THE IMPACT OF AGRICULTURAL MAINTENANCE RESEARCH:
THE CASE OF LEAF RUST RESISTANCE BREEDING IN CIMMYTRELATED SPRING BREAD WHEAT
C. N. Marasas
M. Smale
R.P. Singh
International Maize and Wheat Improvement Center, Apartado Postal 6-641,
06600 Mexico D.F., Mexico
Paper prepared for the International Conference on Impacts of Agricultural Research and
Development, San José, Costa Rica, 4-7 February 2002
ABSTRACT
Leaf rust caused by Puccinia triticina is a wheat disease of major historical and economic
importance worldwide. Genetic resistance, rather than the use of fungicides, remains the
principal means of disease control. We estimated the impact in developing country production of
efforts by the International Maize and Wheat Improvement Center (CIMMYT) to breed leaf rust
resistant spring bread wheat varieties since 1973.
The challenge in estimating these benefits is in dealing with the fact that rust pathogens are able
to rapidly mutate to new races, which are able to infect previously resistant varieties. Various
single genes or gene complexes determine the type, level, and longevity of a variety’s resistance.
Leaf rust resistance breeding is therefore an example of crop maintenance research. Whereas
productivity enhancement is measured in terms of positive yield gains, productivity maintenance
is estimated in terms of the yield losses that would have occurred in the absence of the research
investment. Although its importance has long been argued, there are relatively few economic
analyses of maintenance research, and in particular, breeding for crop disease resistance.
The benefits of CIMMYT’s investment in leaf rust resistance breeding were estimated using an
economic surplus approach adjusted for maintenance research. Gross benefits were modeled as
the surplus generated by avoiding a cost-increasing shift in the supply curve resulting from
changes in the environment caused by evolving leaf rust pathogens. A standard capital
investment analysis was applied to estimate the returns.
1
A sample of the major spring bread wheat varieties grown in the developing world were
classified by their type and level of genetic leaf rust resistance through trials conducted at
CIMMYT. The estimated yield losses of these varieties were compared with those that would
have occurred had all these varieties been fully susceptible. The area to which yield savings
applied were estimated by fitting historical logistic diffusion curves to the study area potentially
affected by leaf rust. The analysis was separated by wheat breeding mega-environment, a
classification developed by the CIMMYT Wheat Program to guide its germplasm enhancement
activities. The full cost of CIMMYT’s wheat improvement effort was included in the analysis. A
range of investment values was elicited by varying assumptions on several parameters.
The results of this study have two major policy implications. They firstly demonstrate that
CIMMYT’s investment in wheat genetic improvement since 1973 has generated substantial
economic returns from leaf rust resistance breeding in spring bread wheat only. In an era
characterized by a global decline in agricultural research investments, the efficient targeting of
scarce resources is becoming increasingly important. Secondly, the results emphasize the
importance of maintenance research, which has often been undervalued in economic analyses.
Studies at CIMMYT indicate that part of the progress in wheat yield gain through the years has
been achieved by protecting this yield potential through disease resistance breeding. Failure to
account for the effects of maintenance research could therefore bias rate of return estimates.
2
INTRODUCTION
The rate of return on investments in agricultural research has often been estimated assuming that
research explains positive productivity growth, and that productivity would remain constant in
the absence of research. However, this assumption ignores the losses that may result from the
physical, biological and economic changes that could render existing technologies less effective.
Most assessments of the returns on crop improvement programs have focused on productivity
enhancement. There are comparatively fewer economic analyses of maintenance research,
particularly for disease resistance breeding. Whereas productivity enhancement is often
measured in terms of positive yield gains, productivity maintenance is estimated in terms of the
yield losses that would have occurred in the absence of the research investment.
In this paper, we draw on the findings of our case study to underscore the importance of
maintenance research in plant breeding programs. Our objective is to estimate the economic
impact in developing country production of efforts by the International Maize and Wheat
Improvement Center (CIMMYT) to breed genetic leaf rust resistance in spring bread wheat from
1973. We initiate our paper by outlining the background and scope of this study, before
presenting the conceptual framework, methodology, results and conclusions.
BACKGROUND AND SCOPE OF THIS STUDY
Leaf rust caused by Puccinia triticina is a wheat disease of major historical and economic
importance worldwide (Roelfs, Singh and Saari, 1992). It is the most widespread of three types
3
of rusts, the other two types being stem rust caused by P. graminis and stripe rust caused by P.
striiformis. Periodic rust epidemics were common in most decades of this century, and some
yield losses to rusts are still suffered in many wheat-producing areas in most years. The
cultivation of resistant varieties remains the principal control method in developing countries,
where fungicides are not often used for this purpose. The development of leaf rust resistance has
therefore been a priority of CIMMYT’s wheat breeding program since its inception.
Varieties could carry different types and levels of genetic leaf rust resistance. Genes conferring
race-specific resistance produce intermediate to major reactions against specific races of the
pathogen. However, the major challenge in breeding for leaf rust resistance is in dealing with the
ability of the pathogen to rapidly mutate to new races, which are able to infect previously
resistant varieties. The effects of race-specific resistance may thus be overcome within a
relatively short time. A severe leaf rust epidemic in northwestern Mexico in 1976-77
dramatically underscored the need for more durable resistance (Dubin and Torres, 1981).
