Front Cover

Title: Assessing the impact of international agricultural research and development programs
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00085394/00001
 Material Information
Title: Assessing the impact of international agricultural research and development programs
Physical Description: Book
Creator: Horton, Douglas E.
 Record Information
Bibliographic ID: UF00085394
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: 14219996 - OCLC

Table of Contents
    Front Cover
        Page 452
        Page 453
        Page 454
        Page 455
        Page 456
        Page 457
        Page 458
        Page 459
        Page 460
        Page 461
        Page 462
        Page 463
        Page 464
        Page 465
        Page 466
        Page 467
        Page 468
Full Text
O/. 6 7

Assessing the Impact of
International Agricultural
Research and Development

Douglas Horton

Reprinted from
World Development
Vol. 14, No. 4, 1986

OI Y Wasinn, D S.

Reprint No. 94

World Developimeu,, Vol. 14, No. 4, pp. 453 468, 1986.
Printed in Great Britain.

(305-750X/86 $31.( 0.00
Pergamon Journals 1.Id.

Assessing the Impact of International Agricultural

Research and Development Programs*

International Potato Center, Lima, Peru

Summary. A number of studies are being conducted to assess the impacts of international
agricultural programs. This "second generation" of impact studies can draw on earlier
assessments of the production impacts of new, high-yielding rice and wheat varieties, particularly
in Asia. New approaches and new data are needed, however, to assess the production impacts of
other types of new technology (e.g., pest control methods, seed systems, and post-harvest
technology) that are used with other crops and livestock, particularly in Africa and Latin
America. Innovative study approaches are also needed to assess the institutional impacts of
international programs. Given the high degree of institutional interdependence which characte-
"rizes agricultural research and extension it is inappropriate to attribute production impacts at the
farm or consumer levels to specific international programs, as has sometimes been done. As an
alternative to the narrow production economics framework employed in most previous impact
studies, an interdisciplinary approach is outlined.


The current environment of increasingly scarce
resources for agricultural research and extension
has spawned a number of "impact studies" which
are intended to provide the donor organizations
of industrial nations as well as policy-making
groups in developing countries with measures of
the value of international programs. The most
ambitious and comprehensive of these is a study
of the impact of the Consultative Group on
International Agricultural Research (CGIAR).'
Other organizations, including Canada's Inter-
national Development Research Centre and the
United States Agency for International Develop-
ment, are also evaluating the impact of their
agricultural projects.
The CGIAR has been in operation for more
than a decade, and numerous bilateral agencies
have supported agricultural programs for at least
this long. Hence, it is understandable that many
people are asking, "What have these programs
accomplished?" "What has been their contri-
bution to agricultural growth and development?"
(ex post assessment) and "What can be expected
in the future?" (ex ante assessment). Results of
impact studies may influence the total allocation
of funds to agricultural research and extension as
well as the distribution of funds among different

programs. Lessons drawn from the experiences
of the multilateral and bilateral programs may
influence the types of institutional strategies that
are used in future agricultural programs. These
institutional lessons may also be useful to groups
outside of agriculture who wish to organize
research networks in such areas as public health
and population control.
The current impact studies which I will term
"second generation studies" can draw on the
methodology and empirical results of a substan-
tial body of economic literature on agricultural
research which includes hundreds of studies of

*This is a revised and expanded version of a paper
presented at the "Workshop on Methodological Prob-
lems in Research on Impact," organized by the CGIAR
Impact Study Team and held at the World Bank,
Washington, D.C., 25-27 April 1984.
tAt the time this paper was written, the author was a
Visiting Research Fellow, International Food Policy
Research Institute (IFPRI), Washington, D.C.; he has
now returned to his post at the International Potato
Center (CIP), Aptdo. 5969, Lima, Peru. Jock Ander-
son, Kenneth Brown, Dana Dalrymple, Robert Herdt,
W. 0. Jones, Robert Rhoades, V. W. Ruttan and
Richard Sawyer provided useful comments on earlier
versions of this paper. The author is particularly
indebted to Gelia Castillo for encouragement and
shared insights during the initial drafting of the paper.


the impact of new technologies.2 However, most
of the earlier, "first generation" studies were
concerned with the development and spread of
new varieties3 and their impact on crop yields,
production, and social welfare. The second
generation studies are also concerned with the
impact of international programs on research and
extension capacity in developing countries.
Methodologies for assessing these institutional
impacts are notable for their absence in the first
generation impact studies. Hence, the current
studies face a considerable methodological chal-
In estimating costs, benefits, and rates of
return, most first generation impact studies
employed rather simple economic models, heroic
assumptions, and questionable data. In this
paper I discuss some problems associated with
these models, assumptions, and data, and
present a case for broader, interdisciplinary
approaches that, while perhaps less elegant or
methodologically rigorous, are likely to produce
results which are more useful for policy makers,
research managers, and donor organizations.
Throughout the paper I take as my point of
reference the current concern for assessing the
impacts of international programs, as contrasted
to national or subnational programs.
Section 2 discusses the two distinct types of
impact that must be assessed: production impact
and institutional impact. Section 3 follows with a
critique of the overly simplistic production eco-
nomics framework and the data used in many
previous impact studies. Section 4 reviews some
difficulties involved in assessing the production
impacts of three specific types of technology: pest
management, seed systems, and storage. In
Section 5, I turn to the most important, and
generally neglected, sphere of impact for inter-
national programs: institutional impact. Section 6
treats the sensitive issues involved in attributing
specific impacts to different international and
national institutions. Section 7 outlines an
interdisciplinary approach used by one inter-
national agricultural research center to assess the
impacts of its research and training programs.


In impact assessment, it is useful to distinguish
between two types of technology that may be
generated by an international agricultural re-
search program production technology and
R & D (research and development) technology
- and the corresponding two types of impact -
production impact and institutional impact.
"Production technology" refers broadly to all

methods which farmers, market agents, and
consumers use to cultivate, harvest, store, pro-
cess, handle, transport, and prepare food crops
and livestock for consumption. "R & D tech-
nology" refers to the organizational strategies
and methods used by research and extension
programs in conducting their work. "Production
impact" refers to the physical, social, and eco-
nomic effects of new cultivation and post-harvest
methods on crop and livestock production,
distribution, and use and on social welfare in
general (including the effects on employment,
nutrition, and income distribution). "Institution-
al impact" refers to the effects of new R & D
technology on the capacity of research and
extension programs to generate and disseminate
new production technology.
Most impact studies to date have assessed the
production impacts associated with new cereal
varieties. While new varieties and complement-
ary production technologies are of undeniable
value, they are not the only, nor are they likely to
be the most important, outputs of international
agricultural programs. This is because production
problems change over time, and national pro-
grams not international programs must
solve most of them. The budgets of national
programs represent about 95% of the total
resources available for agricultural research in
developing countries, and those of international
programs only about 5% (CGIAR, 1985). A
stream of new production technologies is needed
to solve future production problems and main-
tain agricultural growth. Hence, R & D technolo-
gies that improve the capacity of national pro-
grams to generate new production technology
can give international programs a substantial
multiplier effect. Indeed, as Evenson (1977) and
others have shown, the greatest beneficiaries of
international agricultural research are countries
with strong national programs. Because of the
great variability of farming systems and produc-
tion problems, national and subnational
programs have a comparative advantage in gen-
erating production technologies, whereas inter-
national programs have a comparative advantage
in generating R & D technologies.
R & D technologies include scientific pro-
cedures for genetic engineering, screening germ-
plasm, disease identification and eradication, and
rapid multiplication of vegetatively propagated
crops. They also include organizational models,
like the integrated commodity program, and
institutional strategies for program planning and
evaluation, training, "networking,"4 on-farm
trials, and interdisciplinary team research in-
volving social and biological scientists.
The first international agricultural research


centers, CIMMYT and IRRI, produced new
varieties of wheat and rice. However, shipments
of seed the classical, physical "technology
transfer" are now only one of several mechan-
isms used by international programs to distribute
their research output. Even the "new seeds"
produced by international centers are now best
thought of as R & D technologies, rather than
finished production technologies, since they are
usually destined for breeding or screening pro-
grams rather than for immediate use as varieties
by farmers (Dalrymple, 1979).
Perhaps the most important contribution of
international programs is to set the tone for
agricultural research in developing countries. In
this respect, institutional strategies such as the
multidisciplinary commodity program, on-farm
research, and networking for germplasm evalu-
ation, cropping systems studies, and other joint
research ventures all of which were developed
through the collaborative efforts of international
programs and national research centers are
now widely used in the developing countries.
The foregoing presumes that use of new R & D
technologies will, in turn, generate improved
production technologies. Many institution-
building efforts have failed to produce either
better institutional performances or improved
technology. This highlights the need to examine
the strategies and outputs of international pro-
grams and to carefully assess their impacts on
institutional performance as well as production.


