
(d) developing a dialogue necessary for alternative strategy
evaluations and eventual strategy selection, and
(e) considerable effort and time is required to assess the analyses
underway in other Directions and evaluate their results and
importance relative to the total policy-making effort in
general and the sector analysis in particular.
Another set of difficulties attributed to the structure is the
need for coordination between the Directions on the one hand and the
offices on the other. This implies that a "good" policy strategy
derived from an empirical analysis may be very inefficient when
consideration is given to the Offices capacity to implement, monitor
and control the program implied by the strategy. Thus, inorder to
concentrate the analysis on alternative strategies that are feasible in
a program implementation sense, a dialogue must be maintained with these
offices. Thus, since it is not the discretion of the foreign advisor-
technician to modify the organizational structure to better fit his
sector analysis efforts, ASA efforts must be modified to fit the
structure or at least take into consideration these structural constraints.
With this background the issues considered in developing the
sector framework can be presented chronologically. The strategy followed
in developing the framework was that the maturation and institutionalization
of ASA would be determined by the amount of utility it would provide the
Direction (Figure 2), relative to its cost in terms of personnel and
other resources consumed in its development.
The issues perceived as important included: (a) the need to rely,
as much as possible, on data and sources of data collected and obtained
through normal government channels, (b) the construction of an analytical
framework in a manner which maintained, in the "eyes" of our counterparts,
a progression from field data through manipulative stages and finally,
to model coefficients, (c) the specification of a model and selection
of a technique whose complexity did not exceed the capabilities of our
counterparts, (d) the specification of variables whose resolution, time,
space and form (products, inputs) dimensions were as close as possible
to the very instruments and variables of importance to decision making
by the Direction and the Ministry.
To elucidate the above issues, the need to rely on traditional data
sources in addition to the resources required to collect field data was
primarily to ensure that once the model was constructed, its need for
periodic updating would not be constrained by the need to allocate
resources for purposes of obtaining the needed data. However, recent
events seem to show that this was not a problem as the data collection
capacity and quality have been improved considerably. In the case of
(b) above, the progression from field data to model was important as a
training device in model construction and it also provided for consistency
checks. That is, the raw data were converted into regional budgets. This
allowed for consistency checks with other estimates of regional product
and input supplies and also helped provide our counterparts with a better
initial perspective on the agricultural sector.
The specification of the model to reflect the instruments and
constraints which are within the realm of the policy makers authority to
manipulate, is, in retrospect, perhaps one of the most important considerations
determining the eventual complete institutionalization of the method in
the policy-decision making process. It also is perhaps the most difficult.
While the Tunisian ASA model is decomposed regionally and subsectorally,
further decomposition would yield additional utility to the Direction.
This follows because the program implementing functions of several of
the offices rely on instruments and resources that are more decomposed
than those appearing in the model. Furthermore, some of the variables
appearing in the model are not at the discretion of the ministry of
agriculture to-manipulate. For instance, agricultural product prices
and the prices of imported agricultural inputs are instruments at the
discretion of the Ministry of National Economy and the Ministry of Finance.
Consequently, policy analysis based on these instruments tended to engender
little interest on the part of the Ministry of Agriculture officials.
On the other hand, instruments that were later discovered to be at the
discretion of the ministry, such as storage and transportation services,
do not presently appear in the model. This obviously suggests directions
for future modifications of the Tunisian model.
The desirable degree of instrument "resolution" (i.e., the time,
place, form) and control, and the "real" objectives are often even
difficult for policy-decision making authorities to articulate. To
comprehend fully the array of instruments and structural characteristics
at the disposal of the policy makers implies a need to understand the
policy-making process, the capacity and effectiveness of the policy
implementing machinery and to be aware of the human frailities and general
behavior of these institutions. We suggest that this is the importance
of the dialogue between model builder and planner as discussed by Kornai
(9) and the need to consider the institutional constraints as advanced
by Sherbini (7).
It also appears to us that plans designed for making ASA techniques
an integral part of the policy-decision making and program implementing
framework of a ministry delegated with this responsibility, must recognize
and take into consideration that traditional channels for information flow
between policy-decision making authorities and program implementing
agencies have developed over a period of time. Consequently, these
traditional methods and channels have become institutionalized or at
least have developed an inertia in terms of the ways to make decisions
and take action. The more closely the use of ASA techniques can be made
to accommodate and to rely on these traditional structures, the more
quickly it is likely to become part of the planning process. If the
traditional organizational structure is such that ASA techniques cannot
be designed to accommodate and rely on these traditional structures, then
its our notion that they must be associated (in dialogue and office location)
as closely as possible with the decision making authorities themselves.
V. CONCLUSIONS AND IMPLICATIONS FOR FUTURE ASA EFFORTS
This paper focused on factors affecting the institutionalization of
agricultural sector analyses (ASA) as a useful policy-decision making
tool in developing countries. We suggested that the difficulties faced
in institutionalizing sector analysis efforts are probably common to
these efforts. These difficulties are associated with the resource,
structure, and behavior of the host institutions) in the developing country
which should be taken into consideration by the ASA analysts if these
methods are to exert an important influence on the policy decision making
We would also like to raise the following question: since sector
analysis experts are willing and seem interested in incorporating instru-
ments common to policy-decision making problems in their analysis, are
there not perhaps even more important though subtler issues that should
be considered in the design and use of ASA frameworks? Should some
consideration be given to formalizing some aspects of the policy formulation
implementation, feedback and control process into ASA frameworks? After
all, at the farm management level, farm plan formulation and implementation
types of issues are commonly considered by agricultural extension
economists. For example, we suggest the notion that it may be desirable,
in so far as possible, to actually include in a model some of the physical
activities of offices or agencies responsible for program implementation.