Race-nonspecific resistance is based on the interaction of a few or several genes having partial to
additive effects. The genes are effective against several races of the pathogen simultaneously and
result in varying levels of resistance against them. Varieties with race-nonspecific resistance may
suffer comparatively larger yield losses than varieties with effective race-specific resistance.
However, race-nonspecific resistance appears to endure longer. Though CIMMYT’s Wheat
Program emphasizes the selection for race-nonspecific resistance (Rajaram, Singh, and van
Ginkel, 1996), breeders in some national programs might place more emphasis on characteristics
other than race-nonspecific leaf rust resistance. Furthermore, producers often continue to grow
4
varieties with levels of resistance that wheat scientists may no longer consider satisfactory.
CIMMYT-related varieties with both race-specific and race-nonspecific resistance can therefore
be found in wheat fields today.
In their case study of the Yaqui Valley of Mexico, Smale, Singh, Sayre, Pingali, Rajaram, and
Dubin (1998) estimated the yield losses that farmers would have suffered if a breeding strategy
for race-specific rather than race-nonspecific resistance had been employed. The authors
estimated a rate of return of 40% on the investment in race-nonspecific leaf rust resistance
breeding over the period 1970-1990. Detailed information on the genes conferring for resistance
and the longevity of useful resistance for each wheat variety grown in the Yaqui Valley since
1968 was employed. However, the geographical scope of that study was more limited, since it
covered a mere 150,000 ha compared to the estimated 71.5 million ha of spring bread wheat in
the developing world. Similar information on the genetic basis and longevity of useful leaf rust
resistance was not available on a global basis to facilitate our analysis.
Our study therefore encompasses all genetic resistance mechanisms carried by CIMMYT-related
spring bread wheat. We compare the yield losses suffered by varieties with different types and
levels of genetic leaf rust resistance to the yield losses that would have been suffered had all
these varieties been fully susceptible. Our study is limited to the developing world, because the
mandate of CIMMYT’s Wheat Program is to breed advanced lines for the national agricultural
research programs in these countries. We focus on spring bread wheat, because it covers about
two-thirds of the wheat area in the developing world (Heisey, Lantican, and Dubin,
forthcoming). Our analysis is conducted by wheat breeding mega-environment (ME), since this
5
is a classification developed by the CIMMYT Wheat Program to guide its germplasm
enhancement activities (Rajaram, van Ginkel, and Fischer, 1995). We focused on the MEs where
spring bread wheat is grown at low latitudes, and thus included MEs 1, 2, 3, 4a, 4b, 4c, and 5.
More information on these production environments is presented in Marasas, Smale, and Singh
(forthcoming).
With the term “CIMMYT-related” we include those materials selected from advanced CIMMYT
lines by wheat breeders in national agricultural research programs. These varieties generally
descend from the first semidwarf varieties released during the late 1960s. Almost 80% of the
spring bread wheat area in developing countries was sown to CIMMYT-related varieties in 1997
(Heisey, Lantican, and Dubin 1999). A survey of wheat breeders in these countries indicated that
materials from CIMMYT International Nurseries are the most frequently crossed in pursuit of
disease resistance goals (Rejesus, Smale and van Ginkel, 1997). Broad international flow of
CIMMYT-related germplasm with leaf rust resistance has therefore been likely.
CONCEPTUAL FRAMEWORK
The first step in measuring the economic benefits of agricultural research, is to compare the
situation with research to the one with no research, also known as the “with” and “without”
scenarios. In view of the pathogen’s ability to overcome the effects of previously resistant
varieties, we argued that leaf rust resistance breeding is an example of research aimed at
maintaining crop productivity. We applied an economic surplus approach adjusted for
6
maintenance research to estimate the gross benefits of CIMMYT’s investment in leaf rust
resistance breeding since 1973.
The effects of productivity enhancement are often treated as a cost-reducing rightward or
downward shift in the commodity supply function, as shown by S1 in Figure 1. This is assumed
to result from yield increases or cost savings associated with the technology. The “without”
scenario assumes constant supply in the absence of research, as represented by S0. However, the
assumption of a static supply function does not remain valid in the face of evolving leaf rust
pathogens. Once a variety’s resistance has been overcome, its production gains will decline and
result in lower production per unit cost. If not constantly replaced by newly resistant varieties
with similar productivity potential, the “without” scenario would comprise a leftward or upward
shift in the supply curve, shown by S2. In an economic surplus framework, maintenance research
can therefore be defined as the effort required to prevent a cost-increasing shift in the supply
curve, which results from changes in the physical, economic or biological environment (Collins,
1995). The economic surplus thus generated is shown as the shaded area in Figure 1. Though full
adoption and depreciation is assumed in Figure 1, these processes are in fact dynamic and
proceed over a period of time.
In our case, we assume that the “with” scenario represents the actual wheat supply (S0),
generated by the CIMMYT-related spring bread wheat varieties with various leaf rust resistance
categories, grown in the developing world since 1973. The “without” scenario is the supply (S2)
that would have prevailed had all these varieties been fully susceptible.