(a) The production economics framework

Estimates of the impact of new technologies on
production are useful in and of themselves and
also because they are the starting point for
assessments of other economic and social bene-
fits at both the farm and aggregate levels
(Hardaker et al., 1984).
Impact studies to date have generally em-
ployed a production economics framework to
assess the contributions of research to changes in
agricultural output. The general argument, as
stated in a proposal for the CGIAR impact study
(CGIAR, 1984), runs as follows:

The change in production of a particular commodity
or farming system in a specified country over a
defined period of time is measured; this change in
production is to be explained in terms of the
contributions of research and other influences. The
major, distinct production systems, or production
technologies, used by farmers are defined; at least

one of these should he a new "new" technology
the product of agricultural R & I). An estimate of
the relationship between the inputs and outputs
(i.e., a production function) is estimated for each
technology. The proportion of total output pro-
duced using each technology is estimated; this and
the volume of total output imply a corresponding
quantity of inputs used with each production
technology. Accounting for the differences in out-
puts and inputs, the net economic and social
benefits of the technologies can be estimated. The
contribution of research is the difference between
what the net value of output would have been in the
absence of new technologies and what it actually
was. This provides the basis for imputing the
contribution of research to the observed change in

Data required for the analysis include: (a)
estimates of changes in production in regions or
countries over time; (b) knowledge of major
production systems, including "old" and "new"
technologies, and the proportion of total output
generated by each; and (c) knowledge of produc-
tion functions for each system. If distributional
effects of technological change are to be ana-
lyzed, estimates of supply and demand elasticities
are also required.
How adequate is this framework for assessing
the production impact of new technology in
developing countries? Let us look first at the data

(b) Time series data
Each year, national estimates of production,
area, and yield of most agricultural commodities
appear in national statistical publications and in
the FAO Production Yearbook. The quality of
these estimates varies widely among commod-
ities, countries, and variables. Estimates for
developed countries, where sophisticated statist-
ical agencies have operated for years, are gener-
ally far superior to those for most developing
countries. The national statistical offices which
provide figures to the FAO generally estimate
the harvested area and production of each crop,
and calculate yields based on these two variables.
Consequently, errors in estimating area and
production are compounded in the published
yield figures.
It is costly to compile and assure the accuracy
of data on crop production, harvested area, and
yield. Hence, statistical agencies tend to
concentrate their effort on those commodities for
which the information has most value for public
and private sector decision-makers. These are
generally considered to be the "big dollar" crops
and livestock products which are traded on
international markets.


Production and yield estimates for roots and
tubers, vegetables, and other crops which are
grown in remote areas and consumed on the farm
or traded locally tend to be poorer than estimates
for cereals, oilseeds, sugar, and other crops
which are traded on international markets. The
FAO has published manuals with crop-cutting
techniques for estimating yields of the major
cereals, but no "crop digging" methods have
been published for root crops, such as potatoes.
In some cases, errors and revisions of the
statistical series are staggering. FAO estimates of
China's potato production were for example,
revised downward by 70% in 1978 and then
revised upward again 230% in 1983. These
revisions had a substantial impact on estimates of
total potato production in the developing coun-
tries. The 1983 revision amounted to an increase
of 35 million tons of potatoes more than the
total estimated potato production in the other 90
developing countries combined! China is not the
only country with controversial statistics. Simi-
larly differing estimates have been reported for
potato production in Zaire and the Philippines5
(Horton et al., 1984).
Careful analyses of time series estimates for
other crops and livestocks are not available, but
it is likely that careful scrutiny would uncover
similar errors and biases, particularly for non-
traded, "minor" crops (e.g., root crops, veg-
etables, pulses) and livestock products.
Aside from the basic issue of accuracy,
national-level estimates also mask important
variations within countries, between production
zones and different types of producers. Since
farmer adoption and the production response of
new technologies are highly location-specific,6
data which are disaggregated along the lines of
major production zones are necessary for
meaningful assessment of impact but seldom
It seems reasonable to conclude that while
available time series may reflect the approximate
magnitudes and basic trends in crop and livestock
production, they are poor sources of data for
impact assessment.

(c) Production systems

In recent years farming or cropping systems
research has come to be seen by many as a useful
complement, or in some cases an alternative, to
research on component technologies. The
volume of publishing in this field has mush-
roomed (Casement, 1982). Unfortunately, how-
ever, surprisingly few studies actually describe
and explain the logic of how farmers grow their
crops and tend their livestock. General portraits

of study areas, project activities, field methods,
preliminary results, and recommendations are
the rules; description and analysis of technology
the exception.7
Detailed cropping systems reports are avail-
able for rice production in some parts of Asia
(IRRI, 1982), and a few studies are available for
other crops, particularly those grown over large
areas as sole crops. But little has been published
on the technology used for most crops and
livestock grown on small farms, particularly in
marginal, semi-commercial areas. Rarest of all
are analyses of how production systems have
changed over time in response to the introduc-
tion of new technologies. Even in the cases of
rice and wheat production in Asia, detailed
descriptions of farmer's practices before and
after introduction of high-yielding varieties are
rare.8 Hence, many assessments of the impact of
new varieties rely more heavily on experimental
results and economic models than on empirical
studies of changes in farmers' production sys-
In the specific case of root crops, there are still
no reasonably accurate maps indicating where
these crops are grown in the tropics, much less
descriptions of how. Analysis of root crop
production systems is hampered by the fact that
these crops are grown under an extremely wide
range of ecological conditions, and they are often
produced and consumed by minor ethnic groups
in remote areas which are not well integrated into
the market economy (Horton et al., 1984).
National or regional surveys occasionally
provide data which can be used to estimate the
spread and production impact of new technolo-
gies or production systems. A few Asian coun-
tries' statistical agencies report use of "modern"
or "high-yielding" cereal varieties and the result-
ing production. For other areas, crops, and
technologies, however, few "hard" data are
available on which to base aggregate estimates of
the diffusion of new technologies and the share of
total production contributed by old and new
production systems.
Even in the case of cereal varieties, many
estimates of returns to agricultural research are
handicapped by the difficulty of identifying
"new" and "old" varieties. Terms like "IRRI
varieties," "high-yielding varieties (HYV)," and
"modern varieties" although often used are
seldom clearly defined. The principal source of
data on the spread of HYV of rice and wheat is a
series of reports authored by Dana Dalrymple
issued by the United States Department of
Agriculture in cooperation with the Agency for
International Development (sixth edition, 1978).
Dalrymple's estimates are based primarily on


data provided by AID country missions and
USDA agricultural attaches. These data, in turn,
were usually obtained from official reports or
estimates by the countries themselves. As the
author notes, "While the data have been checked
as far as possible, there is really no good way of
knowing how accurate they are" (Dalrymple,
1978, p. 5).
For other crops, the problems of estimating
use of new varieties are greater. Bangladesh, for
example, has recently published official statistics
of farmer use and yields of "HYV of potatoes." It
is generally assumed that HYV of wheat and rice
are in some sense a product of the breeding
programs of CIMMYT and IRRI, and by the
same logic one might assume that the HYV of
potatoes are a product of the breeding program
of the International Potato Center (CIP). How-
ever, this is not the case. What Bangladesh's
statistical reports consider HYV of potatoes
apparently include all varieties recently intro-
duced from abroad, mostly from Europe. How
recently they have been introduced, and the
specific criteria used to distinguish them from
non-HYV are not clear. Hence, the proportion
of potato output generated by HYV's would
clearly overstate the contribution of CIP's breed-
ing program to Bangladesh's potato production.
To the extent that this type of definitional
problem occurs with wheat and rice, previous
impact assessments based on official statistics
might have exaggerated the use and production
impact of varieties associated with IRRI's and
CIMMYT's breeding programs.