This may provide insights into administrative contradictions and incon-
sistencies, and in so doing, provide an additional degree of utility to
the policy-decision making authority.
There is an additional notion that we would like to suggest. The
difficulty of developing an analytical economic model which meets the
tests of explaining observed behavior and predicting future values of
endogenous variables is a difficult task. If we wish to include in our
empirical frameworks the instruments and other variables that are inherent
in the existing policy-decision making implementing apparatus (assuming
that we can't change it to fit our analytical methods) this is likely to
make it yet more difficult to construct models to meet traditionally or
classically established explanatory and predictive criteria.
While this criteria is desirable, its application to analytical
efforts of this type often results in a decision not to model some phenomenon
or to model it in a manner such that the variable it seeks to predict and
structure it seeks to capture are at a "resolution" or degree of aggregation
that is of little utility to a policy-decision makers. Its of little
utility because they are constrained to a given set of instrument whose
dimensions are critical to the implementing agencies of the ministry.
Another validation criterion can be suggested which has more intuitive
than analytical appeal. The criterion advanced in (11) is simply the answer
to the question, does the analytical tool help make "better" decisions
than the previous mix of methods used? "Better" decisions in the sense
that because of the use of the tool, the consequences of the policy
formulation-implementation activity produces results that are in closer
agreement with the targets or objectives for which the policy was
designed than would have otherwise been the case.
H. Klein is an Assistant Professor, School of Management, Temple
University, Philadelphia, Pennsylvania and T. Roe is an Associate Professor,
Department of Agricultural and Applied Economics, University of Minnesota,
St. Paul, Minnesota. Professors Klein and Roe were associated with a
USAID-University of Minnesota Project in Tunisia during the period 1972-74,
This research was funded by the Economic and Sector Planning Division of
the Development Support Bureau, USAID.
(1) Atkin, M. H. and G. E. Rossmiller, Sector Analysis and the General
System Simulation Approach to Agricultural Development Planning,
(paper presented at CENTO Workshop for Agricultural Planners,
Islamabad, Pakistan, 27 Nov. 4 Dec. 1972).
(2) Bassoco, L. M. and Norton, R, D., "A Quantitative Approach to
Agricultural Policy Planning," Annals of Economic and Social
Measurement, 4/4, 1975.
(3) Biggs, S. D., "A Review of Agricultural Sector Models," Agricultural
Development Council Interregional Seminar on Agricultural
Sector Analyses, Singapore, November 1976.
(4) Condox, A., Agriculture Sector Analysis in Tunisia (paper presented
at FAO/SIDA seminar on Agricultural Sector Analysis in the
Near East and Naib Africa, Cairo, 20-26 October 1975).
(5) Duloy, J. H. and Norton, R. D., "CHAC, A Programming Model of
Mexican Agriculture," Goreux and Manne eds., and Multi-Level
Planning Case Studies in Mexico, North-Holland, 1973.
(6) Duloy, J. H. and Norton, R. D., "Prices and Incomes in Linear
Programming Models," AJAE, Nov. 1975, p. 590-600.
(7) El-Sherbini, A.A., Agricultural Sector Analysis in Developing Countries:
Some Institutional Constraints (paper presented at the FAO/SIDA
Seminar on Agricultural Sector Analysis in the Near East.and
North Africa, Cairo, 20-26 October, 1975).
(8) Johnson, G. L., General, Systems, Simulation Models for Sector
Analysis (paper presented at a seminar on the Evaluation of the
Korean Agricultural Sector Simulation Model, Arlie House,
Arlie,~Virginia, 27-29, March 1974).
(9) Kornai, J., "Models and Policy: The Dialogue Between Model Builder and
Planner," in Blitzer, Clark and Taylor eds., Economy-Wide Models
and Development Planning, Oxford University Press, 1975.
(10) Rice, E. B. and E. Glaeser, "Agricultural Sector Studies, An Evaluation
of AID's Recent Experience," AID Evaluation Paper 5, USAID,
(11) Roe, T. L. and Draoui, H., The Tunisian Sector Analysis Model, Initial
Version, and Sector Analysis: Objectives and Constraints to
Framework Design, Bureau du Plan et due Developpement Agricole,
Tunis, Tunisia, 1973.
(12) Roe, Terry, Modeling of Nonlinear Functions into a Linear Programming
Format, Staff Paper, Department of Agricultural and Applied
Economics, University of Minnesota, St. Paul, 1975.
(13) Thorbecke, E., Sector Analysis and Models of Agriculture in Developing
Countries, Department of Economics, Cornell University, 1974.
(14) Vercueil, J., Integration of Agriculture Sector Analysis Models in
the Planning Process (paper presented at the FAO/SIDA Seminar
on Agriculture Sector Analysis in the Near East and North Africa,
Cairo, 20-26 October 1975).