7
P
surplus generated
by productivity
maintenance
research
S2
effect of productivity
maintenance
S0
P2
surplus generated by
productivity enhancement
research
P0
S1
P1
D
Q2
Q0
Q1
Q
S0 supply with maintenance, but no enhancement research
S1 supply with enhancement research and full adoption of the new variety
S2 supply with productivity losses from no investment in maintenance
or enhancement research
Figure 1. General economic surplus approach adjusted for maintenance research
8
Our approach is methodologically simplified, due to standard difficulties in estimating the
impact of maintenance research, estimating the economic impact of agricultural research in
general, and limitations imposed by the data available to us. We apply a capital investment
analysis to estimate the returns instead of a fully developed equilibrium model based on a multimarket world economy. This is partly because the benefits in our analysis are aggregated over a
large number of developing country wheat producers. Losses to leaf rust might have generated a
shift in the short- and long-term wheat supply curve in any one of these countries. However,
these changes would not have been substantial enough to affect the world wheat price in the
presence of the large volumes traded by developed country wheat producers. The demand curve
is therefore completely elastic at the world wheat price in our version of Figure 1. We measure
the wheat supply shift avoided in units on the horizontal axis, valued at the world wheat price,
for each year and wheat-producing environment included in the study. Our supply curve refers to
CIMMYT-related spring bread wheat only.
METHODOLOGY
In our capital investment analysis, the research returns were estimated in terms of the net present
value, internal rate of return, and benefit-cost ratio. The net present value of leaf rust resistance
breeding in CIMMYT-related spring bread wheat can be most generally expressed as:
n
(1)
Net present value =
1
 (1  i)
t 1
t
[(pt yt at) – Ct]
9
Essential parameters are: , the average, annual, farm-level percentage yield loss avoided by
growing varieties with various leaf rust resistance categories; y, the average, annual, farm-level
wheat yield, and a, the area to which yield savings apply. The product of these terms represents
the production savings from leaf rust resistance breeding by genetic resistance category and
wheat breeding environment. The real wheat price p is used to calculate an economic value on
the wheat yield saved. The difference between the gross research benefits and the cost of
research C is calculated for each year t. This begins in 1973 (t1), the year in which the first
variety recognized and promoted for its race-nonspecific resistance was released (Torim 73). It
ends n years later in 2007 (tn), the year the last adoption ceiling predicted in our logistic diffusion
curves is reached. The research benefits are discounted using the interest rate i to obtain the net
present value.
The internal rate of return is estimated by setting the net present value equal to zero in equation
(1) and solving for i arithmetically:
n
(2)
1
 (1  i)
t 1
t
[(pt yt at) – Ct] = 0
The benefit-cost ratio is calculated by dividing the present value of the gross benefits by the
present value of the research costs:
n
(3)
1
 (1  i)
t 1
t
 ( pt yt at ) 


 Ct

10
Estimation of each of the parameters in equations (1) to (3) is described next, with details related
to data sources and assumptions. A summary of parameter assumptions is presented in the Annex
Table 1.
Percentage yield savings
Parameter yt in equations (1) to (3) was estimated as the product of the following three terms:
1) The percentage yield savings of resistant relative to susceptible varieties by genetic resistance
category. For this purpose, we used trial data for a sample of the major spring bread wheat
varieties grown in the developing world, as drawn from CIMMYT’s 1997 Global Wheat Impacts
Survey. Varieties with known CIMMYT origin, released since 1970, grown on more than 500
ha, and for which seed was available in the CIMMYT gene bank, were grown without fungicide
protection in a field trial in El Batán, Mexico. Leaf rust epidemics were established by
inoculating susceptible spreader rows planted adjacent to the trial material. The varieties were
classified by type and level of genetic resistance to the current Mexican population of leaf rust,
based on the modified Cobb-scale (Peterson, Campbell, and Hannah, 1948) (Table 1). Seedling
evaluation tests with selected P. triticina races were conducted in the greenhouse to assess the
presence of effective race-specific genes. The trial data were combined with supplementary data
from previous CIMMYT trials over several years to obtain a sample of 184 varieties. The
percentage infection relative to the susceptible check variety was used to calculate the associated
yield savings for each resistance category.
11
Table 1. Definition of the leaf rust resistance categories used in this study a
Category
1
2
3
4
5
6
a
Percentage leaf rust infection
relative to susceptible check
Type of resistance
80 - 100%
50 - 70%
30 - 50%
10 - 20%
less than 10%
less than 5%
Susceptible
Race-nonspecific, low resistance
Race-nonspecific, moderate resistance
Race-nonspecific, high resistance
Race-nonspecific, high resistance
Effective, race-specific resistance
Based on the modified Cobb-scale (Peterson et al., 1948).
2) The average, annual, farm-level percentage yield lost with susceptible varieties by ME. These
annual yield loss data were not available over the extensive spring bread wheat producing areas
of the developing world. Data on the weather conditions, management practices, and spatial
distributions of pathogen and resistance types were also not available to allow prediction of the
annual disease pressure or the duration of resistance. We then searched various sources of trial
data and historical accounts from the literature for estimates of expected losses from secondary
sources (Marasas et al., forthcoming).
However, experimental estimates were not available for all of the production areas included in
our study, and the available data from small-plot evaluations tend to over-estimate disease losses.
The number and significance of recorded rust epidemics vary, and estimated production losses
have typically been reported anecdotally for the developing world. Even when occurrence of the
disease may be recorded, it is seldom accompanied by data on yield losses, or the relationship to
wheat prices, output levels, and imports. We were furthermore concerned with farm-level yield
losses averaged over several years, large areas, and various production environments in
developing countries. These are clearly lower than the yield losses estimated in zones of high
12
disease pressure, or the losses reported in epidemic years. Yield loss information for developed
countries, such as the comprehensive data from the Cereal Disease Laboratory
(http://www.crl.umn.edu) for the United States of America, is also not appropriate. These
estimates firstly do not represent all the spring bread wheat producing environments included in
our study. They also do not represent the situation in all developing countries, where few farmers
use fungicides to control leaf rust.