(d) Production functions

Production function analysis, now considered
"old hat" by agricultural economists in the
developed countries, has never been widely
applied in most developing countries. Again,
irrigated rice in Asia presents an exception to the
norm. For most other crops and regions little
production function analysis has been done. This
is partly due to the complexity of modeling
farming systems in developing countries and the
lack of necessary data (Anderson and Hardaker,
1979). It is also due to limitations of computer
hardware and software and to the low priority
assigned to farm management-type research. In
Latin America, for example, where the social
sciences, including economics, are macro-
oriented and highly politicized, farm manage-
ment courses are seldom taught. Moreover, there
is little incentive for students or professional
economists to do the fieldwork necessary for
meaningful production function analysis.
Impact assessors may wish to do their own

production function analysis, based on results of
farm surveys or experiments. How adequate are
these sources?

(i) Survey data
Rather than estimates of response surfaces, the
best that can be obtained for most crops and
areas are single-point cost-of-production esti-
mates which may be gleaned from documents of
ministries or agricultural banks.
Unfortunately these often reflect recom-
mended, rather than actual, farming practices.
They may be guestimatess" based on no field-
work whatever. Where they draw on farm
surveys conducted by government employees -
usually extension agents they may be mislead-
ing because extension workers generally inter-
view their clients: larger-than-average farmers in
accessible, commercial areas (Horton, 1984;
Rhoades, 1982a, b). These farmers usually con-
stitute a small minority of the rural population,
whose production practices often differ sharply
from those of most producers. A 1974 survey
conducted by extension agents in highland
Ecuador illustrates this point. Survey results
indicated that over 50% of all potato producers
used credit from the government's agricultural
bank. From the bank's own records and census
reports, however, it was clear that no more than
5% of the area's potato farmers used bank credit.
The reason for this discrepancy was that the
extension workers who conducted the survey had
interviewed only -their "contact farmers" (Val-

derama and Luzuriaga, 1980). Another problem
of official estimates is that they generally include
fictitious costs for such items as minimum wages
and social security payments, which are not, in
fact, paid by farmers.
Survey results from the Mantaro Valley of
Peru's Central Highlands show how markedly
technological coefficients can vary between three
groups of farmers in a single agro-ecological zone
(Table 1). Production cost estimates available
from Peru's Agrarian Bank and Ministry of
Agriculture approximate the survey estimate for
large farmers who constitute a small minority of
potato growers in the area.

(ii) Experimental data
As an alternative to time series and survey
data, experimental results are sometimes recom-
mended as a basis for estimating the impact of
technological change on crop yields. In practice,
available experimental data seldom reflect the
performance of alternative technologies under
actual farming conditions. And for obvious
reasons impact analysts seldom have the luxury
of being able to design the needed experiments


Table I. Input use, yields, and use of output for three size groups of potato
producers in the Mantaro Valley. Peru

Farm size

Unit Large Medium Small

Labor day/ha 175 124 136
Oxen day/ha 4.4 9.4 13.6
Tractor day/ha 5.6 2.6 0.5
Seed ton/ha 1.8 1.2 1.0
Manure ton/ha 7.1 5.4 11.0
Nitrogen kg/ha 330 184 92
Phosphate kg/ha 141 122 5
Potash kg/ha 128 84 30
Soil kg/ha 36 22 74
Foliar liter/ha 4.9 3.0 0.4
Fungicides kg/ha 4.5 16.5 0.8
Yield ton/ha 21.8 9.4 3.3
Percent of
output sold % 75 40 0

Source: Horton et al. (1980).
Note: All farms are in the "Low Zone," a riverine plain along the Mantaro
River. In the production season under study (1977/78) "large farms"
harvested an average of 4.3 ha of potatoes, "medium farms" 1.1 ha, and
"small farms" 0.3 ha.

and wait several months or years until results
become available. Little can be expected from
the first cycle of experiments, aside from useful
training for those who are experimenting (Hor-
ton, 1984). A number of years of interdsciplinary
research, including farm-level trials, farm sur-
veys, and direct observation, are generally
needed to obtain accurate information on
technological change and its true impact on
yields, costs, and returns. The limitations of
available experimental results are illustrated in
the following example, drawn from Fano (1983):

By developing-country standards Peru has a long
history of potato research, dating from the 1940s.
Several thousand potato experiments have been
conducted both on the coast and in the highlands.
Most of these have included new varieties and most
have been on farms, rather than on experiment
stations. Based on these facts, it was assumed that
experimental data could readily be located to
document the effect on yield of changing from an
old to a new variety of potatoes. (The hypothesis
was that the resultant increase in yields would be
greater on the coast, where growing conditions,
particularly irrigation, were more favorable than in
the highlands.) In an exhaustive review of literature
in Peru's Agrarian University and the country's

leading experiment station, a large number of
publications were found which reported the ex-
perimental yields of new potato varieties, but only
four reports were found which compared ex-
perimental yields of both old and new varieties, and
none of these provided an indication of the levels of
non-experimental variables, such as fertilization,
pest control, and seed quality.

Many institutions have recently begun to
conduct specialized experiments on farmers'
fields to quantify the yield differentials associated
with alternative technologies. The International
Agricultural Research Centers have played an
active role in stimulating this work and develop-
ing methodologies for it.9 Results of on-farm
experiments are useful for many purposes, but
they are often misleading for impact assessment,
because they exaggerate the yield increases
which farmers would actually attain if they
switched from their present technology to a new
one (Horton, 1984; IRRI, 1977).
There are four sources of this experimental
bias in favor of new technology. One is that
on-farm experimenters often attempt to prove
they can do better than the farmer. This is
illustrated by a personal observation in the


Mantaro Valley, Peru. An on-farm experiment
was designed to measure the yields, costs, and
returns associated with two sizes of potato seed
tubers: the recommended size and the farmer's
present size (presumably smaller). In a yield
visit, I observed the researcher sorting a farmer's
seed into two size groups: large and small. He
planned to use the larger seed tubers as the
"recommended technology," in the experiment
and the small seed tubers as the "farmer's
technology." Since the farmer would have
planted the entire seed lot, with both large and
small tubers, what the researcher was defining as
the farmer's technology in the experiment was
inferior to the farmer's real technology. In effect,
the researcher was stacking the cards against the
farmer. Based on numerous similar observations,
I believe this is quite common in on-farm
A second, less obvious but nonetheless impor-
tant, source of bias stems from the fact that yields
measured on small experimental plots generally
exceed those on large fields, even when the same
technology is supposedly used. In fact, the
technology applied to experimental plots and
farmers' fields is seldom really the same: more
care is taken and more labor and capital are
generally applied on small experimental plots,
and the harvest tends to be more thorough than it
is on farmers' fields."' Experimenters also tend to
locate experimental plots on better-than-average
land the best parts of the field. If yields on
experimental plots that are designed to simulate
the farmer's and a new technology are both the
same proportion over what they would be on
farmer's fields, the absolute yield gap between
the two technologies is larger on small ex-
perimental plots than on farmers' fields. More-
over, because most experiments are conducted
on better-than-average land, where new tech-
nologies like high-yielding varieties can best
express their production potential, the yield gap
between old and new technologies is likely to be
greater in both absolute and proportional terms.
A third source of bias comes from the methods
used by researchers to reduce sources of vari-
ation among experimental sites by standardizing
the levels of non-experimental inputs at assumed
"normal," "representative," or "average" levels.
More often than not, these standardized levels
are above-average. Hence, all the treatments in
the experiment, including the "farmer control"
treatment, yield far more than they would have if
non-experimental inputs had been left at the
farmer's level. For reasons mentioned above,
elevating the level of non-experimental inputs
generally has the effect of increasing the
measured yield gap between what are considered

to be (but are not, in fact) the farmer's own
technology and the new technology.
A fourth source of bias occurs when re-
searchers do not include a farmer control treat-
ment in their on-farm experiments, but simply
compare the experimental yield of the new
technology against the yield of a "representative"
farmer's crop or the average yield for the area.
Results of this type of comparison generally show
a dramatic, but highly exaggerated, yield advan-
tage in favor of the new technology.
These biases help explain why farmers often do
not adopt technologies that are recommended on
the basis of experimental results. They also imply
that impact assessments based on experimental
results may overestimate the benefits of new
technologies. In this respect, Scobie's (1979)
estimate that adoption of HYV rice and wheat
led to average yield increases of 50 and 100%
respectively, is based on yield comparisons from
both experiment stations and on-farm trials,
which the author gleaned from over 100 second-
ary sources. In view of the substantial "yield
gaps"" between experiment stations, on-farm
trials and farmers' fields, it seems reasonable to
conclude that the experiment station results used
by Scobie almost certainly overstate the yield
increases actually obtained by farmers. The
results of on-farm trials probably did so as well.
Less frequently, an economic analysis based on
experimental results underestimates the profit-
ability and production impact of a new tech-
nology. An example is provided by a seed potato
storage project in the Philippines. Experiments
and economic analysis indicated that the new
storage technology was not likely to be adopted
because it increased yields only by a small
margin. Farmers adopted the technology, how-
ever, because it extended the period over which
they could store their seed and hastened emerg-
ence of the subsequent crop. Hence, because
they did not understand the farmers' motives for
using a new technique, evaluators had under-
rated its potential benefits.
The change in yields resulting from adoption of
HYV has been the central variable in most
impact assessments. Many new technologies,
however, are adopted by farmers not because
they increase yields, but because they reduce
costs, increase quality, or allow crops to be
planted in new areas or at different times of the
year. Such technological improvements may
generate substantial benefits for producers or
consumers even if they are lower yielding.12
The conclusion here is not that costs and
benefits associated with technological change
cannot be captured in economic analyses of
experimental results, but that in many cases they


have not been captured adequately. For impact
assessment ex post and, particularly, ex ante -
more detailed knowledge of farmers' physical
and socioeconomic environments, decision-
making processes, and technologies is needed
than has generally been recognized.

(e) Demand elasticities

Estimates of supply and demand elasticities are
necessary for partitioning, quantitatively, the
benefits of technological change among pro-
ducers and consumers. However, these elastici-
ties have not been estimated in most developing
areas for most foods. Due to the lack of time
series of food prices and consumption and the
sheer complexity of incorporating price variables
into econometric estimations, most demand pro-
jections for agricultural commodities ignore price
effects.13 In the absence of elasticity estimates for
practically all food crops but wheat and rice in
developing countries, there has been a tendency
to assume that the relatively low elasticities
observed in Western Europe and North America
apply in developing countries as well. Since
incomes are lower in developing countries and
food constitutes a much higher share of house-
hold expenditures, this is unlikely. The few
empirical studies which have been done indicate
that price elasticities of demand tend to be
substantially higher in developing countries than
in Western Europe and North America (Timmer
et al., 1983).
Another problem with available elasticity esti-
mates is that in the case of "miscellaneous,"
"starchy," or "minor foods," publications gener-
ally present average estimates for several com-
modities. Lumping such diverse crops as pota-
toes, cassava, yams, sweet potatoes and bananas
together under the single heading, "starchy
foods," has certain advantages in reducing the
size of tables, and the costs of editing, printing,
and distributing publications, but there is no
economic basis for it. These foods often play
quite different roles in the diet. For example, in
many areas where cassava is a low-cost staple
food with inelastic demand, potatoes are a
high-cost vegetable with highly elastic demand
(Horton et al., 1984). Consequently, no single
elasticity can reflect the true association between
price changes and the level of consumption of
any one of the foods in this miscellaneous


Use of production economics to quantify the

impacts of technological change is most straight-
forward where: (a) the physical environment
(e.g., relief, soils, and weather) is uniform across
all observation points; (b) the production process
generates a single output and employs few
variable inputs; (c) inputs and the output are
easily quantifiable and uniform in quality; and
(d) all inputs are purchased and the output is
sold. If the number of variable inputs and outputs
is large, their quality is variable, and production
is partially or wholly subsistence-oriented, im-
pact assessment using a production economics
framework becomes complex and its results are
influenced by a large number of simplifying
assumptions of dubious validity. In some cases -
like the use of experiment station yields as a
proxy for farm-level results simplifying
assumptions lead to overestimation of production
impact. In other cases, such conservative
assumptions may be made, to compensate for
questionable data, that studies underestimate
rates of return.

(a) Varieties vs other technologies

Such variables as yield (as the dependent
variable) and fertilizer level and variety (as
independent variables) fit the data requirements
of production function analysis quite well, par-
ticularly in highly commercialized, irrigated
areas. Yields and fertilizer levels enter the
production function as quantitative variables,
and variety as a dummy variable. Data on use of
new and old cereal varieties, fertilizer levels, and
yields are most readily available in Asia. This is
also the region where university training in
production economics is most advanced. Hence,
it is not surprising that production function
analysis has been used most widely for assessing
the impact of new cereal varieties in Asia. After
variety, fertilizer level and irrigation have re-
ceived most attention in adoption and impact
studies. This is appropriate, since the new,
semi-dwarf varieties are much more responsive
to increased fertilization under irrigation than is
the case of older varieties.
But what of improvements in other tech-
nologies, other crops, and other areas? Have
they not occurred, or have they been over-
looked? Field observations and available statis-
tics indicate that increases in production and
yields of some other crops, such as potatoes,
have been on par with, or not far behind, the
increases for wheat and rice (Table 2). In the case
of potatoes, the crop I know best, technological
improvement has been rapid not only in the use
of new varieties but also in pest control, seed
systems and post-harvest technology.'4 These


Table 2. Value of production of 10 principal food crops in developing countries and average annual
change in yield from 1961/65 to 1981/83

Number of Farm-gate change
producing Production price Value in yield
countries (10* t) ($/S) (106 US $) (%)

Rice 93 404.0 170 68,680 2.1
Wheat 66 179.5 148 26,566 3.4
Maize 115 155.6 119 18,516 1.6
Potatoes 91 82.3 142 11,687 1.8
Cassava 92 127.4 70 8,918 0.5
Sorghum 66 47.3 123 5,818 3.1
Tomatoes 97 21.5 195 4,193 2.0
Millet 51 27.2 144 3,917 1.1
Yams 41 24.0 163 3,912 1.2
Plantains 42 20.0 114 2,280 n.a.

Sources: Number of producing countries: corresponds to 1980, derived from FAO (1981).
Production: average for 1981/83, derived from FAO (1984). Farm-gate price: weighted average for
all developing market economies, provided by FAO Basic Data Unit (unpublished). Value:
production multiplied by farm-gate price. Average annual percent change in yield: corresponds with
1961/65 to 1981/83, derived from FAO (1984 and 1977).

changes have not been assessed for two reasons,
it seems: First, analysts have focused on trade-
able commodities, like wheat and rice, and
assumed that potatoes are unimportant in
developing countries. However, as shown in
Table 2, potatoes are among the most important
crops grown in developing areas in terms of the
number of producing countries and the monetary
value of production. A second reason, elabo-
rated below, is that the impact of changes in pest
management, seed systems, and potato storage
are less well understood and more difficult to
assess than is the impact of a new variety.

(i) Pest management
Pest management practices are changing
rapidly in many farming environments, but data
on changes and their socioeconomic impacts are
difficult to obtain. Farmers' pest management
strategies are extremely complex, and involve
important qualitative variations in both inputs
and outputs. Moreover, detailed understanding
of the biology of pests as well as farmers' control
measures is required for impact assessment.
Before the economic impact of changes in pest
management can be measured, an understanding
of farmers' pest problems, management
strategies, and of whether the influence to
change comes from public or private agencies, or
from neighbors, is needed. This would require
in-depth, multi-disciplinary farm-level research
which has not been done."