In view of all these considerations, we based our upper-bound estimates of the average, annual
percentage yield lost by susceptible varieties at the farm level on those provided by the
CIMMYT Wheat Program by wheat-producing environment (Annex Table 1). Estimates in all
MEs are less than 10% and should be in line with the general global guideline of less than 10%
(Roelfs et al., 1992:2).
3) The average, annual, farm-level yield of CIMMYT-related spring bread wheat varieties by
ME from 1973 to 2007. We generated these time series by combining information from the 1990
CIMMYT Global Wheat Impacts Survey with annual national wheat yields reported by the Food
and Agriculture Organization (FAO) (http://faostat.fao.org). A trend regression was fitted to the
data to project yields to 2007.
13
Areas
Parameter at in equations (1) to (3) was calculated as the product of the following four
components:
1) The percentage areas planted in CIMMYT-related spring bread wheat since 1973 by ME. We
fitted logistic diffusion curves for this purpose, using function parameters from 1977, 1990 and
1997 CIMMYT Wheat Impacts data (Annex Table 1) (CIMMYT, 1989; Byerlee and Moya,
1993: 45; Heisey et al., forthcoming). We assumed that CIMMYT-related varieties released
since 1973 followed similar aggregate adoption paths to those that began to diffuse in 1966. The
year 2007 thus proved to be the latest year predicted by the logistic curves.
2) The annual, average percentage area potentially affected by leaf rust by ME. Estimates were
drawn from the CIMMYT Wheat Program by reviewing a list of production zones corresponding
to the MEs in the countries included in the CIMMYT Global Wheat Impacts surveys.
3) The percentage distribution of area to which yield savings apply by genetic resistance
category and ME. Point estimates were calculated by combining information on the genetic
resistance categories from the sample of varieties tested in trials, with the areas sown to each
variety as recorded in the 1997 CIMMYT Global Wheat Impacts database. Table 2 indicates that
80% of the sample area was protected by genes conferring race-nonspecific resistance
(categories 2 to 5), while only 10% of the area was protected by genes conferring race-specific
resistance (category 6). A further 10% of the area was sown to varieties classified as almost fully
14
susceptible (category 1). Over 80% of the areas in MEs 1, 4a, 4c and 5 were planted in varieties
with race-nonspecific resistance. However, most of the areas in MEs 4b (97%) and 3 (72%), and
a substantial area in ME 2 (34%), were planted in varieties with race-specific resistance.
Characteristics other than race-nonspecific leaf rust resistance might be more important in MEs
2, 3 and 4b.
Table 2. The percentage area by genetic resistance category and mega-environment in the sample
of major CIMMYT-related spring bread wheat varieties grown in the developing world in 1997
Mega-environment
1
2
3
4a
4b
4c
5a
Total sample area
(000 ha)
(Percentage)
a
Genetic resistance category a
1
2
11.8
6.6
1.0
8.0
8.7
0
1.1
2.9
0
0
8.7
5.0
13.0
8.5
3,694
10%
2,342
6%
3
37.7
37.8
7.9
53.6
1.6
36.8
33.2
4
36.1
19.4
11.1
25.2
1.2
41.4
40.9
5
4.1
0
0.3
0
0
4.3
2.5
6
3.7
33.8
72.0
17.1
97.2
3.8
1.9
13,679
37%
12,723
34%
1,222
3%
3,694
10%
Genetic resistance categories are defined in Table 1.
4) The average, annual areas sown to CIMMYT-related spring bread wheat by ME from 1973 to
2007. As for the annual average yields, we generated these time series by combining FAO and
1990 CIMMYT Global Wheat Impacts Survey data. A trend regression was used to project areas
to 2007.
15
The real wheat price
The real wheat price, or pt in equations (1) to (3), was used to the value the production savings
from leaf rust resistance breeding. The free on board (f.o.b.) hard red winter wheat no 2 (at Gulf
port) prices from 1973 to 1997 were obtained from the United States Department of Agriculture,
and deflated to 1990 real US$ terms using the United States Consumer Price Index. A trend
regression was fitted to the data to predict prices towards 2007.
Research costs
CIMMYT’s real research investment from 1967 to 1999, expressed in 1990 US$, and estimated
for a higher and a lower cost scenario, were obtained from Heisey et al. (forthcoming). Costs
were assumed from 1967 onwards to allow for the research lag of those varieties released in
1973. In view of CIMMYT’s shuttle breeding program, it should be reasonable to assume a five
to six year research development period for a new wheat variety. For the higher cost scenario,
CIMMYT’s budget was allocated to wheat genetic improvement by the proportion of the latter
budget to the total budget. For the lower cost scenario, CIMMYT’s budget was allocated to
wheat genetic improvement by the proportion that wheat program senior staff comprises of all
senior staff at CIMMYT.