(ii) Seed systems
Potatoes can be multiplied sexually through
use of true potato seed, but with few exceptions,
farmers the world over plant tubers generally
1-2.5 tons per hectare. In most developing areas,
these "seed tubers" are the single most costly
input in potato production, and poor seed quality
- referring to a constellation of factors, includ-
ing disease level, physiological condition, and
size of tubers is one of the potato crop's most
important yield constraints. Variations ih seed
source often confound results of variety trials and
other experiments, since the impact of seed
quality on yield can be greater than the impact of
variety. As illustrated in the following example
from Monares (1984), assessing the impact of
improved seed systems, which embody both
technical and institutional innovations, calls for a
rare combination of biological, economic, and
institutional knowledge and analytical skill.

Rwanda's National Potato Improvement Pro-
gram (PNAP) established in 1979 uses simple
techniques to supply farmers with improved quality
seed. Without postharvest virus testing facilities.
Rwanda's seed production system depends primari-
ly on field observation of plant vigor and the
proportions of healthy and diseased plants. The
seed program now produces about 250 tons of seed
per year of which about 50% is of new Rwandan
The result is an increasing number of Rwandan
farmers with access to improved seed which gradu-


ally replaces old, degenerated varieties. Production
of improved seed is still significantly lower than
current farmer demand for seed. Yet, reports from
seed projects and preliminary surveys indicate that
about 7,000 hectares, representing nearly 20% of
the total potato area, are now planted with seed
originating from the national seed program.
PNAP conducts on-farm research trials to gauge
the performance of its improved seed. The average
yield increase on farms due to use of improved seed
is estimated to be about 3 tons per hectare a 40%
increase over traditional seed. Newly selected
varieties often yield twice as much as common
farmer seed in on-farm trials.
The program is expected to continue operating at
the same scale, and its costs will remain at about the
same level while benefits increase significantly as
cultivation of new varieties spreads. Given the slow
virus degeneration rate, the multiplier effect of a
small stock of clean seed is great.

An economic assessment of Rwanda's seed
program would require quantitative estimates of:
(a) the performance of improved seed under
existing farming conditions and (b) the seed
degeneration rate (yield reduction over time
from virus infection). An understanding of the
institutions involved in seed production and
distribution would also be needed to estimate
present and future streams of costs and benefits.
Separate assessments of the impacts of new
varieties and improved seed two quite distinct
innovations would require much more de-
tailed information and a more complex analytical

(iii) Storage
Problems of storing table potatoes have pre-
occupied developing country policy makers and
scientists for decades, but seed storage has
received little attention. There are reasons why
table potato storage has received so much atten-
tion. First, violent fluctuations in potato prices
and wide price spreads between farm-gate and
retail prices have been interpreted as evidence of
market imperfections and the need to regulate
the distribution of potatoes via market interven-
tions and storage. Second, since farmers and
processors in developed countries generally store
table potatoes, it has been assumed that their
developing-country counterparts could also
benefit from doing so. Third, it has been assumed
that developed-country storage technology could
readily be "transferred" to developing countries.
In fact, there is little economic basis for storing
consumer potatoes in many developing areas.
This is because harvests occur throughout the
year and there is no assurance that price will rise
after any given harvest. Potato storage in the
tropics is also much more costly and risky than in

temperate zones. As a consequence, most potato
storage schemes in developing countries have
In contrast to the situation with table potatoes,
and despite the general lack of political and
scientific interest in the topic, storage of seed
potatoes is a major concern of potato growers in
most developing areas. This reflects farmers'
desire to avoid purchasing seed tubers, which are
costly and often poor in quality. Purchased seed
often carries seed- and soil-borne disease, some
of which (e.g., bacterial wilt, Pseudomonas
solanacearum) can have a devastating effect on
yields and remain in soil for years. Additionally,
seed tubers in appropriate physiological condi-
tion for planting with their "eyes open" -
may not be available on local markets when
farmers would like to plant. The following Sri
Lankan example, derived from Rhoades (1985)
illustrates the types of diverse impacts which can
result from improvements in seed potato storage.

Potatoes are grown in Sri Lanka during different
months in distinct agro-ecological zones of three
main districts: Jaffna, Badulla, and Nuwara Eliya.
The last two of these, which account for about 80%
of the growing area and 90% of national produc-
tion, have two main production seasons: yala,
referring to the southwest monsoon which lasts from
mid-May to mid-September, and in this paper is
termed "early," and maha, referring to the north-
east monsoon which lasts from October to mid-
January, designated here as "late."
Until 1979, the government allowed importation
of foreign seed for both early and late seasons. Due
to difficulties in obtaining seed, and government
targets of reducing imports, a decision was reached
in 1979 to stop importation from Australia for the
early season. The only importation allowed now is
European seed for the late season, thereby creating
a serious shortage of seed for the early season.
The diffused light storage (DLS) technique is
based on using natural indirect light instead of low
temperatures to prolong seed storage and control
excessive sprout growth and storage losses. A Sri
Lankan scientist and an extensionist were trained in
DLS technology in courses in the Philippines. Upon
their return to Sri Lanka, they set up experiments
and farmer demonstrations in Badulla and Nuwara
From 1979 to 1983 Sri Lankan farmers build more
than 500 diffused light stores, and another esti-
mated thousand farmers modified their existing
storage system. Farmers seldom copied demonstra-
tion models, but modified them to meet their own
needs and budgets. Many farmers integrated dif-
fused light storage principles into existing sheds,
garages, and rooms attached to their houses.
In a 1984 survey, adopting farmers noted the
following benefits of the DLS system: (a) sprouting
is reduced and seed can be stored longer; (b)
storage losses are reduced; (c) pest control is easier;


(d) field emergence of seed stored in diffused light is
earlier, and the growing season is shorter; (e) yields
are higher; and (f) seed is now available for October
planting. In addition, farmers pointed out that seed
tubers stored in light sell for a higher price than
tubers stored in darkness.

As this example illustrates, an economic
assessment of what might appear to be simple
improvements in seed storage may incorporate a
large number of qualitative changes in both
inputs and outputs. The methodological prob-
lems are by no means insurmountable, but they
do require more detailed analysis than is needed
for economic assessment of the impact of a new
Clearly as one moves away from assessing
impact of new cereal varieties, the production
economics framework serves less well, and more
information is needed on farmers' practices, the
nature of changes (in qualitative as well as
quantitative terms), and the institutional
mechanisms responsible for them.


Attribution of the impacts of agricultural
R & D to different institutions is one of the most
sensitive issues of impact assessment. Evenson
(1977) Boyce and Evenson (1975), Evenson and
Kisler (1975) and others have emphasized the
interaction effects of national and international
research. Yet, many developing-country scien-
tists and policy makers feel that impact studies
have failed to recognize important contributions
made by their institutions.
Few, if any, new technologies are generated
solely by international programs or reach farmers
without substantial contributions from national
research and extension programs. Hence, it is
inappropriate to attribute production increases
and associated benefits solely to the work of
bilateral or multilateral programs. This has
generally been acknowledged in the introductory
sections of publications assessing the impacts of
agricultural research (see, e.g., Dalrymple, 1978,
pp. 3-5). Nevertheless, the conclusions of these
same studies have often been interpreted as
overemphasizing the role of international pro-
grams vis-a-vis national programs. Perhaps more
importantly, the popular press has disseminated
the notion that production increases have been
due solely to the good work of international
agencies. This has tarnished the image of inter-
national programs in many developing countries.
Great care is needed to ensure that impact
studies give due credit to the various con-
tributors, which in addition to national programs

in developing countries often include bilateral
agencies, foreign universities, and farmers them-
selves. The following example, derived from
Franco and Schmidt (1985) illustrates how com-
plex the chain of causation can be:

In the early 1970s the Peruvian Ministry of Agricul-
ture requested that CIP help combat a serious
outbreak of bacterial wilt in potato crops in
northern Peru. To this end, CIP obtained resistant
breeding lines from the University of Wisconsin.
These lines had been developed from potato sam-
ples sent to Wisconsin by the Colombia's National
Agricultural Research Institute (Instituto Col-
ombiano Agrario: ICA). Researchers employed by
CIP and the Peruvian Ministry selected potential
new varieties on government experiment stations in
northern Peru. Two resistant varieties were re-
leased by the Ministry of Agriculture in the mid-
1970s. One of these, called "Molinera," is now
among the most widely grown potato varieties in
northern Peru. From advanced variety trials, Peru-
vian farmers also kept and multiplied at least two
other clones which are now grown in the area. One
has gained such importance that it was recently
named and officially released as a Peruvian variety.