We applied the total cost of wheat genetic improvement in the analysis, including the costs of
shipments through international nurseries and testing costs borne by CIMMYT. Only those costs
borne by national programs, such as local screening for rusts and other tests, are excluded. We
16
argued that genetic improvement is an integrated effort involving infrastructure, knowledge and
support that extend across different disciplines and programs at CIMMYT. Leaf rust resistance
breeding can also not be separated from CIMMYT’s other wheat breeding objectives such as
yield, adaptation, and resistance to other pests and diseases. With regard to valuing maintenance
research, this assumption demonstrates the difficulty in separating various pathology, agronomy,
and physiology activities in the production of enhanced germplasm.
As with the other time series data we have employed, costs were projected to 2007. However, the
trend in the cost series is more quadratic than linear in form. The real CIMMYT investment in
wheat genetic improvement increased steadily from 1967 until its peak in 1988, after which it
declined substantially. Rather than predicting either an upward shift or continued downward
pattern in research investment, we chose to hold costs constant at their 1999 level. The research
costs (Ct) were subtracted from the gross benefits to estimate the net benefits.
Discount rate
In view of the debate in the economics literature on choosing appropriate discount rates, we
assumed interest rates of 1%, 5% and 15% to represent different perspectives on the investment
decision. These included: The long-term “social time preference rate” (1%); the current real
interest rate, such as the average interest rate charged by the United States Federal Reserve Bank
over the past 15 years (5%); and the perspective of a private investor such as the World Bank,
with risks or irreversibility incorporated in the interest rate (15%).
17
RESULTS
Discounted gross benefits by genetic resistance category and mega-environment
We firstly present the results in terms of the discounted gross benefits of leaf rust resistance
breeding in CIMMYT-related spring bread wheat by genetic resistance category and ME. This is
because research costs cannot be separated on the same basis. An intermediate discount rate of
5% was assumed for this comparison. When including all genetic resistance categories and MEs,
the discounted real gross benefits since 1973 amounted to 5.6 billion 1990 US$ (Table 3).
Varieties with race-nonspecific resistance (categories 2 to 5) accounted for 91% of the total
benefits. Varieties with race-specific resistance accounted for 7%, while those classified as
almost fully susceptible represented only 2% of the total benefits. Whereas race-nonspecific
resistance generated the major proportion of gross benefits in MEs 1, 2, 4a, 4c and 5, most of the
benefits in MEs 3 and 4b accrued to race-specific resistance. These findings reflected the
assumptions on the percentage cumulative area by ME sown in CIMMYT-related varieties with
different resistance categories (Table 2).
ME 1 accounted for 86% of the gross benefits by ME, which could be explained by at least three
observations. This large wheat breeding environment represented nearly 60% of our study area,
and new wheat varieties have historically been shown to spread at more rapid rates in ME 1.
About two-thirds of this favorable wheat growing environment is found in the irrigated zones of
18
the Asian subcontinent. Since both average yields and potential losses from disease are higher in
these areas, the production savings from resistance are also greater.
Table 3. Discounted gross benefits of genetic leaf rust resistance in CIMMYT-related spring
bread wheat from 1973 to 2007, by mega-environment and genetic resistance type
Megaenvironment
1
2
3
4a
4b
4c
5
All
Gross benefits by resistance type
Environment
(million 1990 US$)
as percentage
RaceRace-specific All
nonspecific
4,423.1
267.4
4,781.0
86
103.0
80.0
183.2
3.3
3.1
15.3
18.6
0.3
3.9
1.2
5.0
0.1
0.3
13.3
13.6
0.2
4.6
0.3
5.0
0.1
540.7
17.0
569.8
10
5,078.7
394.4
5,576.2
100
Notes: “All” include varieties with race-nonspecific and race-specific resistance, and those
classified as almost susceptible, as defined in Table 1. The gross benefits were discounted by
5%.
Returns on the research investment
In the next step, we included research costs to estimate the investment returns in terms of the
internal rate of return, net present value, and benefit-cost ratio. We firstly assumed an
intermediate discount rate of 5% to calculate the net present value and benefit-cost ratio.
The results demonstrate that CIMMYT’s investment in wheat genetic improvement has
generated substantial economic returns from leaf rust resistance breeding only (Table 4). Under
the lower research cost scenario, the internal rate of return was 41% and the net present value
4.02 billion 1990 US$. When higher research costs were assumed, the rate of return was 37%
19
and the net present value 3.96 billion 1990 US$. The benefit-cost ratio of the discounted real
investment was 29:1 under the low, and 20:1 under the high research cost scenario. This implies
that every one US$ invested in wheat genetic improvement at CIMMYT since 1973, has
generated at least 20 times its magnitude in gross benefits by leaf rust resistance breeding alone.
All other wheat breeding benefits are considered as pure benefits. The net benefits were not too
sensitive to the two cost scenarios.
Table 4. Returns on the investment in genetic leaf rust resistance breeding in CIMMYT-related
spring bread wheat from 1973 to 2007, with high and low research cost assumptions
Research costs
Low
High
Internal rate of
return (%)
41
37
Net present value
(billion 1990 US$)
4.02
3.96
Benefit-cost ratio
29:1
20:1
Notes: Estimates include all genetic resistance categories and mega-environments. The net
present value and benefit-cost ratio were calculated with a 5% discount rate.
As for the gross benefits, most of the net benefits were realized in ME 1 (Table 5). Given the
cost streams we have employed, this implies that CIMMYT’s investment in wheat genetic
improvement over the past 30 years would be more than justified by the benefits from leaf rust
resistance breeding in ME 1 only.
20
Table 5. Internal rate of return and net present value of genetic leaf rust resistance breeding in
CIMMYT-related spring bread wheat from 1973 to 2007, by mega-environment and research
cost scenario
Research costs and
mega-environments
Low research cost
ME 1
MEs 2 to 5
All MEs
High research cost
ME 1
MEs 2 to 5
All MEs
Internal rate of
return (%)
Net present value
(billion 1990 US$)
40
18
41
3.43
0.45
4.02
36
15
37
3.36
0.39
3.96
Notes: Gross benefits in ME 1 were charged the full cost of the high research cost scenario.
Gross benefits for MEs 2, 3, 4a, 4b, 4c and 5 were combined and charged the full research
investment in a similar manner. The net present value was calculated with a 5% discount rate.
Estimates include all genetic resistance categories.
Table 6 shows the net present values elicited by varying the discount rate to demonstrate the
effect of the investor’s perspective on the economic returns. As could be expected, the net
present values decreased when discounted by higher interest rates. However, even when a
stringent interest rate of 15% was assumed, the investment still generated a positive and
substantial net present value.
Table 6. Net present value of the investment in genetic leaf rust resistance breeding in
CIMMYT-related spring bread wheat from 1973 to 2007, for various discount rates and research
cost scenarios
Research costs
Low
High
Net present value at different discount rates
(billion 1990 US$)
1%
5%
15%
11.29
4.02
0.47
11.15
3.96
0.45
21
Minimum yield savings necessary to recover the investment
The investment returns in Table 4 were calculated by employing estimates of the expected
annual, average, farm-level yield losses that would have been suffered had all CIMMYT-related
spring bread wheat varieties been fully susceptible. These in turn determined the yield losses
avoided by growing varieties with various leaf rust resistance categories. Considering the
conceptual framework of our analysis (Figure 1), the results are likely to be most sensitive to this
assumption. Yet this parameter was the most difficult to reliably estimate over the large
geographical areas included in our study. We therefore performed a further calculation in our
sensitivity analysis. In addition to the yield loss assumptions in the Annex Table 1, we
arithmetically calculated the minimum annual, average percentage yields that would have had to
been lost to leaf rust by susceptible varieties in ME 1 to recover CIMMYT’s investment in wheat
genetic improvement since 1973. We limited the area included in the calculation to ME 1 to
render our estimates even more conservative, though this environment clearly accounted for the
major share of the benefits from leaf rust resistance breeding.
At the 5% discount rate, the minimum yields that would have had to been lost are 0.2% under the
low research cost scenario and 0.4% under the high research cost scenario (Table 7). Even when
a stringent 15% discount rate was applied, the investment would still have been recuperated
under very low yield loss assumptions. The minimum yield savings that would have been
necessary in ME 1 to recover CIMMYT’s entire investment in wheat genetic improvement is
therefore a mere fraction of those assumed in the Annex Table 1. These minimum estimates
22
would be unusually low for this important wheat disease in this high-yielding zone with heavy
disease pressure. The investment returns presented in Table 4 should therefore be fairly robust.
Table 7. The minimum average annual percentage yield lost by susceptible varieties in megaenvironment 1 to generate a positive net present value on CIMMYT’s investment in wheat
genetic improvement since 1973, for various discount rates and research cost scenarios
Research costs
Low
High
Minimum average annual yield lost by
susceptible varieties (%)
5% discount rate
15% discount rate
0.2
0.6
0.4
0.9
DISCUSSION
In an era characterized by a global decline in agricultural research investments, the efficient
targeting of scarce resources is becoming increasingly important. The results of this study
demonstrate that CIMMYT’s investment in wheat genetic improvement has generated substantial
economic returns from leaf rust resistance breeding in spring bread wheat only. When average
annual yield losses of less than 10%, and the highest research cost scenario were assumed, the
rate of return on the investment from 1973 to 2007 was 37%. When discounted by 5%, the
corresponding net present value was 3.96 billion 1990 US$, and the benefit-cost ratio 20:1. By
arithmetic calculation under the same research cost and discount rate assumptions we further
estimated that, even if the average annual yield lost to leaf rust in ME 1 had been a mere 0.4%,
the net present value of benefits would still cover CIMMYT’s wheat breeding investment since
1973. We included the full cost of CIMMYT’s wheat genetic improvement effort, though we
accounted for the benefits of leaf rust resistance in spring bread wheat only. All other benefits
are considered as pure benefits, such as the increases in yield potential that have been
23
demonstrated over time, and resistance to other biotic and abiotic stresses. Benefits were
primarily generated in ME 1 and by varieties with race-nonspecific resistance to leaf rust.
The economic returns in this study were estimated by comparing the yield losses of CIMMYTrelated spring bread wheat varieties with various genetic resistance categories to the yield losses
that would have been suffered had all these varieties been fully susceptible. These estimates do
not account for the major losses that could be incurred in epidemics, which may result from not
having resistance. These social consequences could be catastrophic, especially for farmers and
societies in the developing world who rely on their wheat crop to a great extent, and for whom
large-scale treatment with fungicides to mitigate the yield losses incurred by epidemics, might
not be feasible. Our economic valuation also did not capture the issue of public risk diseases, as
defined by Brennan, Murray, and Ballantyne (1994). Producers who grow cultivars susceptible
to diseases that can rapidly spread between farms, may place not only their own production at
risk, but also that of other farmers. When considering the pathogen’s ability to spread over long
distances, and to rapidly mutate to new races able to overcome the effects of previously resistant
varieties, rusts are in the high risk category.
The results also emphasize the importance of maintenance research in plant breeding programs.
We have estimated substantial economic returns by only valuing the yield losses that have been
avoided by growing varieties with different leaf rust resistance categories, and assuming all other
wheat breeding benefits as pure benefits. Our findings are in line with other CIMMYT studies
indicating that genetic resistance breeding has generated a substantial proportion of the economic
returns on the investment in international wheat research over the past decades (Bohn and
24
Byerlee, 1993; Byerlee and Moya, 1993; Byerlee and Traxler, 1995; Rajaram et al., 1996; Heisey
et al., 1999). Analyses of trial results confirmed that progress in protecting yield potential
through genetic rust resistance has been greater than advances in yield potential itself (Sayre,
Singh, Huerta-Espino, and Rajaram, 1998).
However, most assessments of the economic returns on investments in wheat research1 have
focused on productivity enhancement, and have often modeled the benefits in terms of positive
yield gains. There are comparatively fewer economic analyses of wheat maintenance research,
and in particular, breeding for pest and disease resistance (Doodson, 1981; Heim and Blakeslee,
1986; Blakeslee, 1987; Priestley and Bayles, 1988; Brennan, Murray, and Ballantyne, 1994;
Morris, Dubin, and Pokhrel, 1994; Collins, 1995; Smale et al., 1998; Marasas, 19992, Marasas,
Smale and Singh, forthcoming).
A review of previous studies, including wheat amongst other enterprises, can be found in
Evenson (1998). Studies more recently conducted in Africa have also been summarized by
Marasas (1999).
1
2
This study is included as an economic valuation of pest resistance in wheat, although the focus
was not exclusively on maintenance research. The author assessed the productivity gains and
cost savings resulting from the use of Russian wheat aphid resistant varieties as opposed to
chemical control.
Although the importance of maintenance research has long been argued (Araji, Sim, and
Gardner, 1978; Plucknett and Smith, 1986; Adusei and Norton, 1990; Bohn and Byerlee, 1993),
the effects are often under-valued in economic analyses. By drawing on data from South Africa,
Townsend and Thirtle (2001) recently illustrated the magnitude of this error by separating the
maintenance effects of animal health research from output increases due to improvement
research. Their estimates suggest a minimum under-estimation of about 50% on the returns on
livestock research when the negative effects of diseases were not explicitly taken into account.
These findings are likely to also apply to returns estimates for plant breeding programs, where
the productivity losses avoided by research have often been ignored. As also pointed out by
Townsend and Thirtle (2001), we do not suggest that these assumptions are made because of a
lack of understanding or effort. The reality is that estimation of these benefits are only too often
restricted by data limitations. However, we conclude that rate of return estimates which assume
that plant breeding explains only positive productivity growth, and that productivity would
remain constant in the absence of research, are bound to be biased downward.
REFERENCES
Adusei, E.O., and Norton, G.W. 1990. The magnitude of agricultural maintenance research in the
USA. Journal of Production Agriculture 3:1-6.
Araji, A.A., Sim, R.J., and Gardner, R.I. 1978. Returns to agricultural research and extension: An
ex ante approach. American Journal of Agricultural Economics 60:164-168.
Blakeslee, L. 1987. Measuring the requirements and benefits of productivity maintenance
research. Pages 67-83 in: Evaluating agricultural research and productivity. Miscellaneous
publication 52-1987. Minnesota Agricultural Experiment Station, University of Minnesota.
Bohn, A., and Byerlee, D. 1993. The wheat breeding industry in developing countries: An
analysis of investments and impacts. CIMMYT World Wheat Facts and Trends 1992/1993, Part
1. International Maize and Wheat Improvement Center, Singapore.
Brennan, J.P., Murray, G.M., and Ballantyne, B.J. 1994. Assessing the economic importance of
disease resistance in wheat. NSW Agriculture, Agricultural Research Institution, Wagga Wagga.
Final report to the Grains Research and Development Corporation, Australia.
Byerlee, D., and Moya, P. 1993. Impacts of international wheat breeding research in the
developing world, 1966-90. International Maize and Wheat Improvement Center, Mexico, D.F.,
135 pp.
Byerlee, D., and Traxler, G. 1995. National and international wheat improvement research in the
post-Green Revolution period: Evolution and impacts. American Journal of Agricultural
Economics 77: 268-278.
27
Cereal Disease Laboratory. http://www.crl.umn.edu
CIMMYT. 1989. The Wheat Revolution Revisited: Recent Trends and Future Directions.
CIMMYT World Wheat Facts and Trends 1987-88. International Maize and Wheat Improvement
Center, Mexico, D.F.
Collins, M.I. 1995. The economics of productivity maintenance research: A case study of wheat
leaf rust resistance breeding in Pakistan. Ph.D. thesis, University of Minnesota, St. Paul,
Minnesota.
Doodson, J.K.1981. The economic contribution of resistant winter wheat varieties. Journal of the
National Institute of Agricultural Botany 15: 41-49.
Dubin, H.J., and Torres, E. 1981. Causes and Consequences of the 1976-77 Wheat Leaf Rust
Epidemic in Northwest Mexico. Annual Review of Phytopathology 19: 41-49.
Evenson, R.E. 1998. Economic impact studies of agricultural research and extension. In:
Handbook of Agricultural Economics. B. Gardner, and G. Rausser, eds. North-Holland,
Amsterdam.
Food and Agriculture Organization. http://faostat.fao.org
28
Heim, M.N., and Blakeslee, L. 1986. Biological adaptation and research impacts on wheat yields
in Washington. College of Agriculture and Home Economics, Washington State University,
Pullman, Washington.
Heisey, P.W., Lantican, M.A., and Dubin, H.J. 1999. Assessing the benefits of international
wheat breeding research: An overview of the global wheat impacts study. Pages 19-26 in:
CIMMYT 1998-99 World Wheat Facts and Trends. Global wheat research in a changing world:
Challenges and achievements. P.L. Pingali, ed. International Maize and Wheat Improvement
Center, Mexico, D.F. ISSN 0257-876X.
Heisey, P.W., Lantican, M.A., and Dubin, H.J. Forthcoming. Assessing the Benefits of
International Wheat Breeding Research in the Developing World: The Global Wheat Impacts
Study, 1966-1997. International Maize and Wheat Improvement Center, Mexico, D.F.
Marasas, C.N. 1999. Socio-economic impact of the Russian wheat aphid integrated control
program. Ph.D. thesis, University of Pretoria, South Africa.
Marasas, C.N., Smale, M., and Singh, R.P. Forthcoming. The global economic impact of
breeding for genetic resistance to leaf rust in CIMMYT-related spring bread wheat. Report by
the International Maize and Wheat Improvement Center, Mexico, D.F.
Morris, M.L., Dubin, H.J., and Pokhrel, T. 1994. Returns to wheat breeding research in Nepal.
Agricultural Economics 10: 269-282.
29
Peterson, R.F., Campbell, A.B., and Hannah, A.E. 1948. A diagrammatic scale for estimating
rust severity on leaves and stems of cereals. Can. J. Res., Sect. C. 26:496-500.
Plucknett, D.L., and Smith, N.J.H. 1986. Sustaining agricultural yields. BioScience 36 (1): 4045.
Priestley, R.H., and Bayles, R.A. 1988. The contribution and value of resistant cultivars to
disease control in cereals. Pages 53-65 in: Control of plant diseases: Costs and benefits. B.C.
Clifford and E. Lester, eds. Blackwell Scientific Publications, Oxford.
Rajaram, S., Singh, R.P., and Van Ginkel, M. 1996. Approaches to breed wheat for wide
adaptation, rust resistance and drought. Pages O2-O30 in: Proceedings of the Wheat Breeding
Society of Australia, eighth assembly, 29 September – 4 October 1996. R.A. Richards, C.W.
Wrigley, H.M. Rawson, G.J. Rebetzke, J.L. Davidson, and R.I.S. Brettell, eds. Wheat Breeding
Society of Australia, Tamworth, Australia.
Rajaram, S., Van Ginkel, M., and Fischer, R.A. 1995. CIMMYT’s wheat breeding megaenvironments. Pages 1101-1106 in: Proceedings of the eighth international wheat genetics
symposium Volume 2, 20-25 July 1993, Beijing, China. Z.S. Li and Z.Y. Xin, eds. China
Agricultural Scientech Press.
30
Rejesus, R., Smale, M., and Van Ginkel, M. 1997. Wheat breeders’ perspectives on genetic
diversity and germplasm use: Findings from an international survey. Plant Varieties and Seeds
9:129-147.
Roelfs, A.P., Singh, R.P., and Saari, E.E. 1992. Rust diseases of wheat: Concepts and methods of
disease management. International Maize and Wheat Improvement Center, Mexico, D.F. 81 pp.
Sayre, K.D., Singh, R.P., Huerta-Espino, J., and Rajaram, S. 1998. Genetic progress in reducing
losses to leaf rust in CIMMYT-derived Mexican spring wheat cultivars. Crop Science 38: 654659.
Smale, M., Singh, R.P., Sayre, K., Pingali, P., Rajaram, S., and Dubin, H.J. 1998. Estimating the
economic impact of breeding nonspecific resistance to leaf rust in modern bread wheats. Plant
Disease 82: 1055-1061.
Townsend, R., and Thirtle, C. 2001. Is livestock research unproductive? Separating health
maintenance from improvement research. Agricultural Economics 25: 177-189.
31
Annex Table 1. Summary of parameters used in this study
Environment
Megaenvironment
Average
annual %
yield lost to
rust
% area
affected by
leaf rust
Cumulative % area CIMMYT-related
wheats a
1977
1990
Adoption lag
1997
Irrigated
1
6
96
83
99
99
0
High rainfall
2
3
92
38
77
81
8
Acid soil
3
3
100
0
60
48
12
Semi-arid, Mediterranean
4a
2
45
5
23
59
9
Semi-arid, Southern Cone
4b
1
100
0
69
91
14
Semi-arid, Subcontinent
4c
1
69
0
25
50
14
Hot, humid
5a b
6
100
83
99
95
0
a
Estimates of the cumulative percentage area planted in CIMMYT-related spring bread wheat in 1997 were obtained from Heisey et
al. (forthcoming), and were assumed as the adoption ceilings in each mega-environment. The diffusion curves were calibrated with the
1977 and 1990 data (CIMMYT, 1989; Byerlee and Moya, 1993).
b
The information for ME 5 refers to ME 5a (Marasas et al., forthcoming).