Such multiple causation seems to be the norm
rather than the exception in agricultural research
and extension. In the examples cited earlier from
Rwanda and Sri Lanka, and in most other
documented cases I know of, similar processes of
multiple causation were responsible for the final
One could perhaps arrive at valid procedures
for separating and quantifying the impacts of
different institutions. But, aside from being
difficult, it would seem more appropriate to
share the credit and emphasize that successful
programs are based on collaboration, rather than
unilateral efforts.


International agricultural programs represent
small but strategic components of the global
agricultural R & D system. As noted in Section 2
of this paper, international programs are sup-
pliers of R & D technology rather than produc-
tion technology per se. They can have the
following types of institutional impact:
1. Strengthen agricultural R & D programs in
developing countries, through training,
supplying research information.
2. Stimulate research institutions in devel-
oped countries to address problems which
are important in developing countries.
3. Link national programs to the global R &
D system, through improved communica-


tions, collaborative research, conferences.
4. Contribute to establishment of priorities in
key research and policy areas.
5. Inform donors and stimulate their con-
tinued support.
The methodological problems of assessing these
various influences and assigning monetary values
to them seem insurmountable, particularly in
view of the resources and time frame of most
impact studies. Before these impacts can be
measured, it is necessary to first understand the
types of impacts being achieved and the institu-
tional strategies used to achieve them. Providing
documentation in these areas is one of the most
valuable potential contributions of the second
generation of impact studies.
There is no analytical framework for assessing
institutional impacts which is analogous to the
production economics framework for assessing
production impacts. The inputs and outputs of
international R & D programs are too numerous
and qualitatively variable to be analyzed within a
single econometric model. Moreover, the policy
environments in which these programs operate
significantly affect their institutional perform-
ance and impact.
Needed are assessments of the strategies used
by different centers and the results obtained in
different settings. Given the diversity of
strategies and types of impact and the lack of
information on these topics, there is no practical
alternative to a case study approach. Case studies
should document impacts in the four areas
identified earlier in this section.


In a study recently conducted by the Inter-
national Potato Center (CIP, 1984) an attempt
was made to avoid some of the conceptual and
methodological problems of earlier impact
assessments. Responsibility for preparation of
the study was given to an interdisciplinary team
of professionals, rather than a team of econo-
mists. Scientists and policy makers from develop-
ing countries were actively involved in preparing
the study, which assessed both production and
institutional impacts. Rather than focus on
econometric estimation of the production impact
of new varieties, the CIP study describes and
illustrates a range of types of impact as well as the
institutional strategies used to achieve them.
Recognizing the significance of collaboration and
multiple causation, the study made no attempt to
attribute specific, farm-level impacts to CIP.
The study's three major sections are:
1. A description of the strategies employed in

program planning and review, inter-
disciplinary team research, institutional
linkages, and training.
2. A review of the Center's research program
and its results to date.
3. Case studies which illustrate how problems
of potato production and use have been
solved and diverse types of institutional and
production impacts achieved.
The case illustrate various types of impact,
ranging from training and institution building, to
yield increases, to the development of an effec-
tive model for interdisciplinary problem-solving.
Preparation of the CIP study was coordinated
by an economist (this author), but involved
taxonomists, breeders, pathologists, entomolo-
gists, nematologists, physiologists, seed special-
ists, research managers, policy makers, and
communication experts, as well as economists,
sociologists, and anthropologists. Major sections
of the final report were drafted by biological
scientists. Three of the seven illustrative cases
were prepared by biologists, two of whom were
national program scientists. Two other cases
were prepared by an anthropologist, and two
were written by economists.
Involvement of scientists and policy makers
from developing countries helped focus the study
on those problems and accomplishments which
CIP's clients considered to be the most impor-
tant. Individuals from several parts of Africa,
Asia, and Latin America, as well as Europe and
North America provided frank and critical
appraisals of the Center's programs, accomplish-
ments, and shortcomings. A selection of their
comments are presented in the study report.
Assessments of institutional impacts required
information on CIP's interaction with national
programs and the resulting effect on national
R & D capacities. Three distinct approaches
were used to obtain and synthesize the needed
information. Initially, country-level information
was recorded by CIP's regional scientists on a
brief questionnaire requiring simple yes/no
answers.'6 The information obtained covered
such topics as: frequency of CIP contacts
(correspondence and visits) with each national
program; participation in various types of train-
ing and seminars; distribution of research re-
ports; collaborative research projects and/or
research contracts in each country; use of CIP-
related technologies in national R & D programs
and on farms.
This country-level information was com-
plemented with more precise, quantitative data
gleaned from the Center's files on training and
collaborative research efforts, and their impacts
on production and institutional capacity.


The third approach involved preparation of
seven case studies which illustrated both the
strategies and impacts. The first case describes
CIP's relations with its host country, Peru,
emphasizing the relationship's reciprocal nature
and the resulting benefits. The second case
documents how, through trial and error, a team
of postharvest technologists and anthropologists
developed an interdisciplinary model, known as
the "Farmer-Back-to-Farmer" model, which
helps researchers identify and solve farmers'
problems in the shortest possible time (Rhoades
and Booth, 1982). The third case, from Vietnam,
illustrates how an international agricultural re-
search center can learn from a developing coun-
try's innovations, and incorporate this knowledge
into its own research and training programs, to
the benefits of other countries. The fourth case
presents an economic assessment of a seed
multiplication program in Tunisia. The fifth case
documents the rapid establishment of a research
and extension program in Rwanda, and presents
an economic assessment of the new program's
impact. The sixth example, from Sri Lanka,
illustrates how innovations in one key element of
the potato production system seed storage -
can have far-reaching effects on planting and
harvesting dates, yields, the intensity of cropping
systems, production costs, foreign exchange ex-
penditures (for seed potatoes) and potato prices.
The seventh case describes the operation of a
regional collaborative research network in Cen-
tral America and its impact on the research
priorities, staffing and performance of the
national potato programs and the linkages and
flow of information between them.
Those who are interested in the broad range of
types of technological and socioeconomic im-
pacts that international and national programs
can achieve and the strategies used to achieve
them should find much of interest in the CIP
study. Those looking for cost-benefit analyses
and rates of return to specific research activities
may be disappointed, since these are presented in
only two of the study's seven cases.


The impact of agricultural research in develop-
ing countries has been assessed primarily in terms
of the production impact: the economic value of
output increases associated with new wheat and
rice varieties in Asia. This was a logical conse-
quence of the great public and scientific interest
in the "Green Revolution." It also reflects the
relative abundance of information on rice and
wheat production in Asia and the relative

simplicity of assessing the economic impact of
new varieties. It seems likely that research land
extension programs have generated many other
types of impacts, but these have, by and large.
been overlooked.
Most impact studies contribute to one or more
of the following biases in the economic literature
on agricultural research:
1. The magnitude of technological change in
wheat and rice production in Asia is ex-
aggerated, relative to the changes which
have taken place with other commodities
in other regions.
2. The significance of new varieties is exagger-
ated, relative to other types of technology.
3. The importance of yield increases is exag-
gerated, relative to other attributes of new
4. Too much credit is given explicitly or
implicitly to international programs, rela-
tive to other sources of change, including
the research and extension programs of
developing countries.
The unique contribution of international pro-
grams to agricultural development is to supply
R & D technology which improves institutional
performance, rather than production technology
which raises productivity at the farm level.
International programs play a strategic role in
disseminating research findings, methodologies,
and institutional strategies that national pro-
grams can use to generate finished technologies
for farmers. Hence, the impact of international
programs should be assessed primarily in institu-
tional terms, not in terms of production increases
at the farm level.
Previous studies have seldom assessed institu-
tional impacts, and the production economics
framework has several limitations in this regard.
Hence, the current generation of impact studies
needs to be particularly innovative in its institu-
tional analysis. Methods such as key informant
interviewing and case studies which are needed
for assessing changes in complex organizations,
are outside the bounds of traditional economics.
Hence, professional contributions are needed
from other disciplines, like anthropology, soci-
ology, and management. To ensure that impact
assessments reflect the values and perceptions of
scientists and policy makers in developing coun-
tries, it is necessary that they be involved, not
merely as sources of data, but as active partici-
pants in the studies' planning, implementation,
and critical review.



1. The CGIAR is an informal association of govern-
ments, international and regional organizations, and
private foundations dedicated to improving agricultural
technology, increasing food production, and raising the
living standards of poor people in developing countries.
Established in 1971, under the joint sponsorship of the
World Bank, the United Nations Development Pro-
gram (UNDP), and the UN's Food and Agriculture
Organizations (FAO), the CGIAR System has ex-
panded to include 13 international agricultural research
centers which employ over 500 scientists and have an
annual budget of approximately US $180 million.
Programs of 10 commodity-oriented centers cover a
range of crops, livestock and farming systems that
provide roughly three-fourths of the developing world's
total food supply. These centers are: CIAT (Inter-
national Center for Tropical Agriculture) in Palmira,
Colombia; CIMMYT (International Center for the
Improvement of Maize and Wheat) in El Batan,
Mexico; CIP (International Potato Center) in Lima,
Peru; ICARDA (International Center for Agricultural
Research in Dry Areas) in Lebanon and Syria;
ICRISAT (International Crops Research Institute for
the Semi-Arid Tropics) in Hyderabad, India; IITA
(International Institute of Tropical Agriculture) in
Ibadan, Nigeria; ILCA (International Livestock Center
for Africa) in Addis Ababa, Ethiopia; ILRAD (Inter-
national Laboratory for Research on Animal Diseases)
in Nairobi, Kenya; IRRI (International Rice Research
Institute) in Los Banos, Philippines; and WARDA
(West African Rice Development Association) in
Monrovia, Liberia. The remaining three centers are
concerned with problems of food policy, national
agricultural research systems, and conservation and
utilization of plant genetic resources. These centers
are: IFPRI (International Food Policy Research Insti-
tute) in Washington, D.C., USA.; ISNAR (Inter-
national Service for National Agricultural Research) in
The Hague, Netherlands; and IBPGR (International
Board for Plant Genetic Resources) in Rome, Italy.
Success of the first two centers, CIMMYT and IRRI,
precipitated what is often termed the "Green Revolu-
tion" with wheat and rice. For further information on
the CGIAR see Ruttan (1982), CGIAR (1980),
Pinstrup-Andersen (1982), and Arndt et al. (1977).
CGIAR (1985) presents a summary of the impact study
of international agricultural research centers.

2. The reader is referred to Pinstrup-Andersen
(1982), Ruttan (1982), Arndt et al. (1977), Scobie
(1979), and the works cited therein.

3. There are exceptions, of course, such as the
recent article on a rice seed multiplication project in
Sierra Leone by Kreul (1984).

4. On the subject of networking in international
agricultural research see Plucknett and Smith (1984).

potato yields in the Baguio area were roughly three
times the official estimate.

6. The location specificity of technology adoption
and response has been well documented in the case of
irrigated versus rainfed rice. A series of "adoption
studies" sponsored by CIMMYT in the 1970s also
indicated that adoption and use of new maize and
wheat varieties was strongly influenced by agro-
ecological conditions. On this point see Barker (1971),
Colmenares (1975), Demir (1976), Gafsi (1976),
Gerhart (1975), Vyas (1975), and Winklemann (1976).

7. Ruthenberg (1980) is a valuable exception. How-
ever, the author presents very little information from
Latin America.

8. The best material has been prepared in conjunc-
tion with IRRI's project on consequences of new rice
technology. See, for example, IRRI (1978a, b).

9. Some of the international centers' experiences
with and methods of on-farm research are discussed in
IRRI (1977a, b; 1984); De Datta et al. (1978); Byerlee
et al. (1980); Perrin et al. (1976); Byerlee et al. (1982);
CIP (1982); Rhoades (1982a or b; 1984); Horton (1984;
1982); Zandstra et al. (1981). Shaner et al. (1982)
outlines methods for on-farm research drawn from a
number of sources.

10. Some of these aspects of on-farm experimentation
are discussed in Davidson et al. (1967) and in Horton

11. See, for example, IRRI (1977a) and Potts (1983).

12. Some illustrative cases are noted in Section 4,
infra, and in CIP (1984).

13. The most substantial body of work on demand
projections is that of the FAO's Commodity Division.
The basic methodology employed is detailed in FAO
(1971) and the Working Papers cited therein.

14. Available statistics on recent changes in area,
production, and yields of potatoes are presented in
Horton and Fano (1984). Types of technological
change and their significance are discussed in Horton
and Sawyer (1985). Van der Zaag and Horton (1983),
Monares (1984), Rhoades (1984), Horton (1983), and
CIP (1984).

15. Economic analysis of different production con-
straints, including pests and diseases, was conducted at
CIAT in the early 1970s. For a presentation of results
for beans in Colombia see Pinstrup-Andersen (1976).
CIP began an assessment of major potato pests and
diseases in 1984.

5. Based on 400 farm-level yield measurements in 16. The questionnaire and summary of results may be
the Philippines, Potts (1983) concluded that actual obtained from the author.



Anderson, J. R., and J B. Hlardaker, "Economic
analysis in design of new technologies for small
farmers," in A. Valdes, G. Scohie, and J. Dillon
(Eds.), Economics and the Design of Small-farmer
Technology (Ames: Iowa State University Press,
Arndt, T. M., D. G. Dalrymple, and V. W. Ruttan
(Eds.), Resource Allocation and Productivity in
National and International Agricultural Research
(Minneapolis: University of Minnesota Press, 1977).
Barker, R., "The evolutionary nature of the new rice
technology," Food Research Institute Studies in
Agricultural Economics, Trade, and Development,
Vol. 10, No. 2 (1971), pp. 117-130.
Boyce, J. K., and R. E. Evenson, National and
International Agricultural Research & Extension
Programs (New York: Agricultural Development
Council, 1975).
Byerlee, D., M. Collinson, et al., Planning Tech-
nologies Appropriate To Farmers Concepts and
Procedures (Mexico City: CIMMYT, 1980).
Byerlee, D., L. Harrington, and D. Winkelmann,
"Farming systems research: Issues in research
strategy and technology design," American Journal
of Agricultural Economics, Vol. 64, No. 5 (1982), pp.
Casement, S., A Farming Systems Research Bibli-
ography of Kansas State University's Vertical File
Materials (1982 Cumulation) (Manhattan: Kansas
State University, 1982.)
Colmenares, J. H., Adoption of Hybrid Seeds and
Fertilizers Among Colombian Corn Growers (Mexico
City: CIMMYT, 1975).
Consultative Group on International Agricultural Re-
search (CGIAR), Consultative Group on Inter-
national Agricultural Research (Washington, D.C.:
CGIAR, 1980).
Consultative Group on International Agricultural Re-
search (CGIAR), "The impact of CGIAR institutes:
A study proposal," Mimeo. (Washington, D.C.:
World Bank, January 1984).
Consultative Group on International Agricultural Re-
search (CGIAR), Summary of International Agricul-
tural Research Centers: A Study of Achievements and
Potential (Washington, D.C.: CGIAR, 1985).
Dalrymple, D. G., Development and Spread of High-
Yielding Varieties of Wheat and Rice in the Less
Developed Nations, Foreign Agricultural Economic
Report 95 (Washington, D.C.: US Dept. of Agricul-
ture, 1978).
Dalrymple, D. G., "The adoption of high-yielding
grain varieties in developing nations," Agricultural
History, Vol. 53, No. 4 (October 1979), pp. 704-726.
Davidson, B. R., B. R. Martin, and R. G. Mauldon,
"The application of experimental research to farm
production," American Journal of Agricultural Eco-
nomics, Vol. 49, No. 4 (1967), pp. 900-907.
De Datta, S. K.. K. A. Gomez, R. W. Herdt, and R.
Barker, A Handbook on the Methodology for an
Integrated Experiment-survey on Rice Yield Con-
straints (Los Banos: IRRI, 1978).

Demir, N., The Adoption of New Bread Wheat
Technology in Selected Regions of Turkey (Mexico
City: CIMMYT, 1976).
Evenson, R. E., "Comparative evidence on returns to
investment in national and international research
institutions," in Arndt, Dalrymple, and Ruttan
Evenson, R. E., and Y. Kisler, Agricultural Research
and Productivity (New Haven: Yale University
Press, 1975).
Fano, H., "Cambio technologico y tendencies de la
production de papa en la region central del Peru
1948-1979," MA thesis (Universidad Nacional
Agraria del Peru, 1983).
Food and Agriculture Organization of the United
Nations, Agricultural Commodity Projections, 1970-
1980, Vol. II (Rome: FAO, 1971).
Food and Agriculture Organization of the United
Nations, FAO Production Yearbook 1976 (Rome:
FAO, 1977).
Food and Agriculture Organization of the United
Nations, FAO Production Yearbook 1983 (Rome:
FAO, 1984).
Franco, E. and E. Schmidt, Adopcion y Difusion de
Variedades de Papa en el Departamento de Cajamar-
ca, Departamento de Ciencias Sociales, Documento
de Trabajo 1985-1 (Lima: CIP, 1985).
Gafsi, S., Green Revolution: The Tunisian Experience
(Mexico City: CIMMYT, 1976).
Gerhart, J., The Diffusion of Hybrid Maize in Western
Kenya (Mexico City: CIMMYT, 1975).
Hardaker, J. B., J. R. Anderson, and J. L. Dilion,
"Perspectives on assessing the impacts of improved
agricultural technologies in developing countries,"
Mimeo. (Armidale: University of New England,
Horton, D. E., Tips for Planning Formal Farm Surveys
in Developing Countries (Lima: CIP, 1982).
Horton, D. E., "Potato farming in the Andes: Some
lessons from on-farm research in Peru's Mantaro
Valley," Agricultural Systems, Vol. 12 (1983), pp.
Horton, D. E., Social Scientists in Agricultural Re-
search: Lessons From the Mantaro Valley Project,
Peru (Ottawa: IDRC, 1984).
Horton, D. E., and H. Fano, World Potato Atlas
(Lima: CIP, 1985).
Horton, D. E., J. Lyman, and H. Knipscheer, "Root
crops in developing countries: An economic
appraisal," in Symposium of the International Society
for Tropical Root Crops, 6th edn (Lima: CIP, 1984).
Horton, D. E., and R. L. Sawyer, "The potato as a
world food crop with special reference to developing
areas," in H. L. Paul (Ed.), Potato Physiology (New
York: Academic Press, 1985).
Horton, D. E., F. Tardieu, M. Benavides, L. Tom-
assini, L., and P. Accatino, Tecnologia de la Pro-
duccion de Papa en el Valle del Mantaro, Peru.
Resultados de una Encuesta Agro-Economica de
Visita Multiple (Lima: CIP, 1980).
International Potato Center, Social Science Research at


the International Potato Center, Report of the Second
Social Science Planning Conference (Lima: CIP,
International Potato Center, Potatoes for the Develop-
ing World (Lima: CIP, 1984).
International Rice Research Institute, Constraints to
High Yields on Asian Rice Farms: An Interim Report
(Los Banos: IRRI, 1977a).
International Rice Research Institute, Proceedings,
Symposium on Cropping Systems Research and
Development for the Asian Rice Farmer (Los Banos:
IRRI, 1977b).
International Rice Research Institute, Economic Con-
sequences of the New Rice Technology (Los Banos:
IRRI, 1978a).
International Rice Research Institute, Interpretive
Analysis of Selected Papers from Changes in Rice
Farming in Selected Areas of Asia (Los Banos: IRRI,
International Rice Research Institute, Report of a
Workshop on Cropping System Research in Asia (Los
Banos: IRRI, 1982).
International Rice Research Institute, Cropping Sys-
tems in Asia: On-farm Research and Management
(Los Banos: IRRI, 1984).
Kreul, W., "Improved seed production a recent
example of a sqed multiplication project in Sierra
Leone," Quarterly Journal of International Agricul-
ture, Vol. 23, No. 1 (1984), pp. 51-64.
Monares, A., Building an Effective Country Program:
Rwanda, Social Science Department Working Paper
1984-83 (Lima: CIP, 1984).
Pinstrup-Andersen, P., Agricultural Research and
Technology in Economic Development (New York:
Longman, 1982).
Pinstrup-Andersen, P., N. de London, and M. Infante,
"A suggested procedure for estimating yield and
production losses in crops," PANS, Vol. 22, No. 3
(1976), pp. 359-365.
Plucknett, D. L., and N. J. H. Smith, "Networking in
international agricultural research," Science, Vol.
225, No. 7 (1984), pp. 989-993.
Potts, M. J. (Ed.), On-farm Research in the Philippines
(Lima: CIP, 1983).

Potts, M. J., L. M. Pacuz, and E. O. Sano, "White
potato yield survey: Benguet Province," Philippine
Agriculturist, Vol. 65 (1982), pp. 385-393.
Rhoades, R., The Art of the Informal Agricultural
Survey, Social Science Department Training Docu-
ment 1982-2 (Lima: CIP, 1982a).
Rhoades, R., Understanding Small Farmers:
Sociocultural Pespectives on Experimental Farm
Trials, Social Science Department Training Docment
1982-83 (Lima: CIP, 1982b).
Rhoades, R., "Changing a post-harvest system: Impact
of diffused light potato stores in Sri Lanka," Agricul-
tural Systems (1985), Vol. 19, pp. 1-19.
Ruthenberg, H., Farming Systems in the Tropics, 3rd
edn (Oxford: Oxford University Press, 1980).
Ruttan, V. W., Agricultural Research Policy (Min-
neapolis: University of Minnesota Press, 1982).
Scobie, G. M., "Investment in international agricul-
tural research: Some economic dimensions, "World
Bank Staff Working Paper No. 361 (Washington,
D.C.: World Bank, October 1979).
Shaner, W. W., P. F. Phillipp, and W. R. Schmehl,
Farming Systems Research and Development: Guide-
lines for Developing Countries (Boulder: Westview
Press, 1982).
Timmer, C. P., W. P. Falcon, and S. A. Pearson, Food
Policy Analysis (Baltimore: Johns Hopkins Univer-
sity Press, 1983).
Valderama, M., and H. Luzuriaga, Produccion y
Utilizacion de la Papa en el Ecuador (Lima: CIP,
Van Der Zaag, D. E., and D. E. Horton, "Potato
production and utilization in world perspective with
special reference to the tropics and sub-tropics,"
Potato Research, Vol. 26, No. 7 (1983), pp. 323-362.
Vyas, V. S., India's High Yielding Varieties Programme
in Wheat, 1966-67 to 1971-72 (Mexico City:
CIMMYT, 1975).
Winkelmann, D., The Adoption of New Maize Tech-
nology in Plan Puebla, Mexico (Mexico City:
CIMMYT, 1976).
Zandstra, H. G., E. C. Price, J. A. Litsinger, and
R. A. Morris, A Methodology for On-farm Cropping
Systems Research (Los Banos: IRRI, 1981).

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs